Nitroxidative, Nitrosative, and Nitrative Stress: Kinetic Predictions of

Douglas D. Thomas , Julie L. Heinecke , Lisa A. Ridnour , Robert Y. Cheng , Aparna H. .... Rebecca S. Silkstone , Maria G. Mason , Peter Nicholls , Ch...
0 downloads 0 Views 592KB Size
1160

Chem. Res. Toxicol. 2006, 19, 1160-1174

Nitroxidative, Nitrosative, and Nitrative Stress: Kinetic Predictions of Reactive Nitrogen Species Chemistry Under Biological Conditions Jack R. Lancaster, Jr.† Departments of Anesthesiology, EnVironmental Health Sciences, and Physiology & Biophysics, Center for Free Radical Biology, The UniVersity of Alabama Birmingham, 901 19th Street, South, Birmingham, Alabama 35294 ReceiVed March 20, 2006

A freely available Windows-based program, RNSim1A, is utilized to predict metal-independent reactive nitrogen species (RNS) chemistry (oxidation, nitrosation, and nitration) under simulated biological conditions and make the following specific predictions. (1) The peak in oxidative reactions that occurs in Vitro with 1:1 fluxes of •NO and O2•- does not occur under biological conditions. (2) By far, the quantitatively dominant (92-99.6%) process in ViVo is oxidation, compared to nitrosation and nitration. (3) Only five of the many possible RNS reactions involving thiol (glutathione, GSH) and tyrosine are quantitatively important biologically. (4) Under inflammatory conditions, approximately 1% of O2•- reacts with •NO to produce ONOO-, with the remainder reacting with SOD. (5) The dominant reaction of tyrosyl radical is a radical swap with GSH, producing the glutathiyl radical and regenerating tyrosine. (6) Nitrosothiol is formed virtually exclusively via radical recombination (RS• + •NO) as opposed to reaction with nitrous anhydride (N2O3). (7) Nitrosothiol is an intermediate, not an endproduct, and responds dynamically to changes in the immediate chemical environment. (8) The formation of a nitroso group on a particular thiol can be considered a marker of increased reactivity of that thiol, and it is likely that other modifications of that thiol (oxidation, glutathiolation) are more abundant than nitrosation and may be the functionally significant modification. (9) Specific chemical mechanisms are proposed for posttranslational protein modification via nitrosation, nitration, glutathiolation, and also dithiol/disulfide exchange, with important roles for the thiolate anion and O2 (suggesting possible mechanisms for O2 sensing) and variable degrees of exposure of cysteine thiol and tyrosine phenolate. (10) Patterns of reactivity are similar for low (20 nM) and high (500 nM) steady-state levels of NO. (11) The dominant reactions are those involving reactants at the highest concentrations (CO2, thiol, O2). Because of the dominance of oxidative processes caused by RNS, the term nitroxidative stress is proposed, emphasizing the oxidative (as opposed to nitrosative or nitrative) stress that dominates RNS actions under biological conditions. Introduction Most of the individual chemical reactions of reactive nitrogen species (RNS1) have been kinetically characterized in isolation, where the composition of the reaction medium can be precisely controlled. However, because of experimental limitations, many of these reactions cannot be studied under biologically relevant conditions. In addition, although all reactions that occur in ViVo also occur in aqueous solution, in the complex environment of the cell, some reactions will be more dominant than others because of competition for reactants and intermediates. In addition, in experiments in Vitro, it is not always appreciated that it is the concentrations of reactants, rather than fluxes of formation, that determine the outcome. Given the kinetic characterization of these individual reactions in isolation, it is, however, possible to predict the dominant reactions, intermediates, and products as they would occur in the biological milieu. I present here a kinetic network incorporating and integrating † To whom correspondence should be addressed. Phone: 1-205-9759673. Fax: 1-205-934-7437. E-mail: [email protected]. 1 Abbreviations: 3-NT, 3-nitrotyrosine; DAF, diaminofluorescein; diTyr, dityrosine; DHR, dihydrorhodamine; EDRF, endothelium-derived relaxing factor; GSH, glutathione; GSNO, S-nitrosoglutathione; GSNO2, S-nitroglutathione; GSOH, glutathionesulfenic acid; GSSG, oxidized glutathione; NOS, nitric oxide synthase; RNS, reactive nitrogen species; SOD, superoxide dismutase; TyrO•, tyrosyl radical; vNO, flux of •NO production; vO2•-, flux of O2•- formation.

three major reaction types of nitrogen oxides, oxidation, nitrosation, and nitration. A Windows-based computer program is presented, which utilizes user-specified initial conditions to predict the rates of reaction and concentrations of intermediates and products of RNS chemistry with time, and is freely available for downloading.

Materials and Methods The simulations presented here were performed on a laptop PC running Windows XP with a Visual Basic program, which is available as the executable program RINSim1A.exe, located at http://nitricoxide.anes.uab.edu/LLab/RNSim/Index.htm. After downloading and unzipping the folder, the program is installed by running Setup. This program utilizes initial parameters supplied by the user to produce a comma-delimited data file that can be imported into any graphics application. The data are in the form of columns, and each column contains the predicted values for concentrations and rates for the specific times in column 1. One thousand equally spaced time points over the time range supplied by the user are written into the data file. The algorithm used for numerical integration of the rate equations is a simple Euler’s method. The scheme used for this program is shown in Scheme 1. Unless indicated otherwise, all rates refer to steady-state. Specific details about using the program are available in a Help screen in the program. Additional information is contained in the Supporting Information section.

10.1021/tx060061w CCC: $33.50 © 2006 American Chemical Society Published on Web 08/18/2006

Predicting Biological ReactiVe Nitrogen Chemistry

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1161

Scheme 1. Oxidation, Nitrosation, and Nitration Reactions of •NO, O2•-, Tyrosine, and Glutathionea

a

See text for explanation. Rate constants utilized are in Supporting Information, Table 1.

Results •NO

Peroxynitrite: Bolus vs Fluxes of and Superoxide. The reaction between •NO and O2•-, well-known by chemists previously (1), provided critical support for the identification of •NO as the endothelium-derived relaxing factor (EDRF) (2). In 1990, Beckman and Freeman et al. (3) revolutionized the fields of free radical biology and nitric oxide biology by demonstrating the potent oxidizing activity of the product of this reaction, peroxynitrite (ONOO-). Many thousands of articles have since implicated peroxynitrite as a key player in cell and tissue injury in a remarkable variety of pathologies, including cardiovascular and neurodegenerative diseases, cancer, infection inflammation and the immune response, and also aging (4). Many studies have appeared examining the effects of ONOO- in biological systems, a great many involving bolus addition of an aliquot of a preformed ONOO- solution. However, under biological conditions, ONOO- is formed from the reaction of •NO with O2•-, each of which is produced as a flux of varying rates, depending on the environment and conditions. Because the reaction to produce ONOO- proceeds with an extremely rapid rate constant (ca. 2 × 1010 M-1 s-1 (5), close to the diffusion limit) under most biological conditions, there will be an excess of either reactant. In this case, it is possible that the excess radical reactant could participate further in reactions downstream from the initial reaction. Initial evidence for such a phenomenon was provided by Rubbo et al. who showed that increasing fluxes of •NO result in an increase and

then decrease in lipid peroxidation from superoxide (6). This bell-shaped behavior has been experimentally examined in detail, using the oxidizable probe dihydrorhodamine (DHR) to detect the formation of oxidizing species (7, 8). Figure 1A is adapted from the study by Miles et al. (7) and presents the oxidation of DHR under conditions of a constant rate of O2•generation (33.3 nM/s, using the system hypoxanthine/xanthine oxidase) with increasing rates of •NO generation (0-150 nM/ s, using the •NO donor SperNO). The results illustrate a phenomenon of critical biological importance, namely, that oxidation is maximal only when the flux of •NO is equal to the flux of O2•-. Additional experiments demonstrated that the same phenomenon occurs when the •NO flux is constant, and O2•flux is increased. The biological implication of this result is that oxidative chemistry only occurs under very stringent conditions (equal fluxes of both reactants), which are probably uncommon in ViVo. To understand whether this is true under biological conditions, however, requires a grasp of the mechanistic basis of this bellshaped response. One possible explanation of this effect is a direct reaction of either •NO or O2•- with ONOO-, thus converting it into the potent nitrosating species nitrous anhydride (N2O3) or nitrogen dioxide (•NO2), respectively (9). However, there is no evidence for a direct reaction of O2•- with ONOO-, and the reaction of •NO with ONOO- is too slow to explain this result (10). An alternative possibility, for which experimental support has appeared (8, 11), is that the products of

1162 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

Lancaster

Figure 1. Oxidation during various fluxes of •NO and O2•-. (A) Oxidation of DHR, adapted from Figure 1 of Miles et al. (7). The rate of superoxide generation vO2•- is 33.3 nM/s. (B) Scheme utilized to simulate oxidation. The rate constants are presented in Supporting Information, Table 1. (C) Steady-state •NO and O2•- concentrations. (D) Predicted steady-state concentrations of •NO2 + •OH and ONOO-, using the conditions in A. (E) Sum of •NO2 and •OH steady-state concentrations with increasing values for k12 and k13. The peak absolute concentrations ([•NO2] + [•OH]) were 2.75 pM (for k12 ) k13 ) 7 × 103 s-1), 0.772 pM (for k12 ) k13 ) 2.5 × 104 s-1), and 0.241 pM (for k12 ) k13 ) 8 × 104 s-1).

