The reaction of arsenite-complexed xanthine oxidase with oxygen. Evidence for an oxygen-reactive molybdenum center.

The effects of arsenite on the reaction of reduced xanthine oxidase with oxygen are determined. The kinetics of the reaction monitoring the return of enzyme absorbance are investigated as are the kinetics and stoichiometries of peroxide and superoxide formation. Although some of the effects of arsenite are qualitatively consistent with expectations based on the known perturbation of the molybdenum midpoint potentials by arsenite, several results cannot be so easily explained. Specifically, arsenite introduces a very rapid phase (kobs = 110 s-1 at 125 microM oxygen) to the oxidative half-reaction which is not observed with the native enzyme. Arsenite also diminishes the amount of superoxide produced and eliminates one-electron reduced enzyme as a detectable kinetic intermediate in the reoxidation pathway. These differences appear to result from the ability of arsenite to greatly enhance the oxygen- and/or superoxide-reactivity of the reduced molybdenum center. This is reflected in the observation that reduced forms of arsenite-complexed xanthine oxidase lacking functional FAD (iodoacetamide-alkylated enzyme and deflavo enzyme) react relatively rapidly with oxygen whereas these reactions are quite slow in the absence of arsenite.

* This work was supported by National Science Foundation Grant PCM-8208240 (to V.M. and R.H.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Supported by a National Science Foundation predoctoral fellowship. Present address: Institute of Molecular Biology, University of Oregon, Eugene, OR 97403.
Supported by a Rackham Faculty Grant from the University of Michigan.
The abbreviations used are: BICINE, NJV-bis(2-hydroxy-ethy1)glycine; MES, 2-(N-morpho1ino)ethanesulfonic acid; AFR, activity to flavin ratio (enzyme activity defined as the absorbance change/min at 295 nm, monitoring conversion of xanthine to uric acid, divided by the enzyme absorbance at 450 nm in the standard assay solution); DCPIP, 2,6-dichlorophenolindophenoI; CCP-I, the Compound I complex of cytochrome c peroxidase with peroxide; TCA, trichloroacetic acid. paper we investigate the effect of these perturbations on the reaction of reduced enzyme with oxygen.
Xanthine oxidase is a complex metalloflavoprotein having three oxidation-reduction centers in addition to the molybdenum mentioned above (FAD, and two distinct ferredoxintype Fe/S centers), and a total of six electrons are required for complete reduction of the enzyme (see Refs. 11 and 12 for reviews). When the reduced native enzyme reacts with oxygen, reducing equivalents are passed to oxygen only via the flavin site (13), producing hydrogen peroxide and superoxide (14)(15)(16). The kinetics and stoichiometry of product formation during the oxidative half-reaction have been analyzed by using cytochrome c peroxidase to trap HzOz and cytochrome c reduction to follow 0; production (15, 16). A scheme (Equation l, below) summarizing much of the available data was first proposed by Olson et ul. (17) and has subsequently been independently confirmed by Hille and Massey (15) and Porras and Palmer (16). XO For each of the partially reduced enzyme intermediates in the scheme, the numbers in parentheses indicate the number of reducing equivalents present. The reaction of reduced xanthine oxidase with oxygen can be interpreted qualitatively and quantitatively in terms of the rapid equilibrium model first proposed by Olson et ul. (17,18). The major assumption made in applying this model to the oxidative half-reaction is that the rate of intramolecular electron transfer among the four redox centers is rapid relative to the rate of electron transfer from reduced enzyme to oxygen. Although recent flash photolysis work has questioned the validity of this assumption (19), the conclusions from conventional kinetic studies of xanthine oxidase are consistent with a rapid equilibration of reducing equivalents within the enzyme (see, e.g. Ref. 20). With this assumption, the optical extinction coefficient and level of FAD reduction in each intermediate may be calculated from the known potentials of these redox centers (18,21). Each intermediate of Equation 1 will therefore react with oxygen at a rate equal to an intrinsic rate constant, k,, for the reaction of oxygen with FADHz multiplied by the fraction of completely reduced flavin in that intermediate (FFADH2). In addition, the reactivity of the flavin semiquinone toward oxygen must be taken into account with an intrinsic rate constant k ' i multiplied by the fraction of flavin semiquinone in that intermediate (fFmH'). Such calculations have been used as the basis for successful simulations of the reoxidation time course of completely reduced enzyme, twoelectron reduced enzyme, and reduced alloxanthine-complexed enzyme (15).
