The free radical formed during the hydroperoxide-mediated deactivation of ram seminal vesicles is hemoprotein-derived.

Prostaglandin synthase is a multi-enzyme complex which catalyzes the oxygenation of arachidonic acid to the various prostaglandins. During the oxygenation, the enzyme is self-deactivated and, on the basis of ESR data, it has been proposed to form a self-destructive free radical. The free radical was suggested to form from the oxygen lost from prostaglandin G2 during its reduction to prostaglandin H2, and the destructive species was therefore thought to be an oxygen-centered free radical, tentatively identified as the hydroxy radical. We have reinvestigated this ESR signal (g = 2.005) and have concluded, with the aid of the known ESR parameters for the hydroxy and other oxygen-centered free radicals, that the free radical formed during the oxygenation is neither a hydroxy nor any known oxygen-centred radical. Prostaglandin synthase is thought to be a hemoprotein, so this unknown ESR signal was compared with the previously observed free radical formed by the reaction of H2O2 with methemoglobin. This comparison indicates that the free radical formed by the reaction of prostaglandin G2 with ram seminal vesicles is hemoprotein-derived and may be formed by the oxidation of an amino acid(s) located near the iron of the heme.

in incubations containing ram seminal vesicle microsomes and arachidonic acid in air, but not in a nitrogen atmosphere; the same radical was detected in incubations containing microsomes and PGGZ, in both air and nitrogen atmospheres. Both phenol and methional (presumed oxygen-centered radical scavengers) inhibited the ESR signal intensity. It was concluded that the free radical obtained during arachidonic acid and PGG2 metabolism by ram seminal vesicles is an oxygencentered species, and that this oxygen-centered free radical caused the irreversible deactivation of cyclo-oxygenase; phenol and methional increased the lifetime of the cyclooxygenase by scavenging the oxygen-centered radical. The observation of the ESR signal from the reaction of PGGz with ram seminal vesicle microsomes originated the concept that the oxygen-centered radical [Ox] was released during the reduction of the 15-hydroperoxy moiety present in PGGz to the hydroxy moiety in PGHz (9).
The same ESR signal, with varying intensity, was observed during the reduction of PGGz to PGH2, 15-hydroperoxy PGEl to PGEI, and 15-hydroperoxy-5,8,11,13-eicosatetraenoic acid to 15-hydroxy-5,8,11,13-eicosatetraenoic acid by the ram seminal vesicle microsomes (5, 6). The same peroxidase activity would even reduce Hz02, concomitantly producing the ESR signal. The HO' radical may be formed from Hz02, so this result supports the identification of the HO' radical as the oxidant [O,]. A wide variety of co-substrates stimulated the hydroperoxide reduction. All the substrates that stimulated the hydroperoxide reduction depressed the ESR signal (5,6).
It was concluded that the oxygen-centered radical released during the hydroperoxide reduction is responsible for substrate co-oxidation, and substrates such as phenol, aminopyrine, and iodide protect the enzyme from deactivation by preferentially scavenging the oxygen-centered radicals (5,6).
Perhaps the strongest evidence that the oxidant released during the conversion of PGG2 to PGHz is responsible for COoxidation comes from the investigation of the co-oxidation of sulindac sulfide (10). Sulindac sulfide is oxidized to its corresponding sulfoxide during the peroxidase-catalyzed reduction of a variety of hydroperoxides. By utilizing the I80-labeled 15hydroperoxy-PGEn as an oxidizing substrate, it was shown that the oxygen atom transferred to sulindac sulfide originated from the hydroperoxide, 15-hydroperoxy-PGE2 (10). Such an 0-0 scission was interpreted to favor the identification of the oxidant as a hydroxy radical (HO') (6).
The same group of workers suggested that the oxidant, possibly HO' or RO', resulting from the hydroperoxide metabolism by ram seminal vesicle microsomes, is capable of destroying prostacyclin synthase (11). Thromboxane AZ synthase is not affected by the oxidant. Since it is well known processes, the investigators proposed that the oxidant (HO' or RO') released during the initial stage of arachidonic acid cascade may play a vital role in the understanding of inflammatory and other pathological processes (11). In essence, the Merck group has shown that the nascent oxidizing agent, an oxygen-centered species which is suggested to be the hydroxy radical (6,9), is responsible for the deactivation of the cyclo-oxygenase (9), the peroxidase component of prostaglandin synthase (5), the co-oxidation of substrates (5,6), and the deactivation of prostacyclin synthase (11). Thus, in order to understand fully these biochemical and biological processes, it is imperative to know the exact nature of the highly reactive oxidant.
In this paper, we have reinvestigated the free radical formed during the reaction of ram seminal vesicles with PGGz. Comparison of the g-value obtained from the ESR spectrum of this free radical with the known g-values of oxygen-centered radicals proves that the free radical released during arachidonic acid or PGG2 metabolism by ram seminal vesicle microsomes is not a known oxygen-centered species and is certainly not a hydroxy or an alkoxy radical. We propose that the free radical is derived from the enzyme and in this respect is quite similar to the species obtained from the oxidation of methemoglobin (12,13) or metmyoglobin (13)(14)(15) with hydrogen peroxide.

