Rapid Kinetics of Tyrosyl Radical Formation and Heme Redox State Changes in Prostaglandin H Synthase-1 and -2*

Hydroperoxide-induced tyrosyl radicals are putative intermediates in cyclooxygenase catalysis by prostaglandin H synthase (PGHS)-1 and -2. Rapid-freeze EPR and stopped-flow were used to characterize tyrosyl radical kinetics in PGHS-1 and -2 reacted with ethyl hydrogen peroxide. In PGHS-1, a wide doublet tyrosyl radical (34–35 G) was formed by 4 ms, followed by transition to a wide singlet (33–34 G); changes in total radical intensity paralleled those of Intermediate II absorbance during both formation and decay phases. In PGHS-2, some wide doublet (30 G) was present at early time points, but transition to wide singlet (29 G) was complete by 50 ms. In contrast to PGHS-1, only the formation kinetics of the PGHS-2 tyrosyl radical matched the Intermediate II absorbance kinetics. Indomethacin-treated PGHS-1 and nimesulide-treated PGHS-2 rapidly formed narrow singlet EPR (25–26 G in PGHS-1; 21 G in PGHS-2), and the same line shapes persisted throughout the reactions. Radical intensity paralleled Intermediate II absorbance throughout the indomethacin-treated PGHS-1 reaction. For nimesulide-treated PGHS-2, radical formed in concert with Intermediate II, but later persisted while Intermediate II relaxed. These results substantiate the kinetic competence of a tyrosyl radical as the catalytic intermediate for both PGHS isoforms and also indicate that the heme redox state becomes uncoupled from the tyrosyl radical in PGHS-2.

peroxidase activity, which transforms prostaglandin G 2 to prostaglandin H 2 . Several oxidized reaction intermediates have been identified (6 -9). As indicated in Scheme 1, PGHS first interacts with a peroxide substrate, such as prostaglandin G 2 , to generate Intermediate I, equivalent to Compound I of horseradish peroxidase. Intermediate I then converts to Intermediate II, equivalent to complex ES of cytochrome c peroxidase, through an intramolecular electron transfer from a tyrosine residue to the oxidized porphyrin. The transient tyrosyl radical in Intermediate II in each PGHS isoform is capable of oxidizing arachidonic acid to generate an arachidonic acid radical and initiate cyclooxygenase catalysis (10,11). Crystallographic data revealed that the presumed site of the tyrosyl radical, Tyr-385 in PGHS-1 (Tyr-371 in PGHS-2), is located between the heme and arachidonate binding sites, well positioned to couple the two enzyme activities (3)(4)(5), as proposed in the original branched-chain mechanism (7, 8) (Scheme 1).
Although the chemical competence of the tyrosyl radical for cyclooxygenase catalysis has been convincingly demonstrated by single-turnover experiments (10,11), the kinetic competence of the tyrosyl radical had been examined only with manual mixing techniques (13). Thus, the relationship between tyrosyl radical kinetics and those of the rapid changes in the heme redox state during PGHS peroxidase reactions remained unresolved. The redox linkage between the heme center and the tyrosyl radical needs to be defined to determine the coupling efficiency between the two enzyme activities.
Several types of tyrosyl radical EPR spectra have been observed in the PGHS isozymes (Refs. 7, 12-16 and Table I). A 33-35-G wide doublet (WD1) and wide singlet (WS1) are observed when PGHS-1 is manually mixed with various hydroperoxides (12). On the other hand, only a 29-G wide singlet (WS2) is found in PGHS-2 for similar manually mixed peroxidase reactions (16,24). It was of interest to know whether the lack of observable WD in PGHS-2 was the result of a fundamentally different tyrosyl radical structure in this isoform or of a more rapid transition to a wide singlet. The peroxidase reaction in PGHS-1 treated with cyclooxygenase inhibitors such as aspirin, indomethacin, and flurbiprofen or with the mutation of Tyr-385 to phenylalanine resulted in the formation of a 25-26-G narrow singlet (NS1a) (12,17,18). A similar 21-G narrow singlet (NS2a) was observed for PGHS-2 treated with the cyclooxygenase inhibitor nimesulide or with the mutation of Tyr-371 (11,16). These observations of a perturbed tyrosyl radical EPR in PGHS-1 and -2 with impaired cyclooxygenase activity were from manually mixed samples. It remained important to determine whether other radicals were generated earlier in the rapid reactions with peroxide.
