Production of Brominating Intermediates by Myeloperoxidase

The existence of interhalogen compounds was proposed more than a century ago, but no biological roles have been attributed to these highly oxidizing intermediates. In this study, we determined whether the peroxidases of white blood cells can generate the interhalogen gas bromine chloride (BrCl). Myeloperoxidase, the heme enzyme secreted by activated neutrophils and monocytes, uses H2O2 and Cl− to produce HOCl, a chlorinating intermediate. In contrast, eosinophil peroxidase preferentially converts Br− to HOBr. Remarkably, both myeloperoxidase and eosinophil peroxidase were able to brominate deoxycytidine, a nucleoside, and uracil, a nucleobase, at plasma concentrations of Br− (100 μm) and Cl− (100 mm). The two enzymes used different reaction pathways, however. When HOCl brominated deoxycytidine, the reaction required Br− and was inhibited by taurine. In contrast, bromination by HOBr was independent of Br− and unaffected by taurine. Moreover, taurine inhibited 5-bromodeoxycytidine production by the myeloperoxidase-H2O2-Cl−- Br− system but not by the eosinophil peroxidase-H2O2-Cl−-Br−system, indicating that bromination by myeloperoxidase involves the initial production of HOCl. Both HOCl-Br− and the myeloperoxidase-H2O2-Cl−-Br−system generated a gas that converted cyclohexene into 1-bromo-2-chlorocyclohexane, implicating BrCl in the reaction. Moreover, human neutrophils used myeloperoxidase, H2O2, and Br− to brominate deoxycytidine by a taurine-sensitive pathway, suggesting that transhalogenation reactions may be physiologically relevant. 5-Bromouracil incorporated into nuclear DNA is a well known mutagen. Our observations therefore raise the possibility that transhalogenation reactions initiated by phagocytes provide one pathway for mutagenesis and cytotoxicity at sites of inflammation.

We previously demonstrated that HOCl generated by myeloperoxidase is in equilibrium with molecular chlorine (Cl 2 ) through a reaction that requires chloride (Cl Ϫ ) and H ϩ (12,20).
HOCl ϩ H ϩ ϩ Cl Ϫ^C l 2 ϩ H 2 O (Eq. 2) Cl 2 generated by this pathway has been implicated in the production of 3-chlorotyrosine and 5-chlorodeoxycytidine by activated neutrophils (12,20). Elevated levels of protein-bound 3-chlorotyrosine and myeloperoxidase are found in human atherosclerotic tissue, strongly suggesting that oxidative reactions involving HOCl damage proteins in this chronic inflammatory disorder (13,21). Chronic inflammation also increases the risk of cancer, raising the possibility that reactive intermediates generated by neutrophils, monocytes, and macrophages might damage nucleic acids and compromise the integrity of the genome (22)(23)(24). Genetic epidemiological studies have revealed that a polymorphism in the myeloperoxidase promoter region alters the risk for various cancers (25)(26)(27)(28)(29)(30). These results suggest that myeloperoxidase may play an important role in carcinogenesis, perhaps by generating mutagenic oxidants during the inflammatory response.
Like HOCl, HOBr oxidizes biomolecules at sites of eosinophilic inflammation (36). DNA may be one important target because eosinophil peroxidase brominates deoxycytidine in vitro. 1 It is noteworthy that schistosomiasis, a chronic inflammatory disease characterized by an intense eosinophilic granulomatous reaction to the eggs of the blood fluke Schistosoma, greatly increases the risk for cancer (reviewed in Refs. [37][38][39]. More than 100 years ago, inorganic chemists proposed the existence of interhalogens, which are combinations of different halogens (XX n Ј). Both binary (BrCl, IBr, and ICl) and ternary (ICl 3 ) interhalogens have since been characterized. One pathway for their formation requires hypohalous acid (HOX) and halide ion (X; Refs. 40 and 41).
HOCl reacts with Br Ϫ by this mechanism to yield molecular BrCl. Anions of interhalogens and polyhalides are also known; they include Cl 3 Ϫ , Br 3 Ϫ , I 3 Ϫ , Br 2 Cl Ϫ , and BrCl 2 Ϫ . Chemically, interhalogens are extremely corrosive species that attack a wide range of other compounds (42).
In the current studies, we show that myeloperoxidase generates reactive brominating species that oxidize nucleobases by a reaction involving HOCl, Br Ϫ , and formation of BrCl, an interhalogen gas. We also found that human neutrophils used myeloperoxidase, Cl Ϫ , and Br Ϫ to brominate deoxycytidine, suggesting that transhalogenation reactions may be physiologically relevant. Our observations suggest that transhalogenation reactions executed by phagocytes may represent one pathway for mutagenesis and cytotoxicity at sites of inflammation.

Materials
Organic solvents, H 2 O 2 , sodium hypochlorite, and sodium phosphate were obtained from Fisher. Bis(trimethylsilyl)trifluoroacetamide and silylation grade acetonitrile were from Regis Technologies, Inc. (Morton Grove, IL). All other materials were purchased from Sigma, except where indicated.