homolysis of peroxynitrous acid, •NO2 and •OH rather than ONOO- or ONOOH, are responsible for DHR oxidation. The bell-shaped result would presumably be due to the reactions of •NO and •OH with the excess •NO and O •-. Because the rate 2 2 constants for these reactions have been determined experimentally, it is possible to model this system of reactions and quantitatively test its competence to predict the experimental results. The scheme in Figure 1B depicts the reactions used to predict these results. The rates of •NO and O2•- formation, set by the experimental conditions, are given by rate constants k1 and k2. The rate constants for all other reactions, with the exception of the rates of disappearance of •NO2 and •OH by reactions other than with •NO or O2•- (k12, k13), are taken from experimental data and presented in Table 1, online Supporting Information. Initial examinations revealed that several other reactions (specifically, the reaction of •NO with ONOO- (10), N2O3 with ONOO- (12), NO with O2 (13), and •OH with ONOO- (14)) can be ignored, but the reaction of NO2- with •OH (15) results in the effective conversion of a substantial amount of •OH into •NO and, therefore, must be taken into consideration. 2 Figure 1C shows the predicted steady-state concentrations of •NO and O2•-, with the fluxes of •NO and O2•- as for Figure

1A, with the values for the rate constants for disappearance of •NO and •OH (k , k ) set at 103 s-1. Under these conditions, 2 12 13 as •NO flux is increased, it reacts with and removes O2•- until the point of equal fluxes, where both species disappear equally. Above this flux, •NO is in excess. Figure 1D shows the concentrations of ONOO- and the sum of •NO2 and •OH. The profile for •NO2 and •OH mirrors quite well the data reported by Miles et al. (7) (Figure 1A) and by Jourd’heuil et al. (8), although the profile for ONOO- increases linearly with •NO flux until the rate is equal to O2•- flux, and above this value, the concentration is constant. This is to be expected because ONOO- formation will be determined by the flux of the reactant produced at the slower rate. This phenomenon is examined in more detail below. These results provide additional verification that the reactive species that oxidizes DHR is not ONOObecause if this were true, then the rate of DHR oxidation would plateau rather than peak and decline. The bell-shaped behavior is, however, consistent with the product(s) of ONOOH homolysis, •NO2 and/or •OH being the oxidizing species2 (8, 11). 2 Simulations including the reaction of ONOO- with CO , as described 2 below, reveal a similar result, except that the oxidizing species is CO3•and/or •NO2 (data not shown).

Predicting Biological ReactiVe Nitrogen Chemistry

For the simulations presented in Figure 1D, the only adjustable parameters are the rates of disappearance of •NO2 and •OH via processes not involving a reaction with •NO or O2•- (k12, k13); all other parameters are either experimentally set or taken from the literature. Figure 1E illustrates the effects on the shape of the [•NO2] + [•OH] profile when these rates (k12, k13) are increased above 103 s-1. An examination of the individual reactions (not shown) reveals that the loss of the bell shape results from a shift in the major routes of disappearance of •NO2 and •OH from the reaction with the excess reactant (•NO, O2•-) to a disappearance via the processes described by k12 and k13. This means that in order to observe the bell-shaped curve, the rates of reaction of •NO2 and •OH with •NO or O2•- must be faster than the reaction of •NO2 and •OH with other targets. Whether this condition holds under biologically relevant conditions is examined in detail below. Validation of the Model: Nitrosation and Nitration. My ultimate goal is to apply this method of analysis to a complex system of reactions under simulated biological conditions in order to predict the most important of the many possible reactions. However, the model must first be validated by showing it can accurately predict experimental results. In addition to oxidation (dealt with in the previous section), the other two major types of RNS radical reaction are nitrosation and nitration, and the experimental results to be modeled are those involving different fluxes of •NO and O2•- formation (as opposed to bolus ONOO- addition). The first refinement of the model to mimic biological conditions is the recognition that the presence of CO2 will alter the mechanism of production of reactive oxidant from ONOO-. As initially pointed out by Lymar and Hurst (16), the rate of reaction between ONOO- and CO2 is sufficiently rapid that it will be a dominant factor in determining the chemistry of ONOO- under biological conditions. It is now known that this reaction involves the formation of an extremely short-lived adduct, nitrosoperoxocarbonate, which homolyzes to form about 33% •NO2 + CO3•-, and the remainder forms NO3- and CO2 (17-19).

ONOO- + CO2 f [ONOOCO2-] f NO2 + CO3•- (33%), NO3- + CO2 (67%) (1)



The rate constant for this reaction (3 × 104 M-1 s-1 (16)) means that under virtually all conditions (unless efforts are made to stringently exclude atmospheric CO2) this reaction will preclude the formation of •NO2 and •OH from ONOOH homolysis, and therefore, for all further simulations, ONOOH homolysis is not considered. Using controlled fluxes of •NO and O2•-, Espey et al. (20) studied the effects of O2•- fluxes on nitrosation with •NO fluxes. Figure 2A is adapted from their work and demonstrates that the stimulation by O2•- of nitrosation (detected using the fluorogenic substrate diaminofluorescein, DAF) above the levels with •NO alone occurs only with O2•- fluxes in a rather narrow range, which peaks below the •NO flux. For a given O2•- flux, as the •NO flux increases, nitrosation peaks, and when the •NO flux exceeds the O2•- flux, the nitrosation rate is similar to that with •NO alone (+O2). In other words, superoxide stimulates nitrosation but is maximal when the O2•- flux is lower than the •NO flux. This phenomenon was attributed to a process dubbed oxidative nitrosylation, whereby the target (in this case, DAF) is first oxidized to a radical species, which is a result of the •NO/O •- reaction, which then combines with •NO to form a 2 nitroso group (21).

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1163

Figure 2B depicts the scheme used to model these data. In this scheme nitrosation occurs both from direct reaction with N2O3 and one-electron oxidation by •NO2 and by CO3•-, followed by radical-radical recombination with •NO. Using the rates of •NO and O2•- fluxes utilized by Espey et al., Figure 2C shows the predicted steady-state rates of total nitrosation for various fluxes of •NO with several fluxes of O2•-, revealing a peak of nitrosation and then a decline, generally consistent with the corresponding experimental data (Figure 2A). Figure 2D presents several parameters for the system with 0.5 µM O2•-/ min and variable fluxes of •NO, which reveal the underlying mechanistic explanations for the lag, peak, and then ramping of total nitrosation. The profile can be divided into three phases (A, B, C), as depicted in Figure 2D. Overall, an important concept is that both nitrosative processes, radical combination and nitrosation via N2O3, require the formation of •NO in excess of that which reacts with O2•-; for the recombination with the radical for the former and for the reaction with •NO2 for the latter. Thus, in phase A, the lack of nitrosation can be attributed to the stoichiometric removal of •NO by reaction with O2•-, as attested to by the presence of excess O2•- as it is titrated by the increasing •NO. As the •NO flux begins to exceed O2•- flux (phase B), the excess •NO reacts with R•, which is formed from oxidation by •NO2. As •NO flux increases further, the reaction of •NO with •NO2 dominates (phase C), resulting in the prevention of radical combination (because •NO2 is the major oxidant for the formation of R• from RH because of the rapid removal of CO3•- via the reaction with •NO; not shown), but now it results in increased nitrosation from N2O3 because this is the product of the •NO/•NO2 reaction. The resulting ramp, thus, is similar to nitrosation from only autoxidation (Figure 2C). The dominance of the N2O3 reaction in this phase is due to the fact that under these conditions the higher •NO concentration outcompetes the target for reaction with •NO2, a condition which as described below is not likely to hold under biological conditions. With this understanding in hand, we can now ask what the origins of the differences between the predicted (Figure 2C) and experimental (Figure 2A) results are. One major difference is the time period over which the reactions take place. Whereas the simulations were computed for the initial 10 min of reaction, the data in Espey et al. were reported after 60 min. An examination of the simulated flux ratios effecting maximal reaction (not shown) reveals that after 10 min as much as 50% of DAF will be nitrosated for the most active conditions, which means that the peak in nitrosation will shift to lower •NO flux rates as the DAF becomes depleted (and fluorescent product maximizes) in the more active conditions. Another factor will be gradual inactivation of the enzymatic generation of superoxide formation over the 60 min incubation period, which will also shift the peak to lower •NO flux rates (22). The next target for modeling is nitration, specifically the formation of 3-nitrotyrosine (3-NT), which is formed from these radical reactions via a two-step process, an initial one-electron oxidation producing the tyrosyl radical (TyrO•) followed by radical recombination with •NO2 (23, 24). An additional product (especially utilizing free tyrosine as the target as opposed to the protein tyrosine residue) is dityrosine (diTyr), resulting from the recombination of two TyrO•. For the modeling, I have selected data reported by Hodges et al. (25) who utilized a novel procedure to vary fluxes of •NO and O2•- with chemical sources under controlled conditions of CO2 concentration. Figure 3A is adapted from Hodges et al. (25) and presents the formation of 3-NT and diTyr with constant O2•- flux (corresponding to 0.5

