Toxicity of Peroxynitrite and Related Reactive ... - ACS Publications

The toxicity of peroxynitrite toward Escherichia coli (expressed as LD50, the concentration ... of LD50 was also pH-independent over the range pH 5.9-...
0 downloads 0 Views 257KB Size
Download PDF
802

Chem. Res. Toxicol. 1997, 10, 802-810

Toxicity of Peroxynitrite and Related Reactive Nitrogen Species toward Escherichia coli James K. Hurst* and Sergei V. Lymar† Department of Chemistry, Washington State University, Pullman, Washington 99164-4630 Received January 23, 1997X

The toxicity of peroxynitrite toward Escherichia coli (expressed as LD50, the concentration required to kill 50% of the bacteria) was found to be independent of bacterial cell densities over a wide experimental range, spanning 106-1010 colony-forming units/mL; the magnitude of LD50 was also pH-independent over the range pH 5.9-8.3. This highly unusual behavior can be quantitatively reproduced by a dynamical model in which (i) ONO2H is identified as the toxic form of the oxidant and (ii) the bulk of the added peroxynitrite decays to nitrate ion under these conditions. From the model, one estimates that 106-107 ONO2H molecules are required to kill a bacterium, indicating a very high intrinsic toxicity (cf. HOCl, for which LD50 ) 107-108 molecules/cell of E. coli). Nearly complete protection was observed when bicarbonate ion was added to the buffer, even when concentrations of peroxynitrite exceeded 50 times the LD50 measured in the absence of bicarbonate. Consistent with previous reports, combinations of H2O2 and •NO and, in weakly acidic media, H2O2 and NO2- were found to exhibit enhanced toxicities relative to the individual reactants. Protection by bicarbonate was utilized to assess the potential role of intermediary formation of ONO2H in bacterial killing in these systems. Approximately 25% protection by bicarbonate was observed for media containing H2O2 and NO2-, consistent with a minor contribution to killing by ONO2H under the experimental conditions. No protection was observed for media containing H2O2 and •NO in both anaerobic and aerobic environments, excluding extracellularly generated ONO2H as a participant in these bactericidal reactions.

Introduction Peroxynitrite (oxoperoxonitrate(1-), ONO2-) is a powerful oxidant (1) that is rapidly formed by addition of •O2and •NO radicals (2). In its acidic form, i.e., as peroxynitrous acid (ONO2H, pKa ) 6.8), it reacts rapidly and selectively at oxidizable sites within biomolecules, generally causing loss or modification of function (3-11). Consequently, it has been suggested that physiological conditions that lead to simultaneous generation of •O2and •NO are potential causes of various debilitating diseases (12, 13) and the oxidative damage that accompanies reperfusion of ischemic tissues (14, 15) and that these reactions may also have a primary role in host cellular defense mechanisms against pathogenic organisms (16-18). These suggestions are supported by several indirect lines of evidence, including the observations that (1) markedly enhanced levels of nitrotyrosine (thought to be a specific marker for peroxynitrite) are found in fluids and postmortem tissues of individuals possessing diseases attributed to peroxynitrite but are not found in biological samples from individuals with unrelated diseases (19-22) and (2) macrophages whose capabilities to generate •NO are compromised either by addition of NO synthase inhibitors (23-27) or by genetic mutation (28-30) have impaired microbicidal capabilities. Although a circumstantial link between peroxynitrite formation and cellular function or dysfunction is suggested by these studies, the extent to which peroxynitrite participates in cellular damage and the actual target sites remain undetermined. Compounding this † Present address: Department of Chemistry, Brookhaven National Laboratory, Upton, NY 11973. X Abstract published in Advance ACS Abstracts, June 1, 1997.

S0893-228x(97)00008-8 CCC: $14.00

problem is our recent observation that ONO2- reacts with CO2 according to the following reaction:

ONO2- + CO2 f ONO2CO2-

(1)

forming a nitrosoperoxycarbonate (ONO2CO2-) adduct (31). There are at least two important consequences of this reaction. First, its rate is sufficiently rapid that, in most physiological fluids,1 virtually all the peroxynitrite that might be formed is expected to react further with CO2 to give the adduct (33). Thus, it is the chemical properties of ONO2CO2-, rather than ONO2-, that are most relevant to biological function. Second, the intrinsic lifetime of ONO2CO2- (1 µs e t1/2 e 1 ms)2 is considerably shorter than the lifetime of ONO2H (t1/2 = 0.7 s) (31). Thus, cellular target sites may be considerably less accessible than previously thought. The chemical reactivity of ONO2CO2- is not well described. In the absence of oxidizable substrates, ONO2CO2- decomposes to give nitrate ion as one of the products.3 Carbon dioxide has been shown to promote nitration of tyrosine and protein tyrosyl groups (11, 34, 35); tyrosine nitration is initiated by one-electron oxidation to tyrosyl radical by ONO2CO2-, indicating that the adduct is a strong oxidant with a redox potential exceeding 1 V (34). These studies also clearly revealed the existence of kinetically distinguishable reactive and unreactive forms of ONO2CO2-. However, formation of 1 Peroxidases are rapidly oxidized to compounds I and II by ONO H 2 (32). On the basis of relative rate comparisons (33), these reactions are expected to be the predominant mechanism of ONO2H disappearance in high-peroxidase environments such as the neutrophil phagosome. 2 These limits were established from the kinetics of aromatic nitration (34).

© 1997 American Chemical Society

Bactericidal Potency of ONO2H and ONO2CO2-

ONO2CO2- is the rate-limiting step in these reactions, effectively precluding direct detection and structural characterization of the intermediates. In this study, we examine the reactions of various peroxynitrite species and potential peroxynitrite-generating systems with Escherichia coli to obtain information on the chemical bases for their toxicities. We confirm previous studies indicating that CO2 protects E. coli from peroxynitrite toxicity (36) in in vitro bactericidal assays, but we differ with other earlier conclusions concerning the identities of the toxic compounds.

