Carbon dioxide stimulates the production of thiyl, sulfinyl, and disulfide radical anion from thiol oxidation by peroxynitrite.

Reaction of peroxynitrite with the biological ubiquitous CO(2) produces about 35% yields of two relatively strong one-electron oxidants, CO(3) and ( small middle dot)NO(2), but the remaining of peroxynitrite is isomerized to the innocuous nitrate. Partial oxidant deactivation may confound interpretation of the effects of HCO3-/CO(2) on the oxidation of targets that react with peroxynitrite by both one- and two-electron mechanisms. Thiols are example of such targets, and previous studies have reported that HCO3-/CO(2) partially inhibits GSH oxidation by peroxynitrite at pH 7.4. To differentiate the effects of HCO3-/CO(2) on two- and one-electron thiol oxidation, we monitored GSH, cysteine, and albumin oxidation by peroxynitrite at pH 5.4 and 7.4 by thiol disappearance, oxygen consumption, fast flow EPR, and EPR spin trapping. Our results demonstrate that HCO3-/CO(2) diverts thiol oxidation by peroxynitrite from two- to one-electron mechanisms particularly at neutral pH. At acid pH values, thiol oxidation to free radicals predominates even in the absence of HCO3-/CO(2). In addition to the previously characterized thiyl radicals (RS.), we also characterized radicals derived from them such as the corresponding sulfinyl (RSO.) and disulfide anion radical (RSSR.-) of both GSH and cysteine. Thiyl, RSO. and RSSR.- are reactive radicals that may contribute to the biodamaging and bioregulatory actions of peroxynitrite.

The fast rate constant of this reaction and the ubiquity of the HCO 3 . /CO2 pair in biological environments indicate that CO 3 .
and ⅐ NO 2 are likely to play relevant roles in peroxynitritemediated oxidations in vivo. Recent detailed studies have established the role of CO 3 . in the potentiating effects of HCO 3 Ϫ /CO 2 on peroxynitrite-mediated tyrosine nitration (20 -23). This process has a zero order dependence on the target, being exclusively dependent on free radicals produced from peroxynitrite decay (20). The higher fluxes of free radicals produced in the presence of HCO 3 Ϫ /CO 2 and the more specific reactivity of the CO 3 . compared with the ⅐ OH radical toward tyrosine result in higher yields of the tyrosyl radical and its recombination products with ⅐ NO 2 (nitrotyrosine) and itself (dityrosine) (21)(22)(23). The effects of HCO 3 Ϫ / CO 2 on the oxidation of targets that compete with CO 2 for the oxidant by reacting with peroxynitrite through second order processes have yet to be elucidated.

EXPERIMENTAL PROCEDURES
Materials-All reagents were purchased from Sigma, Merck, or Fisher and were analytical grade or better. Peroxynitrite was synthesized from 0.6 M sodium nitrite and 0.65 M hydrogen peroxide in a quenched flow reactor; excess hydrogen peroxide was used to minimize nitrite contamination. To eliminate excess hydrogen peroxide, the peroxynitrite solution was treated with manganese dioxide. Synthesized peroxynitrite contained low levels of contaminating hydrogen peroxide (Ͻ1%) and nitrite (10 -30%) which were determined as described previously (29) by the titanyl method and by absorbance measurements at 354 nm (⑀ ϭ 24.6 M Ϫ1 . cm Ϫ1 ), respectively. The concentration of peroxynitrite stock solutions was determined spectrophotometrically at 302 nm using an extinction coefficient of 1.670 M Ϫ1 . cm Ϫ1 (30). The thiol group of bovine serum albumin (fraction V) was blocked by reaction with N-ethylmaleimide as described previously (27). Concentrations of CO 2 were calculated from the added HCO 3 Ϫ concentrations by using pK a ϭ 6.4 (15). Buffers were pretreated with Chelex-100 to remove contaminant metal ions. All solutions were prepared with distilled water purified with a Millipore Milli-Q system.
Oxygen Consumption-Oxygen uptake studies were performed using an oxygen monitor (Gilson 5/6 oxygraphy) at 25°C Ϯ 1°C. The saturation oxygen concentration at this temperature was taken as 250 M.
