Detection of a Higher Oxidation State of Manganese-Prostaglandin Endoperoxide Synthase*

Addition of arachidonic acid or 5-phenyl-4-penten-ylhydroperoxide to manganese-prostaglandin endoperoxide synthase (Mn-PGH synthase) produced a species with an absorbance maximum at 418 nm. This maximum is distinct from those of resting enzyme (372 and 468 nm) or reduced enzyme (434 nm). The formation of the 418 nm-absorbing species was observed immediately after the addition of hydroperoxide to enzyme but only after a 10-s lag period following addition of arachidonate. Mn-PGH synthase exhibited a peroxidase activity that was 0.8% that of Fe-PGH synthase. Addition of peroxidase reducing substrates to the oxidized form of Mn-PGH synthase diminished the absorbance at 418 nm. In the case of N,N,N’,N’-tetra-methylphenylenediamine, reduction of the 418 nm-absorbing species was accompanied by an increase in absorbance at 610 nm due to the oxidized form of the amine. Thus, the spectral and chemical properties of the 418 nm-absorbing species are consistent with its existence as a higher oxidation state of Mn-PGH synthase. Kinetic analysis indicated that formation of the higher oxidation state preceded or was coincident with oxygenation of the fatty acid substrate, eicosa-11,14-dienoic acid. The cyclooxygenase H synthase; Me2S0, dimethyl sulfoxide; EPR, electron paramagnetic resonance; Fe-PGHS, apoprostaglandin H synthase reconstituted with Fe(II1)-protoporphyrin IX; GSH, reduced glutathione; HPLC, high-performance liquid chromatography; MnCCP, manganese cyto- chrome-c peroxidase; MnHRP, manganese horseradish peroxidase; Mn-PGHS, apoprostaglandin H synthase reconstituted with Mn(II1)- protoporphyrin IX chloride; Mn-PPIX, manganese(II1)-protopor-phyrin

Addition of arachidonic acid or 5-phenyl-4-pentenylhydroperoxide to manganese-prostaglandin endoperoxide synthase (Mn-PGH synthase) produced a species with an absorbance maximum at 418 nm. This maximum is distinct from those of resting enzyme (372 and 468 nm) or reduced enzyme (434 nm). The formation of the 418 nm-absorbing species was observed immediately after the addition of hydroperoxide to enzyme but only after a 10-s lag period following addition of arachidonate. Mn-PGH synthase exhibited a peroxidase activity that was 0.8% that of Fe-PGH synthase. Addition of peroxidase reducing substrates to the oxidized form of Mn-PGH synthase diminished the absorbance at 418 nm. In the case of N,N,N',N'-tetramethylphenylenediamine, reduction of the 418 nmabsorbing species was accompanied by an increase in absorbance at 610 nm due to the oxidized form of the amine. Thus, the spectral and chemical properties of the 418 nm-absorbing species are consistent with its existence as a higher oxidation state of Mn-PGH synthase. Kinetic analysis indicated that formation of the higher oxidation state preceded or was coincident with oxygenation of the fatty acid substrate, eicosa-11,14dienoic acid. The cyclooxygenase activity of Mn-PGH synthase was inhibited by the combination of glutathione and human plasma glutathione peroxidase at a glutathione peroxidase concentration 227-fold lower than the concentration that inhibited Fe-PGH synthase. The results suggest that Mn-PGH synthase forms a higher oxidation state following reaction with hydroperoxides added exogenously or generated endogenously from polyunsaturated fatty acid substrates. This higher oxidation state functions in the peroxidase catalytic cycle of Mn-PGH synthase, and its formation appears to be essential for activation of the cyclooxygenase catalytic cycle. PGH synthase' catalyzes the first two steps in prostaglandin and thromboxane biosynthesis (1). Its cyclooxygenase activity catalyzes the bis-dioxygenation of polyunsaturated fatty acids to hydroperoxy endoperoxides and its peroxidase activity catalyzes the reduction of the hydroperoxy intermediates to hydroxy endoperoxides (2)(3)(4)(5). PGH synthase is widely distributed in mammalian tissues, and its levels are sensitive to modulation by a variety of growth factors and glucocorticoids (6). Although it was originally believed that a single gene exists for PGH synthase, recent evidence suggests the existence of a second PGH synthase gene, transcriptionally activated by viral transformation or growth factor treatment (7, 8).