Experimental Section Cultures of E. coli ATCC 25922 were maintained on tryptic soy agar and grown on nutrient broth as previously described (37). Cells were harvested in mid-to-late log phase, washed once in 50 mM phosphate (pH 7.4) containing 0.15 M NaCl (phosphatebuffered saline, PBS) and once in the final reaction buffer. The bacterial pellet was then resuspended in buffer to give a cell density of 106-1011 colony-forming units (cfu)/mL. Stock peroxynitrite solutions were prepared from potassium nitrite and hydrogen peroxide as previously described (34). Typically, for our preparations, [ONO2-] = 25 mM and [NO2-] = 120 mM. Reactions between peroxynitrite and the bacteria carried out at pH g7.4 were initiated by vortex-mixing appropriate volumes of reactants; below pH 7.4, the reactant solutions were flowmixed using a manually driven two-syringe assembly connected to a 12-jet tangential mixer. The latter technique enabled rapid homogeneous mixing to be achieved, thereby minimizing decomposition of peroxynitrite prior to reaction with the bacteria (31). Where desired, CO2 was removed by bubbling argon through the bacterial suspensions for 5-30 min immediately prior to reaction with peroxynitrite. (The time required for effective purging decreased with increasing solution acidities.) Reactions of ONO2CO2- were studied by adding sodium bicarbonate to the bacterial suspension and allowing a few minutes for establishment of the HCO3- + H+ a CO2 + H2O equilibrium before reaction with peroxynitrite; under all experimental conditions, where [CO2] g 5 mM, essentially all of the peroxynitrite converted to ONO2CO2- prior to further reaction (31). For studies involving H2O2 as oxidant, all buffers and reagent solutions were passed through Chelex-100 columns and reaction glassware was rinsed in concentrated HNO3 to remove adventitious metal ions. Our original protocol included washing the bacteria with EDTA-containing buffers, but this step was omitted when it was determined that it caused progressive loss of viability in pH 5.0 buffers. Reactions involving NO2-, H2O2, or combinations of the two oxidants were generally initiated by vortex-mixing bacterial suspensions and oxidant solutions and then incubating at 25 °C. In the complete reaction system, the order of reagent addition was KNO2, NaHCO3, E. coli, and H2O2. Reactions omitting one or more of the reagents generally followed the same order, although reactions between just NO2and E. coli were initiated by adding KNO2 to bacterial suspensions. Reactions were stopped after 30 min by 102-fold dilution into PBS (pH 7.4). For reactions involving nitric oxide, the gas was generated by dropwise addition of 9 M H2SO4 into 250 mL of solution containing 4 M KNO2 and 1 M KI (38); these concentrations are sufficient to generate ∼6 L of •NO at STP. 3 Kinetic analyses indicate that the other product is CO , i.e., this 2 reaction is CO2-catalyzed isomerization to NO3-. However, in the presence of NO2 or other reductants, the immediate product appears to be HCO3- (S. V. Lymar, unpublished observations). The ONO2CO2adduct has sufficient thermodynamic potential to oxidize NO2- by one electron. A plausible reaction sequence is ONO2CO2- f •NO2 + •CO3-; NO2- + •CO3- f •NO2 + CO32-; 2•NO2 + H2O f 2H+ + NO2- + NO3-; H+ + CO32- f HCO3-, yielding as a net reaction: H2O + ONO2CO2f HCO3- + NO3- + H+. Because HCO3- is unreactive toward peroxynitrite and the equilibrium, HCO3- + H+ a H2O + CO2, is only slowly established under physiological conditions, CO2 acts as a stoichiometric reagent under these conditions, i.e., it is “consumed” by ONO2CO2- decomposition.

Chem. Res. Toxicol., Vol. 10, No. 7, 1997 803 Caution: The gaseous products formed from aerobic oxidation of NO are powerful oxidants that can cause major damage to biological tissues, particularly when inhaled; NO-generating reactors should therefore be located in a properly venting fume hood. The evolved gases were passed sequentially through two scrubbing towers containing 10 M NaOH to remove contaminating N2O3 and I2. Septum-stoppered 24 mL test tubes containing 5-10 mL of solution were used as reaction vessels; gases were bubbled through the solutions using 6 in. hypodermic needles to pierce the septa. The reaction buffers were first purged of O2 by bubbling with N2, and then reactor-generated •NO was bubbled through the solutions for 3 min at a rate of 100-200 mL/min to produce a •NO-saturated solution under an atmosphere of •NO. The purity of the gas was examined by determining the amount of NO2- accumulating in the reaction buffers using the sulfanilamide N-(1-naphthyl)ethylenediamine dihydrochloride reaction (Griess reagent) (39). Typically, ∼10 mM NO2- was found in the buffers, which corresponds to ∼0.2% N2O3 impurity in the gas. Oxidants and bacteria were introduced into the reaction buffer from O2-purged stock solutions using Hamilton gas-tight syringes; when less than 1 atm of •NO was desired, measured amounts of the gas were also introduced into the reaction vessel headspace using gas-tight syringes. In these experiments (Table 1), the amount of NO2- introduced into the reaction buffer was markedly less, i.e., 16-80 µM, based upon the analyses of the N2O3 content of the gas. The order of addition in the complete system was •NO, E. coli, HCO3-, and H2O2; reactions omitting one or more of the reagents followed the same order. Reactions were stopped as above by dilution into PBS; for some of the reactions, •NO was removed by bubbling briefly with N2 prior to exposing the solutions to O2. This procedure prevented the otherwise copious formation of •NO gas and acidification of the medium. Nonetheless, this 2 precaution proved unnecessary because surviving cell counts were unaffected by omitting the •NO removal step. For all bactericidal assays, surviving cells were determined using pourplate methods following serial dilutions into PBS (40). Rates of peroxynitrite isomerization were measured by stopped-flow spectrophotometry as previously described (31). Reactions were initiated by mixing unbuffered alkaline peroxynitrite solutions with reaction buffers or supernatants derived from bacterial suspensions.