In the presence of arsenite the distributions of reducing equivalents in partially reduced enzyme molecules are altered in a predictable manner due to the increased Mo(VI)/Mo(V) oxidation-reduction potential (10). In light of the anticipated effects of this perturbation of the oxidation-reduction equilibrium within the enzyme on the oxidative half-reaction, we have undertaken an investigation of the reaction of arsenitecomplexed enzyme with oxygen. Our results indicate that the major effect of arsenite is not the perturbation of this equilibrium but rather the introduction of a new site in the enzyme capable of reacting with oxygen.

MATERIALS AND METHODS~
Details of the experimental procedures used in these studies are given in the Miniprint Supplement (Refs. 22-33).

The Reaction of Arsenite-complexed Xanthine Oxidase with
Oxygen-Previous studies (15-17) of the oxidative half-reaction of xanthine oxidase have established that Equation 1 accurately describes the kinetics and product stoichiometries observed with the native enzyme. The extinction coefficient changes and fractional levels of flavin reduction thus calculated for the kinetic intermediates using the original lowtemperature potentials for the several centers in the enzyme (17) as well as the more recently determined room temperature potentials (10, 34, 35) are given in Table I. The major difference between the two sets of numbers lies in the levels of FADH, and FADH' for XO(2) and XO(1). In the present study, the latter set of potentials has been used throughout in the calculations. Significantly lower levels of FADH2 and higher levels of FADH' are predicted for these two intermediates when the room temperature potentials are used.3 Also shown in Table I are  To adequately fit the observed time course of re-oxidation, one has to assume an intrinsic rate of reaction between oxygen and enzyme-bound flavin semiquinone which is considerably greater (50 s-' compared to 16 s-' at 125 PM oxygen) than has been proposed previously (15, 17). Assuming an intrinsic rate (k;) of 32 s" for the reaction of FADH2 with oxygen and an intrinsic rate (k'J of 50 s-' for the reaction of FADH' with oxygen, the absorbance change observed upon re-oxidation of native enzyme can be simulated. With these values of ki and k',, the relative levels of f F m~~ and fFADH' in XO(2) indicate that most of the XO(2) -+ XO(1) conversion involves 0 2 reacting with enzyme-bound FADH', not with FADH,. This interpretation differs from previous proposals (15, 17) that 0, reacts with XO(2) primarily via FADH,. These earlier interpretations were made before the room temperature potentials became available. The interpretation presented here may provide an explanation for the observation that the reaction of O2 with XO(6) and XO(4) produces Hz02 (0, reacting with FADHZ), whereas the reaction of O2 with XO(2) produces 0, (0, reacting with FADH'). and FADH' predicted for the intermediates of the re-oxidation of arsenite-complexed enzyme using the room temperature potentials for the arsenite-complexed enzyme. Compared with the native enzyme, arsenite-complexed enzyme has considerably lower levels of FADHz and FADH' in the XO(2) and XO( 1) intermediates. If the reaction of reduced arsenitecomplexed enzyme with oxygen occurs in an analogous fashion to the re-oxidation of native enzyme (Equation l), then three phases are predicted for the re-oxidation of completely reduced arsenite-complexed enzyme (as depicted in Equation 2). The first phase would be composed of two discrete steps of approximately the same rate, generating XO(2) from XO(6). Two-electron reduced enzyme in complex with arsenite would then be converted to one-electron reduced enzyme at a rate considerably less than that of the first two steps, and would constitute the second phase. The reaction of oxygen with oneelectron reduced enzyme in complex with arsenite would then proceed in the third phase at a still slower rate because of the extremely low level of FADH' .
