Nitroxyl and its anion in aqueous solutions: Spin states, protic equilibria, and reactivities toward oxygen and nitric oxide

Abstract

The thermodynamic properties of aqueous nitroxyl (HNO) and its anion (NO⁻) have been revised to show that the ground state of NO⁻ is triplet and that HNO in its singlet ground state has much lower acidity, pKa(¹HNO/³NO⁻) ≈ 11.4, than previously believed. These conclusions are in accord with the observed large differences between ¹HNO and ³NO⁻ in their reactivities toward O₂ and NO. Laser flash photolysis was used to generate ¹HNO and ³NO⁻ by photochemical cleavage of trioxodinitrate (Angeli's anion). The spin-allowed addition of ³O₂ to ³NO⁻ produced peroxynitrite with nearly diffusion-controlled rate (k = 2.7 × 10⁹ M⁻¹⋅s⁻¹). In contrast, the spin-forbidden addition of ³O₂ to ¹HNO was not detected (k ≪ 3 × 10⁵ M⁻¹⋅s⁻¹). Both ¹HNO and ³NO⁻ reacted sequentially with two NO to generate N₃O as a long-lived intermediate; the rate laws of N₃O formation were linear in concentrations of NO and ¹HNO (k = 5.8 × 10⁶ M⁻¹⋅s⁻¹) or NO and ³NO⁻ (k = 2.3 × 10⁹ M⁻¹⋅s⁻¹). Catalysis by the hydroxide ion was observed for the reactions of ¹HNO with both O₂ and NO. This effect is explicable by a spin-forbidden deprotonation by OH⁻ (k = 4.9 × 10⁴ M⁻¹⋅s⁻¹) of the relatively unreactive ¹HNO into the extremely reactive ³NO⁻. Dimerization of ¹HNO to produce N₂O occurred much more slowly (k = 8 × 10⁶ M⁻¹⋅s⁻¹) than previously suggested. The implications of these results for evaluating the biological roles of nitroxyl are discussed.

Nitroxyl (HNO, also known as nitrosyl hydride) and its anion, NO⁻, are the simplest molecules with nitrogen in the +1 oxidation state and yet their aqueous chemistry is not well understood. Recent suggestions that these redox neighbors of the biologically important NO radical may play a role in cellular metabolism (1–4) and in aerobic environments may be precursors to cytotoxic peroxynitrite, ONOO⁻, (5, 6) have engendered considerable interest in the chemistry of HNO/NO⁻. The characterization of these species is complicated by their instability with respect to formation of nitrous oxide (7, 8). In most cases where nitroxyl has been invoked as an intermediate, the rate-determining step was its generation, a situation that allows little insight into the properties and reactivities of HNO/NO⁻ themselves. The NO⁻ anion is isoelectronic with O₂ and, like O₂, should have a triplet ground state, whereas the ground state of HNO should be a singlet. Indeed, these ground state assignments have been well established for HNO/NO⁻ in the gas phase (9, 10).

A frequently used source for aqueous HNO/NO⁻ is trioxodinitrate (N₂O, also known as Angeli's anion), whose conjugate acid (H₂N₂O₃) has consecutive pKa values of 2.5 and 9.7 (11). It is widely accepted (7, 8) that slow decomposition of the monoprotonated anion occurs through heterolytic N—N bond cleavage

1

Subsequent addition of O₂ could yield peroxynitrite

2

However, nitrate, which is the peroxynitrite decomposition product, was not detected among the end products of HN₂O decay (12). This result was interpreted as evidence against the occurrence of reaction 2. On the other hand, the same researchers reported peroxynitrite formation during N₂O photolysis in alkaline solution (13). To reconcile the data, it was suggested that thermal reaction 1 followed by deprotonation of HNO produces singlet NO⁻, which is the ground state in water, and that ¹NO⁻ is unreactive toward O₂. In contrast, photochemical cleavage of N₂O was thought to generate the long-lived triplet excited state of NO⁻, which reacted with O₂. However, it seems unlikely that hydration can reverse a gas-phase energy gap of about 70 kJ/mol between the ground state ³NO⁻ and the excited state ¹NO⁻ (10). Moreover, by analogy with ¹O₂, whose lifetime in water is only 4 μs (14), the existence of a long-lived excited state of NO⁻ also appears highly unlikely. Thus, reaction 2 remains obscure, particularly because others have reported O₂ consumption during spontaneous decomposition of HN₂O (2, 15) and detection of small amounts of NO upon completion of the reaction (16).