The mechanism of action of PGH synthase has been the subject of much experimental investigation. The available evidence suggests removal of the 13-pro-S-hydrogen of arachidonic acid is the first step of the cyclooxygenase reaction (9, 10). There appears to be a requirement for activation of resting enzyme by hydroperoxy fatty acids because the combination of reduced glutathione and glutathione peroxidase inhibits cyclooxygenase activity (11-15). PGH synthase is a hemeprotein containing one Fe(II1)-protoporphyrin IX per subunit; native enzyme is homodimer of 70-kDa subunits (16)(17)(18). Both cyclooxygenase and peroxidase activities require heme and higher oxidation states of the prosthetic group have been detected during peroxidase turnover (19)(20)(21)(22). A plausible mechanism for hydroperoxide-dependent activation of the cyclooxygenase activity is formation of a peroxidase higher oxidation state. This implies a synergism between the two activities even though biochemical evidence suggests they are likely to occupy distinct regions of the protein (23,24). Despite several lines of evidence linking peroxidase and cyclooxygenase activities in a functional sense, several different modified enzyme preparations have been reported that exhibit substantially reduced peroxidase activity but only slightly reduced cyclooxygenase activity. These include PGH synthase reconstituted with Mn(II1)-protoporphyrin IX, trypsin-cleaved Fe-PGH synthase and site-directed mutants in which certain histidine residues are replaced by glutamine and alanine (25)(26)(27)(28).

13863
Mn-PGH Synthase Higher Oxidation State ase active site (20,29). The resulting tyrosyl radical acts as the oxidizing agent that removes the bis-allylic hydrogen from substrate fatty acid. Indeed, addition of arachidonic acid or hydroperoxides to Fe-PGH synthase at -12 "C followed by rapid freezing and EPR analysis allowed detection of a protein-derived radical that was unequivocally identified as a tyrosyl radical (29). Power saturation experiments suggested the tyrosyl radical was located 7-12 A from the metal center.
More recent EPR experiments question the catalytic competence of the tyrosyl radical in the cyclooxygenase reaction and imply that its formation may reflect enzyme inactivation (30).
A key observation in the latter study was the absence of tyrosyl radicals following addition of arachidonic acid or hydroperoxides to Mn-PGH synthase, which exhibited a vigorous cyclooxygenase activity (30). A corollary of these observations is that formation of tyrosyl radicals by Fe-PGH synthase is mainly due to its peroxidase activity.
Mn-PGH synthase provides a unique opportunity for investigation of the cyclooxygenase activity without the complications of a highly active peroxidase activity to mask catalytic intermediates. Relatively few investigations of Mn-PGH synthase have been undertaken, possibly because of the difficulty in preparing apoenzyme preparations for reconstitution that were devoid of residual Fe-PGH synthase. We recently reported a straightforward procedure for apo-PGH synthase isolation that routinely yields greater than 99.5% apoprotein (31). In the present manuscript, we describe the use of this material for investigations of the reaction of Mn-PGH synthase with fatty acid and hydroperoxide substrates. We find that Mn-PGH synthase is converted to a spectroscopically detectable higher oxidation state(s) that appears to be an intermediate in a poorly efficient peroxidase activity. The kinetics of formation of this species suggest it may be a catalytically competent intermediate that plays a role in cyclooxygenase catalysis.