EPR Experiments-The EPR fast flow spectra were recorded at room temperature (25 Ϯ 2°C) on a Bruker EMX spectrometer operating at 9.65 GHz and 100 KHz field modulation equipped with a Bruker ER4117 D-MTV dielectric mixing resonator with a 9-mm distance between the mixing cell and the resonator center. Thiol solutions were pre-prepared in appropriated buffers to which sodium bicarbonate was added or not; in the former case, the solutions were left undisturbed for 5 min to permit HCO 3 Ϫ /CO 2 equilibration. Peroxynitrite solutions were prepared with water. Thiol and peroxynitrite solutions were transferred to 60-ml plastic syringes mounted on a syringe infusion pump (Harvard apparatus pump 22). Spectra were recorded 3.5 and 12 ms after mixing at continuous flow of 20 ml/min and 6 ml/min, respectively. The dispensed mixtures were collected for pH measurement at the end of the experiments to detect changes caused by mixing with alkaline solutions of peroxynitrite. The magnetic field was calibrated with 4-hydroxy-2,2,6,6,-tetramethyl-1-piperidinyloxy (g ϭ 2.0056) (15). Computer simulation of spectra was performed using a program written by Duling (31). In the EPR spin trapping experiments, the incubation mixtures were transferred to a flat cell and the spectra recorded at room temperature 1 min after the addition of peroxynitrite.

Oxidation of GSH and Cysteine-To evaluate the effects of HCO 3
Ϫ /CO 2 on peroxynitrite-mediated oxidation of thiols, we monitored cysteine and GSH oxidation at pH 5.4 and 7.4 by both thiol disappearance (Table I) and oxygen consumption (Fig. 2). Oxygen uptake provides a nonspecific measurement of radicals that react fast with oxygen such as RS ⅐ (33,34). The results show that HCO 3 Ϫ /CO 2 inhibited GSH and cysteine thiol disappearance at pH 7.4 (Table I) but in parallel increased the amount of consumed oxygen ( Fig. 2). At pH 5.4, HCO 3 Ϫ /CO 2 had marginal effects on both thiol depletion (Table I) and oxygen consumption (Fig. 2). These results indicate that HCO 3 Ϫ /CO 2 diverts peroxynitrite-mediated thiol oxidation from two-to oneelectron mechanisms as initially hypothesized ( Fig. 1). At pH 5.4 the CO 2 effects were marginal because at this pH most of the thiol oxidation occurs by free radical mechanisms even in the absence of HCO 3 Ϫ /CO 2 . These conclusions are well supported by comparison of the experimental yields of depleted thiol with the yield of products, radicals ( ⅐ OH/ ⅐ NO 2 ; CO 3 . / ⅐ NO 2 ), and thiol oxidized by two electrons (RSOH), which can be estimated from the known rate constants of the main competing reactions occurring under the experimental conditions used (Fig. 1, paths 1-3) (see "Experimental Procedures" and Table  I). The data presented in Table I show that GSH oxidation at pH 5.4 occurs through the radicals produced from peroxynitrite in the presence or absence of HCO 3 Ϫ /CO 2 . The produced radicals are different, but they are expected to be formed in similar yields (Table I), and all of them ( ⅐ OH/ ⅐ NO 2 /CO 3 . ) react quickly with GSH to produce GS ⅐ (Table II). Relevantly, the experimental values of depleted thiol (0.30 and 0.26 mM) were the same as the calculated total radical yields in the presence (0.10 ϩ 0.20 mM) and absence (0.26 mM) of HCO 3 Ϫ /CO 2 , respectively (Table  I). Compared with GSH, a larger fraction of cysteine is oxidized by two-electron mechanisms at pH 5.4 in the absence of HCO 3 Ϫ / CO 2 because of the higher second order rate constant of its reaction with peroxynitrite (24,25). In this case, HCO 3 Ϫ /CO 2 inhibits total thiol disappearance to some extent, but the oxidized fraction should result from the produced radicals whose expected total yield is, again, similar to the measured depleted thiol (Table I). At pH 7.4 both thiols are oxidized mainly by two-electron mechanisms in the absence of HCO 3 Ϫ /CO 2 . In this case, the experimental value of depleted thiol is considerably higher than the calculated one, confirming that the direct reaction between peroxynitrite and RSH consumes more than one thiol (Fig. 1, path 3) (17). At pH 7.4 in the presence of HCO 3 Ϫ /CO 2 , total depleted thiol decreased, but most of it should result from free radical mechanisms because of the similar concentration values of produced radicals and depleted thiols (Table I). In agreement, consumed oxygen increased in the presence of HCO 3 Ϫ /CO 2 , particularly in the case of GSH (Fig. 2).