Results and Discussion Toxicity of Peroxynitrite and ONO2CO2-. 1. Experimental Observations. Peroxynitrite was found to be bactericidal in CO2-depleted media over a wide range of experimental conditions. Dose-response plots of percent survival vs peroxynitrite concentration were sigmoidal-shaped, consistent with death arising from the cumulative effect of reaction at numerous sites (41). These data are summarized in Figure 1, where toxicities are expressed as LD50, the concentration of peroxynitrite required to kill 50% of the bacteria. Independent measurements established that E. coli maintained their viability in minimal media in the absence of oxidants for greater than 3 h, the maximal time period required to complete the viability studies. Remarkably, at cell densities below ∼1010 cfu/mL, the toxicity was independent of the number of bacteria in culture (Figure 1). This behavior contrasts markedly with that observed for stable toxins such as HOCl and H2O2, where the measured LD50 increases with increasing cell densities in suspension (37, 42). Toxicities were also independent of the solution pH over the measured range (pH 5.9-8.2). This observation differs from a previous report based upon a more limited data set wherein toxicity appeared to increase with increasing pH over the physiological range (36). These reactions are also apparently not metal ion-mediated.

804 Chem. Res. Toxicol., Vol. 10, No. 7, 1997

Hurst and Lymar Scheme 1

Figure 1. Toxicity of peroxynitrite toward E. coli in carbonatefree media. Each point represents the LD50 determined from the dose-response curve at the indicated bacterial cell density. Reaction conditions: 0.2 M phosphate, pH 5.9 (b); 0.2 M phosphate, pH 6.6 (1); 0.1 M phosphate, pH 7.4 (9); 0.1 M phosphate, pH 8.3 (2); all at 23 °C. The inset displays the theoretical fit of the pH 5.9 data according to eq 3, treated as described in the text.

Pretreatment of reagents by Chelex-100 chromatography gave no substantive change in the ONO2- dependence of killing at pH 7.4; moreover, deliberate addition of 250 µM Fe(III)-EDTA afforded partial protection, rather than enhanced killing. Addition of 5-25 mM sodium bicarbonate to the reaction medium gave 90-100% protection of the bacteria from 0.4-1.8 mM ONO2- under all reaction conditions. These oxidant concentration levels, which are 3-10 times greater than LD50, were sufficient to reduce cell viabilities to kbuf/ksup, i.e., the correction is smaller than for the indirect mechanism. (In the limit where (1 - R)k3[E. coli] . ksup, LD50′/LD50 ) 1, i.e., no correction is required for scavenging by supernatant components.) From n ) 2b, the value n ) 8 × 106 ONO2H/bacterium is calculated for E. coli at pH 6.0 from the data given in Figure 1. Thus, the same qualitative picture emerges from the kinetic analysis regardless of whether the reaction of the bacteria with ONO2H is considered to be direct (Scheme 2) or indirect (Scheme 1). 3. Mechanistic Conclusions. Three important conclusions can be drawn from these studies. First, the independence of LD50 on cell densities below 1010 cfu/mL indicates that only a small fraction of the added peroxynitrite participates in bacterial killing. Second, the data strongly implicate ONO2H as the bactericidal form of peroxynitrite. This conclusion follows from the observation that the peroxynitrite LD50 is pH-independent throughout the region where it is also independent of the bacterial cell density (Figure 1). For this condition to hold, the rate for reaction of peroxynitrite with E. coli must have the same pH dependence as the rate for the major pathway for peroxynitrite disappearance which, under these experimental conditions, is isomerization to NO3-. Only the protonated form, ONO2H, is unstable with respect to isomerization (49); consequently, ONO2H is the form that reacts with the bacteria. Further, any kinetic scheme that includes direct reaction between ONO2- and bacteria will give rise to an expression for LD50 that is proportional in [H+], in contrast to the observed behavior. Third, the intrinsic toxicity of ONO2H, measured by n, is extremely high. Comparison of n ≈ 5 × 106 ONO2H/bacterium to that of HOCl, for which n =

Figure 2. Enhancement of H2O2 toxicity by nitrite ion. Reaction conditions: H2O2 alone in carbonate-free media (0); H2O2 plus 20 mM NO2- in carbonate-free media (4); H2O2 plus 20 mM NO2- in 20 mM total carbonate (O); all reactions with 3 × 108 cfu/mL E. coli in 50 mM acetate, pH 5.0, 23 °C.

5 × 107 HOCl/bacterium under comparable conditions (42), indicates that on a molar basis ONO2H is ∼10-fold more toxic than HOCl. The latter oxidant, generally regarded as a potent toxin, kills bacteria by selective inhibition of functions associated with energy transduction in the bacterial plasma membrane (50). The greater toxicity of ONO2H implies an even greater selectivity and possibly alternative loci of attack. In particular, the toxicity of ONO2H approaches that estimated for intracellularly generated •OH radical (51), suggesting that the lethal reaction sites may be intracellular. Toxicity of H2O2 plus NO2-. Two reports have appeared indicating that in weakly acidic media the bactericidal activities of H2O2 and NO2- when combined are greater than the activities of the separate oxidants (52, 53). In one of the reports (53), it was suggested from chemical trapping experiments that the basis for this effect is formation of peroxynitrite, i.e., the reaction H2O2 + HNO2 f ONO2H + H2O. However, based upon the measured rate constant for peroxynitrous acid formation (54), the reaction half-time under the experimental conditions is calculated to be t1/2 ≈ 10 h. We have confirmed these calculations by determining spectrophotometrically at 356 nm the rate of NO2- disappearance in our bactericidal assay medium. Second-order reaction conditions were used, for which [NO2-]T ) [H2O2]T; the subscripts T indicate that concentrations refer to total added reactant. The rate constant calculated from -d[NO2-]T/dt ) k2nd[NO2-]T[H2O2]T was k2nd ) 7.3 × 10-4 M-1 s-1 at pH 5.0, 23 °C, nearly identical to literature values for similar reaction conditions (54). The corresponding half-time for the bactericidal conditions for which the cumulative effects of H2O2 and NO2- are maximal (Figure 2) is t1/2 ≈ 13 h. Thus, only small amounts of peroxynitrite can be formed by direct reaction of HNO2 and H2O2 over typical bactericidal assay times comprising 30-60 min exposure to these oxidants. Under our conditions (Figure 2), at most 5% of the added H2O2 will react to form ONO2- within the reaction time. At pH 7.4, the calculated half-time for this reaction is t1/2 ≈ 50 years; consistent with this value, we observed no loss of NO2- in pH 7.4 phosphate buffers for time periods extending to ∼4 h. The extremely long half-time under these conditions reflects the strong acid dependence of the reaction between H2O2 and NO2-.