PGH synthase was isolated from ram seminal vesicle microsomes as previously described and stored at -80 "C in 50 mM Tris-HC1 (pH 8 at 4 "C) containing 300 p~ diethyldithiocarbamate and 0.1% Tween-20 (34). Apo-PGH synthase was prepared via gel filtration procedure with minor variations (31). For each apo preparation, 15-18 mg of enzyme was loaded on a Sephadex G-200 column (180-200 ml). The equilibration and elution buffer was 100 mM Tris-HC1 (pH 8 at 4 "C), 0.1% deoxycholate, 5 mM EDTA, and 5 mM glutathione. Cholate, glutathione, and EDTA were removed by passage through a Sephadex G-25 column (10-20-ml volume) with 80 mM Tris-HC1 (pH 8 at 4 "C) and 0.1% Tween-20. Protein concentration was determined with the Bio-Rad assay using bovine serum albumin (in the same buffer as synthase) to construct the standard curve. The protein was 99.5-100% apo based on cyclooxygenase activities measured without and with the addition of excess heme. The specific activity of the apo-PGH synthase ranged from 50-87 kunits per mg of protein when assayed in 100 mM Tris-HC1 (pH 81, 500 p~ phenol, and 1 p~ hematin at 37 "C. One unit is the amount of PGH synthase necessary to consume 1 nmol of 02/min. at 37 "C with a Gilson model 5/6 oxygraph equipped with a Clark Assay of Cyclooxygenase Activity-O2 consumption was monitored electrode and a thermostatted cuvette. Aliquots of PGH synthase were assayed in a final volume of 1.3 ml containing 100 mM Tris-HC1 reaction was initiated by addition of 13 pl of 10 mM sodium arachi-(pH 81, 500 p~ phenol, and 1 mM hematin. The cyclooxygenase donate (in 10% methanol) for a final concentration of 100 p~ in the oxygraph cell.
Guaiacol Peroxidase Assay-The instrument used was a Hewlett-Packard diode array spectrophotometer (Model 8452A) equipped with both a Hewlett-Packard 9153C hard disk drive and 9000/300 computer. The spectrophotometer was utilized in the kinetic mode. Quartz cuvettes (3 ml) containing star stir bars were used for each reaction. The final reaction volume was 2 ml containing 100 mM Tris-HC1, pH 8, 1 p M porphyrin, 0.5 mM guaiacol, 0.4 mM PPHP, and various amounts of apo-PGH synthase. A thermostatted cell holder (10 "C) equipped with a magnetic stirrer held the cuvette. Air was blown on the cuvette to reduce fogging. The base line was first scanned with only buffer in the cuvette. The porphyrin stocks (500 p~ in Me2SO) were at room temperature, and apoenzyme, PPHP, and the Tris-HC1, pH 8, were on ice throughout the experiment. Porphyrin (4 pl of 500 PM in Me,SO) and apo-PGH synthase were added to the cuvette containing 100 mM Tris-HC1, pH 8, at 10 "C, then 2 min elapsed before addition of 110 p1 of 9 mM guaiacol (in Tris-HC1 buffer). PPHP (20 pl of 40 mM in MeOH) was added to initiate the reaction. The absorbance of the reaction mixture was monitored at 436 nm in the kinetic mode. The first determination was at 0 s. Reactions were analyzed at 1-s intervals for 3 min; PPHP was added between 0 and the linear portion of the curve. Controls were performed with no 1 s. Signal averaging time was 0.5 s. The slope was determined from apoenzyme added to the cuvette; the rates calculated for these reacbefore the peroxidase activities were compared. tions were subtracted from each final rate of the enzymatic reaction in the guaiacol peroxidase assay except that it was in the general UV-Absorption Measurements-The instrument was the same as used visible mode. Quartz cuvettes (3 ml) containing star stir bars were used for each reaction. A thermostatted cell holder (10 "C) equipped with a magnetic stirrer held the cuvette. Air was blown on the cuvette to keep the humidity around the cuvette low. The baseline was first scanned with only buffer in the cuvette. The porphyrin stocks (500 p~ in Me,SO) were at room temperature throughout the experiment. The apoenzyme (in 80 mM Tris-HC1, pH 8, a t 4 "C, 0.1% Tween 20), the 20 mM eicosadienoic acid and 10 mM arachidonic acid (in equimolar NaOH and 10% MeOH), the PPHP (10-40 mM in MeOH), and the Tris-HCI, pH 8, were on ice in a covered bucket. Porphyrin and apo-PGH synthase were added to the cuvette containing Tris-HCl buffer so that the final concentration of each was 1.5 p M in the before recording the first spectrum and subsequently adding sub-2-ml reaction volume. To allow time for reconstitution, 5 min elapsed strate. The absorbance of the reaction mixture was scanned from 190 to 800 nm at specified intervals (usually 5 s). The first scan was at 0 s which was 1 s before substrate addition. Signal averaging time was 0.5 s. All spectra were stored for analysis.