Peroxidase Activity Assay-The purity of myeloperoxidase and eosinophil peroxidase were assessed by peroxidase activity using nondenaturing polyacrylamide slab gel electrophoresis and gel system 8 (46,47). Glycerol (25% w/v) and cetyltrimethylammonium bromide (0.05% w/v) were included in all buffers. Riboflavin (0.024 mg/ml) was used as the polymerization catalyst, and the stacking gel was omitted. Peroxidase activity was visualized by incubating the gel in 400 M tetramethylbenzidine, 10 mM sodium citrate, pH 5, 10 mM EDTA, 5 mM NaBr, and Preparation of Hypochlorous Acid and Taurine Chloramine-Chloride-free NaOCl was prepared by a modification of previously described methods (16). Reagent NaOCl (100 ml) mixed with ethyl acetate (100 ml) was protonated by dropwise addition of concentrated phosphoric acid (final pH Յ 6) with intermittent shaking. The organic phase containing HOCl was washed twice with water, and HOCl was reextracted into the aqueous phase by dropwise addition of NaOH (final pH Ն 9). Residual ethyl acetate in the aqueous solution of chloride-free NaOCl was removed by bubbling with nitrogen gas. The concentration of NaOCl was determined spectrophotometrically (⑀ 292 ϭ 350 M Ϫ1 cm Ϫ1 ; Ref. 48). Taurine monochloramine was prepared by addition of HOCl to taurine (1:100; mol/mol). Taurine monochloramine concentration was determined spectrophotometrically (⑀ 252 ϭ 429 M Ϫ1 cm Ϫ1 ; Ref. 16) Preparation of Hypobromous Acid and Taurine Bromamine-Bromide-free HOBr was prepared as described (49). Briefly, silver nitrate solution was added to ϳ80 mM bromine water in a molar ratio of 1.5:1. The precipitate was removed by centrifugation, and 30 ml of the supernatant was distilled under vacuum using a foil-covered microscale distillation apparatus. The distillate was collected in a foil covered vial at 4°C. Reagent taurine monobromamine was prepared by addition of reagent HOBr to a 100-fold excess of taurine. HOBr concentration was determined spectrophotometrically following formation of taurine monobromamine (⑀ 288 ϭ 430 M Ϫ1 cm Ϫ1 ; Ref. 35).
Oxidation of Uracil and Deoxycytidine-All reactions were performed in gas-tight vials and initiated by addition with a gas-tight syringe of oxidant (H 2 O 2 or HOCl/OCl Ϫ ) through a septum while vortexing the sample. Reactions were terminated by addition of L-methionine to a final concentration of 6 mM. The concentration of H 2 O 2 was determined spectrophotometrically (⑀ 240 ϭ 43.6 M Ϫ1 cm Ϫ1 ; Ref. 50). The pH dependence of 5-bromouracil and 5-bromodeoxycytidine formation was performed using reaction mixtures containing phosphoric acid, monobasic sodium phosphate, and dibasic sodium phosphate (final concentration, 50 mM). The pH of the reaction mixture (which did not contain L-methionine) was determined at the end of the incubation.
Human Neutrophils-Neutrophils were prepared by density gradient centrifugation (51) and suspended in Dulbecco's phosphate-buffered saline supplemented with 1 mg/ml dextrose, 1 mM deoxycytidine, 100 M NaBr, and 100 M DTPA, 2 pH 5.9. Differential cell counts revealed that neutrophil preparations contained 96 -100% neutrophils and 0 -4% eosinophils. Cells (3 ml) were activated with 200 nM phorbol myristate acetate, incubated at 37°C for 60 min, and maintained in suspension with intermittent inversion. The reaction was terminated by addition of methionine to 6 mM and centrifugation of the cells at 400 ϫ g for 10 min. The supernatant was concentrated to dryness under vacuum, dissolved in 0.4 ml of HPLC solvent A, centrifuged at 14,000 ϫ g for 10 min, and the supernatant was subjected to HPLC analysis.
Reverse-phase HPLC of Reaction Products-Uracil and deoxycytidine reaction products were analyzed by reverse-phase HPLC with a C 18 column (Beckman Porasil, 5 m resin, 4.6 ϫ 250 mm) at a flow of 1 ml/min and UV detection at 274 and 295 nm, respectively. Uracil reactions were analyzed by injection of 100 l of reaction mixture onto the column followed by isocratic elution with 20 mM ammonium formate. For analysis of deoxycytidine reactions, 100 l of the reaction mixture was injected onto the column and eluted with a gradient of: 95% solvent A (0.1% trifluoroacetic acid, pH 2.5) and 5% solvent B (0.1% trifluoroacetic acid in methanol, pH 2.5) for 4 min, 5-100% solvent B over 20 min, and then 100% solvent B for 10 min. 5-Bromouracil and 5-bromodeoxycytidine yields were quantified by comparison of integrated peak areas to standard curves generated using commercially available compounds. For mass spectrometric analysis, HPLC fractions were collected and concentrated under vacuum. For NMR analysis, 10-fold concentrated reaction mixtures were fractionated on a semipreparative C 18 column (Porasil; 5 m resin, 10 ϫ 250 mm; Beckman) at a flow rate of 2.5 ml/min with an isocratic gradient consisting of 90% 20 mM ammonium formate, pH 6.3, and 10% methanol. N-Chlorodeoxycytidine (retention time, 10 min) was prepared as described (20) and isolated with an analytical C 18 column (Porasil; 5 m resin, 4.6 ϫ 250 mm; Beckman) column using 5% methanol at a flow rate of 1 ml/min. NMR Studies-Reaction products were isolated by HPLC, solubilized in D 2 O, and analyzed at 25°C with a Varian Unity-Plus 500 spectrometer (499.843 MHz for 1 H) equipped with a Nalorac indirect detection probe. 1 H Chemical shifts were referenced to external sodium 3-(trimethylsilyl)propionate-2,2,3,3-d 4 in D 2 O. Spectra were recorded from 8 transients with a 12-s preacquisition delay over a spectral width of 8000 Hz. Pyrimidine resonances of the brominated deoxycytidine (8.23 ppm; singlet, H 6 ) and uracil (7.79 ppm; singlet, H 6 ) reaction products were essentially identical to those of commercially available 5-bromodeoxycytidine and 5-bromouracil. When compared with substrate, the aromatic region of each product spectrum was notable for the lack of proton resonances at C-5, a downfield shift in the proton resonance at C-6, and conversion of the C-6 proton resonance from a doublet to a singlet, both of which are consistent with substitution of a bromine atom at the C-5 position.