1164 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

Lancaster

Figure 2. Nitrosation during various fluxes of •NO and O2•-. (A) Nitrosation of DAF, adapted from Figure 2A of Espey et al. (20). The published data have been adjusted to reflect the subtraction of the data without O2•- generation from the data with O2•- generation (Espey, M. Personal communication). (B) Scheme utilized to simulate nitrosation. The rate constants are presented in Supporting Information, Table 1. (C) Predicted steady-state rates of nitrosation for the conditions in 2A. (D) Steady-state rates of formation of total RSNO from recombination (GS• + •NO) and from N2O3 reaction and the concentrations of O2•- and •NO. The three phases are depicted.

mM on the abscissa) and increasing •NO flux ([CO2] ) 4 mM). For modeling, reactions and rate constants reported by Goldstein et al. (24) were utilized as depicted in the scheme in Figure 3B. In addition to the formation of dityrosine and 3-NT, Goldstein et al. identified and quantitated (in terms of rate constants) three other reactions, although the products were not identified. Prior studies have suggested the reversible reaction of •NO with TyrO• (26), which may involve the formation of nitrotyrosine via the iminoxyl radical (27). Two additional rapid reactions are between TyrO• and O2•-, producing Products2 (k29) (identified by Jin et al. (28) as an indole carboxylic acid, which may subsequently form tyrosine), and reaction of •NO2 with TyrO•, producing Products3 (k31), which is different from 3-NT. Figure 3C presents the results of the modeling, demonstrating excellent fidelity with the experimental results. An examination of the concentrations of intermediates and rates of individual reactions reveals the existence of three phases, as illustrated in Figure 4A and B. In phase A, as for nitrosation, •NO reacts stoichiometrically with O2•- and is removed as attested to by the excess O2•- (Figure 4A) and the reaction of TyrO• with O2•- (Figure 4B). This removal of TyrO• prevents the formation of either product (3-NT or diTyr). In phase B, •NO is in excess

of O2•-, and therefore, the reaction of O2•- with TyrO• is prevented and now the formation of both products (3-NT and diTyr) peaks (Figure 3C). In phase C, •NO (and, consequently, •NO ) builds up. This increase in •NO tends to decrease TyrO• 2 because of direct reaction (k27, k-27, k28), but the increase in •NO tends to counteract this because the increasing •NO also 2 2 results in increased TyrO• because of the reaction with Tyr. The net result is a plateau in TyrO• and diTyr as well (Figure 3C). The formation of 3-NT, however, increases because a second •NO2 (which is increasing in concentration) reacts with TyrO•. That diTyr is by far the major product quantitatively (over 3-NT) is explained by the fact that the TyrO• concentration is much greater than the •NO2 concentration (Figure 4A) and so a single TyrO• is more likely to react with another TyrO• than with •NO2 (the rate constants for each reaction are comparable in magnitude (Table 1, Supporting Information)). It is not clear that this will be true in ViVo because of the lower concentrations of free compared to protein-bound Tyr. Oxidation, Nitrosation, and Nitration under Simulated Biological Conditions. On the basis of the previous results, it seems reasonable to assume that this approach is capable of accurately predicting the complex chemistry that will take place

Predicting Biological ReactiVe Nitrogen Chemistry

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1165

Figure 4. Concentrations and rates of formation of intermediates during simulated nitration reactions, illustrating the three phases. (A) Steadystate concentrations of •NO, O2•-, •NO2, and TyrO• during the simulations in Figure 3C. (B) Steady-state rates of TyrO• + O2•reaction and formation of dityrosine and ONO-Tyr during the simulations in Figure 3C.

Figure 3. Nitration during various fluxes of •NO and O2•-. (A) Formation of dityrosine and 3-nitrotyrosine (3-NT) for various fluxes of •NO, adapted from Figure 2 of Hodges et al. (25). The •NO concentration refers to the total final concentration of •NO from •NO donor (Sper/NO). The equivalent values for v•NO are shown in Figure 3C. The rate of O2•- formation corresponds to 0.5 mM on the abscissa. (B) Scheme utilized to simulate nitration. The rate constants are presented in Supporting Information, Table 1. (C) Predicted steadystate rates of formation of dityrosine and 3-NT.

in ViVo. It is thus possible to perform a virtual experiment and examine the entire network of these multiple interacting reactions under biologically relevant conditions, a task that is impossible to accomplish experimentally. Scheme 1 presents the scheme utilized, and the rate constants are presented in Supporting Information, Table 1. The three major chemical processes involving nonmetal reactive nitrogen chemistry, oxidation, nitration, and nitrosation, are considered. The two targets are tyrosine and glutathione (GSH), present at 1 and 5 mM concentrations, respectively. The concentrations of O2 and CO2 are maintained at 56 µM (29) and 2 mM, and the initial nitrite concentration is 10 µM (30). The networks for oxidation and for tyrosine reactivity are as that utilized for Figure 3, with three additions: (1) in addition to the nonenzymatic reaction, O2•- dismutation also occurs via SOD (10 µM concentration (31)); (2) the repair of the tyrosyl radical by GSH (k51) (32); and (3) an additional term is included (k49) to account for other pathways for •NO disappearance (e.g., diffusion) (33).

For GSH, rate constants have been measured experimentally for all reactions except certain reactions of the peroxyl radical (reactions described by k45-k47). For these reactions, the rate constants for the equivalent reactions of alkyl peroxyl radicals were used, as reported by Goldstein et al. (34). Thus, this analysis assumes that the reactivity of the thioperoxyl radical toward other radicals is mainly determined by the peroxyl moiety and minimally different between the thiol and alkyl species. Products4 and Products5 were not identified, but O2•is also produced as the product in both cases. Thus, along with the reaction of the disulfide radical anion with O2 (k39), O2•- is produced as a downstream product, which will result in increased O2 consumption and, in the presence of •NO, formation of ONOO-. For these simulations, the rate of •NO formation is fixed, and rates of O2•- formation are increased. The values for these two fluxes (as well as initial concentrations of reactants) are chosen in order to yield desired steady-state concentrations of •NO and O2•-. Two general regimes are examined, where •NO is in the generally accepted ranges of noninflammatory (20 nM) or inflammatory (500 nM) concentrations. The fluxes of O2•- are chosen to provide O2•- concentrations lower than, similar to, and above each of these •NO concentrations. For the initial choice of •NO flux, on the basis of the calculations of Nalwaya and Deen (35), the rate of •NO synthesis in activated macrophages is 2.8 µM/s per cell. Assuming the rate of •NO formation under noninflammatory conditions (e.g., from constitutive NOS) is 1% of this value, the flux of •NO employed is 28 nM/s. Figure 5A presents a simulation with the •NO flux constant at this value with stepwise increases in O2•- flux. With an initial O2•- flux of 0.72 µM/s (0-4 s), after 2 s, the •NO concentration achieves a steady-state value of 20 nM. At 4 s O2•- flux is increased to 12 µM/s to yield similar (and small) concentrations

1166 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

Lancaster

Figure 5. Major reactions under simulated biological conditions and low [•NO]. (A) Concentrations of •NO and O2•- for constant v•NO (28 nM/s) with vO2•- of 0.72 µM/s (0-4 s), 12 µM/s (4-8 s), and 200 µM/s (8-12 s). Network of reactions are as in Scheme 1, using rate constants in Supporting Information, Table 1. (B) Rates of formation of all products of GSH reaction. Three reactions dominate, with all other reactions indistinguishable from the baseline at this ordinate scale. (C) Identification of the three major reactions of GSH. (D) Rates of formation of all products of tyrosine plotted. One reaction dominates, with all other reactions indistinguishable from the baseline at this ordinate scale. (E) Identification of the two major reactions of tyrosine. (F) Rates of formation of GS•. Solid black: formation from CO3•- in the absence of tyrosine. Solid red, blue: net rates of formation in the presence of tyrosine from CO3•- (red) and from the radical swap (blue). Cyan dot: sum of solid red plus blue.

of •NO and O2•-, and then at 8 s, O2•- flux is increased further to 200 µM/s to yield [O2•-] > [•NO]. This analysis thus examines the three possible regimes of relative •NO and O2•concentrations as described above. For initial simulations, the escape of •NO (k49) is not considered but is specifically dealt with below. In Figure 5B are plotted the rates of formation of all products of GSH reaction, and only three (denoted A, B, and C in the scheme (Figure 5C)) account for virtually all (9799.9%) of these reactions. All three of these reactions are oxidation reactions (as opposed to nitrosation or nitration), involving one electron oxidation by •NO2 or CO3•- (A and B) or direct two-electron oxidation via ONOO- (C). Thus, under physiologically relevant conditions, millimolar concentrations of GSH will compete effectively with CO2 for ONOO-. Note that the two major reactions (A and B) produce substantial amounts of O2•- and, for reaction A, a product of unknown structure. The rates of all reactions of tyrosine are shown in Figure 5D, revealing the importance of only three reactions as shown

in the scheme in Figure 5E. By far, the dominant reaction is the reversible radical exchange between TyrO•/Tyr and GSH/ GS• (note difference in scales in the ordinate axes in Figure 5D), which results in a net steady-state flux (difference between A and B) of the reduction of TyrO• by GSH of approximately 5 nM/s. Because the primary oxidant for Tyr is CO3•- rather than •NO2 (not shown), this means that the predominant effect of tyrosine is to effect CO3•--linked oxidation of GSH to GS•. This is illustrated in Figure 5F. The solid black line is the rate of GS• formation by oxidation of GSH by CO3•- in the absence of Tyr, all other conditions being identical. In the presence of Tyr, the total rate of GS• formation from CO3•- is the sum of the formation from direct oxidation (red line) and net formation via the reaction of GSH with TyrO• (Green). This sum (cyan dot) is virtually identical to the rate in the absence of Tyr (black line). Thus, the only effect of Tyr in a system with biologically relevant concentrations of GSH is to divert some CO3•- from direct reaction with GSH; however, there is no effect on net GS• formation from CO3•- because the TyrO• formed from this