In the early 1970s two research groups used pulse radiolysis to produce HNO/NO⁻ by reaction between the hydrated electrons and NO (17, 18). This approach precluded examination of HNO/NO⁻ reactivities toward O₂ because of a very rapid addition of two NO radicals to form N₃O, which then slowly decomposed to nitrous oxide and nitrite

3

A pKa value of 4.7 for HNO was reported, although the spin states of NO⁻ and HNO were not specified (17). The estimates for the reduction potential of the NO/NO⁻ couple based on this value (19) are almost a volt higher than those obtained by electrochemical techniques (20, 21); this discrepancy has never been addressed. Recent ab initio calculations placed the pKa of HNO at 7.2 (22).

In the hope of clarifying the HNO/NO⁻ chemistry, we have used UV laser flash photolysis to produce HNO/NO⁻ species by photochemical cleavage of Angeli's anion and to investigate their reactivities. Here we present evidence that HNO has a much weaker acidity than previously believed. We show that the deprotonation of HNO is a slow spin-forbidden process that controls the observed chemistry in alkaline solutions. The reactivities of HNO and NO⁻ toward O₂ and NO have been investigated and a quantitative mechanistic description of these reactions is presented.

Materials and Methods

Sample Solutions.

All chemicals were of analytical grade and were used as received. Milli-Q purified water was used throughout. Stock solutions of Na₂N₂O₃ (Cayman Chemical, Ann Arbor, MI) in 10 mM NaOH were prepared daily. Relatively stable N₂O [ɛ₂₄₈ = 8,300 M⁻¹⋅cm⁻¹ (23)] sample solutions at pH 11–14.3 were prepared by diluting the Na₂N₂O₃ stock. Unstable HN₂O [ɛ₂₃₇ = 6,100 M⁻¹⋅cm⁻¹ (24)] sample solutions at pH 4–10 were prepared by flow-mixing equal volumes of the Na₂N₂O₃ stock and 0.2 M phosphate, acetate, or borate buffers as described below. Nitric oxide (Matheson) was purified by passing through a scrubbing column with 2 M KOH and then through water. The various NO/Ar and O₂/Ar mixtures were produced by combining the gas streams of NO or O₂ with Ar passed through calibrated flowmeters. All solutions were thoroughly purged with argon before introducing the NO-containing mixtures. The NO and O₂ solubilities at 1 atm (1 atm = 101.3 kPa) were taken as 1.9 and 1.3 mM, respectively. A 100-W xenon arc beam reflected at 45^o from a broadband dielectric mirror (220–300 nm) was used for UV steady-state photolysis. All experiments were performed at ambient temperature (22 ± 2°C).

Laser Flash Photolysis with Flow Premixing.

Transient absorption spectra were recorded by using a computerized kinetic spectrometer system as described (25). Briefly, a 266-nm Nd:YAG laser operating at 20 Hz was used to photolyse the samples in a 0.4 × 1-cm quartz flow cell. To allow for solution replacement between the laser shots and monitoring slower kinetics, the sample excitation frequency was reduced to either 1 or 0.1 Hz by using a computer-controlled electromechanical shutter. Two solutions (e.g., alkaline N₂O and buffer), each saturated with a desired gas mixture, were forced by a positive gas pressure into a 12-jet tangential mixer and then through the flow cell with a flow rate of ≈12 ml/min; the laser excitation occurred within less than 1 s after mixing. The transient absorption was probed along a 1-cm optical path by a light beam from a 75-W xenon arc lamp, which was pulsed for short time scales, and the kinetic traces obtained were averaged over 10–20 laser pulses. For the quantum yield estimates, the laser pulse energies (30–40 mJ/cm²) were measured by a calibrated thermoelectric bolometer.