Bactericidal Potency of ONO2H and ONO2CO2-

Chem. Res. Toxicol., Vol. 10, No. 7, 1997 807 Table 1. Dependence of Toxicity on •NO Concentrations % survivalb P(•NO)

(atm)

0.055 0.165 0.275 1.0

[•NO]a (mM)

0.5 mM H2O2

1.0 mM H2O2

0.11 0.33 0.55 2.0

85 60 11 20

79 56 35 3

a Assuming Henry’s law behavior with [•NO] ) 2.0 mM at 1 atm [Dean, J. A., Ed. (1985) Lange’s Handbook of Chemistry, 13th ed., pp 10-15, McGraw-Hill, New York]. b 3 × 108 cfu/mL in 0.2 M phosphate, pH 7.4, with 30 min exposure.

Figure 3. Enhancement of H2O2 toxicity by •NO. Reaction conditions: H2O2 alone in carbonate-free media (0); saturated •NO (2.0 mM) alone (2); H O plus 2.0 mM •NO in carbonate2 2 free media (O); H2O2 plus 2.0 mM •NO in 24 mM total carbonate (b); all reactions with 3 × 108 cfu/mL E. coli in 0.2 M phosphate, pH 7.4, 23 °C. Note the abscissa scale change following the axis break.

Since CO2 completely blocks the bactericidal action of peroxynitrite, comparison of E. coli survival curves in CO2-containing and CO2-free media affords a means of assessing the contribution of ONO2H to killing. Typical results are given in Figure 2. The data for CO2-depleted suspensions confirm results obtained by Klebanoff (52) under similar conditions that the toxicity of H2O2 plus NO2- is greater than the toxicities of equivalent amounts of either of the oxidants alone. Addition of 20 mM sodium bicarbonate, which gives CO2-saturated suspensions at this pH, afforded slight but reproducible protection (Figure 2). Control experiments were made comparing toxicities of HNO2 or H2O2 alone in argon-purged and aerobic, CO2-saturated suspensions. No systematic differences were noted in the survival curves in these two atmospheres, indicating that CO2 did not modulate the toxicities of either H2O2 or HNO2 alone. Thus, the partial protection by CO2, comprising 20-30% of the killing, is potentially attributable to ONO2H formation. Nonetheless, the bulk of the killing is not CO2-inhibitable (Figure 2), indicating that the predominant bactericidal mechanisms do not involve formation of exogenous peroxynitrite. At pH 7.4, 1 mM H2O2 alone was virtually nontoxic toward 3 × 108 cfu/mL E. coli; further addition of up to 50 mM KNO2 to the bactericidal assay medium caused less than 20% decline in the viable cell count following exposure for 20 min. This behavior contrasts markedly with the extensive killing observed at pH 5 under comparable conditions (Figure 2). Although this result might have been anticipated from the kinetic inertness exhibited between H2O2 and NO2- at pH 7.4, it also establishes that this reaction was not catalyzed by bacterially derived heme proteins. This issue arises because Klebanoff has shown that several mammalian peroxidases can dramatically enhance the toxicity of H2O2 plus NO2- toward E. coli in neutral solutions (52). Toxicity of H2O2 Plus •NO. At physiologically relevant pH values, E. coli is insensitive to relatively high concentration levels of both H2O2 (37) and •NO (55). Typical behavior is illustrated in Figure 3, where exposure for 20 min to 1 mM H2O2 caused no change and to a saturating atmosphere of •NO caused 20% loss in viability. However, it has recently been shown that •NO

can markedly enhance the bactericidal effectiveness of H2O2 toward E. coli in neutral media (56), and similar, albeit smaller, H2O2 toxicity-enhancing effects have been reported for •NO-generating compounds toward hepatoma cells (57). We have examined the •NO-induced toxicity toward E. coli under a wide range of reaction conditions. In Figure 3, the relative toxicity of •NO-saturated solutions is plotted against the H2O2 concentration. The measured LD50 was 200-250 µM H2O2 for a cell density of ∼3 × 108 cfu/mL, as opposed to an LD50 = 10 mM under comparable conditions in the absence of •NO. The extent of killing was unaltered by addition of 24 mM bicarbonate ion to the reaction medium (Figure 3). Nitric oxide alone exhibited modest bactericidal activities, reducing the bacterial count to ∼80% of the values of controls for which no oxidants were added (Figure 3); this result is consistent with reported observations (36, 56). Loss of viability followed approximately first-order kinetics; the bactericidal reaction half-time under the conditions of Figure 3 with [H2O2] ) 1 mM was t1/2 = 7 min, so that killing was ∼97% complete at 30 min, the time when the reaction was stopped. Solutions containing lower concentration levels of •NO were less effective in promoting H2O2 toxicity (Table 1), particularly when the partial pressure fell below about 0.2 atm. Assuming Henry’s law behavior, this pressure corresponds to a solution concentration of [•NO] ≈ 0.4 mM. Deliberate addition of 40-200 µmol of O2 to 1 atm of • NO in the gaseous headspace over anaerobic bacterial suspensions caused immediate formation of the redbrown gas •NO2, which was slowly adsorbed into the aqueous phase upon vortex-mixing the solution. These procedures did not elicit any enhancement of killing, in either the presence or absence of CO2, suggesting that the toxicity-enhancing effects of •NO could not be attributed to formation of •NO2 or other reactive nitrogen intermediates derived from reaction with adventitious O2. Preincubation of H2O2 with •NO for periods of up to 1 h prior to addition of E. coli did not enhance or attenuate the bactericidal properties of the solutions, which is also consistent with the notion that killing does not involve formation of secondary oxidants. The potential influence of NO2- upon bacterial killing by H2O2 plus •NO was examined by syringe-addition of the gas from a collection reservoir to bacterial suspensions containing known amounts of NO2-. This procedure obviated the uncontrolled accumulation of NO2- that could not be avoided when the gas was directly bubbled through the reaction solution (Experimental Section). No systematic changes were observed in the extent of killing when the NO2concentration was varied between 0 and 50 mM. Specifically, over this [NO2-] range, viability losses accompanying exposure of 3 × 108 cfu/mL E. coli in 0.2 M phosphate (pH 7.4) for 20 min to 1 mM H2O2 plus 0.27 or 0.55 mM •NO were within 22 ( 6% and 48 ( 10%, respectively;