Predicting Biological ReactiVe Nitrogen Chemistry

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1167

Figure 6. Major reactions under simulated biological conditions with high [•NO]. The conditions are the same as those for Figure 5 except for v•NO ) 2.8 µM/s and vO2•- ) 5 µM/s (0-8 s), 0.122 mM/s (8-16 s), and 5 mM/s (16-24 s). (A) Concentrations of •NO and O2•-. (B) Rates for all GSH reactions; A, B, and C refer to the reactions in Figure 5C. (C) Rates for all Tyr reactions. A, B, and C refer to the reactions denoted in Figure 5E. (D) Effects of periodic changes in vO2•- (5 µM/s alternating with 0.122 mM/s) with constant v•NO (2.8 µM/s) on concentrations of GSNO, •NO, and 3-NT.

diversion is converted to GS•. This is verified by the finding that all reactions of GSH subsequent to GS• formation (Figure 5C) are essentially identical with or without Tyr (not shown). With regard to reactions of TyrO• not regenerating Tyr, Figure 5D shows that the predominant reaction is with O2•-, producing Products2 (reaction C, Figure 5E); other reactions (including formation of 3-NT) are very minor in abundance. Somewhat surprisingly, the simulations utilizing fluxes of •NO, which yield an initial steady-state •NO concentration of 500 nM (thus presumably mimicking inflammatory conditions), yield patterns of reactivity similar to those above, with 20 nM •NO (Figure 6B and C). The substantive difference (Fibure 6B) is that with excess •NO (0-8 s), there is a cycle whereby GSNO is formed from the radical combination of GS• with •NO (blue) and the breakdown of GSNO via the reaction with GS• (cyan). The net result is an initial increase in GSNO until a steadystate is reached, where the rates of formation and breakdown equilibrate. Thus, with these high •NO concentrations, nitrosothiol is in dynamic kinetic communication with the nitrosative and oxidative chemistry in the immediate environment, as evidenced by the dramatic and rapid decline in GSNO that occurs at t ) 8 s, when O2•- flux is increased such that most of the •NO is removed by reaction (Figure 6A). This is illustrated most dramatically in Figure 6D, where the rate of O2•generation is alternatively increased and decreased, demonstrating the rapid response of nitrosothiol levels to the prevailing chemistry, most especially the free •NO concentration. In contrast, the concentrations of endproducts such as 3NT simply continue to accumulate. A significant advantage of performing in silico experiments is the ability to examine all intermediates during the course of a reaction and determine the precise mechanistic origins of specific results, thus potentially revealing general principles underlying complex networks of interacting rapid reactions. Four

questions become apparent upon inspection of the results in Figure 5. Why Are the Rates of Reactivity Relatively Independent of Excess •NO or Excess O2•-, and What Is the Effect of •NO Diffusion? These questions are mechanistically closely related. For the test tube conditions of Figure 1D, there is a 1:1 titration of •NO and O2•- and a linear increase in oxidative reactions as the flux of the limiting reactant is increased until the equivalence point, where the concentrations of each reactant are very small; above this point, the excess reactant accumulates, and oxidative reactions decline. In contrast, under the biologically simulated conditions considered here, oxidative reactions are essentially independent of whether O2•- or •NO is in excess (Figure 5B and D). Why is this the case? For this analysis, it is helpful to consider a simplified model (Figure 7A), which, when the equivalent values for fluxes and rate constants are utilized, produces results that are virtually identical for the complete model, as described below. Assuming the following steady-state conditions

[•NO] ) k1/(k5 + k3[O2•-])

(2)

[O2•-] ) k2/(k4 + k3[•NO])

(3)

I consider first the reactions of O2•- and ask, in the context of eq 3, the following question. What would be the concentration of •NO required for the reaction with O2•- (given by k3[•NO]) to become appreciable compared to that for the reaction with SOD (given by k4, which is assumed to be 1 × 104 s-1 (Supporting Information, Table 1))? This value is given by

[•NO] ) k4/k3 ) 104 s-1/1.9 × 1010 M-1 s-1 ) 530 nM (4)

1168 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

Lancaster

Figure 7. Simplified model for predicting relative fluxes. (A) Scheme of reactions. The rate constants used were the same as those for the same reactions in Scheme 1, with k5 ) k49 (Scheme 1), k4 ) k50 (Scheme 1). (B) Steady-state rates of the disappearance of •NO and of O2•- for simulations in Scheme 1 with v•NO ) 28 nM/s and indicated values for vO2•-. (C) Steady-state •NO and O2•- concentrations for Scheme 1 (×) and simplified scheme (Figure 7A) (O) with •NO and O2•- fluxes as in Figure 7B. (D) Steady-state ONOO- concentrations for Scheme 1 (×) and simplified Scheme (Figure 7A) (O). (E) Steady-state rates of ONOO- formation for the simplified model (Figure 7A) with rates of •NO disappearance 20 s-1 (black) or 4000 s-1 (red). Green line: the rate of ONOO- formation predicted by eq 10 for k5 ) 4000 s-1.

Thus, under noninflammatory conditions ([•NO] , 530 nM), it can safely be assumed that cellular O2•- disappears predominantly via the SOD reaction, and under these conditions, the steady-state O2•- concentration can thus be described by •-

[O2 ] = k2/k4

(5)

For •NO disappearance under these conditions ([•NO] , 530 nM), combination of eq 5 with eq 2 yields

[•NO] ) k1/(k5 + k2k3/k4])

(6)

I examine first the case where •NO disappears predominantly via the reaction with O2•-. This is in fact the situation for the simulations in Figure 5, as shown in Figure 7B. Here, all steadystate rates of disappearance of •NO (left ordinate) and of O2•(right ordinate) with increasing fluxes of O2•- and •NO flux fixed as in Figure 5 (vNO ) 2.8 × 10-8 M/s). •NO disappears predominantly via the reaction with O2•- and O2•- disappears predominantly via SOD. When •NO disappears predominantly via the reaction with O2•-, from eq 6

k2k3/k4 > k5

(7)

and

k2/k5 > k4/k3

(8)

Assuming the rate of O2•- disappearance of Figure 5 (k4 ) 104 s-1 and k3 ) 1.9 × 1010 M-1 s-1), the two adjustable parameters here are thus k2 (the rate of O2•- production) and k5 (the net rate of •NO disappearance from the cell, via a combination of efflux and influx). Thus, the two ways to accomplish the condition of eq 8 is to either increase k2 or decrease k5. As recently pointed out by Quijano et al. using a modeling approach similar to that used here (33), the effect of variations in •NO diffusion on this network of reactivity can be incorporated in modeling by varying this rate of •NO disappearance (k5), corresponding to k49 in Scheme 1. Using the lowest value for k2 in Figure 7A, it is thus possible to calculate the limiting value of k5 to ensure this condition (that •NO disappears predominantly via the reaction with O2•-):

k5 < (7.2 × 10-7 M/s)(1.9 × 1010 M-1 s-1)/ 104 s-1 ) 1.37 s-1 (9) To demonstrate the ability of the simplified model (Figure 7A) to predict the comprehensive model (Figure 5), Figure 7C

Predicting Biological ReactiVe Nitrogen Chemistry

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1169

Figure 8. Concentrations and rates of formation of intermediates and products under simulated biological conditions. (A) Rates of reaction of CO3•- with all reactants during the simulations in Figure 5. The rates of reaction with all reactants other than Tyr and GSH are indistinguishable from the baseline at this ordinate scale. (B) Rates of reaction of •NO2 with all reactants during the simulations in Figure 5. The rates of reaction with all reactants other than GSH are indistinguishable from the baseline at this ordinate scale. (C) Steady-state rates of oxidation of GSH and Tyr by CO3•- and •NO2 with constant v•NO (28 nM/s) and the indicated vO2•-. (D) Rates of formation of GSNO (from GS• + •NO (black) and from reaction with N2O3 (red)) and 3NT (green) with conditions the same as those in Figure 5.

and D show the steady-state •NO and O2•- concentrations (7C) and rates of ONOO- formation (7D) for the comprehensive model (× symbols) and the simplified model (circles, with k5 ) 0.137 s-1). Additionally, for k3[O2•-] > k5, eq 2 dictates that the rate of ONOO- formation will be equal to k1 (in this case, 2.8 × 10-8 M/s) and unaffected by k2 (as long as k2 is large enough so that [O2•-] > k5/k3), both of which are also verified by Figure 7D. Figure 7E illustrates the effect of increasing the rate of •NO disappearance relative to the reaction with O2•- with parameters identical to those in Figure 7D except with k5 ) 20 s-1 (black) and k5 ) 4000 s-1 (red). In the latter case, with •NO disappearing predominantly via k5 (diffusion, for example, as in Quijano et al. (33)), k5 > k2k3/k4 and from eqs 2 and 3 [•NO] ) k1/k5 and [O2•-] ) k2/k4 (for [•NO] < 530 nM); thus, the rate of ONOOformation is given by

d[ONOO-]/dt ) k3[•NO][O2•-] ) k1k2k3/k4k5

(10)