Gas Chromatography-Mass Spectrometry (GC/MS)-After nucleobases were dried under vacuum, residual water was removed by forming an azeotrope with 50 l of pyridine and again drying the suspension under vacuum. DNA bases were converted to trimethylsilyl derivatives with excess bis(trimethylsilyl)trifluoroacetamide ϩ 1% trimethylchlorosilane in acetonitrile (3:1 v/v) at 100°C for 60 min. Aliquots (1 l) were analyzed in the positive electron ionization mode using full mass scanning on either a Hewlett Packard 5890 Series II gas chromato-graph (Santa Clarita, CA) interfaced with a Hewlett Packard 5972 Series Mass Selective Detector or a Varian Star 3400 CX gas chromatograph (Walnut Creek, CA) interfaced with a Finnegan SSQ 7000 mass spectrometer (San Jose, CA). Each gas chromatograph was equipped with a 12-m DB-1 capillary column (inner diameter, 0.2 mm; film thickness, 0.33 m; J&W Scientific, Folsom, CA). Injector and interface temperatures were 250 and 280°C, respectively. The initial GC oven temperature was 70°C for 2 min, followed by a 60°C/min increase to 180°C and a final 10°C/min ramp to 220°C. Derivatizing agent injections were analyzed between samples to ensure that no residual analyte remained in the injection port.
Cyclohexene addition products (0.2 l) were analyzed by selected ion monitoring of ions of m/z 195-201 using an initial GC temperature of 50°C for 2 min, followed by a 20°C/min increase to 100°C and a final 60°C/min ramp to 220°C.
Electrospray Ionization-Mass Spectrometry-Aliquots from HPLC fractions were analyzed on a Waters Alliance 2670 HPLC equipped with a C 18 column (Porasil, 5 m resin, 2.1 ϫ 150 mm; Beckman) interfaced with a Finnigan LCQ. Sample (5 l) was injected at a flow rate of 200 l/min. Solvent C was 1% acetic acid in 4% methanol, and solvent D was 1% acetic acid in 95% methanol. The sample was eluted from the column by a discontinuous gradient of solvent D. The gradient was: 0% solvent D for 2 min, 0 -100% solvent D over 8 min, and then 100% for 10 min. Mass spectrometric analysis was carried out in full mass scanning, zoom scanning, and low energy collisionally activated dissociation modes as described (20). Helium was used as a damping gas and collision activation partner. The temperature of the heated capillary was 220°C. In the full scan mode (m/z 150 -300), each scan consisted of three 300-ms microscans. Full scan mass spectra consisted of 10 signalaveraged scans with subtraction of the carrier solvent background from the same number of scans.
Mass Spectrometric Analysis of Oxidation Products-The electrospray positive ion mass spectra of authentic 5-bromodeoxycytidine and the product from myeloperoxidase, HOCl ϩ Br Ϫ , and HOBr yielded the same major [M ϩ H] ϩ ion at m/z 306. All compounds also exhibited a prominent ion at m/z 308. The relative abundances of the ions at m/z 306 and 308 reflected that of the natural isotopic abundance of 79 Br and 81 Br, strongly suggesting that the deoxycytidine oxidation product was monobrominated. The collisionally activated dissociation tandem mass spectrum of the m/z 306 ion generated a product ion at m/z 190, which is consistent with cleavage of the N-glycoside bond of 5-bromodeoxycytidine to yield 79 Br-substituted cytosine. The collisionally activated dissociation tandem mass spectrum of the m/z 308 ion likewise generated a product ion at m/z 192, consistent with cleavage of the brominated nucleoside to yield 81 Br-substituted cytosine. The electrospray ionization MS/MS spectrum of the myeloperoxidase product indicates that the bromine substitution site resides on the cytosine base of deoxycytidine.