Although the yield of depleted thiol by free radical mechanisms was roughly similar to total radical yields (Table I), it is not possible to infer that RS ⅐ are decaying mainly by reaction with themselves (Reaction 1) because a fast consumption of oxygen was associated with thiol oxidation by peroxynitrite under all experimental conditions tested (Fig. 2) (see also Ref. 25). Because oxygen consumption was lower than oxidized thiol ( Fig. 2 and Table I), it is likely that RS ⅐ is decaying by at least three competing routes, i.e. dimerization, reaction with oxygen, and reaction with excess thiolate (RS Ϫ ) (Reactions 1-3).
Other decay routes also occur as indicated by the detection of low levels of GSNO 2 (41) and GSNO (42) in incubations of GSH with peroxynitrite. The importance of Reactions 2 and 3 arises from the production of sulfonyl (RSOO ⅐ ) and disulfide radical anions (RSSR ⅐ Ϫ ), respectively (Table II). In previous studies of thiol oxidation by peroxynitrite, we detected RS ⅐ radicals by EPR spin trapping and obtained indirect evidence for their conversion to both RSOO ⅐ and RSSR ⅐ Ϫ (25). Here, we present direct EPR evidence for the formation of these species and for the stimulatory effects of HCO 3 Ϫ /CO 2 on their yields. EPR Detection of Thiyl, Sulfinyl, and Disulfide Radical Anion-We have demonstrated previously that fast flow EPR of concentrated solutions of peroxynitrite and CO 2 produces detectable concentrations of CO 3 . (15). In the presence of 5 mM cysteine (Fig. 3) or 5 mM GSH (Fig. 4), the one-line EPR signal of CO 3 . produced from 5 mM peroxynitrite and 5 mM CO 2 became barely detectable (indicated by q in Figs. 3 and 4) and new EPR signals appeared, particularly at pH 7.4. Cysteine oxidation produced a three-line signal (a 2H ϭ 9.3 G; line width ϭ 3.5 G; g ϭ 2.0107) (Fig. 3) which has been characterized previously as the corresponding sulfinyl radical (CysSO ⅐ ) by fast flow EPR studies of cysteine oxidation by Ti(III)-H 2 O 2 (43). GSH oxidation produced a four-line signal (a H ϭ 7.1 G and a H ϭ 10.7 G; line width ϭ 2.6 G; g ϭ 2.0109) whose EPR parameters are also consistent with the corresponding sulfinyl radical (GSO ⅐ ) (Fig.  4). Both CysSO ⅐ and GSO ⅐ have two ␤-methylene hydrogens, but in the latter, a hindered rotation of the methylene group that is adjacent to a chiral center is likely to result in the magnetic nonequivalence (44,45) reflected in the different hyperfine splitting constants (a H ) obtained for its two hydrogens (Fig. 4B). Direct EPR detection of RSO ⅐ during peroxynitrite-mediated oxidation of thiols is consistent with the initial formation of RS ⅐ that react with oxygen to produce RSOO ⅐ which, in turn, as metastable intermediates, yield RSO ⅐ (Reaction 2) (46). Thiyl radicals cannot be detected by direct EPR in aqueous solutions at room temperature because of the large anisotropy in their g tensors which broadens the EPR signal beyond detection (43). They are detectable in aqueous solutions by EPR spin trapping. Indeed, addition of the spin trap DMPO to the fast flow mixtures containing the thiols, HCO 3 Ϫ /CO 2 and peroxynitrite led to the substitution of the RSO ⅐ spectra by those characteristic of DMPO/ ⅐ SCys (a N ϭ 15.2 G; a H ϭ 17.4 G) (Fig. 3E) and DMPO/ ⅐ SG (a N ϭ 14.9 G; a H ϭ 15.6 G) (Fig. 4E) radical adducts (47), confirming that RS ⅐ are the RSO ⅐ precursors (Reaction 2).