The green line in Figure 7E shows this relationship, further verifying the validity of these kinetic relationships. Finally, this analysis allows the estimation of the relative proportion of O2•- that reacts with •NO compared to that in the reaction with SOD under simulated biological conditions. Assuming that the major route for •NO disappearance is via processes (e.g., diffusion) other than the reaction with O2•-(k5 ) 400 s-1) and that O2•- disappears mainly via SOD (k4 ) 104 s-1) with k3 ) 1.9 × 1010 M-1 s-1, eq 10 predicts the following.

d[ONOO-]/dt ) k1k2 × 4.75 × 103 s M-1

(11)

For a rate of •NO formation (k1) by an activated macrophage of 2.8 × 10-6 M-1 s-1 (35)

d[ONOO-]/dt ) k2 × 0.013

(12)

This indicates that under inflammatory conditions, approximately 1% of O2•- reacts with •NO to produce ONOOwith the remainder reacting with SOD. This proportion will obviously increase if the •NO concentration increases to the point where coreaction with O2•- becomes an appreciable proportion of the route(s) of NO disappearance, estimated here to be greater than approximately 500 nM. In summary, a linear increase in ONOO- formation will occur if the rates of formation of •NO and/or O2•- are small such that the disappearance of •NO and O2•- occurs by processes other than coreaction; with increasing fluxes, a plateau in the rate of ONOO- formation can occur when the rate of coreaction becomes the major route of disappearance. Why Is there No Bell-Shaped Curve for Oxidation? As described above, the origin of this previously described behavior (6, 7) lies in the fact that under defined conditions both CO3•and •NO2 react with excess •NO or O2•- when the fluxes are not equivalent. However, under biological conditions, other targets, if present at sufficient concentrations, will outcompete both •NO and O2•- for CO3•- and •NO2. This is verified by the results presented in Figure 8, which utilize the complete model (Scheme 1). Figure 8A presents all the rates of reaction of CO3•-, demonstrating that its reactions with Tyr and GSH account for virtually all its reactivity. Likewise, Figure 8B shows that •NO2 reacts predominantly with GSH (the rate constant for its reaction with GSH is much higher than its reaction with Tyr; Table 1, Supporting Information). This is demonstrated further in Figure 8C, where steady-state rates of oxidation of GSH and Tyr by CO3•- and by •NO2 are shown when •NO flux is held constant (at 28 nM/s) and O2•- flux is increased, analogous to that in Figure 1. No bell-shaped behavior is observed. Why Is the Complex Network of Reactions, Involving Multiple Reactions with Very Rapid Reaction Constants, Dominated by Only Five Reactions? All of the reactions depicted in Scheme 1 occur with quite rapid rate constants. However, an inspection of the entire network of reactions reveals that a critically important determinant of the importance of a

1170 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

Lancaster

Table 1. Relative Contributions of Oxidation, Nitrosation, and Nitration to Products of Reactive Nitrogen Species Chemistry Under Biological Conditions [•NO]

ss

) 20 nM

[•NO]ss ) 500 nM

vO2•- < v•NO

vO2•- ) v•NO

vO2•- > v•NO

98.6% 0.2% 1.2% 92.0% 3.3% 4.8%

99.3% 0.1% 0.6% 98.6% 1.4% 0.1%

99.6% 0.1% 0.4% 99.1% 0.9% 0.002%

oxidation nitration nitrosationa oxidation nitration nitrosationa

a As described in the text, GSNO, the major product of nitrosation, is an intermediate; therefore, its concentration does not change at steady-state, and thus, its relative proportion of total products will decline with time. The percentages are for 4 s ([•NO]ss ) 20 nM) and 8 s ([•NO]ss ) 500 nM).

particular reaction is the concentration of the reactants. Reactions involving only intermediates that are present at relatively low concentrations (such as •NO2 and •NO) are not as important as reactions involving species that are present at higher concentrations (O2, GSH, and Tyr). The concentrations of •NO2 and •NO are in the nanomolar range, whereas O2, Tyr, and GSH are present in the range of tens of micromolar to millimolar concentrations. What Are the Relative Patterns of Nitrosation and Nitration? It is important to note that although nitrosation and nitration may be minor processes quantitatively, they may be major processes biologically. In addition, highly sensitive and selective methods exist for their detection, and in ViVo formation (and correlation with endogenous RNI formation) is well established. Thus, it is of interest to examine the rates of their product formation and the effects of varying fluxes of •NO and O2•-. Figure 8D shows the rates of formation of 3-NT and GSNO (by the two different mechanisms, N2O3 and radical combination) as vO2•- is incrementally increased. Four important conclusions are as follows: (1) nitrosation occurs virtually exclusively by radical combination as opposed to via the reaction with N2O3; (2) nitrosation is much more prevalent than nitration; (3) nitrosation is virtually eliminated when [O2•-] g [•NO]; and (4) nitration is relatively independent of excess •NO versus excess O2•-. Increasing vO2•- over vNO removes free •NO (which is required for both nitrosative processes) but has relatively little effect on 3-NT formation because the steadystate levels of both TyrO• and •NO2 are determined by reaction with GSH (Figure 5C andE), which does not appreciably change in concentration.

Discussion Although the reactions of reactive nitrogen and oxygen species are kinetically well characterized, there has been no comprehensive approach to identify which of the many reactions possible in the test tube are most important under biological conditions. The general approach applied here is to utilize a user-friendly computer program to design and then validate (using predictions of published experimental data) a comprehensive network of interacting reactions of biological importance. It is then possible to perform a virtual experiment, where initial conditions can be selected to mimic the biological milieu, and progress curves of each intermediate and reaction can be obtained for the chosen time course. The executable computer program is freely available for downloading at http:// nitricoxide.anes.uab.edu/LLab/RNSim/Index.htm. There are numerous caveats for the results reported here, in most cases because of the unavailability of experimentally measured rate constants. The critically important roles of transition metals in reactive nitrogen species (36) as well as the effects of heterogeneous phases (e.g., membranes) on the concentrations and reactivities of intermediates in different phases (37, 38), cellular antioxidants (e.g., ascorbate, vitamin

E (19)), and reactions with lipid radical species (39) are not included. It is possible, for example, that partitioning of peroxynitrous acid into the hydrophobic interior of membranes enhances homolysis and, thus, the localized formation of •OH even in the presence of substantial CO2 concentrations. These simulations also do not consider the effects of the dramatic differences in diffusibility of individual reactants, which will greatly influence the processes that occur at a specific location with a heterogeneous spatial distribution of reactants, targets, sources, and sinks (40, 41). The only attempt to incorporate spatial diffusibility is in the concentration of •NO (Vide infra), which is undoubtedly the major reactive species that diffuses appreciable distances from the site of formation. In addition, the rate of production of O2•- is adjusted in order to obtain steady-state superoxide concentrations of 20 or 500 nM. If O2•disappears predominantly via SOD, the required rates of O2•formation (given by [O2•-] × k4, eq 3) are 200 µM/s or 4 mM/ s, which are almost certainly unrealistically high. I note, however, that these rates are determined by the concentration of SOD, and lower (and more realistic) rates will be required at locations where the SOD concentration is lower. In any event, these simulations are done for comparison and are physiologically untenable. Another caveat is that the only thiol considered is GSH, and as discussed in more detail below, the reactivities of other cellular thiols (such as in proteins) are known to be quite diverse. Another assumption is that the reactivity of thioperoxyl radicals is similar to that of alkylperoxyl radicals. Although it seems reasonable to assume that the chemical properties of these species will be determined mainly by the peroxyl radical moiety, this remains to be established. For example, unlike alkylperoxyl radicals, thioperoxyl radicals can relatively rapidly form sulfinyl radicals (42). The rationale for this assumption is that although other reactions of thioperoxyl radicals can occur the most rapid should be reactions with other radical species, and the rate constants are, thus, assumed to be similar to those for alkylperoxyl radicals as reported by Goldstein et al. (34). Whether this assumption is valid will come only from the determination of the rates for thioperoxyl radicals, which has not been performed. In spite of these caveats, however, a number of new insights into reactive nitrogen chemistry are obtained, as summarized below. A concept central to the findings reported here is presented in Table 1. Under all conditions of varying •NO and O2•- fluxes and high compared to low steady-state •NO levels, by far the quantitatively dominant (g92%) process is oxidation as opposed to nitrosation or nitration. The term oxidative stress is commonly used to denote the injurious cellular effects of reactive oxygen species. The terms nitrosative stress and (less commonly) nitrative stress are often used to denote injury from reactive nitrogen species, but unfortunately, these terms oftentimes do not refer specifically to nitrosation or nitration. In an attempt to clarify this confusion in terminology and to recognize the importance of oxidative chemistry in the cellular actions of