Results and Discussion

Energetics and Spin States.

To facilitate the analysis of the kinetic data, we begin with a thermodynamic description of the intermediates that can be generated by photolysis of the trioxodinitrate solutions. Fig. 1 presents the free energies of formation, Δ_fG⁰, for the NO/HNO/NO⁻ species in aqueous solution, estimated by using published data. The accurate value for Δ_fG⁰(NO_aq) = 102 kJ/mol has been derived (19). The enthalpy of ¹HNO formation in the gas phase has been reviewed and the value of 107 kJ/mol has been recommended (26). From this value and the tabulated entropy S⁰(¹HNO_gas) = 220.7 J/(mol·K) (27), we calculate Δ_fG⁰(¹HNO_gas) = 120 kJ/mol. The free energy of HNO hydration is unknown, but is expected to be small by analogy with neutral molecules of similar dimensions and compositions, e.g., HCN and H₂CO, for which Δ_hydrG⁰ are approximately −5 and −1.7 kJ/mol, respectively. The value Δ_fG⁰(¹HNO_aq) ≈115 kJ/mol is, therefore, a reasonable estimate, which is also close to a 109 kJ/mol value derived previously (19).

Figure 1.

Energy diagram for NO/HNO/NO⁻ species in aqueous solution at 298 K and 1 mol/kg standard states. Note the energy axis break at 120 kJ/mol. The spectroscopic designations for the electronic states are given in parenthesis. Only the lowest excited states are shown for HNO and NO⁻.

The reduction potential of NO measured by the photoelectrochemical technique has been reported as E⁰(NO/NO⁻) = −0.81 V vs. NHE without specifying the NO⁻ spin state (21). This value is consistent with the upper limit E⁰(NO/NO⁻) < −0.7 V vs. NHE that can be inferred from the onset of the irreversible NO reduction wave observed in controlled-potential coulometry (20). In both experiments the reducing electrons were supplied by the metal electrodes, a process for which there is no spin prohibition regardless of the product spin state. Assuming, therefore, that these measurements pertain to the NO reduction in the ground ³NO⁻ state, Δ_fG⁰(³NO) ≈180 kJ/mol can be estimated. Finally, the values for Δ_fG⁰(³HNO_aq) ≈190 kJ/mol and for Δ_fG⁰(¹NO) ≈248 kJ/mol can be assigned under a reasonable presumption that hydration does not appreciably alter the gas-phase energy gaps of 75 kJ/mol between ³HNO and ¹HNO (9) and 68 kJ/mol between ¹NO⁻ and ³NO⁻ (10).

From the energy diagram, we estimate pKa(¹HNO/³NO⁻) ≈11.4 and pKa(¹HNO/¹NO⁻) ≈23, i.e., the acidity of HNO is very low. This result attests to the closer chemical similarity between HNO and an aldehyde than between HNO and an oxyacid. In its triplet state, HNO becomes a strong acid, pKa(³HNO/³NO⁻) ≈−1.8.

Peroxynitrite Formation upon Addition of O₂.

Alkaline solutions.

In strongly alkaline solutions trioxodinitrate is completely deprotonated and stable for hours. Steady-state UV photolysis of N₂O in air-equilibrated solutions resulted in the rapid disappearance of the N₂O absorption band at 248 nm and in the simultaneous appearance of characteristic absorption of the peroxynitrite anion (ONOO⁻) around 300 nm and nitrite at λ < 230 nm (see Fig. 6, which is published as supporting information on the PNAS web site, www.pnas.org). This result is in qualitative accord with the previous communication (13). The same spectral changes were observed in O₂-saturated solutions. The presence of two well-defined isosbestic points at 225 and 283 nm indicated the constancy of reaction stoichiometry during the photolysis. From the spectral changes, a chemical yield of peroxynitrite (defined as −Δ[ONOO⁻]/Δ[N₂O]) of 0.95 was calculated, i.e., within the uncertainties in molar absorptivities used for this estimate, the stoichiometry of photochemical reaction was

4

Purging O₂ from the solution did not alter the rate of N₂O decay, but completely suppressed ONOO⁻ formation. These observations suggest that ONOO⁻ is produced by the reaction of O₂ with the nascent products of the N₂O photochemical decomposition, rather than by direct scavenging of the N₂O excited state by O₂.

