Mechanisms of Protection of Catalase by NADPH

NADPH is known to be tightly bound to mammalian catalase and to offset the ability of the substrate of catalase (H2O2) to convert the enzyme to an inactive state (compound II). In the process, the bound NADPH becomes NADP+ and is replaced by another molecule of NADPH. This protection is believed to occur through electron tunneling between NADPH on the surface of the catalase and the heme group within the enzyme. The present study provided additional support for the concept of an intermediate state of catalase, through which NADPH serves to prevent the formation (rather than increase the removal) of compound II. In contrast, the superoxide radical seemed to bypass the intermediate state since NADPH had very little ability to prevent the superoxide radical from converting catalase to compound II. Moreover, the rate of NADPH oxidation was several times the rate of compound II formation (in the absence of NADPH) under a variety of conditions. Very little NADPH oxidation occurred when NADPH was exposed to catalase, H2O2, or the superoxide radical separately. That the ratio exceeds 1 suggests that NADPH may protect catalase from oxidative damage through actions broader than merely preventing the formation of compound II.

Interest in the disposal of reactive oxygen species stems from the growing evidence that these molecules are active or participating agents in mutagenesis and aging and in the cellular damage from a wide variety of environmental and endogenous stresses (1)(2)(3). One of the more highly studied cells under oxidative stress is the human erythrocyte with, and without, a genetic impairment in maintaining NADP in the reduced state (NADPH) (4 -10). The impairment results from the most common of the potentially lethal human enzyme defects, that of glucose-6-phosphate dehydrogenase. The susceptibility of such cells to oxidative stress was initially attributed to the need for NADPH to remove hydrogen peroxide (H 2 O 2 ) via glutathione reductase and glutathione peroxidase (4). Attempts to identify a soluble protein that was binding NADPH within human erythrocytes led to the discovery that one molecule of NADPH is tightly bound to each of the four subunits of catalase (H 2 O 2 : H 2 O 2 oxidoreductase, EC 1.11.1.6) of mammals (11). Studies of purified catalase revealed that NADPH effectively protects cat-alase against H 2 O 2 at physiologically realistic concentrations of both NADPH and H 2 O 2 (12). These and other findings (13) with highly purified catalase confirmed studies that led earlier investigators to notice the same effect in hemolysates (14). This action of NADPH solved a decades-old puzzle as to the nature of reducing equivalents that serve to keep catalase active in vivo. Later studies revealed that, within human erythrocytes, the role of NADPH in keeping catalase active was more important than the role of NADPH in the glutathione reductase/ peroxidase pathway (7)(8)(9)(10). Evidence has been presented that NADPH protects catalase by preventing and reversing the formation of an inactive form of catalase, compound II, which differs within the heme group from active catalase (12). Work in over five laboratories has provided information on how NADPH could accomplish this task (12,(15)(16)(17)(18)(19)(20). Among the difficulties in explaining the protection of catalase by NADPH is that NADPH is bound on the surface of catalase, some 20 Å from the heme, that the channel to the heme is too narrow to accommodate NADPH, and that conversion of compound II back to native catalase is a one-electron reduction step, whereas NADPH is traditionally regarded as a two-electron reducing substance. The present study provides kinetic and stoichiometric observations on the mechanism of this action by NADPH and suggests how a current model of this action will need to be revised to accommodate these new findings.