Predicting Biological ReactiVe Nitrogen Chemistry

reactive nitrogen species, it seems reasonable to propose the term nitroxidative stress (43). This term specifically distinguishes oxidative chemistry from nitrosative or nitrative chemistry because they induce biological injury. In addition, it emphasizes the different major players for each process (e.g., O2•-, 1O2, H2O2, and •OH for oxidative stress compared to ONOO-, CO3•-, and •NO2 for nitroxidative stress)3. •NO and O •- will react extremely rapidly in ViVo, and the 2 product ONOO- will also react very rapidly with either thiol targets or CO2. The reactive products of CO2 reaction, CO3•and •NO2, also disappear rapidly. The concentration of any of these intermediate reactive species will, thus, be quite low and difficult to measure or estimate. Thus, the common practice of the addition of bolus large amounts of ONOO- may well result in patterns of reactivity that are not physiologically relevant because of extremely high (although transient) levels of reactive intermediates (especially •NO2 and CO3•-) produced by rapid reaction with CO2. Therefore, the proper method to mimic the biological milieu is simultaneous generation of •NO and O2•at controlled flux ratios. Employing this approach in systems in Vitro has revealed a distinct optimal flux ratio for oxidation, below and above which reactivity is suppressed (6, 7). As suggested previously (8, 11), the results here demonstrate that this effect can be ascribed to the scavenging of downstream reactive species by excess •NO or O2•-. The simulations suggest in addition that the oxidizing species under these in Vitro conditions is not ONOO- but the reactive products of its reaction with CO2 and rapid homolysis (•NO2 and CO3•-). Importantly, the simulations also reveal that this bell-shaped behavior is observed because the dominant reactions of •NO2 and CO3•are with excess O2•- and •NO. With a constant flux of one reactant (A, either •NO or O2•-), as the flux of the other reactant (B) increases, the rate of ONOO- production will initially increase linearly. A point is reached, however, where ONOO- formation plateaus and is unchanged with increasing flux of B. In the absence of pathways for disappearance of A or B other than the reaction with each other, this is a simple titration with excess concentration of A below the point of equal fluxes and excess B above this point (e.g., see Figures 1C, 2D, and 4A). This plateau of ONOOformation also will occur in the presence of other mechanisms for disappearance of either or both reactants, but the plateau is now attained where because of increasing concentration of B, the rate of coreaction exceeds the rate(s) of disappearance of A. In this case, as illustrated by the modeling results under simulated biological conditions (Figure 5A and B), it is possible that the plateau of ONOO- (and, thus, also of subsequent oxidative reactions) is attained at fluxes of •NO and O2•- that are not equal and not necessarily at the flux ratio where both •NO and O •- concentrations are minimal. 2 Given the critical dependence of these patterns of reactivity on the concentrations and rates of production and disappearance of •NO and O2•-, a central question then becomes what the most reasonable choices of these fluxes and concentrations are for modeling the biological milieu. Simulations here utilize an experimentally based flux of •NO formation (28 nM/s, based on data of Nalwaya and Deen (35)). The flux of O2•- is adjusted to yield low and high nM steady-state concentrations of •NO and O2•-. In these simulations, no allowance is made for additional pathways of •NO disappearance such as diffusion, which will have the sole effect of decreasing •NO concentration. Quijano et al. (33), using a modeling approach similar to that 3 Recent studies, however, have revealed potential sources of CO •3 independent of •NO (44, 45).

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1171

used here, demonstrate that the inclusion of a term for rapid disappearance of •NO (400 s-1, an estimate for diffusion away from an isolated •NO-producing cell) dramatically lowers the •NO concentration under their set of conditions, from 12 µM to 40 pM. The inclusion of SOD also dramatically lowers the steady-state O2•- concentration (from 200 nM to 0.8 pM). As pointed out above, these very low •NO and O2•- concentrations dictate that the major route of disappearance for each species will be via processes not involving coreaction, and therefore, the rate of ONOO- production will be linear with increasing flux of either reactant with no plateau. This means that no bellshaped behavior will be observed because oxidation will increase with increasing fluxes. Also, at these very low concentrations, it is highly unlikely that the reaction of •NO or O2•- with downstream oxidants (specifically, •NO2 and CO3•-) will compete with the reaction with other cellular components, which as I show above is the major explanation for the bell-shaped behavior in experiments in Vitro. The modeling here shows that under simulated biological conditions even with low to high nanomolar concentrations of •NO and O2•- there is no bellshaped behavior, even with a plateau of ONOO- formation. This is because of the prevention by cellular targets (principally thiol) of reaction of •NO and O2•- with the oxidizing species •NO and CO •-. 2 3 Although nitrosation and nitration are minor processes in terms of total flux of nitrogen oxide reactions (Table 1), this does not necessarily minimize the biological importance of these processes. Even if only a minority of total reactions, nitrosation, or nitration could still result in substantial biological actions through, for example, posttranslational protein modification (46). In addition, under normal conditions, oxidative processes may have relatively less biological impact compared to nitrosation/ nitration because of more extensive protective/repair mechanisms. These relationships are complicated by the fact that these three processes are not entirely independent. There are targets (e.g., thiols) that are common for all three, and in fact, both nitrosation and nitration require oxidation for initiation. There are important implications of the results here for the molecular mechanistic bases of post-translational regulation of protein activity via reactive oxygen and nitrogen species, including nitrosation (46), nitration (47), glutathiolation (48), and thiol/disulfide oxidation/reduction (49). The results in Figure 8D suggest that the exclusive mechanism of nitrosothiol formation under biological conditions involves one-electron oxidation of thiol followed by radical recombination (due to the much more effective competition of thiol over •NO for reaction with •NO2, as suggested previously (50)). Thus, after the initial one-electron oxidation, the reaction of the thiyl radical with either O2 or thiolate anion is much more likely than the reaction with •NO (Figure 5B), and therefore, the extent of nitrosothiol formation will reflect only a very small fraction of thiol modification from reactive nitrogen species. Although nitrosation has been most often cited as the explanation for NOeffected thiol modification, many studies do not distinguish nitrosation from other possible •NO-dependent modifications such as oxidation. Especially relevant is the common assumption that ascorbate restoration of thiol functionality is a specific indicator for nitrosothiols (the basis for the biotin switch assay (51)), which is not necessarily valid (52). For example, it seems quite possible that ascorbate could restore thiol by the reduction of products produced from the thioperoxyl radical (Scheme 1). An additional insight from these simulations is that unlike 3-NT, nitrosothiol is, strictly speaking, an intermediate and not an endproduct (Figure 6D). This may provide a mechanism

1172 Chem. Res. Toxicol., Vol. 19, No. 9, 2006

whereby the levels of an individual RSNO responds to changes in its reactive environment, thus perhaps providing a sensing mechanism. In the case of tyrosine, Figure 5D shows that once formed, the major reaction of TyrO• is with thiol, which produces thiyl radical and regenerates Tyr. This effect, with consequent increases in thiol nitrosation and decreases in tyrosine nitration, has recently been observed experimentally using synthetic peptides (53). It is important to note that Tyr residues in proteins may be more inaccessible to free thiol species (GSH, cysteine), and therfore, the TyrO•/RSH reaction may be hindered resulting in the selective reactivity of TyrO• with small molecular reactant species (O2•-, •NO, and •NO2). Figure 5E shows that the predominant reaction in this case will be with O2•-, possibly resulting in the regeneration of Tyr. The reaction with •NO2 to produce 3-NT is still only a minor reaction quantitatively. With regard to protein glutathiolation and also thiol/disulfide oxidation/reduction, Figure 5C shows that one of the most dominant reactions of thiol in the biological milieu is the thiyl/thiolate anion reaction, followed by disulfide formation with concomitant production of superoxide. This suggests a mechanism for both glutathiolation from protein cysteine and GSH and also disulfide formation from vicinal cysteine dithiols. Both processes will be directly proportional to oxygen concentration (required for oxidation of the disulfide radical anion), which could provide a mechanism for sensing oxygen. As with protein-bound tyrosine, the specific reactions of cysteine thiols will be greatly influenced by accessibility to the various reactants. In particular, relatively inaccessible thiol groups will be less susceptible to reaction with thiolate and consequent disulfide formation. In the case of both Tyr and Cys residues, it seems reasonable to assume that if these groups, although inaccessible to glutathione or other cytosolic thiol, are accessible to •NO2 and/or CO3•-, forming the initial one-electron oxidized species (GS• and TyrO•). These resulting radicals should also be accessible to a second reaction with the small reactive species •NO, •NO2, O2•-, and O2. In the case of thiyl, undoubtedly the most important of these is O2 because of its much higher concentration, which means that the formation of thioperoxyl radical will be much more dominant than the formation of nitrosothiol or an other product. This highlights the need for further understanding of the reactions of the thioperoxyl radical. However, it seems plausible that the positioning of a glutathione binding site adjacent to a critical thiol residue could in fact provide a mechanistic basis for glutathiolation at a specific site via these mechanisms. Whatever the prevailing chemistry within the microenvironment of a protein thiol, an important realization from the results here is that the formation of nitrosothiol would appear to be an only minor proportion of the modifications expected from the further reactions of the thyl radical, and thus, the formation of nitrosothiol could serve as a valid marker for the increased reactivity of an individual thiol, but other modifications could be more abundant and may be the ones that actually effect modification in protein functional activity. As shown in Figure 5C, two of the three major reactions of reactive nitrogen species produce O2•-. This will serve to spatially confine RNS reactivity, both by acting as a sink for NO and also as a feedback mechanism, whereby the location of this O2•- formation is also a source for more very shortlived reactive species (CO3•-, •NO2, and ONOO-), which in the presence of thiol, generates more O2•-. If the targets of these reactions are spatially adjacent to an •NO source and the rate of •NO formation is relatively small compared to the rates of reaction, this may provide a mechanism for the apparent spatial