808 Chem. Res. Toxicol., Vol. 10, No. 7, 1997

these data have been corrected for the small amount of toxicity caused by H2O2 plus NO2- in the absence of •NO (preceding section). Thus, there is no evidence of significant modulation of H2O2-•NO toxicity by NO2-. Physiological Implications. 1. ONO2H. The high intrinsic toxicity of ONO2H is masked under the usual in vitro bactericidal assay conditions because, as discussed above, the bulk of the added peroxynitrite decomposes to nitrate. Whether or not this also occurs in biological microcompartments such as the phagosomes of phagocytic cells is probably moot because the predominant reactions of peroxynitrite in physiological environments appear to be with CO2 to form ONO2CO2- (reaction 1) or with peroxidases (32, 33). Within the context of the mathematical development presented above, CO2 protection arises because the k4 term becomes predominant in ksup. The chemical basis for the extreme protection afforded by CO2 cannot presently be ascertained because too little is known about the properties of ONO2CO2-. Alternative possibilities are (i) that ONO2CO2- is less reactive toward the vulnerable sites than ONO2H, (ii) that ONO2CO2- is physically excluded from the vulnerable sites, or (iii) that the ONO2CO2- lifetime is too short to allow appreciable reaction with the bacterium. Alternative i appears unlikely, however, based upon observations that some reactions of ONO2H with biomolecules in homogeneous solution, e.g., tyrosine nitration (34), are catalyzed by CO2. Physical exclusion might be important if the vulnerable sites are intracellular. Because the ONO- group is more strongly electron withdrawing than the hydrogen atom, ONO2CO2H is expected to be a stronger acid than carbonic acid, whose pKa ≈ 3.6. Consequently, the adduct should be anionic in all physiological environments and, barring the presence of a transport system, should penetrate the bacterial plasma membrane much more slowly than the neutral ONO2H molecule (36), whose pKa ) 6.8. As previously discussed (33, 51), if the basis for in vitro protection is the short lifetime of ONO2CO2-, it is erroneous to conclude that because CO2 is protective in in vitro bactericidal systems it would also protect in cellular microcompartments such as phagosomes. This issue cannot be addressed until the ONO2CO2- lifetime is determined. 2. H2O2 Plus NO2-. It has been recently hypothesized from sequence data that the product of the Nramp1 gene, which confers resistance to certain microbes in mammals (58), is a NO2- transport protein (59). This implies an active role for NO2- in microbicidal action. Almost certainly, this putative role does not involve direct reaction with phagocyte-generated H2O2. Above pH ∼6, the uncatalyzed rate of reaction between H2O2 and NO2is far too low to permit formation of any substantive amounts of ONO2- on the time scale of phagocytic killing. Likewise, intraphagosomal acidification, at least within neutrophils (60, 61), appears to occur too slowly to achieve lower pH values within this time period. Under optimal conditions (i.e., pH 5), enhancement of the toxicity of H2O2 toward E. coli required exceptionally high NO2- concentrations and, even then, only a minor fraction of the killing was attributable to peroxynitrite (Figure 2). It is doubtful that these conditions could be achieved in physiological environments; in this case, formation of a nitrogen-derived toxin from H2O2 and NO2-, if it occurs, must be catalyzed. The recent observation by Klebanoff (52) that the neutrophil peroxidase, myeloperoxidase, renders neutral solutions containing H2O2 and NO2- toxic to E. coli suggests that catalysis of

Hurst and Lymar

NO2--derived bactericidal agents could indeed occur at sites of infection. An alternative sequence of reactions involving NO2- that is consistent with the known chemistry of reactive nitrogen species is aerobic oxidation of • NO forming N2O3, followed by its reaction with H2O2 to give ONO2- (62) (discussed below), then reaction with CO2 to give the ONO2CO2- adduct (31), and finally reaction of the adduct with NO2- to give the strongly oxidizing •NO2 radical as an immediate product.3 However, bacterial suspensions containing ONO2CO2- and NO2- at concentrations as high as 11 and 54 mM, respectively, were not toxic; also, deliberate introduction of O2 into anaerobic suspensions containing H2O2, •NO, and NO2- did not promote toxicity. Thus, these studies give no evidence for involvement of •NO2 in bactericidal reactions. Nonetheless, as we have previously demonstrated (51, 63), reactions of extremely short-lived oxidants with bacteria are promoted by compartmentation. Hence, it is not possible in these cases to reliably extrapolate results obtained in in vitro assay systems to intracellular environments, and the possibility exists that analogous chemistry could contribute to cellular death within phagosomes. 3. H2O2 Plus •NO. As previously discussed, at the acidity of these experiments (pH 7.4), direct reaction of H2O2 with any NO2- introduced into the medium by N2O3 impurities in the •NO gas is negligible. Conflicting data exist concerning the reactivity of H2O2 toward •NO. Based upon recent kinetic studies (62), it appears that there is no appreciable direct reaction between these redox reagents. However, in the presence of O2, peroxynitrite can be formed by an indirect pathway comprising intermediary formation of N2O3, which then nitrosates H2O2 (62). E. coli contains relatively high concentrations of catalases (64) which could provide a source of O2, even when rigorously anaerobic suspensions of the bacteria are used. Thus, one possible explanation for the strong synergism between the toxicities of H2O2 and •NO is the formation of toxic levels of ONO2H. However, bicarbonate did not protect the cells (Figure 3), indicating that extracellularly generated ONO2H, at least, was not the source of the enhanced killing in the presence of •NO. In the absence of any evidence for enzyme-catalyzed reactions between H2O2 and •NO, a plausible alternative explanation is that •NO renders E. coli more susceptible to oxidative damage by inactivating the normal cellular defenses against strong oxidants, which might include H2O2, N2O3, or peroxynitrite-derived species (33, 47, 6567). Consistent with this scenario, it has been recently demonstrated that •NO elicits an oxidative stress response in E. coli (68), that both bovine catalase (69) and bovine glutathione peroxidase (70) are inactivated by • NO, and that glutathione levels in E. coli exposed to a •NO-releasing agent were depleted (56). The reported inhibition constant for catalase is Ki ) 0.18 µM (69), so that under the conditions of our experiments, where [•NO] g 110 µM (Table 1), bacterial catalase activity should have been negligible. It remains to be established whether cellularly generated •NO can achieve the high concentrations of •NO necessary for expression of the bactericidal synergism observed with H2O2 in vitro.