Further structural characterization of the modified nucleobase generated from deoxycytidine and uracil was achieved using GC/MS to obtain an informative electron ionization mass spectrum and GC retention time from material isolated by HPLC. This procedure yields information only about the deoxycytidine nucleobase because the N-glycoside bond of the nucleoside is hydrolyzed during the derivatization reactions. The mass spectrum of the reaction products exhibited low abundance ions consistent with the molecular ion [M ϩ⅐ ] of the bistrimethylsilyl derivatives (m/z 333 and 334 for brominated deoxycytidine and uracil, respectively). The major ion from each analyte was consistent with loss of CH 3 ⅐ as expected for trimethylsilyl derivatives. Ϫ HBr] ϩ or [M Ϫ CH 4 Ϫ Br Ϫ ] ϩ fragments lacked the prominent Mϩ2 fragments as expected for compounds that lack bromine. The positive ion mass spectra and GC retention times of the trimethylsilyl derivatives of the reaction products were essentially identical to those obtained for commercially available 5-bromodeoxycytidine or 5-bromouracil.

5-Bromodeoxycytidine Is the Major Product when the Myeloperoxidase-H 2 O 2 -Br Ϫ -Cl Ϫ System Oxidizes Deoxycytidine at
Plasma Concentrations of Halide-We previously demonstrated that the myeloperoxidase-H 2 O 2 -Cl Ϫ system (containing 100 mM Cl Ϫ ) oxidizes deoxycytidine to 5-chlorodeoxycytidine (20). When we supplemented this system with a plasma concentration of Br Ϫ (100 M), HPLC analysis detected a peak of material that migrated with a retention time distinct from that of 5-chlorodeoxycytidine (Fig. 1A). The new oxidation product was isolated by HPLC and identified as 5-bromodeoxycytidine on the basis of its HPLC retention time, ultraviolet absorption spectrum, GC retention time and positive ion mass spectrum, electrospray ionization tandem mass spectrum, and 1 H NMR spectrum (see "Methods").
Under these reaction conditions, 5-bromodeoxycytidine was also the major product when eosinophil peroxidase or lactoperoxidase oxidized deoxycytidine. It was therefore important to determine whether our myeloperoxidase preparation was contaminated by other peroxidases, even though we isolated the enzyme from HL-60 cells, a promyelocytic cell line that is not known to express other peroxidases. Myeloperoxidase was apparently pure as assessed by its heme spectrum and by denaturing polyacrylamide gel electrophoresis. Moreover, it yielded a single band of peroxidase activity that migrated with a retention time distinct from that of eosinophil peroxidase on nondenaturing gel electrophoresis. These observations indicate that the reactive intermediates that brominated deoxycytidine at plasma concentrations of Cl Ϫ and Br Ϫ resulted from the action of myeloperoxidase.

5-Bromouracil Is the Major Product When the Myeloperoxidase-H 2 O 2 -Br Ϫ -Cl Ϫ System Oxidizes Uracil at Plasma Concentrations of Halide-
To determine whether the enzyme can halogenate other pyrimidines, we incubated uracil with the myeloperoxidase-H 2 O 2 -Cl Ϫ system. In the absence of added Br Ϫ , a new peak of material (retention time, 13.7 min) was detectable in the reaction mixture by HPLC analysis (Fig. 1B).
In the presence of 100 M Br Ϫ , we observed a second peak (retention time, 18 min). We isolated both peaks of material by HPLC and determined their structures by ultraviolet absorption spectroscopy, GC, and positive ion mass spectrometry, electrospray ionization tandem mass spectrometry, and 1 H NMR spectroscopy (see "Methods"). We identified the early and late eluting materials as 5-chlorouracil and 5-bromouracil, respectively.
Production of 5-bromouracil by myeloperoxidase was dependent on the concentration of Br Ϫ in the reaction mixture (Fig. 2). Increasing [Br Ϫ ] from 0 to 100 M in the presence of 100 mM Cl Ϫ reduced uracil chlorination but increased uracil bromination. The apparent K m values for Cl Ϫ and Br Ϫ binding by myeloperoxidase are 175 and 2 mM (46), respectively, suggesting that the enzyme did not directly oxidize Br Ϫ to generate 5-bromouracil. The relative yields of the halogenated pyrimidines also differed; under optimal conditions, the yield of 5-bromouracil was 5-fold greater than that of 5-chlorouracil.
We used HPLC to investigate the reaction requirements for generation of 5-bromodeoxycytidine by the myeloperoxidase and eosinophil peroxidase-H 2 O 2 -Cl Ϫ -Br Ϫ systems (Table I). Both systems required enzyme and H 2 O 2 and were blocked by catalase, a scavenger of H 2 O 2 . The product yield of the reaction was nearly quantitative relative to H 2 O 2 . Omitting Br Ϫ did not completely prevent bromination, perhaps because the NaCl used as the salt contained up to 0.01% NaBr (up to a final [Br Ϫ ] of 10 M). Two heme-enzyme inhibitors, cyanide and aminotriazole, also inhibited product formation. Both myeloperoxidase and eosinophil peroxidase brominated deoxycytidine in the absence of added Cl Ϫ , which is consistent with the ability of myeloperoxidase to convert Br Ϫ into HOBr (46). These results demonstrate that bromination of deoxycytidine by myeloperoxidase or eosinophil peroxidase requires active enzyme, Br Ϫ , and H 2 O 2 .
The two enzymes differed in one important respect, however. Bromination by myeloperoxidase but not by eosinophil peroxidase was inhibited by taurine (Table I), which can react with HOCl or HOBr to form chloramines or bromamines (14,35). This observation suggests that the enzymes brominate pyrimidines by different reaction pathways.