We have used rapid-freeze EPR to measure the tyrosyl radical kinetics during reaction of PGHS-1 and -2 with EtOOH, both with and without pretreatment of the enzymes with cyclooxygenase inhibitors. Parallel stopped-flow measurements were done to determine the kinetics of Intermediate I and II formation. The results indicate that very rapid formation of WD species coincides with formation of Intermediate II in both PGHS-1 and -2, consistent with the proposed mechanism (Ref. 7 and Scheme 1), and the formation of the narrow singlet EPR coincides with Intermediate II formation in the isozymes treated with cyclooxygenase inhibitors.

EXPERIMENTAL PROCEDURES
Phenol, indomethacin, flurbiprofen, nimesulide, and hemin were purchased from Sigma. EtOOH was purchased as a 5% aqueous solution from Polysciences Inc. (Warrington, PA) or from Accurate Chemical and Scientific Corp. (Westbury, NY) as a 10% solution. EtOOH concentrations were determined from the absorbance at 230 nm using an extinction coefficient of 43 M Ϫ1 cm Ϫ1 . Tween 20 was purchased from Pierce.
PGHS-1 was prepared as the apoenzyme from ram seminal vesicles, with 5 mM glutathione included in the isoelectric focusing step (9,20). Recombinant human PGHS-2 was expressed in a baculovirus system using Sf9 or High Five (Invitrogen) insect cells and purified as described previously (21). PGHS holoenzymes were formed by adding excess heme to the apoenzyme followed by treatment with DE52 cellulose (Whatman) and gel filtration on a Bio-Rad 10-DG column (for PGHS-1) or gel filtration alone (for PGHS-2). A minimal amount of phenol (50 M) was added to PGHS-2 during reconstitution to avoid loss of activity. The concentration of PGHS holoenzyme was based on the absorbance at 411 nm for PGHS-1 and at 408 nm for PGHS-2 (165 mM Ϫ1 cm Ϫ1 ).
Optical stopped-flow kinetic measurements were conducted using a Bio-SEQUENTIAL DX-18MV stopped-flow instrument (Applied Photo-physics, Leatherhead, UK). Wavelength pairs of 428 and 524 nm (isosbestic points between resting PGHS-1 and Intermediate I (8,9)) were used to monitor changes in Intermediate II in PGHS-1. In the case of PGHS-2, the large value of k 2 (conversion of Intermediate I to Intermediate II) resulted in a minimal accumulation of Intermediate I, and so there was hardly any observable accumulation of Intermediate I (26). We thus chose to monitor Intermediate II kinetics in PGHS-2 at 406 and 428 nm.
Rapid-freeze experiments were conducted using a System 1000 chemical/freeze quench apparatus with a Model 1019 syringe ram, a Model 715 ram controller, and an 0.008-in nozzle (Update Instrument, Inc. Madison, WI). A low temperature isopentane bath was used for pre-chilling the packing assembly containing isopentane and later for cooling during sample packing via a pressure filtration process developed recently in our laboratory (19). The bath temperature was kept at 125-130 K, bringing the temperature of isopentane inside the funnel to 130 -135 K. The temperatures were monitored with a silicon diode temperature sensor (Lakeshore model 200 monitor with a DT-471-PA1 probe). Unless otherwise mentioned, the ram velocity was 2 cm/s in all rapid-mixing experiments. Single-push mixing was used for reactions of less than 160 ms; longer reactions used a push-push mixing syringe program. The dead time of the apparatus, measured with myoglobinazide reactions, is 4 -5 ms (19). EPR spectra of samples collected from rapid-freeze experiments were recorded with a Varian E-6 spectrometer. The EPR conditions were as follows: modulation amplitude, 3.2 G; time constant, 1 s; power 1 mW; and temperature, 96 K. Radical concentrations were determined by double integration of the EPR signal using a copper standard (12) and a packing correction factor of 0.45 determined earlier for this system (19).