An understanding of the experiments that follow requires knowledge of the terminology and pathways for the interconversion of the various forms of catalase. Fig. 1 consists of the additions of Lardinois (21) to the traditional scheme (22) for those interconversions. The role of NADPH in the scheme is described under the "Discussion." Ferricatalase has a protoporphyrin IX-iron(III) complex as its active site and is the native, free, or resting state of the enzyme. Compound I, which is the other active form of catalase, contains an atom of oxygen gained from step 1 (see Fig. 1 for steps [1][2][3][4][5][6][7][8][9], leaving the protoporphyrin-iron group at 2 oxidation equivalents above that of ferricatalase (19 . (19). Compound II arises by one-electron reduction of compound I and is considered to be an iron(IV) oxo-ligated porphyrin (23). The one-electron reduction can result from certain reducing substances of relatively small molecular size, such as ferrocyanide, or from a poorly understood "endogenous donor" within the structure of catalase (17,22). Compound III, also called oxycatalase, is regarded as having similarity to the oxy compounds of myoglobin and hemoglobin (21). Although the rate constant for step 2 is much higher than that for step 9, ethanol can be added at a concentration greatly exceeding the steady-state concentration of H 2 O 2 that is gen-erated by glucose oxidase (22). The resulting severe reduction in concentration of compound I has been used to demonstrate that compound I is a precursor to compound II (22 Continual presence of some of the catalase as compound II or compound III, however, leads gradually to irreversible inactivation of the enzyme through step 8 (21).

EXPERIMENTAL PROCEDURES
All incubations were with commercially available, extensively purified enzymes. Sigma was the source of the manganese superoxide dismutase from Escherichia coli. As with the studies of Kono and Fridovich (23), manganese superoxide dismutase was used for most of this study because, unlike Cu,Zn superoxide dismutase, it is not inactivated by H 2 O 2 . Roche Molecular Biochemicals (Mannheim, Germany) was the source of catalase from bovine liver, xanthine oxidase from bovine milk, Cu,Zn superoxide dismutase from bovine erythrocytes, glucose oxidase from Apergillus niger, and glucose-6-phosphate dehydrogenase from yeast. The buffer used for most of this study was 0.1 mM EDTA, 50 mM Na 2 HPO 4 -KH 2 PO 4 buffer, pH 7.4 (Na-K phosphate buffer). A second buffer was KR-Tes, 1 which is a Krebs-Ringer/Tes solution containing, in the following final millimolar concentrations: NaCl 119; KCl 4.7; CaCl 2 2.5; KH 2 PO 4 1.2; MgSO 4 1.2; and (sodium) Tes, pH 7.4, 22.6. Crystalline bovine liver catalase was dissolved in KR-Tes, concentrated on CF-25 ultrafiltration cones (Amicon), and then washed on the cones (12) with the buffer to be used in each experiment. All other enzymes were both dissolved and washed with the buffer to be used in each experiment.
The solution of each enzyme was assayed for protein concentration (24) and activity. The activity of catalase was determined from the first-order rate constant of the rate of disappearance of H 2 O 2 , at an initial concentration of 10 mM, as measured by absorbance at 240 nm with a recording spectrophotometer (25). By this assay, the bovine liver catalase had an activity of 23.7 s Ϫ1 per micromolar catalase concentration (5.9 s Ϫ1 per micromolar heme concentration). The rate of formation of H 2 O 2 by glucose oxidase in the presence of 5 mM glucose was determined with the assay for H 2 O 2 of Green and Hill (26). The glucose oxidase catalyzed the generation of H 2 O 2 at an average rate of 2,100 mol min Ϫ1 per mol of glucose oxidase. The rate of generation of O 2 . by xanthine oxidase in the presence of xanthine (100 M) was measured in 50 mM Na-K phosphate buffer, pH 7.4, by the rate of reduction of ferricytochrome c (10 M) at 550 nm, using the extinction coefficient at 550 nm of 21,000 M Ϫ1 cm Ϫ1 , according to the method of McCord and Fridovich (27). The replotting method of Sawada and Yamazaki (28) provided assurance that the spontaneous dismutation reaction was second-order with respect to O 2 . and revealed a value of 9.14 ϫ 10 Ϫ4 M s for k d /k c 2 , in which k d is the rate constant for the spontaneous dismu-tation reaction and k c is the rate constant for the reduction of cytochrome c by O 2 . . The factor k d /k c 2 provided a means for correcting for the small fraction of O 2 . that undergoes spontaneous dismutation before the O 2 . can reduce cytochrome c. The rates of generation of O 2 . , as measured by the rate of cytochrome c reduction, were