We used HPLC to establish the optimal reaction conditions for bromination of uracil (Fig. 3) and deoxycytidine (Fig. 4) by the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system. Bromination of both substrates was proportional to [H 2 O 2 ] from 0 to 100 M. The yields of both products increased with rising [Br Ϫ ] over a physiologically relevant range. The pH dependence demonstrated a distinct optimum of 4.5-5.0 for 5-bromodeoxycytidine generation (Fig. 4C) and a broader range of 4.0 -6.0 for 5-bromouracil generation (Fig. 3B). Both reactions were largely complete after 60 min.
Reagent HOCl Generates 5-Bromouracil and 5-Bromodeoxycytidine by a Reaction That Requires Br Ϫ -Previous studies have indicated that myeloperoxidase preferentially oxidizes Cl Ϫ to HOCl at plasma concentrations of Br Ϫ and Cl Ϫ . In contrast, eosinophil peroxidase selectively oxidizes Br Ϫ to an HOBr-like species. To determine whether HOCl could be an intermediate when myeloperoxidize generates 5-bromouracil (Fig. 5) or 5-bromodeoxycytidine (Fig. 6), we compared the Br Ϫ , oxidant, and pH dependences of product production by reagent HOBr and HOCl. When we used HOBr, product yield was independent of [Br Ϫ ] (Figs. 5A and 6A). In striking contrast, HOCl required Br Ϫ to brominate deoxycytidine.
Production of 5-bromouracil and 5-bromodeoxycytidine increased linearly with [HOBr] (Figs. 5B and 6B). In contrast, the yield of brominated products reached a plateau at around 75 M of HOCl, a concentration similar to that of Br Ϫ (100 M) in the reaction mixture. The pH dependences of product generation by HOCl and HOBr were similar under acidic and neutral conditions. However, the yields of 5-bromouracil differed under alkaline conditions: at pH 8, HOBr but not HOCl generated large amounts of product (Fig. 5C). The decline in uracil bromination by HOCl at pH 8 was accompanied by an increase in uracil chlorination (data not shown). These results indicate that HOCl readily oxidizes Br Ϫ to generate a brominating agent at physiologically plausible concentrations of halide.
Hypochlorous Acid and Myeloperoxidase Generate the Interhalogen Gas BrCl at Physiologic Concentrations of Bromide and Chloride-Previous studies indicate that reagent HOCl   FIG. 2. The effect of Br ؊ on the generation of chlorinated and brominated uracil by the myeloperoxidase-H 2 O 2 -Cl ؊ system. The reaction was initiated by adding H 2 O 2 (50 nmol) to 1 ml of buffer B containing 1 mM uracil, 10 nM myeloperoxidase, and the indicated NaBr concentrations. After a 60-min incubation at 37°C, the reaction was terminated by adding 6 mol of L-methionine, and products were quantified by reverse-phase HPLC.  oxidizes Br Ϫ to BrCl (40,41).
To determine whether HOCl generates BrCl under physiologically plausible conditions, we added 50 M of HOCl to a reaction mixture containing 100 mM Cl Ϫ and 100 M Br Ϫ , sparging the mixture continuously with nitrogen gas. The gas then was bubbled through cyclohexene, which contains a cis-double bond analogous to those found in pyrimidines and biological lipids. Because cyclohexene is an aprotic, nonpolar solvent that would prevent the proton and halide-dependent formation of BrCl shown in Equation 5, direct reaction of cyclohexene with BrCl from the reaction mixture would be the most likely source of any 1-bromo-2-chlorocyclohexane that appeared in this experiment (Scheme 1 and Ref. 52).
Mass spectrometric analysis of cyclohexene exposed to the gas sparged from the HOCl reaction mixture revealed a major peak of material with the mass-to-charge-ratio (Fig. 7A) and isotopic pattern expected for the molecular ion of 1-bromo-2chlorocyclohexane. This material exhibited a retention time midway between that of authentic 1,2-dichlorocyclohexane and that of 1,2-dibromocyclohexane. It was not detectable when we omitted HOCl from the reaction mixture (Fig. 7A). When we replaced HOCl-Cl Ϫ -Br Ϫ with the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system, we also observed a peak of material with the retention time and characteristic isotopic pattern of 1-bromo-2-chlorocyclohexane (Fig. 7B) Addition of Cl Ϫ should drive this equilibrium toward the formation of BrCl. We therefore examined the effect of [Cl Ϫ ] on the formation of 5-bromodeoxycytidine by Br Ϫ and chloridefree HOCl (Fig. 8). The yield was doubled by plasma concentrations of Cl Ϫ (100 mM) and tripled by the highest concentration we examined (1,000 mM Cl Ϫ ). This increase in yield was not observed when chloride was replaced with perchlorate (ClO 4 Ϫ ), ruling out ionic strength as a cause of enhanced yield. Control experiments demonstrated that contamination of NaCl with NaBr also was not responsible for the increased yield of 5-bromodeoxycytidine. These observations support the hypothesis that BrCl is an important intermediate when myeloperoxidase brominates deoxycytidine.