This conclusion is in accord with the kinetics of peroxynitrite formation, which was monitored by flash photolysis of N₂O in O₂-saturated solutions. The prompt bleaching of the N₂O spectrum at λ < 300 nm induced by the laser flash was followed by the absorption growth around 300 nm, corresponding to ONOO⁻ formation. At pH ≥12, the rise of absorbance conformed well to a single exponential growth, ΔA_t = ΔA_∞{1 − exp(−k_formt)}. The values of k_form, determined from kinetic traces recorded at different pH values, sharply increased with the solution alkalinity as shown in Fig. 2. At pH 12.5 and above, k_form was essentially proportional to [OH⁻]. At the same time, the peroxynitrite signal amplitude at the end of the kinetic run (ΔA_∞) showed a sigmoidal dependence on pH (Fig. 2). A salient kinetic feature of the ONOO⁻ formation is that, at a fixed pH, its rate constant was independent of oxygen concentration throughout the experimentally accessible range (Fig. 7, which is published as supporting information on the PNAS web site). Thus, the rate-determining step of this process is the reaction between the N₂O photolysis product and OH ⁻; the reaction with O₂ occurs in a step that is not rate limiting.

Figure 2.

Dependencies on the solution pH of the peroxynitrite formation rate constant (k_form, ○) and the normalized absorbance amplitude recorded at 315 nm after completion of the reaction (□). Conditions were: 0.15 mM N₂O in O₂-saturated solution; pH ≥ 11 were maintained with NaOH, pH < 11 with 0.1 M borate. The solid lines show theoretical dependencies with k₇ = 4.9 × 10⁴ M⁻¹⋅s⁻¹, k₉ = 8 × 10⁶ M⁻¹⋅s⁻¹, and [¹HNO]₀ = 0.05 mM, calculated as described in the text and Note 1.

Collectively, the results described are consistent with the reaction sequence that begins with heterolytic photochemical cleavage of N₂O, producing NO⁻ in its singlet state

5

As described above, ¹NO⁻ is an extremely strong base (pKa ≈23) and, therefore, should be protonated by water nearly instantaneously

6

In fact, the lifetime of ¹NO⁻ is expected to be so short that not even its diffusion-controlled reactions with O₂ should be able to compete with reaction 6. Here we suggest that the rate-determining step in the peroxynitrite formation is the deprotonation of ¹HNO to ³NO⁻, which is a much weaker base (pKa ≈11.4) than ¹NO⁻

7

Completing the process is the rapid, spin-allowed addition reaction

8

This mechanism accounts for the major reactivity features. Specifically, it predicts the independence of k_form of [O₂], as well as the linear dependence of k_form on pH. The deviation from linearity of the latter and the signal amplitude decrease in solutions with pH <13 (Fig. 2) suggest the existence of an additional channel for ¹HNO disappearance, which competes with reaction 7. We could not satisfactorily account for these effects by introducing a first-order channel for the ¹HNO decay. Moreover, a falloff in the ONOO⁻ yield with decreasing pH analogous to that shown in Fig. 2 was not observed under steady-state photolysis, where the transitory concentration of ¹HNO is much smaller than in the flash photolysis experiments. We therefore conclude that the reaction competing with reaction 7 is a bimolecular recombination of ¹HNO

