Oxidizing intermediates generated in the Fenton reagent: kinetic arguments against the intermediacy of the hydroxyl radical.

It has long been recognized that the aqueous mixture of hydrogen peroxide and ferrous ion, known as the Fenton reagent, generates powerful oxidants. Furthermore, the chemical intermediates and reaction pathways of the type generated by this reagent have been implicated in oxidative damage in biological systems. Although the subject continues to be debated, the hydroxyl radical (.OH) is generally invoked as the predominant oxidizing intermediate formed by the Fenton reagent. However, recent results from this laboratory have demonstrated that the principal pathway for the Fenton-mediated oxidation of N-nitrosodimethylamine does not involve .OH, but instead must involve the intermediacy of another strongly oxidizing species. This conclusion was based on stopped-flow spectrophotometric observation of a transient, A, believed to be an iron(II) nitrosyl adduct, which was formed at a rate five-fold faster than that predicted for formation of .OH. Subsequent experiments have shown that methanol and other organic compounds can quench the formation of A. This quenching procedure provides a unique spectrophotometric probe with which to examine the relative reactivities of putative Fenton-type oxidizing intermediates toward organic substrates. Presented here are the results of several such quenching studies, plus an overview of our mechanistic investigations of the Fenton reaction.


Introduction
The reaction between ferrous ion and hydrogen peroxide in acidified aqueous solution was shown by Fenton, in the late 1800s (1), to generate a powerful oxidizing intermediate. The chemistry of the "Fenton reagent" has since been a model upon which metal-mediated oxidation by peroxide in biological media has been based (2). For years, the nature of the specific reactive intermediate(s) involved has been debated. For example, in the 1930s, Haber and Weiss (3) argued that the hydroxyl radical (0OH) is the principal oxidizing species, while Bray and Gorin (4) proposed that a metallo-oxo species, such as solvated ferryl ion, was the predominant oxidant. Later, Merz and Waters (5,6) suggested that the involvement of OH was supported by the stoichiometry of the reactions taking place in the presence of Fenton reagent. In 1952, Cahill and Taube concluded, on the basis of experiments utilizing 01 8-labeled peroxide, that the oxidant present under acidic conditions was most likely a metallo-oxo species (7). The hydroxyl radical has taken the more predominant role, dating from a 1975 review by Walling (8) which surmised that the lack of effect of differing ionic strength on the rate of reaction of the oxidizing intermediate with Fe2+ and methanol at different perchlorate concentrations was evidence that -OH constituted the predominant oxidizing species generated under acidic conditions (Eq. 1). Fe2+ + H202 4 Fe3+ + -OH + OH- [1] Since the early 1980s, the involvement of oxygen radicals has been proposed in many degenerative conditions, including cancer, senility, autoimmune disorders, etc. (9)(10)(11)(12). The cytotoxic and cytostatic actions of cells of the immune system were also thought to result from similar chemical events (13,14). As a result of the increasingly recognized physiological importance of this chemistry, discussion of mechanistic aspects of the Fenton reaction has intensified. Iron complexes with chelating ligands such as EDTA have been examined under physiologically relevant conditions. Electron spin resonance (ESR) spin-trap studies revealed the existence of the 5,5dimethyl-1-pyrroline N-oxide (DMPO)-OH adduct, and this has been used as evidence of the generation (and, presumably, the intermediacy) of OH in oxidation reactions occurring in the Fenton reagent (15). Competitive kinetic studies have been performed to compare the reactivity of the oxidizing intermediates generated in the Fenton reaction with authentic OH generated by radiolysis of water or photolysis of H202. Rahhal and Richter (16) examined Fe"(EDTA) oxidation, and suggested that an oxidant other than 0OH was generated in this system. Koppenol and Rush (17), after studying a number of chelated iron complexes using stopped-flow spectrophotometry, concluded that a metallo-oxo species was generated in neutral solutions, while OH was predominant in acidic solutions of nonchelated iron. Sutton et al. (18)  faster than that for formation of ferric ion (k =76 M_ sec I Figure 1). That the (A) Temporal absorbance changes at k=625 appearance of an oxidizing intermediate :ored by stopped-flow spectrophotometry, in also occurred at a rate faster than formainitially containing 5 mM Fe2 20 mM H202, tion of Fe3 was shown using methylene s 0, 25, 50, or  (kA=400M sec). As previously discussed, the formation of OH from fertalization for this discrepancy, in rous ion also must involve concomitant is argued that under certain condi-formation of ferric ion (Equation 1); thus ie metallo-oxo species or -OH can transient A does not result from the oxirated in both systems (9)(10)(11)(12)(13)(14)(15)(16)(17)(18). A dation of NDMA by -OH (20). tudy, based on the assumption that To investigate further the possible role l-OH adducts are formed solely of -OH in the formation of A, the reac-)H, has suggested that there is tion of this radical with NDMA was Lan one type of oxidizing interme-examined using authentic -OH, generated -esent, and that the ratio between by pulse radiolysis (20). A reaction ount of -OH and metallo-oxo between OH and NDMA did occur, but depends on the chelated ligand did not lead to formation of A. Moreover, the second-order rate constant for attack his context, we have used the of OH on the nitrosamine was -flow kinetics of the oxidative den-3.3 ± 0.4 x 108M-seC considerably difrn of the carcinogen N-nitrosodi-ferent than required for the behavior mine (NDMA) as a probe of the shown in the stopped-flow kinetics study.
ism of the Fenton reagent under The NDMA concentration at which the onditions (20). These studies have extent of transient formation in the ,trated that the principal pathway Fenton reaction was half-maximal (i.e.,