As shown in Fig. 1, when arsenite-complexed xanthine oxidase is completely reduced by dithionite (at approximately 3 mol of dithionite/mol of FAD) and then mixed with oxygenated buffer, a triphasic reaction is in fact observed. This can be seen more clearly in the semilog plots accompanying the observed absorbance changes at 370 nm (A-C), 450 nm (D-F), and 550 nm (G-I). The rates of the three phases comprising the re-oxidation of arsenite-complexed enzyme are notably more rapid than expected on the basis of the above discussion, however, and are in fact greater than the corresponding rates associated with oxidation of the native enzyme in the absence of arsenite. For example, at 62 FM oxygen and 550 nm rates of 110 s" (32% of AA,,,), 8 s" (50% of A A , , 1 ) , and 1.5 s-l (18% of A A m , l ) are observed with arsenite-complexed enzyme, compared with rates of 10 s" (63% of AA,,I) and 0.6 s-' (37% of aA,,I) for the native enzyme under the same conditions. At all wavelengths (including 370 nm where the re-oxidation of the molybdenumarsenite complex contributes appreciably to the observed spectral change; Ref. lo), all of the spectral change associated with re-oxidation was complete within 8 s, indicating complete re-oxidation of all centers, including the Mo-arsenite, during this interval. The rates of each of the three phases observed with arsenite-complexed enzyme appear to exhibit hyperbolic dependences on oxygen concentration, as shown in the double reciprocal plots of Fig. 2. The observed rate constants and the relative spectral contributions of the three phases depend on the wavelength of observation. At 370 and 550 nm, three distinct phases are observable at all oxygen concentrations employed, and the relative contributions of these phases are essentially constant over this range of oxygen concentrations (Fig. 2). At 450 nm, however, the fastest phase is observable only at the lower oxygen concentrations (31 and 62 p~) .
At oxygen concentrations greater than 125 pM, the fastest phase cannot be readily resolved from the second phase at 450 nm (see top panel of Fig. 2 A ) . As discussed below, this behavior   (first) phase can be obtained from the semilog plot (-) in panel C. An analogous procedure was used to analyze appears to arise from the different oxygen dependences of the first two phases and from the relatively small per cent of the total spectral change associated with the first phase at 450 nm. At 31 and 62 p~ oxygen (where the rate of the first phase can be determined at 450 nm), the rate of the fastest phase at 450 nm appears to be considerably slower than the corresponding rates at 370 and 550 nm (Fig. 2 A ) . Such observations are not unusual in investigations of the oxidative half-reaction of xanthine oxidase (15) because the four redox centers make different relative contributions to the observed spectral change at different wavelengths, and the rates of oxidation of these centers are somewhat different even though they cannot be kinetically resolved as discrete phases (see Miniprint Supplement).

At460 f F A D H 2 f F A D H -
The hyperbolic dependences of the three phases suggest that each phase involves formation of a reduced enzymeoxygen complex that proceeds to give a less reduced enzyme molecule and reduced forms of oxygen ( 0 2 or H z W . The observed rate of each phase should therefore be given by the following equation: The fastest (first) phase has a limiting rate of approximately 150 s" and a Kd of about 50 p~. This phase accounts for 30-40% of the total spectral change at 370 nm and for 35-45% of the total spectral change at 550 nm. At 450 nm, 20% of the total spectral change occurs in the first phase at oxygen concentrations lower than 62 p~. At higher oxygen concentrations, however, the first phase cannot be resolved from the second phase, and the entire time course of re-oxidation at 450 nm is adequately described as the sum of two exponentials. The intermediate (second) phase has a limiting rate of approximately 50 s" and a Kd of 250 pM. This phase accounts for 30-40% of the total absorbance change at 370 nm and for 40-50% of the total spectral change at 550 nm. At 450 nm the spectral contribution of this phase increases from 60% at 31 p~ oxygen to about 80% at 625 p M oxygen. This increase results from the inability to resolve the fastest two phases at 450 nm.4 The slowest (third) phase of the re-oxidation reaction in the presence of arsenite exhibits rates which are about 10-fold slower than those for the second phase. A Kd of about 300 p~ is indicated by the double reciprocal plot of Fig. 2C. From the data at 450 and 550 nm a limiting rate of 14 s" is obtained, while the data at 370 nm indicate a limiting rate of 7 s-l (see Miniprint Supplement for a more complete discussion of the wa.velength dependence of these kinetic phases).
' It is not surprising that the first and second phases of the reaction cannot be resolved at 450 nm when one considers that: 1) the expected rate constants for these phases differ only by a factor of three or four, 2) the second phase contributes three to four times more to the absorbance change at 450 nm than does the first phase, and 3) the rate of the first phase is not particularly sensitive to oxygen concentration above 125 pM, owing to its low K d whereas the rate for the second phase remains quite sensitive to oxygen concentration in this range.