MATERIALS AND METHODS
Arachidonic acid was purchased from Nuchek, and indomethacin, methemoglobin, and tryptophan were obtained from Sigma. All other chemicals were of reagent grade and obtained from standard suppliers. Ram seminal vesicles were obtained from a local slaughterhouse and stored at -80°C. The microsomal protein was prepared from ram seminal vesicles as described previously (16), stored at -80°C and used within 1 week.
Oxygen uptake studies were carried out with an oxygen electrode (Yellow Springs Instrument). The stimulatory effects of the peroxidase substrates were determined from the increase in slopes of the oxygen uptake curves obtained upon the addition of arachidonic acid (400 p~) in ethanol (0.25% of total volume) to 1.5 ml of Tris buffer (air saturated, pH 7.5) containing 2.0 mg/ml of microsomal protein.
ESR measurements were made with a Varian Century series E-109 spectrometer equipped with a TM,w cavity. The time-swept ESR signal was obtained by adjusting the magnetic field to the maximum resonance of the ESR signal and recording the signal with the magnetic field sweep switched off, using the instrument's time-based recorder. Low temperature ESR spectra were obtained with the standard Varian variable temperature accessory or liquid nitrogen dewar. For the low temperature work, the incubation mixtures were transferred to a 3-mm cylindrical quartz tube, freeze-quenched within 20 s in liquid nitrogen, and kept in liquid nitrogen until further use. gvalue measurements were carried out with an E-232 dual sample cavity using Fremy's salt as ag-standard (17).

RESULTS
The ESR spectrum of the free radical formed in an incubation mixture containing ram seminal vesicle microsomes and arachidonic acid was obtained according to published procedures (5,9). If such an incubation is frozen in a 3-mm quartz sample tube containing a capillary of DPPH dispersed in KC1, the spectrum in Fig. 1 results. The g-value of the broad spectrum so obtained is clearly very near the g-value of DPPH (2.0037). More accurate g-values of this unknown free radical were obtained by means of a dual cavity using Fremy's salt as the g-standard (Table I). This technique enables the separate recording of the ESR spectrum of the unknown free radical and the g-value standard so that the g-value of the sample of the unknown free radical can be determined without corrections for the overlap of two spectra (17). Table I1 lists g-values of the known oxygen-centered free radicals. g-values of the unknown free radical at room temperature and at -170 "C are listed in Table I. Comparison of the g-value for the free radical under investigation with g-values given in Table I1 rules out peroxy, alkoxy, or hydroxy radicals as possible candidates. The free radical of the ozonide type is the only species which matches the g-value of the free radical under investigation (Table 11). The "ozonide" free radical is thought to be formed from the reaction of oxygen with an alkoxy radical (20), but such a species can be excluded, since the ESR spectrum shown in Fig. 1 could be obtained by incubating the ram seminal vesicle microsomes with hydroperoxide PGGz under anaerobic conditions (9). In addition, the ESR spectrum of the unknown free radical can be obtained at room temperature (with a slightly lower g-value) where highly reactive species, such as HO' or RO' are not observed. If the unknown radical is a small molecule, then its observation at room temperature requires that the g-value measured at the point at which the spectrum crosses the base line be the isotropic g-value. For this reason, we have compared the g-value of the unknown free radical with the isotropicg-values of the oxygen-centered free radicals.
Although the hydroxyl radical has been observed with ESR in ice, its spectrum has never been detected in the liquid phase (19)(20). According to Symons (20), this is largely due to the fact that the unpaired electron is in an orbitally degenerate state, which causes the hydroxyl radical g-tensor to vary from 0 to 4. In the absence of an asymmetric hydrogen bonding (such as in ice) to lift this degeneracy, the ESR spectrum is so broad that the hydroxyl radical cannot be detected in a fluid state. Similar reasoning has been advanced to explain the failure to detect alkoxy radicals in solution (20, 21).
From the arguments presented above, it is clear that the free radical generated during the reaction of hydroperoxides with ram seminal vesicles is not a freely rotating oxygencentered radical, but must be derived from the ram seminal vesicle microsomes, presumably from their protein component. A g-value in the range of 2.004 to 2.005 suggests that an organic free radical, possibly from one or more aromatic amino acids, is responsible for the ESR signal of the unknown free radical. However, the absence of hyperfine structure or apparent g-anisotropy precludes the identification of the radical structure with ESR under these conditions. Incubations in DzO buffer did not improve the spectral resolution. Henceforth, the unknown free radical is referred to as the ram seminal vesicle free radical. Once it is realized that the ESR signal arises from the ram seminal vesicles and not from a