Fenton-mediated oxidation of this half that observed at infinite [NDMA])
,e cannot involve -OH, but instead was calculated from the negative X-inter-Lvolve the intermediacy of another cept (-Ix) of Figure IA (inset) to be oxidizing species. In the present 92 ± 10 mM. At this concentration, half we describe various quenching of the active oxidant is being consumed of the oxidizing intermediates, dis-by reaction with NDMA and the remainpossible roles of two such inter-ing half by reaction with other compoes, and further demonstrate that nents of the medium, i.e., tually plays little or no role in the  4.3 x 10, 2.7 x 10, and 6.9 x 10 M sec-t (22), respectively, indicating that kni trosamine could be no greater (under the conditions described in Figure 1) than 2.7x107 M'Isect1if the primary Fenton oxidant was -OH. This value is at least an order of magnitude smaller than that observed in the pulse radiolysis of NDMA (3.3 x 108 M-1 sec-'), (20), thus the oxidizing intermediate under the present Fenton reagent conditions is shown by both its spectrophotometric and kinetics behavior to be different from -OH. We have also used kinetics simulation techniques to evaluate a more complex model invoking formation of OH, which included a feedback loop via Fe3+reduction of the hydroperoxide radical; however, this simulation was unable to duplicate the experimentally observed kinetics (Appendix 1). In this context, the simple model for the oxidation of NDMA indicated in Equations 3 to 7 was proposed to account for the appearance of transient A via the reaction of NDMA with an oxidizing intermediate X.
Fe + H202 kX k3 X + NDMA -*> A Abs [3] [4]  (20). This simplified model was successful for the qualitative simulation of the temporal absorbance behavior as illustrated by Figure   3, but leaves unanswered questions such as the nature of the putative intermediates X, Y, and transient A. Notably, the absorption spectrum of the transient displays maxima at 590 and 450 nm, features which are observed also in the spectrum of the iron(II) nitrosyl complex (H20) Fe(NO)2+ ( Figure 1B). Thus, it appears ?ikely that transient A is a closely related species. As we have shown previously, the formation of A displays a substantial deuterium isotope effect when the substrate is NDMA-d6, thus C-H bond cleavage would appear to be directly involved in the rate-limiting process.
To ascertain the relationship between formation of transient A and of the denitrosation products, a competitive kinetics study was performed in the presence of MeOH. When  -IX(MeOH)) was shown to be 17 mM.
However, since the bulk of nitrate formation would have occurred via reaction of NDMA with Y, this implies that the reaction of MeOH with Y is at least a factor of 17 less extensive than the reaction of NDMA with Y. However, the rate constant for the reaction of MeOH with OH has been determined to be 9.7x 10 M sec-, a factor of 3 larger than that for the reaction of NDMA with OH (see above). Thus, one can conclude that Y is not the hydroxyl radical.
The above observations, with NDMA as substrate, clearly show the presence of two oxidizing intermediates under the Fenton reagent conditions, neither of which shows the competition reactivity characterized previously for the hydroxyl radical in aqueous solution. The specific natures of X and Y are as yet undetermined. However, the fact that X is formed (apparently reversibly) in a 1/1 stoichiometry from Fe2+ and H202 leads one to speculate that this is simply a ferrous hydrogen peroxide complex, i.e., Fe(H20)5(H202)2+. Furthermore, since the formation of Y from X is apparently unimolecular, a ferryl species, e.g., FeO 2+, should certainly be considered for the former. The fact that the FeO2+ cation has recently been shown to be relatively unreactive toward alcohols (16) is at least consistent with the apparent reactivity of Y as shown by the competition studies.