General Conclusions These studies establish that peroxynitrous acid is extremely toxic to E. coli. This becomes particularly evident when toxicity is expressed in terms of stoichio-

Bactericidal Potency of ONO2H and ONO2CO2-

metric ratios required for killing. This high inherent toxicity is partially masked in in vitro bactericidal assays because most of the ONO2H decays by isomerization to nitric acid. In marked contrast, no evidence for killing by ONO2CO2-, an oxidant with comparable oxidizing capabilities (33, 34), could be found, even at very high oxidant concentrations. This capacity of bicarbonate to completely protect bacteria from ONO2H was utilized to demonstrate that approximately one-fourth of the toxicity expressed by weakly acidic solutions containing H2O2 and NO2- is attributable to ONO2H formation, whereas no evidence for the intermediacy of this oxidant could be found under a variety of conditions for bactericidal solutions containing H2O2 and •NO. Whether or not biochemically derived CO2 ameliorates oxidative damage by peroxynitrite or functions to enhance its bactericidal properties in physiological environments depends heavily upon the as-yet-undefined chemical properties, lifetime, and decomposition products of ONO2CO2- (33).

Acknowledgment. This research was supported by the National Institute of Allergy and Infectious Diseases under Grant AI15834.

References (1) Koppenol, W. H., Moreno, J. J., Pryor, W. A., Ischiropoulos, H., and Beckman, J. S. (1992) Peroxynitrite, a cloaked oxidant formed by nitric oxidant and superoxide. Chem. Res. Toxicol. 5, 834842. (2) Huie, R. E., and Padmaja, S. (1993) Reaction of NO with O2-. Free Radical Res. Commun. 18, 195-199. (3) Radi, R., Beckman, J. S., Bush, K. M., and Freeman, B. A. (1991) Peroxynitrite oxidation of sulfhydryls. J. Biol. Chem. 266, 42444250. (4) Radi, R., Beckman, J. S., Bush, K. M., and Freeman, B. A. (1991) Peroxynitrite-induced membrane lipid peroxidation. Arch. Biochem. Biophys. 288, 481-487. (5) Moreno, J. J., and Pryor, W. A. (1992) Inactivation of R1proteinase inhibitor by peroxynitrite. Chem. Res. Toxicol. 5, 425431. (6) Beckman, J. S., Ischiropoulos, H., Zhu, L., van der Woerd, M., Smith, C., Chen, J., Harrison, J., Martin, J. C., and Tsai, M. (1992) Kinetics of superoxide dismutase- and iron-catalyzed nitration of phenolics by peroxynitrite. Arch. Biochem. Biophys. 298, 438445. (7) Bauer, M. L., Beckman, J. S., Bridges, R. J., Fuller, C. M., and Matalon, S. (1992) Peroxynitrite inhibits sodium uptake in rat colonic membrane vesicles. Biochim. Biophys. Acta 1104, 87-94. (8) Hausladen, A., and Fridovich, I. (1994) Superoxide and peroxynitrite inactivate aconitases, but nitric oxide does not. J. Biol. Chem. 269, 29405-29408. (9) Castro, L., Rodriguez, M., and Radi, R. (1994) Aconitase is readily inactivated by peroxynitrite but not by its precursor nitric oxide. J. Biol. Chem. 269, 29409-29415. (10) Ischiropoulos, H., and Al-Mehdi, A. B. (1995) Peroxynitritemediated oxidative protein modifications. FEBS Lett. 364, 279282. (11) Berlett, B. S., Friguet, M. B., Chock, P. B., and Stadtman, E. R. (1996) Peroxynitrite-mediated nitration of tyrosine residues in Escherichia coli glutamine synthetase mimics adenylylation: relevance to signal transduction. Proc. Natl. Acad. Sci. U.S.A. 93, 1776-1780. (12) Beckman, J. S., and Crow, J. P. (1993) Pathological implications of nitric oxide, superoxide and peroxynitrite formation. Biochem. Soc. Trans. 21, 330-334. (13) Graham, A., Kalyanaraman, B., O’Leary, V., Darley-Usmar, V., and Moncada, S. (1993) Peroxynitrite modification of low-density lipoprotein leads to recognition by the macrophage scavenger receptor. FEBS Lett. 330, 181-185. (14) Beckman, J. S. (1991) The double-edged role of nitric oxide in brain function and superoxide-mediated injury. J. Develop. Physiol. 15, 53-59. (15) Szabo, C., Salzman, A. L., and Ischiropoulos, H. (1995) Peroxynitrite-mediated oxidation of dihydrorhodamine 123 occurs in early stages of endotoxic and hemorrhagic shock and ischemiareperfusion injury. FEBS Lett. 372, 229-232.