Kinetics of PGHS-1 Tyrosyl Radicals-
To observe the earliest possible events in tyrosyl radical formation, we reacted PGHS-1 with two equivalents of EtOOH at 0°C and freezetrapped the sample at various reaction times. The sample trapped at 4 ms showed a typical wide-doublet tyrosyl radical (WD1) with a peak-to-trough width of 34 G and a well resolved 19-G center splitting (Fig. 1). A sample trapped after 5 s of reaction still showed a doublet spectrum but with some decay of the center splitting. Comparison of the freeze-trapped samples with the wide-doublet EPR predicted from the parameters obtained in electron nuclear double resonance (ENDOR) meas-SCHEME 1. Branched-chain radical mechanism of PGHS. AA, arachidonic acid; PGG 2 , prostaglandin G 2 ; PGH 2 , prostaglandin H 2 ; ROOH and ROH, hydroperoxide and corrresponding alcohol. urements 2 indicates that the center splitting in the WD1 EPR had already declined by ϳ30% at 4 ms (Fig. 1).
Increasing the ratio of EtOOH/PGHS-1 to 5 and raising the temperature permitted observation of a more complete transition from WD1 to a wide singlet tyrosyl radical (WS1) (Fig. 2,  top). This transition was essentially complete within 50 s. The initial formation of tyrosyl radical was rapid (k ϭ 16 s Ϫ1 ) and reached a plateau at ϳ0.6 spin/heme within 300 ms (Fig. 2,  bottom). A slow decay in radical intensity (k ϭ 0.047 s Ϫ1 ) began after about 500 ms (Fig. 2, bottom). The kinetics of Intermediate II, monitored at A 524 , closely tracked those of the tyrosyl radical. The rate of Intermediate II formation was 24 s Ϫ1 (Fig.  2, bottom). These results indicate a tight redox linkage between the heme center and the tyrosyl radical in PGHS-1. The slight difference in formation rates between tyrosyl radical and Intermediate II probably originates in the different methods used for their measurement, but it might also imply an additional step occurring between formation of Intermediate II and tyrosyl radical. Stopped-flow observations at 428 nm paralleled those at 524 nm in the early phase but showed a more rapid decay (data not shown), indicating that additional events, such as self-inactivation, were also detected at this wavelength (25).
Effects of Cyclooxygenase Inhibitors on PGHS-1 Tyrosyl Radical Kinetics-The EPR spectra of samples trapped at various times during reaction of indomethacin-treated PGHS-1 and EtOOH (5 eq) all showed a 25-26-G narrow singlet (NS1a) with discrete hyperfine features (Fig. 3, top) just as was found earlier in manual mixing experiments (12,17). There was no indication of initial formation of either WD1 or WS1 signals. Formation of the NS1a radical was rapid (k ϭ 47 s Ϫ1 ) and reached a plateau at ϳ0.35 spin/heme within 200 ms, followed by a slow relaxation (Fig. 3, bottom). As in the case of PGHS-1 itself, the kinetics of Intermediate II (29 s Ϫ1 for the initial phase) paralleled those of the NS1a radical, indicating that binding of indomethacin did not disrupt the redox linkage between the heme and the tyrosyl radical. Separate experi- ments using flurbiprofen instead of indomethacin gave very similar results for both the tyrosyl radical and Intermediate II kinetics (data not shown).
Kinetics of PGHS-2 Tyrosyl Radicals-Substantial formation of tyrosyl radicals was evident at the earliest stage (5 ms) of reaction of PGHS-2 with 5 eq of EtOOH (Fig. 4, top). The EPR spectrum of the initial sample exhibited a complex, 29 -30-G wide signal with a clearly discernible splitting, whereas the spectrum of the sample at 50 ms was essentially a simple singlet (WS2), indicating a very rapid transition from doublet to singlet in PGHS-2. Subtraction of 75% of the signal intensity of the 50-ms sample from the 5-ms sample EPR yielded a doublet with a peak-to-trough width of 30 G (Fig. 4, top), indicating the presence of a substantial wide doublet tyrosyl radical in the very early stages of the PGHS-2 peroxidase reaction.
The formation of tyrosyl radical in PGHS-2 was rapid (ϳ140 s Ϫ1 ). The radical intensity reached a plateau near 0.8 spin/ heme by ϳ50 ms and remained at this level for at least 100 ms (Fig. 4). Formation of Intermediate II, monitored by absorbance changes at 406 or 428 nm, was also rapid (ϳ240 s Ϫ1 ). In contrast to the persistent radical intensity, Intermediate II quickly relaxed back to the ferric heme (17-19 s Ϫ1 ). This difference in behavior between the heme center redox state and the tyrosyl radical was observed for PGHS-2 but not PGHS-1 (compare Figs. 2 and Fig. 4).