The effects of HCO 3 Ϫ /CO 2 in increasing the concentrations of RSO ⅐ radicals detected from cysteine and GSH oxidation by peroxynitrite are shown in Figs. 3 and 4. In agreement with the conclusion that thiols are oxidized mainly by two-electron mechanisms at pH 7.4 in the absence of HCO 3 Ϫ /CO 2 ( Fig. 2 and Table I), EPR signals were not detectable under these condi-

FIG. 2. Effects of HCO 3 ؊ /CO 2 on oxygen consumption during the oxidation of 1 mM GSH (A) and 1 mM cysteine (B) by 0.5 mM peroxynitrite at 25°C at pH 7.4 (full bars) and pH 5.4 (open bars).
The reactions were started by the addition of peroxynitrite, which triggered an extremely fast consumption of oxygen. Concentrations of CO 2 were calculated from the added HCO 3 Ϫ by using pK a ϭ 6.4. tions. Although the calculated total yields of peroxynitritederived radicals at pH 7.4 in the presence of HCO 3 Ϫ /CO 2 were similar to those at pH 5.4 in any condition (Table I), the detectable radical concentrations varied considerably (Figs. 3,  B-D, and 4, B-D). This is because EPR flow experiments detect instantaneous radical concentrations whose values depend not only on radical yields but also on the observation time and rates of radical formation and decay. Rates of radical formation become more important at the short observation time used to detect RSO ⅐ (3.5 ms) (48). Consequently, the higher RSO ⅐ concentrations detected in the presence of HCO 3 Ϫ /CO 2 at pH 7.4 than at pH 5.4 (Figs. 3, B and D, and 4, B and D) are likely to be due to the usually higher rate constants of thiol oxidation at alkaline pH values (25,36,49). Detection of RSO ⅐ was possible in incubations containing equimolar concentrations of peroxynitrite and RSH (Figs. 3 and  4). Increases in thiol molar ratios led to composite EPR spectra whose partial characterization was possible with a 5 molar excess thiol. Such excess, low oxygen tensions, and the presence of HCO 3 Ϫ /CO 2 led to the detection of EPR spectra dominated by the spectra of GSSG ⅐ Ϫ (Fig. 5B) or CysSSCys ⅐ Ϫ (Fig.  5D). These spectra have been characterized previously in flow mixtures of horseradish peroxidase/H 2 O 2 /acetaminophen and the corresponding thiols (50). The spectra shown in Fig. 5 were obtained with four times lower concentrations of thiols and consequently were less resolved than those published previously (50). Computer simulation (noiseless lines in Fig. 5) of the experimental spectra obtained in mixtures of excess thiol, peroxynitrite, and HCO 3 Ϫ /CO 2 confirmed that GSSG ⅐ Ϫ (a 2H ϭ 6.9 G; a 2H ϭ 7.0 G) and CysSSCys ⅐ Ϫ (a 2H ϭ 6.6 G; a 2H ϭ 7.7 G) are the predominant species produced under these conditions. Again, RSSR ⅐ Ϫ attained detectable concentrations only in the presence of HCO 3 Ϫ /CO 2 (Fig. 5). Oxidation of Albumin Thiol-The main target of peroxynitrite in albumin is its free thiol group (17,51). It has been reported that HCO 3 Ϫ /CO 2 partially inhibits albumin thiol oxidation by peroxynitrite (19). This result was confirmed under our experimental conditions where 0.5 mM peroxynitrite at pH 7.4 depleted 0.68 Ϯ 0.02 mM and 0.56 Ϯ 0.02 mM albumin thiols