Bromamines and Chloramines Produce 5-Bromodeoxycytidine-HOCl can react with amines to form N-chloro compounds, which then could react with Br Ϫ to generate brominating species (52). Alternatively, chloramine formation might compete with Br Ϫ , inhibiting the formation of brominating species by HOCl. Our observation that taurine inhibits bromination by HOCl and Br Ϫ (Table I) is consistent with the latter possibility. To distinguish between these two possibilities, we determined whether different haloamines can produce 5-bromodeoxycytidine (Fig. 9). Both N-chlorotaurine and N-bromotaurine generated high concentrations of 5-bromodeoxycytidine below pH 5 in the presence of 100 M Br Ϫ and 100 mM Cl Ϫ . Over the pH range 5.5-7, however, only N-bromotaurine generated significant levels of 5-bromodeoxycytidine.
Unlike N-chlorotaurine, N-chlorodeoxycytidine failed to convert cytosine to 5-bromocytosine except under strongly acidic conditions (ϽpH 3), and the yield was low even at pH 2 (Fig. 9). The chloramine of deoxycytidine therefore is unlikely to be an intermediate when myeloperoxidase brominates deoxycytidine. These observations indicate that primary haloamines can brominate pyrimidines in the presence of Br Ϫ . They also suggest that Br Ϫ oxidation by HOCl is more rapid than nucleoside chloramine formation.
Activated Human Neutrophils Generate 5-Bromodeoxycytidine and 5-Bromouracil at Plasma Halide Concentrations-To determine whether oxidants generated by human neutrophils can brominate nucleobases, we stimulated the cells with phorbol myristate acetate in physiological salt solution supplemented with 100 M NaBr and 1 mM deoxycytidine (Table II) or uracil (data not shown). HPLC and mass spectrometric analysis detected substantial quantities of 5-bromodeoxycytidine and 5-bromouracil in the medium of the activated cells. Bromination required stimulation of the cells with phorbol ester and was inhibited by catalase and heme poisons, implicating H 2 O 2 and a heme protein in the reaction (Table II). Omitting supplemental Br Ϫ significantly reduced but did not eliminate 5-bromodeoxycytidine production, most likely because the medium was contaminated with Br Ϫ . Superoxide dismutase failed to affect the reaction.
Generation of 5-bromodeoxycytidine (Fig. 10, upper panel) by activated neutrophils was strongly affected by the pH of the culture medium. Acidification to the range observed in inflamed tissue and the phagolysosome significantly enhanced the cellular production of 5-bromodeoxycytidine. The progress curve of the reaction was essentially complete 30 min after the cells were activated with phorbol ester (Fig. 10, lower panel).
Human Neutrophils Generate Brominating Intermediates by a Pathway Involving HOCl-In vitro studies demonstrated that N-bromotaurine, but not N-chlorotaurine, brominates deoxycytidine at pH 5.9 in the presence of 100 M Br Ϫ and 100 mM Cl Ϫ (Fig. 9). This observation suggests that reagent HOBr, but not HOCl, should brominate deoxycytidine in the presence of taurine under these conditions. Indeed, this reaction occurred when mM concentrations of taurine were included in the reaction mixture (Fig. 11A). In contrast, taurine inhibited bromination by HOCl under the same conditions, probably because it consumed the oxidant to form N-chlorotaurine, which is unreactive at pH 5.9. Taurine was also inhibitory (Fig. 11A) when the myeloperoxidase-H 2 O 2 system replaced HOCl. In contrast, it had little affect on bromination by the eosinophil peroxidase-H 2 O 2 system. It is noteworthy that, in the absence of taurine, product yields (relative to oxidant) of 5-bromodeoxycytidine were similar with reagent HOCl, reagent HOBr, myeloperoxidase, and eosinophil peroxidase. These observations indicate that reactive species generated by HOCl or HOBr brominate deoxycytidine with similar efficiency under these conditions. They also demonstrate that use of taurine, a potent scavenger of hypohalous acids, can distinguish between bromination reactions mediated by HOBr and HOCl.
We used taurine to determine whether human neutrophils use HOCl or HOBr to brominate deoxycytidine (Fig. 11B). We activated the cells with phorbol ester, incubated them for 60 FIG . 7. Generation of 1-bromo-2-cyclohexane by gas evolved  from reaction mixtures containing HOCl (A) or myeloperoxidase (B). A, HOCl (100 nmol) was added to 2 ml of 100 mM Cl Ϫ , 100 M Br Ϫ , and 50 mM sodium phosphate (pH 5.9) at 25°C. The reaction mixture was sparged for 60 min with nitrogen gas that was subsequently bubbled through 2 ml of cyclohexene. At the end of the incubation, the cyclohexene solution was concentrated by evaporation under nitrogen and analyzed by electron ionization GC/MS. A, selected ion chromatogram of ions of m/z 198 of cyclohexane exposed to nitrogen gas passed through a reaction mixture containing HOCl and Br Ϫ (ϩ HOCl) or Br Ϫ alone (Ϫ HOCl). B, mass spectrum of material eluting at 4.5 min from cyclohexene exposed to gas evolved from the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system. The molecular ion of 1-bromo-2-chlorocyclohexane is predicted to have an m/z of 196. Note that the ions at m/z 196, 198, and 200 display the isotopic abundances expected from a compound containing Br Ϫ and Cl Ϫ .