All time courses were plotted from net absorbance changes determined from difference spectra. To construct difference spectra, each spectrum was subtracted from the spectrum of Mn-PGH synthase alone. Net absorbance change was determined by subtracting the absorbance of a reference wavelength from each maximum's or minimum's absorbance on the same difference spectrum. A reference wavelength of 510 or 520 nm was chosen because no peaks or troughs were observed in this region throughout the course of the experiments.
Reduction of Mn-PGH Synthase-A 2-ml solution of 1.5 p M Mnwas placed inside a Thunberg cuvette with 3.9 mg of sodium dithio-PGH synthase in 90 mM Tris-HC1, pH 8, at ambient temperature, nite. The cuvette remained on ice as it was made anaerobic by evacuating with a vacuum line, then flushing with argon that had been passed through an oxygen trap. This cycle was repeated five times. Once anaerobic, the cuvette was inverted several times to add and dissolve the dithionite in the enzyme solution. The absorption spectrum was recorded at room temperature.

RESULTS
Mn-PGH synthase displayed absorbance maxima at 372, 468, and 554 nm. Upon incubation with arachidonate at either 10 "C or ambient temperature, a new maximum appeared at 418 nm (Fig. 1). The appearance of the 418-nm absorbance was accompanied by a decrease in the 372-nm absorbance of Mn-PGH synthase, a decrease and shift in the absorbance at 468-466 nm, and a shift of the 554-nm maximum to 548 nm. When Mn-PGH synthase was preincubated with the cyclooxygenase inhibitor indomethacin, arachidonic acid-dependent formation of the 418-nm absorbance was prevented. Formation of the 418-nm peak was also prevented by heat inactivation of Mn-PGH synthase before addition of arachidonate. Addition of the organic hydroperoxide PPHP to Mn-PGH synthase produced changes in the Soret absorbances of Mn-PGH synthase identical to those produced by arachidonate (Fig. 2). This suggests the formation of the 418-nm peak was due to a hydroperoxide-dependent reaction at the metal center of Mn-PGH synthase. To determine if these spectral changes were due to reduction, Mn-PGH synthase was reduced with   sodium dithionite under anaerobic conditions. The resulting species exhibited an absorbance maximum at 434 nm (Fig. 2).
The results of these experiments imply that Mn-PGH synthase reacts with exogenous or endogenously generated hydroperoxides to produce higher oxidation states of its metal center. Such higher oxidation states are known to be intermediates of peroxidase catalytic cycles (35). Their detection with Mn-PGH synthase is surprising because the manganese enzyme is believed to be devoid of peroxidase activity (25). Therefore, the peroxidase activity of Mn-PGH synthase was measured with guaiacol as reducing substrate and PPHP as hydroperoxide substrate (Table I). The rate of guaiacol oxidation by Mn-PGH synthase was 0.8% of the rate egbibited by Fe-PGH synthase. Guaiacol peroxidase activity by Mn-PGH synthase was 45 times greater than the activity of apo-PGH synthase plus protoporphyrin IX and 36 times greater than the apparent activity of Mn(II1)-PPIX. Thus, the peroxidase activity of Mn-PGH synthase was not an artifact due to small amounts of Fe-PGH synthase in the apoprotein or to a nonspecific peroxidase activity of Mn-PPIX. HPLC analysis of the products of PPHP metabolism by Mn-PGH synthase in the presence of p-methoxythioanisole as reducing substrate revealed the formation of 5-phenyl-4-pentenylalcohol and 5-phenyl-4-pentenylaldehyde. ' The spectra displayed in Figs. 1 and 2 appear to represent steady-state mixtures of resting enzyme and an oxidized enzyme. Therefore, difference spectroscopy was employed to study the kinetics of appearance and disappearance of individual peaks. A typical difference spectrum recorded after addition of arachidonic acid to Mn-PGH synthase is presented in Fig. 3A. A peak was apparent at 418 nm and troughs at 374 and 472 nm. A similar difference spectrum was recorded following addition of PPHP to Mn-PGH synthase (Fig. 3B).