9

This reaction has been invariably invoked previously to explain the stoichiometry and the isotopic composition of end products of the spontaneous decay of Angeli's salt (7, 8, 28). The competition between reactions 7 and 9 is readily amenable to analytical solution, which gives dependencies of the relative amplitude and the ONOO⁻ formation rate constant upon k₇[OH⁻] and k₉[¹HNO]₀ (see Note 1, which is published as supporting information on the PNAS web site). The initial concentration [¹HNO]₀ ≈0.05 mM was determined from the peroxynitrite signal amplitudes around pH 14, where it was assumed that [¹HNO]₀ = [ONOO⁻]_∞. Fitting the theoretical dependencies to the data in Fig. 2 yielded k₇ = (4.9 ± 0.5) × 10⁴ M⁻¹⋅s⁻¹ and k₉ = (8 ± 3) × 10⁶ M⁻¹⋅s⁻¹. The value of k₇ closely corresponds to the slope of the k_form vs. pH curve at pH >13 in Fig. 2 and is rather insensitive to the value of k₉. In contrast, the position of the inflection point in the calculated relative amplitude curve is sensitive to the magnitude of k₉; this, together with the significant scatter of the data points, results in the relatively large uncertainty in the derived value for k₉.

Our suggestion of ¹HNO deprotonation (reaction 7) as the rate-determining step of ONOO⁻ formation is further supported by the observation of the H/D isotope effect on the reaction rate. The ONOO⁻ formation rate constants were measured in H₂O and D₂O under conditions of Fig. 2 at 0.25 M alkali adjusted with NaOH and NaOD, respectively. At this alkalinity, reaction 7 dominates reaction 9 and k_form ≈ k₇[OH⁻]. The ratio k_form(H₂O)/k_form(D₂O) = 3.0 ± 0.1 was obtained, which is consistent with proton involvement in the rate-determining step.

The quantum yield of ONOO⁻ upon 266-nm laser pulse irradiation of N₂O in O₂-saturated 0.25 M NaOH was estimated to be ≈0.5. The stoichiometry of peroxynitrite formation given by reaction 4 shows that there are no channels for the N₂O photochemical decomposition other than reaction 5. Thus, the quantum yield of the N₂O photochemical cleavage is also close to 0.5.

Neutral solutions.

In neutral solutions, the HN₂O monoanion is the predominant trioxodinitrate species. The transient absorption spectra recorded after laser flash photolysis of O₂-saturated HN₂O solution showed a characteristic peroxynitrite band around 300 nm (Fig. 3). This absorption decayed in seconds, which is consistent with peroxynitrite lifetime at pH 7. To further verify that the 300-nm band in Fig. 3 belongs to peroxynitrite, we resorted to a highly specific test based on the rapid decay of peroxynitrite in bicarbonate solutions. At 20 mM HCO and pH 7, the half-life of the 300-nm band decreased to 20 ms, which corresponds to the half-life of authentic peroxynitrite under these conditions (29).

Figure 3.

Transient absorption spectra measured at the indicated delay times after laser pulse excitation of 0.6 mM HN₂O at pH 7 (0.1 M phosphate). Open symbols: O₂-saturated solution; closed symbols: Ar-saturated solution. The solid line is calculated by using Eq. 13 with y = 0.25 and ΔC = 0.052 mM; the dashed lines are visual aids only. (Inset) Kinetic traces recorded at 315 nm for O₂- and Ar-saturated solutions; the short negative-going signal at the time origin is caused by laser stray light.

The kinetics of peroxynitrite formation in nearly neutral solutions differed from those in alkali in two major ways. First, the formation rate constant (k_form) did not depend on pH in the 6–9 range and was orders of magnitude greater than at alkalinities above pH 11. Second, k_form linearly depended on the O₂ concentration (Fig. 7), suggesting that the rate-determining step of peroxynitrite formation in neutral solutions involves a direct reaction between a nascent product of the HN₂O photolysis and O₂. Apparently, a mechanism given by reactions 5–8 cannot account for the observations when the light-absorbing species is HN₂O.