similar whether catalase (2 M) was present or absent. At pH 7.4 and 37°C, the xanthine oxidase had an average specific activity of 115 mol min Ϫ1 per mol of enzyme. The total rate of production of H 2 O 2 by xanthine oxidase was determined from the rate of production of urate, as measured by the rate of increase in absorbance at 295 nm (29), and was confirmed with the assay for H 2 O 2 of Green and Hill (26) on incubations containing xanthine oxidase, xanthine, and superoxide dismutase. The rate of oxidation of NADPH was determined from the amount of 6-phosphogluconate formed, as described previously (12). All reaction mixtures had a final volume of 1.0 ml. All incubations were at pH 7.4 and 37°C except for one experiment in which the conditions were otherwise stated. The reaction components were at the final concentrations indicated within parentheses when the concentrations were not specified and were as follows: catalase and, when present, xanthine oxidase, xanthine (100 M), superoxide dismutase, NADP ϩ (2 M), glucose 6-phosphate (1 mM), glucose-6-phosphate dehydrogenase (10 g/ml), glucose (5 mM). For following the kinetics of reactions, readings of absorbance from the spectrophotometer were taken on each group of cuvettes at intervals of 1 min per group and were automatically stored in a computer for later statistical analysis. For obtaining spectra of catalase, absorbance readings were taken and stored at intervals of 1 nm. The absorption spectra were obtained with a Beckman DU-7 spectrophotometer at a recording speed of 1,200 nm min Ϫ1 . Prof. Peter Nicholls kindly provided the absorbances of equimolar concentrations of ferricatalase (the resting or free form of catalase) and compounds I, II, and III at intervals of 1 nm between 350 and 750 nm. The plots from these absorbances were similar to previously published spectra (22,30). The spectra for ferricatalase, compound I, and compound II were confirmed in the following manner. One ml of 100 mM potassium Ches buffer, pH 8.6, was added to 0.5 ml of the bovine liver catalase from the bottle (20 mg/ml) to dissolve all enzyme crystals. The dissolved enzyme was further diluted in 50 mM Na-K phosphate buffer, pH 6.5, to a concentration of about 10 M. One ml of enzyme preparation was then transferred to a cuvette. After a preincubation period of 5 min at 37°C, the spectrum of ferricatalase was obtained. After the first spectrum was obtained, 5 l of 3% peracetic acid were added to the cuvette, and the spectrum was immediately re-determined (compound I). One microliter of 60 mM potassium ferrocyanide was then added to the same reaction mixture, and after 10 min of incubation at 37°C, the spectrum was determined again (compound II). The absorbances of equimolar concentrations of the four forms of catalase were used to determine the amount of each form in H 2 O-treated catalase by a least squares method, as follows. The minimum was found for the sum (from 501 to 750 nm) of the squares of the difference between U n and f f A n ϩ f I B n ϩ f II C n ϩ f III D n , in which n is the wavelength, U n is the absorbance of the treated catalase at wavelength n, and f f through f III represent the unknown fractions of the treated catalase that are in the form of ferricatalase, compound I, etc. A n through D n represent the known absorbances of equimolar concentrations of ferricatalase through compound III at wavelength n. The value for each f was obtained from the solution of the resulting simultaneous, linear equations for the fs. A visual check of the validity of the result was obtained with a spreadsheet/graphing program that allowed a comparison of the observed spectrum with the spectrum from any specified combination of the four forms of catalase. The formation of compound II is usually followed at 435 nm, the isosbestic point between ferricatalase and compound I (22). The difference in extinction coefficient between ferricatalase and compound II was considered to be 32 mM Ϫ1 cm Ϫ1 (31). As observed also by Chance (30), the extinction coefficient of compound III at 435 nm was found to be similar to that of compound II. The increases in absorbance at 435 nm were therefore considered to represent the combined rate of formation of both compounds. conditions and under peroxidative stress, respectively (9). Under these conditions, the concentrations of H 2 O 2 and O 2 . were too low to be determined by present methods, but they could be determined indirectly. Whether generated by glucose oxidase or xanthine oxidase, H 2 O 2 rises to a steady-state concentration at which the rate of removal in the presence of catalase (Reaction 1) equals the rate of generation.