Chem. Res. Toxicol., Vol. 10, No. 7, 1997 809 (16) Denis, M. (1994) Human monocytes/macrophages: NO or no NO? J. Leuk. Biol. 55, 682-684. (17) Kuhl, S. J., and Rosen, H. (1995) Killing of microorganisms by 'professional phagocytes'. Curr. Opin. Infect. Dis. 8, 181-185. (18) DeGroote, M. A., and Fang, F. C. (1995) NO inhibitors: antimicrobial properties of nitric oxide. Clin. Infect. Dis. 21, S162S164. (19) Beckman, J. S., Ye, Y. Z., Anderson, P. G., Chen, J., Accavitti, M. A., Tarpey, M. M., and White, C. R. (1994) Extensive nitration of protein tyrosines in human atherosclerosis detected by immunohistochemistry. Biol. Chem. Hoppe-Seyler 375, 81-85. (20) Haddad, I. Y., Pataki, G., Hu, P., Galliani, C., Beckman, J. S., and Matalon, S. (1994) Quantitation of nitrotyrosine levels in lung sections of patients and animals with acute lung injury. J. Clin. Invest. 94, 2407-2413. (21) Kaur, H., and Halliwell, B. (1994) Evidence for nitric oxidemediated oxidative damage in chronic inflammation. Nitrotyrosine in serum and synovial fluid from rheumatoid patients. FEBS Lett. 350, 9-12. (22) Kooy, N. W., Royall, J. A., Ye, Y. Z., Kelly, D. R., and Beckman, J. S. (1995) Evidence of in vivo peroxynitrite production in human acute lung injury. Am. J. Respir. Crit. Care Med. 151, 1250-1254. (23) Dawson, V. L., Dawson, T. M., London, E. D., Bredt, D. S., and Snyder, S. H. (1991) Nitric oxide mediates glutamate neurotoxicity in primary cortical cultures. Proc. Natl. Acad. Sci. U.S.A., 88, 6368-6371. (24) Mulligan, M. S., Hevel, J. M., Marletta, M. A., and Ward, P. A. (1991) Tissue injury caused by deposition of immune complexes is L-arginine dependent. Proc. Natl. Acad. Sci. U.S.A., 88, 63386342. (25) Nowicki, J. P., Duval, D., Poignet, H., and Scatton, B. (1991) Nitric oxide mediates neuronal death after focal cerebral ischemia in the mouse. Eur. J. Pharmacol. 204, 339-340. (26) Matheis, G., Sherman, M. P., Buckberg, G. D., Haybron, D. M., Young, H. H., and Ignarro, L. J. (1992) Role of L-arginine-nitric oxide pathway in myocardial reoxygenation injury. Am. J. Physiol. 262, H616-H620. (27) Karupiah, G., Xie, Q., Buller, M. L., Nathan, C., Duarte, C., and MacMicking, J. D. (1993) Inhibition of viral replication by interferon-γ-induced nitric oxide synthase. Science 261, 14451448. (28) Stenger, S., Thuring, H., Rollinghoff, M., and Bogdan, C. (1994) Tissue expression of inducible nitric oxide synthase is closely associated with resistance to Leishmania major. J. Exp. Med. 180, 783-793. (29) MacMicking, J. D., Nathan, C., Hom, G., Chartrain, N., Fletcher, D. S., Trumbauer, M., Stevens, K., Xie, Q., Sokol, K., Hutchinson, N., Chen, H., and Mudgett, J. S. (1995) Altered responses to bacterial infection and endotoxic shock in mice lacking inducible nitric oxide synthase. Cell 81, 641-650. (30) Wei, X., Charles, I. G., Smith, A., Ure, J., Feng, G., Huang, F., Xu, D., Muller, W., Moncada, S., and Liew, F. Y. (1995) Altered immune responses in mice lacking inducible nitric oxide synthase. Nature 375, 408-411. (31) 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, 8877-8868. (32) Floris, R., Piersma, S. R., Yang, G., Jones, P., and Wever, R. (1993) Interaction of myeloperoxidase with peroxynitrite. A comparison with lactoperoxidase, horseradish peroxidase, and catalase. Eur. J. Biochem. 215, 767-775. (33) Lymar, S. V., and Hurst, J. K. (1996) Carbon dioxide: physiological catalyst for peroxynitrite-mediated cellular damage or cellular protectant? Chem. Res. Toxicol. 9, 845-850. (34) Lymar, S. V., Jiang, Q., and Hurst, J. K. (1996) Mechanism of carbon dioxide-catalyzed oxidation of tyrosine by peroxynitrite. Biochemistry 35, 7855-7861. (35) van der Vliet, A., O’Neill, C. A., Halliwell, B., Cross, C. E., and Kaur, H. (1994) Aromatic hydroxylation and nitration of phenylalanine and tyrosine by peroxynitrite. Evidence for hydroxyl radical production from peroxynitrite. FEBS Lett. 339, 89-92. (36) Zhu, L., Gunn, C., and Beckman, J. S. (1992) Bactericidal activity of peroxynitrite. Arch. Biochem. Biophys. 298, 452-457. (37) Elzanowska, H., Wolcott, R. G., Hannum, D. M., and Hurst, J. K. (1995) Bactericidal properties of hydrogen peroxide and copper or iron-containing complex ions in relation to leukocyte function. Free Radical Biol. Med. 18, 437-449. (38) Johnston, H. L., and Giauque, W. F. (1929) The heat capacity of nitric Oxide from 14 K to the boiling point and the heat of vaporization. J. Am. Chem. Soc. 51, 3194-3225. (39) Greenberg, A. E., Conners, J. J., and Jenkins D., Eds. (1981) Standard Methods for the Examination of Water and Wastewater, 15th ed., American Public Health Publications, Washington, DC.