The PGHS-2 preparation used above included 50 M phenol as a protectant, and therefore the possible effect of phenol on the PGHS-1 tyrosyl radical and Intermediate II kinetics was examined. In the presence of phenol the rates of tyrosyl radical and Intermediate II formation (18 and 26 s Ϫ1 , respectively) were similar to those for phenol-free PGHS-1 reactions (Figs. 2 and 5). The maximal tyrosyl radical intensity in the presence of phenol was ϳ0.4 spin/heme, somewhat less than in the experiment done without phenol (Figs. 2 and 5). The presence of phenol resulted in a slower transition from WD1 to WS1, evident from the center inflection present in the EPR even after 330 ms (Fig. 5, top). This finding contrasts with the almost complete collapse of the center splitting pattern in the phenolfree PGHS-1 reactions (Fig. 2, top). A comparison of the PGHS-1 and -2 results in the presence of phenol indicated that the transition from wide doublet to wide singlet and the relaxation of Intermediate II were both slower in PGHS-1 (Figs. 4  and 5).
Tyrosyl Radical Kinetics in Nimesulide-treated PGHS-2-Reaction of nimesulide-treated PGHS-2 with 5 eq of EtOOH resulted in rapid formation of both the tyrosyl radical (176 s Ϫ1 ) and Intermediate II (ϳ 350 s Ϫ1 from A 428 observations) (Fig. 6). Intermediate II later decayed at a rate of 33 s Ϫ1 even though the tyrosyl radical intensity remained relatively constant (Fig.  6, bottom). The divergent decay kinetics of oxidized heme and tyrosyl radical were thus seen with both inhibitor-treated and native PGHS-2 (Figs. 4 and 6). The maximal intensity of the tyrosyl radical achieved in nimesulide-treated PGHS-2, ϳ0.4 spin/heme, was substantially lower than that found for PGHS-2 itself. EPR of samples collected throughout the reaction showed only the 21-G narrow singlet (NS2a), with no significant change in line shape (Fig. 6). There was no sign of earlier formation of either a wide doublet or wide singlet, a situation similar to that with indomethacin-treated PGHS-1 (Figs. 3 and 6).

DISCUSSION
The rapid formation of WD and WS tyrosyl radicals observed in both PGHS-1 and -2 parallels the formation of the oxyferryl heme species (Intermediate II). This finding is just as predicted by the branched-chain radical mechanism originally proposed by Ruf and colleagues (7,8). The kinetic competence of the tyrosyl radical for the proposed coupling of the peroxidase and cyclooxygenase activities in both of the PGHS isoforms is thus firmly established.
We previously examined the time-dependent changes of tyrosyl radical concentration and EPR line shape on a time scale of seconds using manual mixing at lower temperatures (-10 to 0°C) (12,13,17). The slower kinetics observed in these earlier experiments probably reflects a previously unsuspected inefficiency in mixing the enzyme and peroxide in the narrow EPR tube by agitation with a nichrome wire loop. Earlier results from single-turnover experiments firmly established the chemical competence of the tyrosyl radical to oxidize arachidonic acid to a fatty acid radical (10,11,22,23). It is clear from the present kinetic results that tyrosyl radicals in both of the PGHS isoforms are formed very rapidly, in concert with heme redox changes, just as expected from the proposed mechanism (Scheme 1). This result provides strong support for the mechanism and indicates that it accurately describes the events leading from the reaction with peroxide to the initiation of cyclooxygenase catalysis in both PGHS-1 and -2.