Effect of Superoxide Dismutase on the Re-oxidation Reaction-The time course for re-oxidation of native xanthine oxidase (no arsenite) is not noticeably altered by the presence of superoxide dismutase (17), indicating that the superoxide produced on oxidizing two-electron reduced enzyme is not readily utilized as an oxidant for one-electron reduced enzyme. With arsenite-complexed enzyme, however, the later stages of the oxidation reaction are markedly slower in the presence of superoxide dismutase, as shown in Fig. 3. Dismutase affects only the rate of the slowest phase: the relative contributions of the three phases to the overall absorbance changes are not altered significantly by the dismutase, nor are the rates of the first and second phases. It thus appears that arsenite-complexed enzyme can utilize superoxide as an electron acceptor.
Kinetics and Stoichiometry of Superoxide and Peroxide Production-We have monitored production of superoxide by using ferric cytochrome c as a trapping reagent, as described previously for the native enzyme (14)(15)(16). An experimentally determined operational extinction coefficient difference of Atssssss = 20 mM" cm" was used as described in Ref.

15.
The results of these experiments are summarized in Table 11.
As shown in Fig. 4, with arsenite-complexed enzyme approx-imately 0.4 O;/FAD were produced in a biphasic manner; considerably less than the 1.6 O;/FAD observed with the native enzyme under the same conditions (16): The reaction between cytochrome c peroxidase and H2O2 quantitatively produces a stable Compound I (CCP-I) (36,37). Formation of this complex results in a characteristic spectral change and is rapid ( k = 3-4 X lo7 M-' s-' , Ref. 33), and this reaction has been used successfully to monitor Hz02 production during the re-oxidation of completely-reduced, native xanthine oxidase (16). Using similar conditions, we have compared peroxide production in the oxidative halfreaction of native and arsenite-complexed xanthine oxidase (Fig. 4). The results of these experiments are also summarized in Table 11. Our experimental results with the native enzyme ( Fig. 4) are in qualitative agreement with previous results (16): A total of approximately 3 H2O2/FAD is detected 2.4 H202/FAD at a rate of 15 s" and 0.6 H202/FAD at 0.9 s-'. With arsenite-complexed enzyme (Fig. 4), approximately 1.8 H2O2/FAD are produced at a rate of 24 s-', 0.8 H202/FAD at a rate of 2 s-', and 0.2 H2OZ/FAD at a rate of 0.2 s-'. This A small fraction (5-10%) of the enzyme used in these experiments was inactive desulfo enzyme. As discussed in the Miniprint Supplement, the oxidative half-reaction of desulfo xanthine oxidase is not affected by arsenite and therefore produces two O;/FAD. The stoichiometry quoted is thus a slight overestimate for the superoxide yield obtained with arsenite-complexed enzyme, and the true stoichiometry may be estimated to be approximately 0.2 O;/FAD. This indicates that less than 10% of the reducing equivalents initially present in the arsenite-complexed enzyme produce superoxide. It is therefore not surprising that the process giving rise to superoxide is difficult to observe in following the re-oxidation of enzyme. It should be noted that the superoxide dismutase and cytochrome c which scavenge the superoxide in these experiments also slow the third and slowest phase of the enzyme re-oxidation, as demonstrated in the previous section. The slow process (0.3 s-') producing 0; may well be the result of superoxide itself reacting with enzyme in a prior step and may thus not be a part of the re-oxidation scheme for arsenitecomplexed enzyme in the presence of dismutase or cytochrome c.
While the kinetics and stoichiometry of peroxide production which we observe for the native enzyme in the absence of arsenite agree with the results expected for the accepted pathway for the reoxidation of xanthine oxidase (Equation l), they differ somewhat from previously reported results (16). In this earlier work no slow phase for peroxide production was observed, for example. It was suggested that a slow reaction of CCP-I by partially reduced enzyme was responsible for eliminating the slow phase of Hz02 production, and such a process was demonstrated to take place with a t% of approximately 10 s (16). In our hands, a roughly similar tlh of 30-40 s is observed.