Lancaster

confinement of the actions of •NO in some biological systems (54). Theoretical calculations show that only reactions as fast as the •NO/O2•- reaction can sufficiently compete with the diffusion of free •NO to prevent its escape (not shown). Although nitrosation has been most often cited as the explanation for this effect, most studies do not distinguish nitrosation from other possible •NO-dependent thiol modifications such as oxidation. In summary, several specific predictions emerge from these simulations. (1) Under inflammatory conditions (e.g., an activated macrophage) with [•NO] < 500 nM, approximately 1% of O2•produced will react with •NO to produce ONOO- with the remainder being removed via reaction with SOD. This proportion increases with [•NO] greater than 500 nM. (2) As suggested previously (16), •OH is not a viable oxidant from the •NO/O2•- reaction because of the predominance of the reaction of ONOO- with CO2. (3) The major oxidizing species are CO3•-, •NO2, and ONOO-, the latter involving direct two-electron oxidation of thiol. (4) In contrast to isolated chemical systems, there is no bellshaped behavior of oxidative reactivity as •NO or O2•- fluxes are increased because of the much more effective competition of cellular targets (principally thiol) for reactive intermediate oxidants (CO3•-, •NO2) than that for the excess •NO or O2•-. (5) With increasing fluxes of one reactant (O2•- or •NO) and constant flux of the other, the formation of ONOO- (and, thus, the downstream oxidizing species) will increase linearly until the concentrations of the two reactants becomes appreciable enough so that coreaction becomes the dominant route for the disappearance of either reactant; increasing the flux further does not increase ONOO- formation. (6) By far, the most abundant modification of thiol and tyrosine by these reactions is oxidation, prompting the suggestion of the use of the term nitroxidative stress (43) to distinguish the biological effects of this process from the other two processes (nitrosation and nitration) of reactive nitrogen chemistry. (7) The most dominant reaction of the tyrosyl radical is a radical swap with GSH forming GS• and regenerating tyrosine. (8) For both glutathione and tyrosine, the two most rapid reactions under biological conditions (other than the radical swap) produce products of unknown structure (Products2 and Products5, Figure 5), suggesting that the identification of these products may provide new markers for nitroxidative stress4. (9) As suggested previously (50), nitrosothiol (GSNO) is formed virtually exclusively from radical combination (GS• + •NO) as opposed to a reaction with N O . 2 3 (10) The extent of nitrosothiol formation is a marker for conditions where •NO is in excess of O2•-, whereas 3-NT formation occurs relatively independently of the •NO/O2•- ratio. (11) Nitrosothiol is an intermediate not an endproduct and is in dynamic communication with the reactive chemistry in the immediate environment. (12) The formation of a nitroso group on a particular thiol can be considered a marker of increased reactivity of that thiol, and it is likely that other modifications of that thiol (oxidation, glutathiolation) are more abundant than nitrosation and could be, in fact, the functionally significant thiol modification actually effecting the modification of protein function. 4 These products, however, may well be tyrosine regenerated by O •2 reduction (32) and GSSG, respectively. In addition, there are good reasons to speculate that at least some of the apparent products of the glutathione reaction (GSNO2, GSOH) form GSSG upon reaction with GSH (55).

Predicting Biological ReactiVe Nitrogen Chemistry

(13) These patterns of reactivity are similar for low (noninflammatory) and high (inflammatory) concentrations of •NO and O2•-. (14) The dominance of two reactions of thiyl radical (reaction with thiolate anion and with O2) suggests specific chemical mechanisms for posttranslational protein thiol modification involving nitrosation, oxidation, glutathiolation, and vicinal dithiol/disulfide formation. (15) The dominant role of O2 in determining thiol reactivity provides possible chemical mechanisms for O2 sensing involving thiol modification. (16) Within proteins, the accessibility of cysteine thiol or tyrosine phenolate to small thiols (GSH, cysteine) greatly affects reactivity to produce different products. (17) The dominant reactions in a network of interacting rapid reactions are those involving reactants that are present at the highest concentrations (e.g., O2 and thiol). Acknowledgment. This work was supported by NIH Grants DK46935, HL074391, and HL71189. Supporting Information Available: Table of rate constants used in this article and a tabular key to the output columns from the program RNSim1a. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Blough, N. V., and Zafiriou, O. C. (1985) Reaction of superoxide with nitric oxide to form peroxonitrite in aqueous solution. Inorg. Chem. 24, 3502-3504. (2) Gryglewski, R. J., Palmer, R. M., and Moncada, S. (1986) Superoxide anion is involved in the breakdown of endothelium-derived vascular relaxing factor. Nature 320, 454-456. (3) Beckman, J. S., Beckman, T. W., Chen, J., Marshall, P. A., and Freeman, B. A. (1990) Apparent hydroxyl radical production by peroxynitrite: implications for endothelial injury from nitric oxide and superoxide. Proc. Natl. Acad. Sci. U.S.A. 87, 1620-1624. (4) Estevez, A. G., and Jordan, J. (2002) Nitric oxide and superoxide, a deadly cocktail. Ann. N. Y. Acad. Sci. 962, 207-211. (5) Kissner, R., Nauser, T., Bugnon, P., Lye, P. G., and Koppenol, W. H. (1997) Formation and properties of peroxynitrite as studied by laser flash photolysis, high-pressure stopped-flow technique, and pulse radiolysis. Chem. Res. Toxicol. 10, 1285-1292. (6) Rubbo, H., Radi, R., Trujillo, M., Telleri, R., Kalyanaraman, B., Barnes, S., Kirk, M., and Freeman, B. A. (1994) Nitric oxide regulation of superoxide and peroxynitrite-dependent lipid peroxidation. Formation of novel nitrogen-containing oxidized lipid derivatives. J. Biol. Chem. 269, 26066-26075. (7) Miles, A. M., Bohle, D. S., Glassbrenner, P. A., Hansert, B., Wink, D. A., and Grisham, M. B. (1996) Modulation of superoxide-dependent oxidation and hydroxylation reactions by nitric oxide. J. Biol. Chem. 271, 40-47. (8) Jourd’heuil, D., Jourd’heuil, F. L., Kutchukian, P. S., Musah, R. A., Wink, D. A., and Grisham, M. B. (2001) Reaction of superoxide and nitric oxide with peroxynitrite. Implications for peroxynitrite-mediated oxidation reactions in vivo. J. Biol. Chem. 276, 28799-28805. (9) Miranda, K. M., Espey, M. G., Jourd’heuil, D., Grisham, M. B., Fukuto, J. M., Feelisch, M., and Wink, D. A. (2000) The Chemical Biology of Nitric Oxide. In Nitric Oxide Biology and Pathobiology (Ignarro, L. J., Ed.) pp 41-55, Academic Press, San Diego, CA. (10) Merenyi, G., and Lind, J. (1998) Free radical formation in the peroxynitrous acid (ONOOH)/peroxynitrite (ONOO-) system. Chem. Res. Toxicol. 11, 243-246. (11) Glebska, J., and Koppenol, W. H. (2003) Peroxynitrite-mediated oxidation of dichlorodihydrofluorescein and dihydrorhodamine. Free Radical Biol. Med. 35, 676-682. (12) Goldstein, S., Czapski, G., Lind, J., and Merenyi, G. (1999) Effect of *NO on the decomposition of peroxynitrite: reaction of N2O3 with ONOO-. Chem. Res. Toxicol. 12, 132-136. (13) Ford, P. C., Wink, D. A., and Stanbury, D. M. (1993) Autoxidation kinetics of aqueous nitric oxide. FEBS Lett. 326, 1-3. (14) Goldstein, S., Saha, A., Lymar, S. V., and Czapski, G. (2005) Oxidation of peroxynitrite by inorganic radicals: a pulse radiolysis study. J. Am. Chem. Soc. 120, 5549-5554.