To explain the data, we suggest two channels of the heterolytic photochemical cleavage of HN₂O, one of which results in the formation of HNO in its triplet state, i.e., the reactions

10

11

The protonation of N₂O reduces the N—N bond order from 2 to 1, because an H atom in HN₂O resides on a nitrogen atom (7). This difference in the N—N bonding may conceivably modulate the primary photochemical cleavage pathways. Branching into multiple pathways has been well documented for the photochemical decompositions of oxyanions; moreover, many of them very rapidly expel ³O as a result of the formally spin-forbidden heterolytic bond cleavage (30–32). It is also possible that reaction 10 is, in fact, spin-allowed. The triplet state of NO is only 2.3 eV above the ground state (33) and the 266-nm photon energy (4.7 eV) is sufficient to produce ³HNO and ³NO. As we argued previously, ³HNO is a strong acid and, therefore, should dissociate very rapidly in a buffered solution

12

Peroxynitrite formation via reaction 8, which becomes the rate-limiting step, then follows. It should be noted that any alternative chain of events that leads to ³NO⁻ within a few nanoseconds upon laser light absorption will explain the data.

The kinetics of absorbance increase at 315 nm in O₂-saturated solution consisted of a fast rise, which could not be time resolved, and a slower growth with first-order rate constant k_form (Fig. 3 Inset). The transient spectrum at the end of the fast rise (0.05 μs in Fig. 3) could be assigned to the superposition of absorption by ³NO⁻ and bleaching of HN₂O apparent below 280 nm. Indeed, very similar spectra appeared on this time scale upon flash photolysis of Ar-saturated HN₂O solutions (Fig. 3), where no peroxynitrite was formed. The slower absorption growth at 315 nm corresponded to reaction 8; from the linear dependence of k_form on [O₂] the rate constant k₈ = (2.7 ± 0.2) × 10⁹ M⁻¹⋅s⁻¹ was obtained (see Fig. 7).

The transient absorbance amplitude at 315 nm upon the completion of peroxynitrite formation was independent of oxygen concentration in the 0.3 to 1.3 mM range, which indicates that ³NO⁻ was quantitatively scavenged by O₂. The spectrum at the end of peroxynitrite accumulation can be described by the equation

13

where ΔC is the concentration of HN₂O decomposed by the laser flash and y = ([ONOO⁻] + [ONOOH])/ΔC is the stoichiometric yield of peroxynitrite. The effective peroxynitrite spectrum, ɛ_λ(ONOO⁻/ONOOH), was constructed from the spectra of ONOO⁻ and ONOOH (pKa = 6.6) according to their contribution at pH 7. The peroxynitrite yield y ≈0.25 was estimated by fitting this equation to the data in Fig. 3. Thus, the yield of ³NO⁻ is also about 0.25 and the yield of ¹HNO should be close to 0.75. The estimated quantum yield of peroxynitrite (≈0.2) is nearly the same as its stoichiometric yield.

Formation of N₃O by Reactions of HNO/NO⁻ with NO.

Neutral solutions.

In neutral solutions saturated with NO, the prompt bleaching of HN₂O below 280 nm was followed by the growth of an absorption band at 380 nm (Fig. 4), which subsequently decayed with a 2.5-ms half-life (see Fig. 8, which is published as supporting information on the PNAS web site). Both the absorption spectrum and the lifetime were characteristic of the N₃O anion reported previously (17, 18, 34); reaction 3 was suggested for the N₃O formation and decay in these reports.

Figure 4.

Transient absorption spectra recorded at the indicated delay times after flash photolysis of 0.3 mM HN₂O in NO-saturated solution at pH 7 (0.1 M phosphate). (Inset) Kinetic trace recorded at 380 nm and showing formation of N₃O; note the time axis scale change.

The kinetics of N₃O formation consisted of two exponential components well separated in time (Fig. 4 Inset). The pseudofirst-order rate constants for both components (k_fast and k_slow) increased linearly with NO concentration. We have attributed the fast process to the production of N₃O by the rapid sequential additions of two NO radicals to the ³NO⁻ anion generated in reactions 10 and 12, i.e., the reactions

14

15

The effective bimolecular rate constant of (2.3 ± 0.2) × 10⁹ M⁻¹⋅s⁻¹, which approximately corresponds to the slower of these reactions, was determined from the slope of the k_fast vs. [NO] plot (see Fig. 9, which is published as supporting information on the PNAS web site). The relative contributions of the fast and slow pathways to the N₃O formation measured from the amplitudes of these components (Fig. 4 Inset) were about 25% and 75%, respectively. These values are in agreement with the yields of ³NO⁻ and ¹HNO estimated above for the reaction sequence 10–12 from the data on peroxynitrite formation. We have, therefore, concluded that the rate-determining reaction in the slow process of the N₃O formation is the addition of NO to ¹HNO