Ram Seminal
Vesicle Free Radical  (peak-2 milliwatts at -196 "C 25 G at -196 "C 18 G at -196 "C 10-16 G/ at -196 "C to-peak) 9 Gg at 23 "C 15 Gn at 23 "C ' This free radical was obtained upon the addition of arachidonic acid to ram seminal vesicles according to the literature (5,9). This free radical was obtained upon the addition of 1 m g / d of methemoglobin to a solution of 100 p~ H202 in Tris buffer (pH 7.5).
The free radical was either observed at 23 "C where it decayed in a few minutes or frozen in liquid nitrogen after 30 s of reaction. e This g-value is taken from Ref. 14, where the error was reported as k0.0006. A similar error is estimated for the other g-values. The spectrum of this radical in Ref. 13 indicates that this g-value is gl where gll = 2.027.
PmaX is the microwave power at which the signal amplitude is the small molecule, it is clear that our measured g-value will not necessarily correspond to an isotropic g-value even at room temperature, because motional averaging cannot be assumed.
Microwave power saturation studies at -196 "C show that the ESR signal of the ram seminal vesicle radical saturated at higher powers (Fig. 2), but that it is more difficult to saturate than the methemoglobin free radical. This result is consistent with the ESR signal arising from an amino acid(s) free radical, because paramagnetlc transition metal spectra usually cannot be saturated at -196 "C. On the other hand, aromatic organic free radicals, structurally related to aromatic amino acid free radicals, saturate much more easily than does the ram seminal vesicle radical. If a paramagnetic transition metal such as iron is near the organic free radical, then the free radical signal becomes more difficult to saturate, as has been shown for paramagnetic metal bound to melanin (22). These results probably indicate that the organic free radicals of these hemoproteins are near the heme iron (12).
Not only is the ram seminal vesicle free radical more difficult to saturate than the methemoglobin free radical, but its line width is significantly different (Table I). These results indicate that these hemoprotein free radicals are distinct, and exclude the possibility that the ram seminal vesicle free radical is due to methemoglobin contamination of the ram seminal vesicles.
There is a considerable amount of evidence in the literature that the cyclo-oxygenase and peroxidase activities of pure prostaglandin synthase are fully dependent on heme for their enzymatic activity (23-26). Hemoproteins such as methemoglobin or metmyoglobin substitute for hemin in restoring the maximum. e P I / Z is the microwave power at which the signal amplitude is half of what it would be if no saturation occurred (22). 'This line width increased from IO G at 30 s to 16 G at 400 s of reaction time (13). The error i n these line width measurements is about 1-2 G. P A modulation amplitude of 0.8 G was used for this line width measurement. A modulation amplitude of 3.2 G was used for the line width measurements at -196 "C.
This line width is an estimate taken from Ref. 14 where hyperfine and/or g-value structure was observed.  Fig. 1. ., the microwave power saturation characteristics of the radical obtained upon the addition of 1 mg of methemoglobin to a I-ml solution of 100 p~ H202. The incubation mixture was freezequenched within 20 s and kept in liquid N2. The spectrometer conditions used to obtain both curves were as follows: modulation amplitude = 3.2 G, time constant = 1 s, scan time = 1 min. The spectrometer gains used to obtain the curves were 1.25 X lo4 and 6.3 X lo3, respectively. synthase activity (23). Recently, Marnett and Reed (4) reported that hemin, methemoglobin, and metmyoglobin greatly increase the synthase-dependent peroxidatic benzo(a)pyrene co-oxidation. Another recent report suggests that destruction of the hemin-binding site or removal of hemin may be the general mechanism of the irreversible inhibition of the prostaglandin synthase (26). In light of these findings, two models for the ram seminal vesicle free radical are proposed.
The first model was reported in 1958 by Gibson et al. who found that a free radical is formed during the oxidation of metmyoglobin with HzOz (13)(14)(15). The second model is the methemoglobin radical, which also forms from the oxidation of the hemoprotein with Hz02 (12,13). We have reinvestigated the latter species and determined its g-value and other spectral parameters (Table I). Literature values of the spectral parameters of the metmyoglobin free radical are also shown in Table I. Although significant differences in P,,, (the microwave power which maximizes the signal) and the peak-topeak line width are apparent, the g-values of the three hemoprotein free radicals are nearly indistinguishable. This near identity in the g-values suggests that all three species are closely related. The metmyoglobin radical also forms as a result of reaction with either organic hydroperoxides or HZ02 (13).