Chem. Res. Toxicol., Vol. 19, No. 9, 2006 1173 (15) Coddington, J. W., Hurst, J. K., and Lymar, S. V. (1999) Hydroxyl radical formation during peroxynitrous acid decomposition. J. Am. Chem. Soc. 121, 2438-2443. (16) Lymar, S. V., and Hurst, J. K. (1995) Rapid reaction between peroxonitrite ion and carbon dioxide: implications for biological activity. J. Am. Chem. Soc. 117, 8867-8868. (17) Goldstein, S., Lind, J., and Merenyi, G. (2002)The reaction of ONOOwith carbonyls: estimation of the half-lives of ONOOC(O)O- and O2OOC(O)O-. J. Chem. Soc., Dalton Trans. 808-810. (18) Squadrito, G. L., and Pryor, W. A. (2002) Mapping the reaction of peroxynitrite with CO2: energetics, reactive species, and biological implications. Chem. Res. Toxicol. 15, 885-895. (19) Augusto, O., Bonini, M. G., Amanso, A. M., Linares, E., Santos, C. C., and De Menezes, S. L. (2002) Nitrogen dioxide and carbonate radical anion: two emerging radicals in biology. Free Radical Biol. Med. 32, 841-859. (20) Espey, M. G., Thomas, D. D., Miranda, K. M., and Wink, D. A. (2002) Focusing of nitric oxide mediated nitrosation and oxidative nitrosylation as a consequence of reaction with superoxide. Proc. Natl. Acad. Sci. U. S. A. 99, 11127-11132. (21) Goldstein, S., and Czapski, G. (1996) Mechanism of the nitrosation of thiols and amines by oxygenated •NO solutions: the nature of the nitrosating intermediates. J. Am. Chem. Soc. 118, 3419-3425. (22) Wang, C., and Deen, W. M. (2004) Peroxynitrite delivery methods for toxicity studies. Chem. Res. Toxicol. 17, 32-44. (23) Van der Vliet, A., Eiserich, J. P., O’Neill, C. A., Halliwell, B., and Cross, C. E. (1995) Tyrosine modification by reactive nitrogen species: a closer look. Arch. Biochem. Biophys. 319, 341-349. (24) Goldstein, S., Czapski, G., Lind, J., and Merenyi, G. (2000) Tyrosine nitration by simultaneous generation of (.)NO and O-(2) under physiological conditions. How the radicals do the job. J. Biol. Chem. 275, 3031-3036. (25) Hodges, G. R., Marwaha, J., Paul, T., and Ingold, K. U. (2000) A novel procedure for generating both nitric oxide and superoxide in situ from chemical sources at any chosen mole ratio. First application: tyrosine oxidation and a comparison with preformed peroxynitrite. Chem. Res. Toxicol. 13, 1287-1293. (26) Petrouleas, V., and Diner, B. A. (1990) Formation by NO of nitrosyl adducts of redox components of the photosystem ii reaction center. I. NO binds to the acceptor-side non-heme iron. Biochim. Biophys. Acta. 1015, 131-140. (27) Gunther, M. R., Sturgeon, B. E., and Mason, R. P. (2002) Nitric oxide trapping of the tyrosyl radical-chemistry and biochemistry. Toxicol. 177, 1-9. (28) Jin, F., Leitich, J., and von Sontagg, C. (1993)The superoxide radical reacts with tyrosine-derived phenoxyl radicals by addition rather than by electron transfer. J. Chem. Soc., Perkin Trans. 2 1583-1588. (29) Filho, I. P., Leunig, M., Yuan, F., Intaglietta, M., and Jain, R. K. (1994) Noninvasive measurement of microvascular and interstitial oxygen profiles in a human tumor in SCID mice. Proc. Natl. Acad. Sci. U.S.A. 91, 2081-2085. (30) Rodriguez, J., Maloney, R. E., Rassaf, T., Bryan, N. S., and Feelisch, M. (2003) Chemical nature of nitric oxide storage forms in rat vascular tissue. Proc. Natl. Acad. Sci. U.S.A. 100, 336-341. (31) Liochev, S. I., and Fridovich, I. (2002) Superoxide and nitric oxide: consequences of varying rates of production and consumption: a theoretical treatment. Free Radical Biol. Med. 33, 137-141. (32) Kirsch, M., Lehnig, M., Korth, H. G., Sustmann, R., and de Groot, H. (2001) Inhibition of peroxynitrite-induced nitration of tyrosine by glutathione in the presence of carbon dioxide through both radical repair and peroxynitrate formation. Eur. Chem. J. 7, 3313-3320. (33) Quijano, C., Romero, N., and Radi, R. (2005) Tyrosine nitration by superoxide and nitric oxide fluxes in biological systems: modeling the impact of superoxide dismutase and nitric oxide diffusion. Free Radical Biol. Med. 39, 728-741. (34) Goldstein, S., Lind, J., and Merenyi, G. (2004) Reaction of organic peroxyl radicals with •NO2 and •NO in aqueous solution: intermediacy of organicperoxynitrate and peroxynitrite species. J. Phys. Chem. A 108, 1719-1725. (35) Nalwaya, N., and Deen, W. M. (2005) Nitric oxide, oxygen, and superoxide formation and consumption in macrophage cultures. Chem. Res. Toxicol. 18, 486-493. (36) Thomas, D. D., Miranda, K. M., Colton, C. A., Citrin, D., Espey, M. G., and Wink, D. A. (2003) Heme proteins and nitric oxide (NO): the neglected, eloquent chemistry in NO redox signaling and regulation. Antioxid. Redox Signaling 5, 307-317. (37) Liu, X., Miller, M. S., Joshi, M. S., Thomas, D. D., and Lancaster, J. R., Jr. (1998) Accelerated reaction of nitric oxide with O2 within the hydrophobic interior of biological membranes. Proc. Natl. Acad. Sci. U.S.A. 95, 2175-2179. (38) Goss, S. P., Singh, R. J., Hogg, N., and Kalyanaraman, B. (1999) Reactions of *NO, *NO2 and peroxynitrite in membranes: physiological implications. Free Radical Res. 31, 597-606.

1174 Chem. Res. Toxicol., Vol. 19, No. 9, 2006 (39) Bloodsworth, A., O’Donnell, V. B., and Freeman, B. A. (2000) Nitric oxide regulation of free radical- and enzyme-mediated lipid and lipoprotein oxidation. Arterioscler., Thromb., Vasc. Biol. 20, 17071715. (40) Lancaster, J. R., Jr. (2000) The Physical Properties of Nitric Oxide. Determinants of the Dynamics of NO in Tissue. In Nitric Oxide: Biology and Pathobiology (Ignarro, L. J., Ed.) pp 209-224, Academic Press, San Diego, CA. (41) Mikkelsen, R. B., and Wardman, P. (2003) Biological chemistry of reactive oxygen and nitrogen and radiation-induced signal transduction mechanisms. Oncogene 22, 5734-5754. (42) Sevilla, M. D., Becker, D., Swarts, S., and Herrington, J. (1987) Sulfinyl radical formation from the reaction of cysteine and glutathione thiyl radicals with molecular oxygen. Biochem. Biophys. Res. Commun. 144, 1037-1042. (43) Lancaster, J. R., Jr. (2004) Nitroxidation: the dominant process of reactive nitrogen species chemistry under biological conditions. Free Radical Biol. Med. 37, S98. (44) Bonini, M. G., Miyamoto, S., Di Mascio, P., and Augusto, O. (2004) Production of the carbonate radical anion during xanthine oxidase turnover in the presence of bicarbonate. J. Biol. Chem. 279, 5183651843. (45) Karunakaran, C., Zhang, H., Joseph, J., Antholine, W. E., and Kalyanaraman, B. (2005) Thiol oxidase activity of copper, zinc superoxide dismutase stimulates bicarbonate-dependent peroxidase activity via formation of a carbonate radical. Chem. Res. Toxicol. 18, 494-500. (46) Hess, D. T., Matsumoto, A., Kim, S. O., Marshall, H. E., and Stamler, J. S. (2005) Protein S-nitrosylation: purview and parameters. Nat. ReV. Mol. Cell Biol. 6, 150-166. (47) Radi, R. (2004) Nitric oxide, oxidants, and protein tyrosine nitration. Proc. Natl. Acad. Sci. U.S.A. 101, 4003-4008.

Lancaster (48) Klatt, P., and Lamas, S. (2000) Regulation of protein function by S-glutathiolation in response to oxidative and nitrosative stress. Eur. J. Biochem. 267, 4928-4944. (49) Ghezzi, P., Bonetto, V., and Fratelli, M. (2005) Thiol-disulfide balance: from the concept of oxidative stress to that of redox regulation. Antioxid. Redox Signaling 7, 964-972. (50) Ford, E., Hughes, M. N., and Wardman, P. (2002) Kinetics of the reactions of nitrogen dioxide with glutathione, cysteine, and uric acid at physiological pH. Free Radical Biol. Med. 32, 1314-1323. (51) Jaffrey, S. R., and Snyder, S. H. (2001) The biotin switch method for the detection of S-nitrosylated proteins. Sci. STKE. 2001, L1. (52) Zhang, Y., Keszler, A., Broniowska, K. A., and Hogg, N. (2005) Characterization and application of the biotin-switch assay for the identification of S-nitrosated proteins. Free Radical Biol. Med. 38, 874-881. (53) Zhang, H., Xu, Y., Joseph, J., and Kalyanaraman, B. (2005) Intramolecular electron transfer between tyrosyl radical and cysteine residue inhibits tyrosine nitration and induces thiyl radical formation in model peptides treated with myeloperoxidase, H2O2, and NO2-: EPR SPIN trapping studies. J. Biol. Chem. 280, 40684-40698. (54) Hare, J. M. (2004) Spatial confinement of isoforms of cardiac nitricoxide synthase: unravelling the complexities of nitric oxide’s cardiobiology. Lancet 363, 1338-1339. (55) Zhang, H., Squadrito, G. L., Uppu, R. M., Lemercier, J. N., Cueto, R., and Pryor, W. A. (1997) Inhibition of peroxynitrite-mediated oxidation of glutathione by carbon dioxide. Arch. Biochem. Biophys. 339, 183-189.

TX060061W