Steady
The bovine liver catalase was found to have a specific activity (see "Experimental Procedures") that would result in a steadystate nanomolar concentration of Results of the cytochrome c assay indicated that 1 M manganese superoxide dismutase reduced the net rate of generation of O 2 . from 40 nM xanthine oxidase (Reaction 5) by 97%. Determinations of Compounds I, II, and III-Changes in the absorbance at 435 nm are traditionally used to follow the kinetics of compound II formation and removal. Changes at this wavelength, however, also reflect changes in the concentration of compound III. Computerized analysis of the absorption spectrum of bovine liver catalase at various times in the incubation gave the percentage of catalase that was in each of the four states of the enzyme (see "Experimental Procedures") and thereby revealed the extent to which absorbance changes at 435 nm were essentially measures of compound II alone. At a xanthine concentration of 100 M, the generation of O 2 . and H 2 O 2 by 40 nM xanthine oxidase ended between 10 and 20 min after the start of the reaction, when the xanthine was depleted (Fig. 2). During this exposure, the percentage of catalase in the native state (ferricatalase) fell to a minimum of 49%. The percentage as compound III rose initially more rapidly than did compound II but reached a maximum of only 9% (Fig. 2). In contrast, compound II rose steadily to a maximum of 33% at 20 -24 min. The increase in absorbance at 435 nm provided an estimate of the combined percentage of compound II and compound III (see "Experimental Procedures"), and this estimate was in general agreement with the combined percentage as determined by the computerized analysis of the absorption spectrum from 501 to 750 nm (Fig. 2). This agreement served to confirm that the difference in molar (heme) specific absorbance between ferricatalase (or compound I) and compound II at 435 nm was 32,000, that compound III had a similar molar specific absorbance at 435 nm, and that the bovine liver catalase used in this study had four functional heme groups. This information, in turn, allowed comparison of the loss in activity of bovine liver catalase and the formation of compounds II by both H 2 O 2 and O 2 . . In results not given, this comparison confirmed that compound II has no catalase activity at pH 7.4 and at 37°C, as had been demonstrated earlier under other conditions by Chance (33). The percentage of catalase in each of its four forms was determined at intervals for a variety of reactions. Table I gives the results only at either 8 or 60 min of reaction time. The ratio of compound I to ferricatalase was lower when the catalase was exposed to the action of xanthine oxidase (Table I, row e) than when it was exposed to the action of glucose oxidase (Table I, rows a and c). The percentage as compound II was lower in the four reactions containing NADPH than in the corresponding reactions without added NADPH. At 8 min, less compound III was present in those reactions in which only H 2 O 2 was generated (Table I,  . was also generated (Table I,   ͓compound II͔ ϭ V II ͑1 Ϫ e Ϫkt ͒/k (Eq. 1) for a reaction in which the formation of compound II, at a constant rate of V II , is offset by the first-order decay of compound II, with a decay constant of k. The visual curve-fitting method was used to determined provisional values for V II and k, then the final values were determined from iteration for a least squares fitting of the increases in absorbances at 435 nm between 1 and 9 min after the start of the reaction. These calculations indicated that NADPH reduced the rate of compound II formation (V II ) from H 2 O 2 by an average of 82%, when the H 2 O 2 was generated by glucose oxidase in three experiments of the type shown in Fig. 3A, and by an average of 83% when the H 2 O 2 was generated in five experiments by xanthine oxidase in the presence of superoxide dismutase, as in Fig. 3, B and C. The difference between the two curves for the xanthine oxidase reaction, with and without superoxide dismutase (Fig.  3B), was considered to represent the contribution to the rate of formation of compound II that was attributable only to O 2 . . The difference between the two curves in Fig. 3B was similar to the difference between the two curves in Fig. 3C, indicating that NADPH had very little effect on the rate of generation of compound II that could be attributed to O 2 . alone. In five experiments of the type shown in Fig. 3, B and C, NADPH decreased the difference between the VIIs (with and without superoxide dismutase) by an average of 11 Ϯ 12% (ϮS.E. of the mean). Results similar to those of Fig. 3, B and C, were obtained with 40 nM xanthine oxidase and Cu,Zn superoxide dismutase and, at a 5-fold slower rate of O 2 . generation, with 8 nM xanthine oxidase and manganese superoxide dismutase (results not given). In contrast to the large effect of NADPH in lowering the rate of compound II formation by H 2 O 2 , NADPH only mildly affected the rate of decay of compound II (Table II).