The time courses of the spectral changes in the Soret region associated with addition of arachidonate or PPHP to Mn-PGH synthase at 10 "C are summarized in Fig. 4. A 10-s lag time existed between the addition of arachidonate to Mn-PGH synthase and the onset of spectral changes. The maximal increase in absorbance at 418 nm was observed after approximately 70 s of reaction at 10 "C. As the 418-nm peak increased, the troughs at 374 and 472 nm became deeper. Arachidonate concentrations from 1.5 to 100 /IM produced similar time courses (data not shown). Addition of PPHP to Mn-PGH synthase resulted in the immediate appearance of the absorbance at 418 nm (Fig. 4B). The maximum absorbance at 418 nm was attained within 20 s. In the absence of reducing agents, the species produced by addition of arachi-* G. R. Reddy, unpublished results. donate or PPHP to Mn-PGH synthase persisted. The 418nm peak was detectable (albeit with low intensity) up to 2 h after arachidonate addition to Mn-PGH synthase at 10 "C.

Mn-PGH
The effect of peroxidase-reducing substrates on the 418 nm-absorbing species was assessed. After a 40-s reaction between 1.5 p~ Mn-PGH synthase and an excess of arachidonate (100 p~) , addition of 5 PM hydroquinone immediately decreased the absorbance at 418 nm (Fig. 5). With the arachidonate concentration of 100 pM, the 418-nm maximum slowly regenerated after being quenched by 5 PM hydroquinone. When a 12-fold higher concentration of hydroquinone was used, the 418-nm peak did not reappear. The two troughs in the difference spectra of the steady-state Mn-PGH synthase mixture (376 and 472 nm) increased in absorbance as the 418nm absorption decreased following addition of hydroquinone. Another PGH synthase-reducing substrate, TMPD, also reduced the 418-nm absorbance of Mn-PGHS. Using difference spectra, absorbance changes were simultaneously monitored at 610 nm to detect oxidized TMPD and at 418 nm for oxidized Mn-PGH synthase. With each TMPD addition, oxidized TMPD was detected as the 418-nm maximum diminished, directly linking oxidation of a peroxidase-reducing substrate with reduction of Mn-PGH synthase's higher oxidation state (Fig. 6).
It was of interest to compare the time course of appearance of the 418-nm peak to the time course of cyclooxygenase activity. The rapidity of the spectral changes necessitated the use of a continuous assay for cyclooxygenase activity, preferably spectrophotometric. The major product of arachidonic acid oxygenation by Mn-PGH synthase is prostaglandin Gz which does not possess a chromophore (25). Several hydroxy fatty acids are also produced but they are minor products, and the extent of their formation depends upon the reaction conditions. However, Fe-PGH synthase oxygenates eicosa-11,14-dienoic acid to 1l-hydroperoxyeicosa-12,14-dienoic acid, which exhibits an intense UV absorbance at 236 nm ( t  -24,000) (36). Thus, by performing difference spectroscopy over the range of 200-550 nm, it was possible to monitor simultaneously the time course of substrate oxygenation and prosthetic group oxidation in the same incubation mixture. Fig. 7 presents the results of a typical experiment. The absorbance maximum of the oxidized enzyme increased rapidly and reached a steady-state level within 10-20 s. Thus, the lag phase in the production of the 418-nm peak was much shorter following addition of eicosadienoic acid than arachidonic acid. The concentration of the higher oxidation state decreased only slightly from 20 to 120 s. Substrate oxygenation, as judged by absorbance at 236 nm, proceeded in a linear fashion for approximately 40 s then began to slow. The oxygenation of eicosadienoic acid represented stoichiometric substrate conversion and as a result, the 236-nm absorbance continued t o slowly grow even as the steady-state levels of the 418-nm absorbance declined. In these experiments, an absorbance maximum at 286 nm was also detected, and the rate of its formation paralleled that of the 236-nm peak (Fig. 7). The 286-nm absorbing species is probably a keto fatty acid produced by decomposition of the hydroperoxy fatty acid intermediate. The absorbance at 286 nm could be seen not only in the difference spectra, but also in the absorbance spectra at all time points.