Reactivity of X with Oxygen Radical Scavengers
Formation of the absorbing transient, A, solely from intermediate X provides a unique opportunity to examine the competitive kinetics of the reaction of X with additional substrates. Figure 5 illustrates this for a quenching experiment carried out with initial concentrations: 200 mM NDMA, 20 mM H202, and 0.5 mM Fe2+. When various amounts of t-butanol were added, the amount of transient A observed decreased, with simultaneous generation of a new product displaying an absorption maximum at 500 nm ( Figure 5). The absorbance due to this new product was dependent on the concentration of tbutanol. While the exact nature of that species remains uncharacterized, it most likely results from the formation of tbutanol radical and the subsequent reaction of this radical with the nitroso function of NDMA as previously reported (23), followed by decay to a stable product. In analogy to the quenching by methanol, the -IX(ROH) value (the con- values for different substrates allows the direct comparison of the rate constants for the reactions of NDMA and t-butanol and other substrates with X (Scheme 1). Table   1 summarizes the -IX(substrate) values for benzyl alcohol, DMPO, and dimethylsulfoxide (DMSO) and compares the relative reactivities to those previously reported for -OH. The relative reactivity of X toward benzyl alcohol is ten times greater than that predicted for -OH. Similarly, the selectivity for DMSO is five times different than the predicted selectivity of OH for DMSO. Little difference is apparent in the case of the substrates t-butanol, methanol, and DMPO. Many of the previous investigations of Fenton chemistry have based the conclusion as to the presence of -OH or a metallo-oxo species solely on the competitive kinetics studies involving one or two quenching agents. Based solely on relative reactivities toward the substrates described above, it is quite possible that OH could have been mistaken for X. Another interesting aspect is the high affinity of DMPO for X. Under conditions normally used in spin-trapping experiments (19), nearly 90% of the DMPO would react with the intermediate, suggesting that DMPO-OH adduct could be formed from this intermediate.

Conclusions
The denitrosation of NDMA, studied with a combination of stopped-flow spectrophotometry and product analysis, has provided a unique opportunity to elucidate the reactivity of novel oxidizing intermediates generated in the Fenton reaction. It now appears that -OH plays little or no role in the oxidation of substrates under acidic conditions. There are instead two reactive intermediates, X and Y; and we speculate that X is [Fe(H202)]2+ and Y is [FeO]2+. For certain substrates, intermediate X displays a reactivity pattern sufficiently similar to that of OH that this species may have been mistakenly identified in previous studies as -OH. However, the use of the reaction with NDMA allows relative reactivity comparisons with a broad range of substrates and such comparisons are clearly incompatible with the behavior predicted for intermediacy of OH.

Appendix
The arguments presented, based on the reaction rate constants for -OH, may not  [15] Environmental Health Perspectives provide a comprehensive analysis of this potentially complicated reaction. For instance, the formation of superoxide might result in a back reaction with ferric ion to form ferrous ion as shown by Bielski et al. (24). It could be argued that some back reaction could be involved to give the rapid formation of transient A. To provide a comprehensive determination of the involvement of OH in the formation of transient A, the increase in absorbance at 625 nm characteristic of the transient was simulated by numerical integration of the elementary reaction pathways described in Scheme 2, using a 4th-order Runge-Kutta method (25). As indicated in these equations, the reaction of Fe2+ with peroxide yields *OH and ferric ion. The *OH then could react with NDMA to form the alkyl radical, which then may undergo denitrosation to form the Schiff base (CH2 = NCH3) and NO (21). NO then could react with Fe2+ to produce transient A (postulated to be a Fe +-NO complex, Figure IB).
Transient A then could further react with peroxide to produce NO-. Hydroxyl radical could also react with ferrous ion to generate ferric ion, or react with peroxide to yield the perhydroxyl radical (HO2). HO2 then could react with ferric ion to form ferrous ion, or react with ferrous ion to yield ferric ion and peroxide. The rate constants for these steps have been reported previously in the literature. A simulation based on this model, where k,= 76 M1sec-1, did predict some accumulation of absorbance attributed to transient A (not shown). However, a plot of (Abs) l versus [NDMA]F1 (simulation) yielded (-Ix)' of 5.9 mM, nearly 16 times lower than that experimentally observed (92 mM). The amount of the ultimate nitrogenous product of NDMA denitrosation (nitrate) formed in this reaction was [NDMA] dependent. The predicted negative (X-intercept)-l from the model involving OH as oxidant was 3.4 mM, while the actual observed value was 20 mM. The simulation results confirm that the relative selectivity of OH is similar to that indicated by the simpler model described above.