Stoichiometry of NADPH Oxidation and
Compound II Formation-NADP (2 M) was kept in the reduced form (NADPH) at a steady-state concentration of 2 M in the present study by the addition of a relative surplus of glucose 6-phosphate and glucose-6-phosphate dehydrogenase. When NADPH is generated in this way, the amount of 6-phosphogluconate present at the end of an incubation serves as a measure of how much NADPH was oxidized. In Fig. 4, the amount of NADPH oxidized during the first 10 min of the reaction is compared with the amount of compound II and compound III formed over the same interval. The amount of compounds II and III formed, and the amount of NADPH oxidized, was only a small fraction of the amount of H 2 O 2 and O 2 . generated over the 10-min period in the various experiments of Fig. 4 (approximately 72 and 40 nmol ml Ϫ1 , respectively). Conversely, the amount of NADPH oxidized was greater than the amount of compounds II and III that was formed in the absence of NADPH and was decidedly greater than the amount by which the presence of NADPH decreased the formation of compounds II and III (Fig. 4). In contrast, the amount of NADPH oxidized was less than the amount by which compounds II and III decayed in the absence of H 2 O 2 and O 2 . (but more than the amount by which NADPH increased the decay of compounds II and III) (Fig. 4). As with the rates of formation of compound II in Fig. 3, B and C, the differences in micromolar concentrations of compounds II and III at 10 min, caused by superoxide dismutase (Fig. 4), had a value (1.02 M), when NADPH was absent, that was similar to the corresponding difference (1.07 M) when NADPH was present. Superoxide dismutase, however, reduced the amount of NADPH oxidized in 10 min (Fig. 4). The reduction was from 4.05 to 3.17 nmol ml Ϫ1 , which was significant at the p ϭ 0.01 level. The results of row i of Table I indicated that essentially  Fig. 1) was therefore evaluated by observing the effect on catalase of the xanthine oxidase reaction in the presence of ethanol (Fig. 4). Before conversion to nmol ml Ϫ1 in 10 min, the increase in absorbance at 435 nm was corrected for the drop in absorbance resulting from the addition of the starting solution of xanthine, as observed in a control solution to which an equivalent volume was added as water. Although low, the concentration of compound III was less when NADPH was present. Without added NADPH, the concentration of compound III reached a (heme) concentration 0.103 Ϯ 0.027 M (mean Ϯ S.D.) at 10 min, corresponding to 1.3% of the catalase. The concentration of compound III at 10 min with added NADPH was 0.027 Ϯ 0.021 M. The difference was significant (4 degrees of freedom, t ϭ 3.9) at the level of 0.01 Ͻ p Ͻ 0.025. A second method for determining the stoichiometry between NADPH oxidation and compound II formation is that of Hillar and Nicholls (16) of adding different amounts of NADPH, measuring the duration of the lag before the generation of compound II begins, as measured by absorbance at 435 nm, and determining the slope that follows the lag. Because of differences in interpretation (see "Discussion"), an experiment of this type by Hillar and Nicholls (16) was repeated (Fig. 5). The concavity at the beginning of the upsweep of curves b, c, and d was due to the presence of NADP ϩ . Although NADP ϩ is a weak competitor of NADPH (12), the ratio of NADP ϩ to NADPH is very large just before the NADPH is exhausted. The duration of the lag of reactions b, c, and d of Fig. 5 indicated that NADPH was being oxidized at an average rate of 0.288 nmol ml Ϫ1 min Ϫ1 . The average rate of compound II formation in reactions b, c, and d was 0.073 nmol ml Ϫ1 min Ϫ1 , giving a ratio of NADPH oxidation to compound II formation of 3.9. The same experiment at pH 7.4 gave a ratio of 5.1 (figure not shown).