These kinetic experiments are consistent with the possibility that the higher oxidation state@) of Mn-PGH synthase plays an important role in cyclooxygenase turnover. In the case of Fe-PGH synthase, support for the importance of oxidized forms of the enzyme in cyclooxygenase catalysis is provided by the observation that the combination of glutathione (GSH) and glutathione peroxidase (GSH peroxidase) inhibit cyclooxygenase activity (11). This inhibition is proposed to be due to lowering of the level of hydroperoxide activators of enzyme activity (13). Therefore, we conducted experiments to determine the sensitivity of Mn-PGH synthase to GSH/GSH peroxidase. The GSH peroxidase used for these experiments was the human plasma enzyme purified in these laboratories (33). Fig. 8 compares the concentration dependence of cyclooxygenase inhibition of Mn-PGH synthase and Fe-PGH synthase. It is evident that the manganese enzyme is much more sensitive to inhibition than the iron enzyme. Under the conditions of these experiments, 50% inhibition of cyclooxygenase activity was detected at GSH peroxidase levels of 0.28 and 63.7 units for equal amounts of the manganese and iron enzymes, respectively. Thus, the manganese enzyme is approximately 227-fold more sensitive to inhibition than the iron enzyme. As observed with the iron enzyme, Mn-PGH synthase is not inhibited by GSH or GSH  peroxidase alone, or GSH in the presence of heat-inactivated GSH peroxidase.

DISCUSSION
Approximately 10 s after arachidonate addition to Mn-PGH synthase, a species that absorbs at 418 nm is detectable as a part of a steady-state mixture of resting enzyme and oxidized Mn-PGH synthase ( Figs. 1 and 4). The immediate appearance of this species following the addition of a hydroperoxide substrate to Mn-PGH synthase suggests it is a higher oxidation state (Figs. 2 and 4). Compounds known to be peroxidase-reducing substrates for Fe-PGH synthase not only quench the 418 nm-absorbing species of Mn-PGH synthase, but are simultaneously oxidized by it (Figs. 5 and 6 and Table  1) (37). When PPHP is used with guaiacol in a peroxidase assay, Mn-PGH synthase exhibits low but reproducible peroxidase activity that does not result from trace amounts of Fe-PGH synthase in the apo preparation or from Mn-PPIX alone (Table I). Accurate comparison of the peroxidase activities of Fe-PGH synthase and Mn-PGH synthase is difficult because the reaction with Fe-PGH synthase proceeds extremely rapidly. When PPHP and guaiacol are the substrates, the initial rate of guaiacol oxidation by Fe-PGH synthase declines within 5 s at 10 "C. In contrast, under identical assay conditions, the rate of guaiacol oxidation by Mn-PGH synthase remains linear for at least 80 s. The persistence of the Mn-PGH synthase peroxidase activity compared to the iron enzyme is also characteristic of their respective cyclooxygenase activities. Upon reconstitution of apo-PGH synthase with Mn-PPIX chloride, cyclooxygenase activity typically proceeds three times longer and consumes three times more oxygen than apo-PGH synthase reconstituted with hematin even though the initial velocity of oxygen uptake by Mn-PGH synthase is lower than that of Fe-PGH synthase.