There are several differences in our experimental conditions compared with those of the earlier work. First, we routinely used approximately half the xanthine oxidase and peroxidase concentrations, which would have the effect of slowing the second order reaction between CCP-I and partially re-oxidized xanthine oxidase. Second, most of our peroxide-monitoring experiments utilized the large spectral change at 563 nm associated with the formation of CCP-I. While the spectral change is larger at 424 nm, where the reaction was observed in the earlier work, the background absorbance was so great (at least in the instrument used in the present work) that poor signal to noise was observed. It is thus possible that the slow reaction was made more difficult to observe in the previous work for this reason as well. Finally, the xanthine oxidase used in the present studies was considerably more active than that in the previous study, having only trace amounts of the inactive desulfo form of the enzyme. Any complications arising from the inability of the desulfo molybdenum center to participate in the rapid equilibrium during the course of the oxidative half-reaction (17,38) are thus avoided in the present work. Also, as the desulfo molybdenum center has an appreciably lower midpoint potential than the functional center, it is possible that this results in some indirect effect on the rate of peroxide production in the last stages of the enzyme re-oxidation when inactive enzyme is present. Hz02/FAD are produced. A semilog plot of the data of Panel C is shown in Panel D (-) and indicates three phases. The contribution of the third (slowest) phase is quite small (8% of AAmd, i.e. 0.2 HzOz/FAD). When this small contribution is subtracted from the data, a biphasic semilog plot (-.-) of the remaining spectral change is obtained. This plot indicates that 0.8 HZOJFAD are produced at a rate of 2 s-'. When the contribution of this phase is subtracted, the semilog plot of the remaining absorbance change (---, Panel D ) indicates that the remaining 1.8 Hz02 are produced in a single phase (k,h = 24 s-'). This phase exhibits a gradually accelerating time course (or a noticeable lag) which may be due, in part, to the fact that two sequential steps of approximately the same rate comprise this phase (see Refs. 17 and 18).

H202/FAD.
There is no peroxide produced at a rate compachange. This observation is discussed further below.
rable to the rate of the fastest phase (120 s" at 125 pM Effect of Arsenite on Re-oxidation of Reduced Enzyme Conoxygen) observed when monitoring the enzyme absorbance taining Alkylated FAD-When completely reduced enzyme is

TABLE I1 Stoichiometry and kinetics of superoxide and hydrogen peroxide production during the reaction o f oxygen with various forms of reduced xunthine oxidase
Reactions were carried out as described in the text with 125 p~ oxygen in 0.1 M BICINE at pH 8.3 and 25 "C.

0;
HzO, treated with iodoacetamide, the FAD is alkylated at the C(4a) centers are affected by this treatment. Thus, although alposition, rendering the reduced flavin redox-inactive and inkylated enzyme has less than 1% of the original xanthine:02 capable of reducing oxygen (27): none of the other redox reductase activity, it retains its original xanthine:2,6-dichlo-rophenolindophenol reductase activity (39). When alkylated enzyme is reduced with approximately 2 eq of dithionitel active site and then is mixed with oxygenated buffer, the reoxidation reaction proceeds very slowly and with multiple phases (Fig. 5, A and B ) . Neither the rates themselves nor their dependence on the oxygen concentration correspond to those observed for the oxidative half-reaction of native enzyme containing functional FAD. Because the rates observed for alkylated enzyme are so slow, the flavin-independent reoxidation pathway appears to play no role in oxidation of reduced native enzyme. However, when arsenite-complexed alkylated xanthine oxidase is completely reduced and then reacted with oxygen, most of the absorbance change occurs relatively rapidly (Fig. 5, C-E). For example, in the presence of 2 mM arsenite and 125 pM oxygen, over 50% of the total absorbance change at 470 nm occurs during the first 80 ms, and almost 80% of the total absorbance change at 470 nm is complete within 0.8 s of mixing. In contrast with these relatively rapid rates are the results for alkylated enzyme in the absence of arsenite: at the same oxygen concentration, less than 1% of the total absorbance change at 470 nm occurs during the first 80 ms, and almost 40 min are required for the reaction to go to completion. The dependences of the rates of the three, well-resolved phases on oxygen concentration for alkylated, arsenite-complexed enzyme are given in Fig. 6. The rates of the first two phases are sufficiently fast that the flavin-independent re-oxidation pathway could well play a major role in the re-oxidation reaction of arsenite-complexed enzyme, even when the enzyme contains functional FAD. For example, the double reciprocal plot of Fig. 6A indicates a limiting rate of 180 s" and a K d of 70 pM for the fastest phase of the reaction of reduced iodoacetamide-alkylated, arsenitecomplexed enzyme with oxygen. These values are comparable to those observed for the fastest phase of the oxidation of arsenite-complexed enzyme containing functional FAD (Fig.  2). The rate of the second phase for the alkylated arsenitecomplexed enzyme (7-8 s-') is also sufficiently rapid that it zite-complexed Xanthine Oxidase 8899 could play a role in the re-oxidation reaction even when functional FAD is present. The insensitivity of the rate of the second phase in the reaction of alkylated arsenite-complexed xanthine oxidase with oxygen to the oxygen concentration suggests that it is due to utilization of 0; generated in the very rapid 02dependent first phase rather than 0 2 as an electron acceptor. This possibility is supported by the ability of superoxide dismutase to alter the time course of the re-oxidation occurring during the second phase of the re-oxidation of alkylated arsenite-complexed enzyme (Fig. 7 A ) . In the absence of dismutase, the second phase contributes approximately one-third of the total observable spectral change at 470 nm. As shown in the semilog plots of Fig. 7 , B-D, in the presence of dismutase the amplitude of the second phase is cut almost by half, although its rate is approximately the same. An additional slow phase (kobs = 0.3 s-') which is not present in the absence of dismutase accounts for the diminished amplitude of the second phase. Neither the rates nor the amplitudes of the first phase and the final phase are altered by dismutase.