810 Chem. Res. Toxicol., Vol. 10, No. 7, 1997 (40) Klebanoff, S. J., Waltersdorph, A. M., and Rosen, H. (1984) Antimicrobial activity of myeloperoxidase. Methods Enzymol. 105, 399-403. (41) Dertinger, H., and Jung, H. (1970) Molecular Radiation Biology, Springer-Verlag, New York. (42) Albrich, J. M., and Hurst, J. K. (1982) Oxidative inactivation of Escherichia coli by hypochlorous acid. Rates and differentiation of respiratory from other sites. FEBS Lett. 144, 157-161. (43) Leung, H. B., and Schramm, V. L. (1980) Adenylate degradation in Escherichia coli. The role of AMP nucleosidase and properties of the purified enzyme. J. Biol. Chem. 255, 10867-10874. (44) 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. (45) Inoue, S., and Kawanishi, S. (1995) Oxidative damage induced by simultaneous generation of nitric oxide and superoxide. FEBS Lett. 371, 86-88. (46) Yermilov, V., Rubio, J., and Ohshima, H. (1995) Formation of 8-nitroguanine in DNA treated with peroxynitrite in vitro and its rapid removal from DNA by depurination. FEBS Lett. 376, 207-210. (47) Pryor, W. A., and Squadrito, G. L. (1995) The chemistry of peroxynitrite: a product from the reaction of nitric oxide with superoxide. Am. J. Physiol. 268, L699-L722. (48) Goldstein, S., and Czapski, G. (1995) Direct and indirect oxidations by peroxynitrite. Inorg. Chem. 34, 4041-4048. (49) Edwards, J. O., and Plumb, R. C. (1994) The Chemistry of Peroxonitrites. Prog. Inorg. Chem. 41, 599-635. (50) Hurst, J. K., and Barrette, W. C., Jr. (1989) Leukocyte oxygen activation and microbial oxidative toxins. CRC Crit. Rev. Biochem. Mol. Biol. 24, 271-328. (51) Lymar, S. V., and Hurst, J. K. (1995) Role of compartmentation in promoting toxicity of leukocyte-generated strong oxidants. Chem. Res. Toxicol. 8, 833-840. (52) Klebanoff, S. J. (1993) Reactive nitrogen intermediates and antimicrobial activity: Role of nitrite. Free Radical Biol. Med. 14, 351-360. (53) Kono, Y., Shibata, H., Adachi, K., and Tanaka, K. (1994) Lactatedependent killing of Escherichia coli by nitrite plus hydrogen peroxide: a possible role of nitrogen dioxide. Arch. Biochem. Biophys. 311, 153-159. (54) Anbar, M., and Taube, H. (1954) Interaction of nitrous acid with hydrogen peroxide. J. Am. Chem. Soc. 76, 6243-6247. (55) Brunelli, L., Crow, J. P., and Beckman, J. S. (1995) The comparative toxicity of nitric oxide and peroxynitrite to Escherichia coli. Arch. Biochem. Biophys. 316, 327-334.

Hurst and Lymar (56) Pacelli, R., Wink, D. A., Cook, J. A., Krishna, M. C., DeGraff, W., Friedman, N., Tsokos, M., Samuni, A., and Mitchell, J. B. (1995) Nitric oxide potentiates hydrogen peroxide-induced killing of Escherichia coli. J. Exp. Med. 182, 1469-1479. (57) Ioannidis, I., and DeGroot, H. (1993) Cytotoxicity of nitric oxide in Fu5 rat hematoma cells: evidence for co-operative action with hydrogen peroxide. Biochem. J. 296, 341-345. (58) Nathan, C. (1995) Natural resistance and nitric oxide. Cell 82, 873-876. (59) Vidal, S., Malo, D., Vogan, K., Skamene, E., and Gros, P. (1993) Natural resistance to infection with intracellular parasites: isolation of a candidate for Bcg. Cell 73, 469-485. (60) Segal, A. W., Geisow, M., Garcia, R., Harper, A., and Miller, R. (1981) The respiratory burst of phagocytic cells is associated with a rise in vacuolar pH. Nature 290, 406-409. (61) Cech, P., and Lehrer, R. I. (1984) Phagolysosomal pH of human neutrophils. Blood 63, 88-95. (62) Goldstein, S., and Czapski, G. (1996) Formation of peroxynitrite from the nitrosation of hydrogen peroxide by an oxygenated nitric oxide solution. Inorg. Chem. 35, 5935-5940. (63) Wolcott, R. G., Franks, B. S., Hannum, D. S., and Hurst, J. K. (1994) Bactericidal potency of hydroxyl radical in physiological environments. J. Biol. Chem. 269, 9721-9728. (64) Heimberger, A., and Eisenstark, A. (1988) Compartmentalization of catalases in Escherichia coli. Biochem. Biophys. Res. Commun. 154, 392-397. (65) Rubbo, H., Darley-Usmar, V., and Freeman, B. A. (1996) Nitric oxide regulation of tissue free radical injury. Chem. Res. Toxicol. 9, 809-820. (66) Tamir, S., Burney, S., and Tannenbaum, S. R. (1996) DNA damage by nitric oxide. Chem. Res. Toxicol. 9, 821-827. (67) Beckman, J. S. (1996) Oxidative damage and tyrosine nitration from peroxynitrite. Chem. Res. Toxicol. 9, 836-844. (68) Nunoshiba, T., deRojas-Walker, T., Wishnok, J. S., Tannenbaum, S. R., and Demple, B. (1993) Activation by nitric oxide of an oxidative-stress response that defends Escherichia coli against activated macrophages. Proc. Natl. Acad. Sci. U.S.A., 90, 99939997. (69) Brown, G. C. (1995) Reversible binding and inhibition of catalase by nitric oxide. Eur. J. Biochem. 232, 188-191. (70) Asahi, M., Fujii, J., Suzuki, K., Seo, H. G., Kuzuya, T., Hori, M., Tada, M., Fujii, S., and Taniguchi, N. (1995) Inactivation of glutathione peroxidase by nitric oxide. Implication for cytotoxicity. J. Biol. Chem. 270, 21035-21039.

TX970008V