Manganese derivatives of other peroxidase enzymes form higher oxidation states upon addition of hydrogen peroxide (38-40). Characterization of these peroxide compounds by electronic absorption and electron paramagnetic resonance spectroscopies suggests that the higher oxidation state of manganese horseradish peroxidase (MnHRP) contains manganese (IV), whereas the higher oxidation state of manganese cytochrome-c peroxidase (MnCCP) contains Mn(V) (38-40). Electronic absorption spectra of both the resting enzymes and the higher oxidation states of MnHRP and MnCCP are similar to the spectra of Mn-PGH synthase, although there are some differences in the absorbance maxima in the 500-700-nm region. Upon addition of hydrogen peroxide, the 564nm maximum of MnHRP disappears and maxima at 509 and 605 nm are evident (40). In contrast, the MnCCP maximum at 568 nm shifts only slightly to 555 nm upon formation of

Mn-PGH Synthase
Higher Oxidation State its higher oxidation state (38, 39). Both of these MnCCPderived maxima display small shoulders at slightly longer wavelengths. Mn-PGH synthase exhibits a maximum at 554 nm that shifts to 548 nm upon addition of arachidonate or PPHP (Fig. 1). However, there appears to be considerable nonspecific binding of Mn-PPIX to PGH synthase at the stoichiometry of Mn-PGH synthase used in these experiments, which tends to obscure the peroxide-induced spectral changes in the visible region. Strieder et aL3 have concluded from similar experiments that the oxidation state of the metal in the higher oxidation state of Mn-PGH synthase is Mn(1V). Reactions with PPHP corroborate the initial formation of a Mn(V) oxidation state in Mn-PGH synthase. To produce 5-pheny1-4-pentenylalcoho1, two-electron reduction of PPHP is required. This suggests the enzyme is oxidized by two electrons; it seems likely that both electrons were provided by the metal. Since the spectrally detectable higher oxidation state is Mn(IV), the initially formed Mn(V) must be rapidly reduced by one electron.
Higher oxidation states of Mn-PGH synthase appear to play a role in cyclooxygenase catalysis. By monitoring absorbances changes in the 200-500-nm region during the reaction of Mn-PGH synthase with eicosadienoic acid, it is possible to determine simultaneously the kinetics of formation of the higher oxidation state and the oxygenation of the cyclooxygenase substrate. The data in Fig. 7 suggest that formation of oxidized enzyme precedes fatty acid oxygenation which is what one would expect for a catalytically competent intermediate. Presumably, formation of the higher oxidation state results from reaction of Mn-PGH synthase with hydroperoxide impurities in the eicosadienoic acid.
It is important to note that the time courses in Fig. 7 represent different types of chemical events. The formation of the higher oxidation state represents the approach to and maintenance of a steady state, which contains some fraction of the resting enzyme. In contrast, the increase in absorbance at 235 nm represents conversion of substrate molecules to products and is an integrated response. Using a molar absorptivity of 24,000 M-', one calculates that after 120 s, 45 p M product is formed (41). Thus, 30 molecules of substrate are oxidized per molecule of enzyme (assuming all the enzyme is active). The time course for oxygenation of the first substrate molecules is much faster than the time course for accumulation of 235 nm-absorbing material in Fig. 7 and is probably coincident with the time course of formation of the enzyme higher oxidation state.
The likelihood that the higher oxidation state plays a role in fatty acid oxygenation is also suggested by the effect of GSH/GSH peroxidase on the cyclooxygenase activity of Mn-PGH synthase. Plasma GSH peroxidase is a potent inhibitor of the oxygenation of arachidonate by Mn-PGH synthase.
Direct comparison of the same apo-PGH synthase preparation reconstituted with either manganese or iron indicates that the manganese enzyme is 227-fold more sensitive than the iron enzyme to inhibition by GSH/GSH peroxidase. The difference in sensitivity to GSH peroxidase is similar to the difference in peroxidase activities of the manganese and iron enzymes (123-fold). Kulmacz (42) has reported a 5-fold higher peroxide concentration requirement for activation of Mn-PGH synthase as opposed to Fe-PGH synthase. The 227-fold difference in GSH peroxidase sensitivity of the two preparations suggests there are also significant differences in rate coefficients for reaction of hydroperoxide with the metal center. This is borne out by the difference in peroxidase  . Biol. Chem. 267,13870-13878. activities at saturating hydroperoxide concentrations.
As stated earlier, removal of the 13-pro-S-hydrogen is the initial step in cyclooxygenase catalysis. The present results strongly suggest that the higher oxidation state of the metal prosthetic group is involved in the generation of the fatty acid oxidizing agent. One possibility is that the oxidized metal is the agent that removes the hydrogen from the fatty acid. However, the fatty acid cannot serve as a reducing substrate for a typical manganese peroxidatic cycle because the rate of peroxidase turnover is much less than the rate of cyclooxygenase turnover. This could be accommodated by invoking reversible redox changes between Mn(V) and Mn(1V) or Mn(1V) and Mn(II1) in the cyclooxygenase reaction. Once the higher oxidation state is produced it could undergo one electron reduction by the fatty acid and the reduced metal could be reoxidized by the prostaglandin G peroxyl radical in the last step of catalysis. Thus, the higher oxidation could be maintained without traversing a typical peroxidase catalytic cycle. However, there is no literature precedent for such a series of redox transformations in peroxidase enzymology.