Reaction of Reduced Arsenite-complexed Deflavo Xanthine
Oxidase with Oxygen-We have also examined the effect of arsenite on the re-oxidation kinetics of another form of xanthine oxidase lacking functional FAD, the deflavo enzyme. In the absence of arsenite, the re-oxidation of dithionite-reduced deflavo enzyme is extremely slow, exhibiting a rate constant of about 0.02 s" at 125 p~ oxygen in pyrophosphate buffer at pH 8.5 and 25 "C (13). As shown in Fig. 8 6) and with the values for arsenite-complexed enzyme containing functional FAD (Fig. 2 A ) . As was observed with the alkylated enzyme in complex with arsenite, the deflavo arsenite-complexed enzyme must be completely reduced for the enhancement of oxygen reactivity to be observed. Further results are given in the Miniprint Supplement.

DISCUSSION
From a comparison of the rates of re-oxidation for alkylated or deflavo xanthine oxidase in the presence and absence of arsenite ( Figs. 5 and 8, respectively), it is clear that the inorganic reagent enhances the oxygen reactivity of the enzyme. Arsenite does not appear to alter the properties of either Fe/S center (4-6), but there is abundant evidence that arsenite interacts directly with the molybdenum center of xanthine oxidase (2-10). Therefore, it is not unreasonable to propose that arsenite changes the oxidative half-reaction by making the molybdenum center capable of reducing oxygen and superoxide. Although there have not been many reports of reactions between reduced forms of molybdenum and oxygen, Wilshire et al. (40) have reported a mononuclear Mo(V)catechol compound capable of reducing O2 to 0;. In addition, the molybdenum center of sulfite oxidase appears to be capable of reducing oxygen (41) in a steady-state system, although the oxidative half-reaction with O2 has not yet been well-characterized. The possibility that arsenite affects the oxidative half-reaction by altering the oxygen-reactivity of Fe/S I cannot be ruled out completely on the basis of work reported in this paper, although we regard this as unlikely in view of the ability of xanthine, salicylate, or nitrate (reagents all known to interact with xanthine oxidase at the molybdenum center) to block the rapid reaction of oxygen with the Several characteristics of the flavin-independent re-oxidation pathway suggest that this process occurs even when a functional FAD is present in arsenite-complexed xanthine oxidase. The rates of the first two phases observed in the reoxidation of alkylated arsenite-complexed enzyme are certainly sufficiently rapid to compete with re-oxidation via the FAD. Whether the arsenite-complexed enzyme contains normal or alkylated FAD, the rate and hyperbolic oxygen dependence of the first phase is about the same, suggesting that at least this very rapid flavin-independent reaction results from the reaction of O2 with the Mo(1V)-arsenite center. The sensitivity to superoxide dismutase of the third phase observed in the re-oxidation of arsenite-complexed enzyme (but not native enzyme) also suggests direct participation of the molybdenum center in the slow phase of the oxidation halfreaction. The rate of this phase exhibits a roughly hyperbolic dependence on oxygen concentration (Fig. 3 0 . While this could be interpreted as indicating reversible formation of a complex between XO(2)-arsenite and O2 prior to electron transfer, it is also possible that oxygen reacts with the FADH' of XO(2)-arsenite in a second order reaction that generates superoxide which is then utilized as an electron acceptor by the XO(1)-arsenite (possibly at the Mo(V)-arsenite site) at a more rapid or comparable rate which would be independent of oxygen concentration. If the reaction between superoxide and XO(1)-arsenite is sufficiently rapid, re-oxidation of XO(S)-arsenite would appear as a single phase. This sequence of reactions can also account for the observed saturation of k3 as oxygen concentration is increased. Using the scheme shown in Equation 3, we have been able to simulate the observed k3 rates at oxygen concentrations from 62 to 625 phi.