16

which is followed by rapid reaction 15. The rate constant k₁₆ = (5.8 ± 0.4) × 10⁶ M⁻¹⋅s⁻¹ was obtained from the k_slow vs [NO] dependence (see Fig. 9). At a constant concentration of NO, k_slow was independent of pH in the 4 to 10 range (Fig. 5 Inset).

Figure 5.

Dependence of the N₃O formation rate constant (k_app) on the solution pH (adjusted with NaOH at pH ≥ 12 and 0.1 M borate at pH 10). (Inset) pH independence of the slow step rate constant (k_slow) for N₃O formation in buffered solutions. Both solid lines correspond to Eq. 17 with a = 1.1 × 10⁴ s⁻¹ and b = 5.3 × 10⁴ M⁻¹⋅s⁻¹. All data are for NO-saturated solutions.

Alkaline solutions.

In alkaline solutions, the fast process of N₃O formation was not observed and only the slow N₃O accumulation occurred as a single exponential step. Both the yield of N₃O and the rate of its subsequent decay were pH-independent above pH 10 and remained the same as at pH 7. In contrast, the apparent pseudo-first-order rate constant for N₃O formation (k_app) steeply increased with alkalinity (Fig. 5) in a manner similar to that shown in Fig. 2 for ONOO⁻ formation. The pH dependence of k_app conformed to the expression

17

with a = (1.1 ± 0.1) × 10⁴ s⁻¹ and b = (5.3 ± 0.5) × 10⁴ M⁻¹⋅s⁻¹. The second term describes the effect of alkalinity and can be attributed to the generation of ³NO⁻ by deprotonation of ¹HNO (reaction 7). Indeed, within the experimental accuracy, the b value is identical to the k₇ value determined above from the peroxynitrite formation experiments. The formation of ³NO⁻ is followed by very rapid reactions 14 and 15, leading to N₃O. Likewise, the term a corresponds to reaction 16, i.e., k₁₆[NO] = a in the NO-saturated solutions. Under these conditions, side reaction 9 is too slow to play a role. Thus, the flash photolysis data for both O₂- and NO-containing solutions could be accounted for on the basis of the common reactions 5–7 and 10–12.

Mechanistic Considerations.

The mechanistic information obtained in this work is summarized in Scheme S1, where reactions are numbered as in the text. Their common feature is that at least one of the reaction partners is an uncharged species; consequently, the effects of medium ionic strength on the rate constants reported here are expected to be small. The interconversion between ³NO⁻ and ¹HNO (reaction 7) is of special interest because it gives a unique example of a spin-forbidden (nonadiabatic) proton transfer between a pair of small molecules. The equilibrium constant K₇ = 400 M⁻¹ can be calculated from pKa(¹HNO/³NO⁻) = 11.4. Using the experimental value for k₇ = 4.9 × 10⁴ M⁻¹⋅s⁻¹, we estimate the rate constant for the ³NO⁻ decay through the spin-forbidden protonation by water

-7

to be k₋₇ = 1.2 × 10² s⁻¹, i.e., the ³NO⁻ lifetime in water is several milliseconds. This long lifetime allows the ³NO⁻ anion to participate in reactions with O₂ and NO. Without the spin prohibition, anions with similar basicity equilibrate with water much more rapidly. Recently, the spin-forbidden protonation of ³NO⁻ has been directly observed in the gas phase, albeit for much stronger acids than water (35). The independent estimates for the upper limit of pKa(¹HNO/³NO⁻) < 14.3 and for the lower limit of E⁰(NO/³NO⁻) > −1 V, which are in agreement with the values in Fig. 1, can be made by using our kinetic data (see Note 2, which is published as supporting information on the PNAS web site).

Scheme 1.