in the absence and in the presence of 1 mM CO 2 , respectively. Although it inhibited total thiol depletion, 1 mM CO 2 increased about two times the yield of albumin thiyl radical that was trapped by POBN (Fig. 6, A and B). All POBN/ ⅐ protein radical adducts have similar EPR spectra, a broad triplet characteristic of high molecular weight nitroxides (52). Thus, to prove that albumin-thiyl is the main radical produced under the experimental conditions employed, parallel experiments were performed with albumin pretreated with N-ethylmaleimide. EPR signals were barely detectable from albumin whose thiol group was previously blocked with N-ethylmaleimide (Fig. 6, C and  D). Under conditions of high molar excess of peroxynitrite over albumin thiol, other protein-derived radicals are produced and detected (53), as expected from the reactivity of peroxynitritederived radicals. DISCUSSION Presently, most investigators agree that CO 2 is likely to be the major sink of peroxynitrite in most physiological environments (2, 5, 10). Carbon dioxide deactivates about 65% of peroxynitrite because it catalyzes isomerization of the oxidant to the innocuous nitrate, although the remaining products (about 35%) are CO 3 . and ⅐ NO 2 (Fig. 1, path 2). The high rate constant of the reaction between ONOO Ϫ and CO 2 and the resulting deactivation of 65% of the oxidant may lead to the conclusion that HCO 3 Ϫ /CO 2 inhibits the oxidation of targets that react with peroxynitrite directly, competing with CO 2 for the oxidant. Thiols are examples of such targets (Fig. 1, path 3) (17, 24 -26). Here we demonstrate that although HCO 3 Ϫ /CO 2 inhibits the total yield of GSH and cysteine oxidized by peroxynitrite at pH 7.4 (19,26), the oxidized fraction produces free radicals (Figs. 1-6 and Table I). Carbon dioxide inhibits thiol oxidation because it outcompetes the thiols for the direct reaction with peroxynitrite. As a consequence, thiols are oxidized to RS ⅐ by CO 3 . and ⅐ NO 2 that escape solvent cage ( Fig. 1 and Table   I). Subsequent reactions of RS ⅐ (Reactions 1-3) produce the detectable RSO ⅐ (Figs. 3 and 4) and RSSR ⅐ Ϫ (Fig. 5). Undoubt-edly, the effects of HCO 3 Ϫ /CO 2 in diverting peroxynitrite reactivity from two-to one-electron mechanisms may apply to other important biological targets that react directly with the oxidant such as hemoproteins (54,55). The demonstration that at acid pH values peroxynitrite acts as an one-electron thiol oxidant in the presence and in the absence of HCO 3 Ϫ /CO 2 (Figs. 2-4) was also relevant. Acid pH values are important for phagocyte function and ischemic tissue pathology but have been overlooked in studies addressing peroxynitrite reactivity in biological environments. Because the three radicals that can be produced from peroxynitrite are strong ( ⅐ OH, E ϭ 2.3 V (24); and CO 3 . , E ϭ 1.8 V (15, 56 -58) 2 or moderately strong ( ⅐ NO 2 , E ϭ 0.99 V) (24) one-electron oxidants, they are likely to mediate some of the biological damage attributed to peroxynitrite such as nitration of protein tyrosine residues (2, 5, 20 -23) and protein thiol oxidation (27,59,60).