DISCUSSION
Previously Proposed Model of the Action of NADPH-Soon after the discovery that each subunit of bovine catalase has one tightly bound molecule of NADPH (11), the x-ray crystallographic data on the three-dimensional structure of bovine liver catalase was re-evaluated, and the NADPH was found to be on the surface of the enzyme, approximately 20 Å from the heme group (15). Because NADPH is too large to reach the latter through the relatively narrow channel leading to the heme group, electron tunneling was assumed to be the mechanism by which NADPH is able to reduce compound II to ferricatalase (18). Olson and Bruice (19) used computerized calculations, along with information from the known three-dimensional structure of bovine liver catalase, to estimate the probable route of the electron tunneling. Others have also obtained information on the probable route of electron tunneling between the bound NADPH and the heme center of bovine or Proteus mirabilis catalase (18,34,35). Using time-resolved x-ray crystallography and single crystal microspectrometry, Gouet et al. (20) determined the structure of compound I and compound II of NADPH-dependent catalase from Proteus mirabilis, including the formation and transformation of the ferryl groups. An unexpected result of their study was the finding that compound I acquired an anion at a site near neither the heme iron nor the NADPH-binding site. The most likely candidate for the anion seemed to be O 2 . (20). Hillar et al. (16,17) proposed that NADPH reduces neither compound I nor com-TABLE II Rate constants for disappearance of compound II Compound II was generated at 37°C and pH 7.4 by the addition of xanthine to a reaction mixture containing xanthine oxidase and bovine liver catalase. Concentrations immediately after the addition were xanthine, 100 M, xanthine oxidase, 4 nM, and catalase, 2 M. Measurements of urate production indicated that the xanthine had been consumed by 15 min of reaction time. Additions, to the final concentrations indicated below, were made at 15 min of reaction time. The first-order rate constant, k, for compound II disappearance was calculated from the absorbance at 435 nm between 28 and 38 min after the start of the reaction. The means (ϮS.D.) are from five replicates. pound II but rather a postulated intermediate between the two (Fig. 6). The findings of Ivancich et al. (36), who used EPR and rapid mix freeze-quench techniques, provided some support for the presence of such an intermediate. NADPH functions in its traditional role as a two-electron reducing substance in step 10 of the model of Hillar et al. (17) (Fig. 6), whereas Almarsson et al. (18) propose that NADPH leads to one-electron reduction of compound II via electron tunneling, the other electron going elsewhere, such as to O 2 to form O 2 . .

Support for the Intermediate State-
The present authors (12) have said that NADPH both decreases the formation of compound II and increases the rate of removal of compound II. In contrast, Hillar et al. (16,17) state that NADPH decreases the rate of formation compound II but does not increase its rate of removal. Both actions of NADPH were observed in the present study, but decreasing the rate of formation of compound II was the predominant action. NADPH only mildly increased the rate of spontaneous decay of compound II (Table II). The decay constant of compound II in the presence of NADPH was 1.4 times the constant without NADPH (Table II). The least squares fitting of the absorbances of Fig. 3A (upper curve) to Equation 1 provided an estimate of the decay constant and initial rate of increase in concentration of compound II. Equation 1 revealed that the concentration of compound II at 10 min of reaction time would have been reduced by only 12% by increasing the decay constant by a factor of 1.4. In contrast, NADPH actually decreased the concentration by 80% (Fig. 3A). Moreover, reconsideration of our results at pH 6.5 (12) in the same manner indicates that the action of NADPH at that pH was also one of preventing the formation of compound II by H 2 O 2 , rather than increasing the rate of removal of compound II. That NADPH prevents the formation of compound II by H 2 O 2 , rather than increases the decay of compound II, is supported also by the unexpected finding, in the present study, that NADPH has a limited ability to offset the formation of compound II that arises from the primary action reacts directly with compound I to produce