FIG. 8. Chloride dependence of 5-bromodeoxycytidine generation by reagent HOCl. Chloride-free HOCl (50 M) was added to a rapidly stirred reaction mixture at 4°C containing 1 mM deoxycytidine, 100 M NaBr, 50 mM sodium phosphate, pH 5.9, and the indicated concentration of NaCl. After mixing for 5 s, the reaction was stopped by addition of L-methionine (6 mM). 5-Bromodeoxycytidine was quantified by reverse-phase HPLC. Values are the means of duplicate experiments and are representative of the results found in three independent experiments. FIG. 9. Effect of pH on the oxidation of deoxycytidine by taurine monobromamine, (q), taurine monochloramine (E), and  N-chlorodeoxycytidine (N-chloro-dC, ). The reaction was initiated by adding the haloamine (50 nmol) to 1 ml of buffer A containing 1 mM deoxycytidine and 100 M NaBr. After a 60-min incubation at 37°C, the reaction was terminated by adding 6 mol of L-methionine, and products were quantified by reverse-phase HPLC. The pH of the reaction mixture was varied using a mixture of phosphoric acid and sodium phosphate. The pH of the reaction mixture was determined at the end of the incubation without adding methionine.

TABLE II
Requirements for conversion of deoxycytidine into 5bromodeoxycytidine by human neutrophils The complete system consisted of 1 ϫ 10 6 cells/ml in Dulbecco's phosphate-buffered saline solution (pH 5.9) supplemented with 100 M Br Ϫ , 1 mM deoxycytidine, and 1 mg/ml dextrose. Incubations were performed as described in the legend to Fig. 10 min at pH 5.9 in buffer containing 100 M NaBr, 100 mM NaCl, and 1 mM deoxycytidine, and used HPLC to determine whether 5-bromodeoxycytidine was produced. Taurine markedly inhibited the brominating ability of neutrophils (Fig. 11B), with an IC 50 similar to that of the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system (Fig. 11A). These observations strongly support the hypothesis that the major pathway by which human neutrophils generate reactive brominating species involves HOCl production followed by oxidation of Br Ϫ .

DISCUSSION
Our observations demonstrate that the myeloperoxidase system of human neutrophils generates brominating oxidants that convert uracil into 5-bromouracil and transform deoxycytidine into 5-bromodeoxyuridine (Scheme 2). Importantly, these pyrimidines become halogenated at plasma concentrations of Br Ϫ (100 M) and Cl Ϫ (100 mM), suggesting that this pathway may be physiologically relevant.
Hypochlorous acid is thought to be the major product when myeloperoxidase is exposed to plasma halide concentrations. In contrast, eosinophil peroxidase preferentially converts Br Ϫ to HOBr. It was therefore important to determine whether myeloperoxidase was oxidizing Cl Ϫ or Br Ϫ in our experiments. We obtained several lines of evidence to support the proposal that myeloperoxidase initially generates HOCl, which then is converted into reactive brominating species. First, reagent HOCl brominated uracil and deoxycytidine as effectively as HOBr. The HOCl-mediated reaction required Br Ϫ and was optimal under acidic conditions. Second, taurine, which rapidly reacts with HOCl to form N-chlorotaurine, inhibited bromination of deoxycytidine by HOCl-Br Ϫ or the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system. Importantly, kinetic studies indicate that chlorination of taurine by myeloperoxidase may involve an enzymatic intermediate, suggesting that taurine scavenges chlorinating species before they can diffuse out of the active site of the enzyme (14). In contrast, taurine had little effect on bromination by HOBr or the eosinophil peroxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system. Taurine thus selectively inhibits bromination mediated by HOCl under these experimental conditions. Finally, human neutrophils converted deoxycytidine to 5-bromodeoxycytidine by a reaction requiring H 2 O 2 and Br Ϫ . Bromination was optimal under acidic conditions and was inhibited by taurine and heme poisons, implicating HOCl and myeloperoxidase in the reaction pathway. Collectively, these observations indicate that activated human phagocytes use myeloperoxidase to brominate pyrimidines at plasma concentrations of halide by reactions that initially require HOCl.
A key question is the mechanism by which HOCl then generates reactive brominating species. Based on the chemistry of the interhalogen compounds (40,41), we propose that BrCl is  Fig. 2. B, reactions with activated human neutrophils (1 ϫ 10 6 /ml) were performed as described in the legend to Fig. 10. All buffers contained 100 mM Cl Ϫ , 100 M Br Ϫ , 1 mM deoxycytidine, and the indicated final concentration of taurine. After a 60-min incubation, reactions were terminated by adding 6 mM L-methionine. Cells were removed by centrifugation. 5-Bromodeoxycytidine was quantified by reverse-phase HPLC. Values are the means of two independent experiments with duplicate determinations (enzymes and hypohalous acid) or the means of three independent experiments with duplicate determinations (neutrophils). one potential intermediate in the pathway. BrCl is a stronger brominating oxidant than Br 2 or HOBr (42), suggesting that it halogenates pyrimidines. Alternatively, bromination may be mediated by HOBr or Br 2 , which are in equilibrium with BrCl.