Formation of RS ⅐ radicals during the oxidation of low and high molecular weight thiols by peroxynitrite has been reported in the literature by our group and others (6,25,(27)(28)(29). However, this report provides the first direct EPR detection of RSO ⅐ (Figs. 3 and 4) and RSSR ⅐ Ϫ (Fig. 5) formed from the initially produced RS ⅐ (Reactions 1-3). All low molecular weight radicals, GSO ⅐ , GSSG ⅐ Ϫ , CysSO ⅐ , and CysSSCys ⅐ Ϫ were unambiguously characterized by the parameters of their corresponding EPR spectra (see "Results"). Moreover, it was shown that the presence of HCO 3 Ϫ /CO 2 stimulated the production of RS ⅐ and of all the radicals derived from them (Figs. 3-6). Among the identified radicals those that may have greater physiological consequences are GSSG ⅐ Ϫ (Fig. 5B) and protein-S ⅐ (Fig. 6). Formation of protein-S ⅐ and protein-SO ⅐ may alter protein structure and function (61,62), and relevantly, proteinthiol oxidation is a recurrent event in cell signaling cascades (63). On the other hand, because GSSG ⅐ Ϫ is produced under conditions of excess GSH over peroxynitrite, it is likely to be the prevalent radical produced from peroxynitrite attack on thiols in aqueous intracellular environments where oxidant concentration is expected to be much lower than the GSH and HCO 3 Ϫ / CO 2 concentration. The GSSG ⅐ Ϫ radical reacts fast with oxygen, producing superoxide anion (Table II), which can further react with thiols, although not particularly fast, to produce RS ⅐ (64). Consequently, intracellular GSH oxidation by peroxynitrite-derived radicals may trigger an oxygen-dependent free radical chain reaction, as was shown to occur in vitro (25). This possibility argues against GSH being particularly effective in directly counteracting the oxidant properties of peroxynitritederived radicals. In contrast, an antioxidant such as ascorbate which is oxidized to a radical that reacts very slowly with oxygen is expected to be more effective. Moreover, ascorbate reacts faster with CO 3 . than GSH (65).Consequently, in aqueous environments, ascorbate may be more important than GSH in counteracting the oxidant action of free radicals. This has been shown to occur in the case of the acetaminophen-derived radical (50) and more recently, in the case of peroxynitritederived radicals (65).
In conclusion, our results demonstrate that HCO 3 Ϫ /CO 2 diverts low and high molecular weight thiol oxidation by peroxynitrite from two-to one-electron mechanisms particularly at neutral pH. At acid pH values, thiol oxidation to free radicals predominates in the presence and absence of HCO 3 Ϫ /CO 2 . The produced thiol-derived radicals were identified as RS ⅐ , RSO ⅐ , 2 The reduction potential of CO 3 . has usually been assumed to be 1.6 V. The determined value was 1.58 Ϯ 0.02 V(56, 58) and corresponds to the pair CO 3 . /CO 3 2Ϫ . At pH 7.0, the relevant pair is CO 3 . C,H ϩ /HCO 3 Ϫ (pK a of HCO 3 Ϫ ϭ 10.32) whose reduction potential can be calculated as 1.78 V (E CO 3 . ,Hϩ/HCO 3 Ϫ ϭ E CO 3 . /HCO 3 2Ϫ ϩ 0.059 ϫ (pK a Ϫ pH).
FIG. 5. EPR continuous flow spectra of GSSG ⅐؊ and CysSS-Cys ⅐؊ radical anions produced from mixing 5 mM peroxynitrite and 25 mM GSH or cysteine with or without 5 mM CO 2 at pH 7.4. A, GSH; B, GSH in the presence of CO 2 ; C, cysteine; D, cysteine in the presence of CO 2 . The specified concentrations are those in the final incubation mixture. The flow rate was 6 ml/min. Instrumental conditions: microwave power, 2 mW, time constant, 81.9 ms; scan rate, 0.6 G/s; modulation amplitude, 2 G; gain, 1.78 ϫ 10 6 . The noiseless lines correspond to computer simulation of the GSSG ⅐ Ϫ (a 2H ϭ 6.9 G; a 2H ϭ 7.0 G) and CysSSCys ⅐Ϫ (a 2H ϭ 6.6 G; a 2H ϭ 7.7 G) spectra.
FIG. 6. EPR spectra of the POBN/ ⅐ S-albumin radical adduct obtained from a 1-min incubation of 1 mM albumin with 25 mM POBN and 0.5 mM peroxynitrite at pH 7.4. A, in the absence of added CO 2 ; B, in the presence of 1 mM CO 2 ; C, the same as A but with albumin whose thiol group was blocked; D, the same as B but with albumin whose thiol group was blocked. Instrumental conditions: microwave power, 20 mW, time constant, 81.9 ms; scan rate, 0.6 G/s; modulation amplitude, 1G; gain, 8.93 ϫ 10 5 .