compound II, bypassing the intermediate in the process (Fig. 6). Hillar et al. (16,17) found that ferrocyanide increased the rate of formation of compound II through what they assumed to be the oneelectron reduction of the intermediate by ferrocyanide at step 3b (see Fig. 6). Since they found that NADPH greatly reduced the rate of compound II formation by ferrocyanide, it is unlikely that the action of O 2 . is at step 3b. In their two articles on the concept of an intermediate, and in support of that concept, Hillar et al. (16,17) claimed that the rate of oxidation of NADPH was similar to the rate at which compound II was formed in the absence of NADPH. When corrected for the molar specific absorbances of NADPH and compound II, however, the rate of NADPH oxidation can be shown to be approximately 3 times the rate of compound II formation in their first article (16) and even higher in their second report (17). Fig. 5, in fact, is a repeat of the experiment of Hillar and Nicholls (16) and reveals a ratio of 3.9. When the rate of NADPH oxidation is measured by the rate at which 6-phosphogluconate is formed by the NADPH-generating system, the ratio of NADPH oxidation to compound II formation is 3 to 4 at pH 7.4 (Fig. 4) and pH 6.5 (12) and at different rates of H 2 O 2 generation (12). Thus, a ratio of 3 or more has been demonstrated at two pH values, by two methods, and at different rates of H 2 O 2 production. As indicated in the legend of Fig. 4, very little NADPH was oxidized when H 2 O 2 and O 2 . were generated in the absence of catalase. For detection of an effect of substances of low molecular weight, such as ions of trace metals that might accompany the catalase, these control incubations contained an ultrafiltrate of the catalase. The volume of the ultrafiltrate was the same as the volume of catalase solution added to the incubation mixtures. Moreover, these control incubations provide an overestimate of the ability of H 2 O 2 to oxidize NADPH in the absence of catalase, since the H 2 O 2 would reach concentrations well above those in the presence of catalase. Very little oxidation of NADPH occurred when catalase was exposed to H 2 O 2 and O 2 . in the presence of ethanol (Fig. 4) and therefore at extremely low concentrations of compound I (Table I). Reversibility of step 3a of Fig. 6 could cause the rate of NADPH oxidation to exceed the rate at which compound II would be formed in the absence of NADPH. Specifically, the reversibility would need to result in a relatively rapid interconversion of compound I and the intermediate. Computer simulations of the schemes of Figs. 1 and 6, however, indicate that this reversibility would cause the curve for compound II formation to have two distinct slopes and therefore to fail to fit Equation 1.
Although the steady-state concentration of compound III is low, the concentration is even lower in the presence of NADPH, even under conditions when essentially no compound I or compound II is present (Table I and Fig. 4). Almarsson et al. (18) point out that compound I is an oxidant and also an unstable species, tending to engage in the side reaction of becoming compound II when the encounter between compound I and H 2 O 2 is delayed. Such a delay occurs under physiological conditions, when the rate of generation of H 2 O 2 is low. That compound I is a strong oxidant is underscored by the fact that compound I oxidizes H 2 O 2 in the normal cycle of H 2 O 2 disposal (Fig. 1). Compound I leads to the oxidation of a reductant within the structure of catalase, the so-called endogenous donor, causing compound I to become compound II. We wish to FIG. 6. Sequence for prevention of compound II formation by NADPH. This scheme replaces the reactions shown on the right in Fig.  1. Except for step 3, the representations and reactions are those of Hillar et al. (17). AH represents a postulated oxidizable amino acid near the heme group (17). An intermediate, formed by step 3a, is reduced by NADPH via electron tunneling.
suggest that NADPH may protect catalase from oxidative damage through actions broader than merely preventing the formation of compound II. NADPH may lead to the reduction of oxidizing states and internal groups of catalase other than the intermediate, possibly including a small percentage of compound I, itself. This broader action of NADPH could account for the oxidation of NADPH at a rate exceeding by severalfold the rate at which compound II would otherwise be formed.