To directly detect BrCl, we sparged a reaction mixture containing HOCl-Br Ϫ or the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system with nitrogen gas that we subsequently passed through cyclohexene. Mass spectrometric analysis of the resulting cyclohexene solution detected an ion with the expected mass-tocharge ratio, GC retention time, and isotopic abundance of 1-bromo-2-cyclohexane. Cyclohexene is an aprotic, nonpolar solvent that should not contain halide ions under these conditions. Thus, bromination of cyclohexene cannot involve a backsided nucleophilic attack of a bromonium ion intermediate by free Cl Ϫ in solvent. Instead, it is likely to involve the concerted attack on the double bond by [Br ϩ -Cl Ϫ ] derived from molecular BrCl (Scheme 1). Detecting 1-bromo-2-cyclohexane therefore provides strong evidence that HOCl and myeloperoxidase generate the interhalogen gas BrCl.
Our demonstration that [Cl Ϫ ] increased the yield of 5-bromodeoxycytidine generation by reagent HOCl and Br Ϫ is consistent with this proposal. Chloride ion is likely to stimulate bromination by reacting with HOBr to form BrCl (Equation 6). These observations further support the hypothesis that HOBr and HOCl are in equilibrium with BrCl under acidic conditions in the transhalogenation pathway (Scheme 3).
Another important question is whether the acidic conditions that are optimal for myeloperoxidase to generate brominating intermediates are likely to exist in vivo. We suggest that the acidic conditions that result from infection or tissue hypoxia at sites of inflammation may represent such an environment (54,55). Also, cultured activated macrophages form phagocytic compartments in which the pH falls below 4 (56). Moreover, strongly acidic environments exist in the phagolysosome and stomach, where myeloperoxidase-derived HOCl (or chloramines) might generate brominated pyrimidines. Human neutrophils use a Cl 2 -like species to generate 3-chlorotyrosine from tyrosine in vitro, and we have shown that 3-chlorotyrosine levels are markedly elevated in human atherosclerosis, a chronic inflammatory condition (13,21). Therefore, inflammation may generate acidic environments in which myeloperoxidase can produce brominating oxidants.
Chronic inflammation is an important risk factor for cancer, and oxidants generated by phagocytic cells mutate bacteria and transform cultured mammalian cells (57,58). Moreover, recent genetic epidemiological studies have found a relationship between cancer risk and a polymorphism in the promoter region of the myeloperoxidase gene. People with a polymorphism that increases myeloperoxidase expression were at increased risk for promyelocytic leukemia (26,27). In contrast, those with polymorphisms that lower myeloperoxidase expression were at decreased risk for lung and laryngeal cancers (25, 28 -30). These observations suggest that nucleobase halogenation might provide one mechanism for mutagenesis and cytotoxicity in vivo. Indeed, 5-bromouracil, one product of the myeloperoxidase-H 2 O 2 -Cl Ϫ -Br Ϫ system, is incorporated into nuclear DNA as 5-bromodeoxyuridine, a known mutagen (59 -63). Moreover, myeloperoxidase converts deoxycytidine to 5-bromodeoxycytidine, and we have shown that 5-bromodeoxycytidine generated by eosinophil peroxidase is incorporated into the genomic DNA of cultured mammalian cells as 5-bromodeoxyuridine. 1 Mutagenesis by oxidants is generally thought to occur via direct damage to DNA. However, our results suggest an additional mechanism: halogenation of nucleobases within a precursor pool. In this scenario, activated phagocytes generate halogenating intermediates, which react with nucleotides or nucleotide precursors in the extracellular and intracellular milieu. Consequently, potentially cytotoxic and mutagenic deoxynucleotide derivatives become incorporated into the genomes of daughter cells. If they find their way into tumor suppressor genes, genes for DNA repair, or potential oncogenes, they might increase the risk for cancer. A similar incorporational mechanism of mutagenesis has been suggested for the MutT system in bacteria, which cleanses the deoxynucleotide pool of 8-oxodGTP (64). When MutT is genetically inactivated, the spontaneous mutation rate increases 100 -10,000fold (65)(66)(67).
We found that two primary haloamines, N-bromotaurine and N-chlorotaurine, also brominate deoxycytidine at plasma concentrations of halide. However, the reaction with N-chlorotaurine required more acidic conditions than bromination by HOBr or HOCl; bromination by N-bromotaurine and N-chlorotaurine were half-maximal at pH 5.8 and 4.5, respectively. Also, the pyrimidine ring of cytosine contains an exocyclic amino group and two nitrogens, raising the possibility that N-chlorocytosine is an intermediate in the bromination reaction (20,68). However, we found that N-chlorodeoxycytidine was unable to generate 5-bromodeoxycytidine under our standard experimental conditions. These observations indicate that primary chloramines and bromamines, but not N-chlorocytosine, can be intermediates in the bromination of deoxycytidine. Haloamines are relatively stable compounds that can diffuse long distances and cross plasma membranes. Because hypohalous acids rapidly react with primary amino groups and extracellular fluids contain high concentrations of low molecular weight amines, it is possible that haloamines represent one mechanism for brominating nucleobases under acidic conditions.
Chronic inflammation is a risk factor for cancer, and reactive chlorinating, brominating, and nitrating intermediates generated by myeloperoxidase damage nucleobases in vitro (20,69). These observations suggest that phagocytes may constitute a physiologically important pathway for oxidative damage to DNA. This hypothesis would be strongly supported by detection of brominated or chlorinated pyrimidines in vivo, with important implications for the pathogenesis of tissue injury and perhaps mutagenesis during inflammation.