Another potentially important spin-forbidden reaction could lead to peroxynitrite

18

Although this reaction should be exoergonic by 100 kJ/mol, we found no evidence for its occurrence, which allowed an estimate of the upper limit for k₁₈. From the relative amplitudes in Fig. 2 it follows that k₁₈[O₂] ≪ k₉[¹HNO]₀ and, therefore, k₁₈ ≪ 3 × 10⁵ M⁻¹ ⋅s⁻¹. A more plausible mode of oxygen reactivity is hydrogen atom transfer

19

which is not only energetically favorable by 25 kJ/mol, but is also spin-allowed. Both reactions 9 and 19 have been studied in the gas phase and the ratio (k₉/k₁₉)_gas ≈2 × 10⁵ at 298 K can be calculated from the published data (36). If the ratio were the same in water, reaction 9 would always dominate under our experimental conditions, making reaction 19 undetectable. For more on water vs. gas rate comparison see Note 3, which is published as supporting information on the PNAS web site.

The mechanistic interpretation of the experimental data based on Fig. 1 and shown in Scheme S1 provides a novel view on the interplay between the spin states and reactivities in the HNO/NO⁻ system. At the same time, it challenges the interpretation of the early pulse radiolysis work (17, 18). The disagreements pertain to pKa of HNO, to the existence of rapid equilibrium N₂O ⇋ NO + NO⁻, and to the rate of N₃O formation from NO⁻ in NO-saturated solutions. It appears that reexamination of the related experiments in the literature is required to resolve these matters. Similarly, the pKa(¹HNO/³NO⁻) value of 7.2 derived from the quantum mechanical calculations (22) is incompatible with our data, suggesting that further computational refinements may be necessary. Finally, a value that is almost 1,000 times greater than determined in present work was suggested for k₉ (28). However, this value is based largely on the early pulse radiolysis results that we are now questioning.

Biological Implications.

If nitrogen(+1) is indeed produced in biological systems, it must be present in the form of ¹HNO at physiological pH. As a small neutral molecule, ¹HNO should be able to freely traverse biological membranes, a property that can make a physiological role of ¹HNO very significant. We estimate the reduction potentials E⁰(NO, H⁺/¹HNO) ≈ −0.14 V and E⁰(¹HNO, 2H⁺/NH₂OH) ≈ 0.7 V; the corresponding values for pH 7 are −0.55 V and 0.3 V, all vs. NHE. With these potentials, ¹HNO can engage in redox reactions with a number of diverse biological reductants and oxidants. However, in NO-rich environments, consecutive addition of NO to form N₃O (reactions 16 and 15) can be envisioned as the major sink for ¹HNO. In this respect the spontaneous decomposition of Angeli's salt provides a convenient method to generate ¹HNO in the absence of NO. At the same time, our results underscore the importance of using freshly prepared alkaline stocks of Angeli's salt to avoid contamination with peroxynitrite through reactions 7 and 8 upon aging of the stock.

Can peroxynitrite be produced through intermediacy of ³NO⁻ and ¹HNO in oxygenated tissues? Although our results do not rule out this possibility, they impose limitations on the reaction conditions that may lead to peroxynitrite. We have shown that addition of O₂ to ³NO⁻ is essentially diffusion controlled and gives peroxynitrite. However, generation of ³NO⁻ requires a very strong reductant, which is unlikely to survive other reactions in biological milieus, unless embedded in a hydrophobic environment with limited access to dissolved species. Such environments may include hydrophobic pockets in proteins and biological membranes. It is much more probable that biological systems can generate ¹HNO rather than ³NO⁻ because it requires only a moderate reductant. Direct addition of O₂ to ¹HNO that could yield peroxynitrite is spin-forbidden and, as we have shown, cannot be rapid, if it occurs at all. It appears that the possibility of peroxynitrite formation depends greatly on the magnitude of the rate constant for this reaction, which remains unknown. In this regard, coordination to paramagnetic metal ions (including those in proteins) could conceivably facilitate formally spin-forbidden peroxynitrite formation from ¹HNO and O₂.