The present results also make it clear that a wide doublet tyrosyl radical (WD2) species is formed early in the reaction of peroxide with PGHS-2 just as with PGHS-1. The initial wide doublet undergoes a transition to a wide singlet in both isozymes, with a much faster rate for the transition in PGHS-2. Our ENDOR studies 2 support earlier reports (14,15) that the WS1 of PGHS-1 is not a distinct species but is rather a mixture of the WD1 and a narrow-singlet (NS1b) associated with selfinactivated enzyme (13). This singlet is roughly similar to that seen with PGHS inhibited by anti-cyclooxygenase agents FIG. 6. Tyrosyl radical and ferryl heme kinetics during reaction of nimesulide-treated PGHS-2 with EtOOH. Nimesulidetreated PGHS-2 (8.4 M) was reacted with 5 eq of EtOOH in the same buffer and under the same conditions as described in Fig. 4. Top panel, EPR spectra were acquired for freeze-trapped samples at the indicated reaction times. Bottom panel, tyrosyl radical intensities (solid symbols) were determined by double integration of the EPR spectra and are presented along with the kinetics of ferryl heme monitored in stoppedflow measurements at 428 nm in parallel reactions. The smooth line represents single-exponential fit to the tyrosyl radical intensity.
(NS1a) but lacks the latter's detailed hyperfine features (13). Conditions that decrease PGHS-1 self-inactivation, such as decreasing the amount of EtOOH (Fig. 1) or including a reducing co-substrate (Fig. 5), significantly slowed the transition from WD to WS, consistent with the assignment of the NS1b signal to a tyrosyl radical in self-inactivated PGHS-1.
The structural interpretation for the WS observed in PGHS-2 (WS2) is ambiguous despite the general EPR similarities with the WS1 in PGHS-1. Definitive ENDOR data are not available for any of the PGHS-2 radical species. The narrower EPR line width of WS2 than of WS1 allows simulation of WS2 by simply increasing the dihedral angle between the p z orbital of C1 and the C-H bond at the ␤ carbon, corresponding to a change in coupling constant from 21 to 14 G for the ␤ proton (16). This interpretation is consistent with WS2 arising from a distinct structural entity. However, it remains possible that WS2 is a mixture of a wide doublet and a narrow singlet, similar to WS1 in PGHS-1. ENDOR analysis of WS2 will be needed to decide between these two alternative structural interpretations and to understand the basis for the WD 3 WS transition observed in PGHS-2 (Fig. 4).
The present rapid-freeze EPR measurements did not detect formation of either WD or WS signals, even in the earliest stages of the peroxidase reaction, in either indomethacin-PGHS-1 or nimesulide-PGHS-2 ( Figs. 3 and 6). This clear difference with the native isoforms strongly suggests that inhibitor treatment changes the pathway of intramolecular electron transfer in Compound I, resulting in oxidation of a surrogate tyrosine (exhibiting NS1a or NS2a EPR) rather than Tyr-385 (Tyr-371 in PGHS-2) during electron transfer to the heme porphyrin radical, as we previously proposed (18). The identity of this surrogate tyrosine residue is not known for either of the PGHS isoforms, although the similarity in power saturation between the NS1a and WD1 signals (12) and between the NS2a and WS2 signals (16) indicates similar environments for radicals on the surrogate tyrosine and on Tyr-385/Tyr-371.
The present kinetic data have revealed unexpected differences between the two isozymes in the redox association between the heme center and the tyrosyl radical (Figs. 2 and 4). Although formation of Intermediate II heme paralleled formation of tyrosyl radicals in both isozymes, the rate for PGHS-2 was faster than for PGHS-1. The faster rate of Intermediate II/tyrosyl radical formation for PGHS-2 fits with our recent comparative kinetic studies of these two isozymes, which showed PGHS-2 to have a much faster conversion from Intermediate I to II (26). Surprisingly, the redox relaxation of the tyrosyl radical and the heme center parallel one another in PGHS-1 but diverge markedly in PGHS-2 ( Figs. 2 and 4). A similar difference was observed with the NS radicals in PGHS-1 and -2 treated with inhibitors ( Figs. 3 and 6), with the redox relaxation of the tyrosyl radical and the heme center paralleling one another in PGHS-1 but diverging markedly in PGHS-2 ( Fig. 2 and 4). This uncoupling of the tyrosyl radical redox state from the heme redox state in PGHS-2 may be physiologically significant because it tends to prolong the lifetime of the catalytically active tyrosyl radical and thus better sustain propagation of cyclooxygenase activity. This difference in redox coupling between oxyferryl heme and the tyrosyl radical in the two isoforms may contribute to the greater efficiency of hydroperoxide activation in PGHS-2 than in PGHS-1 (21).