H z 0 2 I These simulations used the rapid equilibrium predictions for the rates of oxygen reacting with X0(2)-arsenite via the FADH' and gave the best fit to both time courses when a second order rate constant of approximately lo3 M" s" was assigned to the reaction of superoxide with XO(1)-arsenite.
At 125 p~ oxygen the first phase of the re-oxidation of arsenite-complexed xanthine oxidase proceeds at a rate of approximately 110 s-', and the above discussion suggests that the reduced Mo-arsenite site participates in this reaction. There is, however, no superoxide or peroxide released from the enzyme at a rate corresponding to this phase. This process could involve a single electron transfer step generating bound superoxide that is subsequently reduced to Hz02 in a slower reaction. It will be remembered, however, that very little spectral change is predicted for the X0(6)-arsenite to XO(5)arsenite conversion because this primarily represents oxidation of the lowest potential site, Mo(1V)-arsenite (10). To explain the observed spectral change one would have to propose that bound superoxide raises the Mo(V)-arsenite/ Mo(1V)-arsenite potential sufficiently to make electron transfer from the site of next lowest potential (Fe/S I) thermodynamically favorable. Alternatively, the first phase could represent conversion of X0(6)-arsenite to X0(4)-arsenite producing H202 which is then released into solution at a slower rate (comparable to that for the second phase of the reaction). Results with the native enzyme (no arsenite) presented in this paper and in previous publications (15, 16) have demonstrated that the rates of 0; and HzOz release from the flavin center correspond well with the rates determined by monitoring enzyme absorbance changes. However, the first phase of the oxidative half-reaction with arsenite-complexed xanthine oxidase appears to involve reduction of oxygen at the Moarsenite site, so there is no requirement that the rate constant for product release correlate with the rate constant describing enzyme absorbance change. The spectral change associated with the first phase in the re-oxidation of reduced arsenitecomplexed enzyme (see Miniprint Supplement) is in fact quite close to that expected for the two-electron alternative based on rapid equilibrium considerations (see Miniprint Supplement).
The rate constant for the rapid phase of peroxide generation (24 s-'), on the other hand, matches well the rate constant of the second phase of enzyme-arsenite re-oxidation (20 s-'). Approximately 2 H202/FAD are produced in this phase, and the slightly accelerating time course shown in the semilog plot (Fig. 4, inset) is to be expected for two sequential reactions taking place at similar rates (approximately 30 s" on the basis of computer simulations), each generating a single peroxide molecule/flavin. A qualitative model describing the re-oxidation of arsenitecomplexed xanthine oxidase is presented in Equations 4 through 6. represents the first phase of this reaction; it exhibits a hyperbolic oxygen dependence and does not result in product release. This phQe appears to involve O2 reacting with the Mo(1V)-arsenite center. Equation 5 summarizes the reactions of the second phase which generate XO(2)-arsenite and produce 2 H202/FAD. Re-oxidation may occur via both the FAD and the Mo-arsenite site in this phase. The two steps indicated for this phase are not necessarily sequential as indicated, since both the FADH2 and the flavin-independent re-oxidation site are presumably capable of functioning simultaneously. Equation 6 represents the third phase of the oxidative half-reaction, re-oxidation of two-electron reduced enzyme. In this phase, oxygen reacts with XO(2)-arsenite (probably via the FADH-) to generate superoxide. This superoxide is subsequently utilized relatively rapidly by XO(1)arsenite to generate completely oxidized arsenite-complexed enzyme. Some of the superoxide generated in the third phase is accessible to scavengers such as superoxide dismutase and cytochrome c. Such a scheme would adequately account for the very low yield of superoxide anion in the reaction of arsenite-complexed enzyme with oxygen.