Another possibility is that the higher oxidation state oxidizes a protein residue that removes the hydrogen from substrate fatty acid. Although protein-derived radicals are detectable on reaction of Mn-PGH synthase with arachidonic acid, the time course of their formation significantly lags behind the time course of fatty acid oxygenation (30). In fact, a t -12 "C, most of the fatty acid oxygenation is complete before a protein radical is detectable. Furthermore, reaction of Mn-PGH synthase with PPHP produces no protein radicals that are detectable by EPR spectroscopy (30). However, it is difficult to exclude the possibility that a protein radical is produced that is so broadened by interaction with the paramagnetic metal and the slow tumbling of the protein that it is not detectable by EPR spectroscopy. In fact, Strieder et a1. 3 have proposed that an unspecified amino acid reduces Mn(V) to Mn(1V) thereby generating a protein radical that oxidizes substrate fatty acid. Support for such a scheme to explain activation of cyclooxygenase activity is provided by the recent finding that the protein radical of iron cytochromec peroxidase is reduced independently of electron transfer to the metal center (43). Thus, once formed, the protein radical and the Mn(1V) prosthetic group could be considered separate and distinct oxidizing agents.
Finally, one should consider oxidized fatty acid molecules as potential substrate-oxidizing agents. Lands and associates (12) first noted the similarity of the kinetics of the cyclooxygenase reaction to the kinetics of lipid peroxidation. Both reactions exhibit lag phases under certain conditions that are abolished by addition of hydroperoxides. Once initiated, the cyclooxygenase reaction and lipid peroxidation rapidly attain a maximal rate. Hamberg and Samuelsson (9, 10) proposed many years ago that the cyclooxygenase reaction is a free radical oxygenation, and strong supportive evidence for this hypothesis is now available (Scheme I). The initiating agent(s) in lipid peroxidation is still a matter of controversy but there seems to be agreement that an oxidized metal complex (probably iron) is responsible (44). Once peroxidation is initiated, it is maintained by peroxyl radicals formed by coupling oxygen to fatty acid carbon radicals (45). If the peroxyl radical precursor to prostaglandin G oxidizes another fatty acid molecule, this would propagate the cyclooxygenase reaction independent of any redox changes at the metal center. Since the metal center would mainly function to initiate the cyclooxygenase via the formation of a higher oxidation state, multiple cyclooxygenase turnovers could occur on an enzyme with low peroxidase activity. The role of the protein would be to control the stereochemistry of oxygenation by binding the fatty acid in the conformation displayed in Scheme I. Such a scheme requires multiple fatty acid binding sites on the protein and translocation of the fatty acid from a site where it is oxidized to a site where it is the oxidizing agent. It also does not explain why there are differences in V,,, between fully activated Mn-PGH synthase and Fe-PGH synthase.
Regardless of the precise role of the oxidized metal in the cyclooxygenase reaction, the present results demonstrate the utility of Mn-PGH synthase for mechanistic studies. Its higher oxidation state exhibits spectral properties distinct from resting enzyme [Mn(III)] or reduced enzyme [ (Mn(II)]. This is not the case for Fe-PGH synthase because resting enzyme and its peroxidase compound I have the same X , , , but different molar absorptivities (19). Mn-PGH synthase is two orders of magnitude more sensitive to inhibition by GSH/ GSH peroxidase than Fe-PGH synthase. This not only provides insight into the mechanism of the cyclooxygenase reaction but should be exploitable in kinetic studies of the higher oxidation state. Finally, the application of rapid scan diode array spectroscopy to the analysis of the reaction of Mn-PGH synthase with polyunsaturated fatty acids provides an opportunity rare in enzymology: the ability to simultaneously monitor substrate conversion and enzyme intermediates in a single reaction vessel.