Ram Seminal
The methemoglobin radical has been observed at room temperature (12), but the ram seminal vesicle radical had been reported only at temperatures near that of liquid nitrogen (5,6,9). Observation of the ram seminal vesicle free radical at room temperature shows it to increase for over 10 s after the addition of arachidonic acid, which corresponds to the time when oxygen is being consumed rapidly by the incubation. The signal then reaches a plateau and decays from this maximum over a period of 2 to 3 min. Under the conditions used to measure the line width of this spectrum, the doublet of the ascorbate semiquinone free radical is resolved (aH = 1.8 G); therefore the ram seminal vesicle free radical is not due to the ascorbate radical. Although the room temperature g-value of the ram seminal vesicle radical was little changed from its value at liquid nitrogen temperature, it is important to determine if the room temperature species responds to co-substrates in the same way as does the species observed at low temperatures (5, 6, 27).
With few exceptions, the compounds co-oxidized by the peroxidatic activity of prostaglandin synthase are known substrates of more thoroughly studied peroxidases. Phenol is one of the classic horseradish peroxidase substrates (28, 29), and is also a substrate for the peroxidatic activity of the methemoglobin/H202 system (30). Not surprisingly, the room temperature ram seminal vesicle radical signal is depressed by phenol (Table 111) just as is the methemoglobin free radical signal (12). Iodide is almost unique in being a two-electron donor to horseradish peroxidase (28) and consequently does not form a free radical intermediate, unlike phenol and most other substrates of the horseradish peroxidase/H202 system (28, 29). Iodide destroys the methemoglobin free radical (14), and the effect of iodide on the prostaglandin synthase free radical ESR signal is shown in Table 111.
Oxygen consumption is a good index of cyclo-oxygenase activity (9), and the inverse relationship between prostaglandin and radical formation has been interpreted as co-substrate scavenging of the free radical oxidant [O,], thereby protecting the enzymatic activity from free radical destruction (5, 6,9). However, the simplest explanation is an electron transfer from the electron-donating substrates, thereby oxidizing the cosubstrates and reducing the ram seminal vesicle free radical to its original state, which may be the active form of cyclooxygenase (24). Alternatively, since the metmyoglobin radical oxidizes its own aromatic amino acids, in the absence of alternate substrates (13), a similar process could account for the deactivation of prostaglandin synthase and its associated peroxidase activity.
Tryptophan also depresses the room temperature ram seminal vesicle ESR signal and concomitantly increases the arachidonic acid-induced oxygen consumption (Table 111). Tryptophan protects prostaglandin peroxidase activity against inactivation by an interaction with heme without being oxidized stoichiometrically (31). It should be noted that the  related compound indoleacetate is a much-studied horseradish peroxidase substrate (28).
In any case, the effect of co-substrates on the ram seminal vesicles at room temperature is very similar to the decreases in signal amplitude reported for phenol (5,9) and iodide (6,27) at -185 "C. The corresponding increases in oxygen consumption caused by these substances are also similar to those previously reported. Indomethacin, a classic cyclo-oxygenase inhibitor, completely inhibited the formation of the room temperature ram seminal vesicle radical, presumably by preventing PGG2 formation (9).

DISCUSSION
Once the similarity between the ram seminal vesicle free radical and the methemoglobin and metmyoglobin radicals is realized, a role for this free radical in the peroxidase activity of prostaglandin synthase can be proposed.
It is possible that the prostaglandin synthase radical is formed by the reaction of hydroperoxides, including hydrogen peroxide, with the hemoprotein.
The sp+ecies X: is the free radical described in this work. Note that Xis the ultimate site of oxidation and may not be the primary free radical. The chemical reactivity of the hydroperoxide-induced enzyme free radical is essentially identical with that of the compound I intermediate of horseradish peroxidase (28), the metmyoglobin radical (32 and references therein), or the ES complex of cytochrome c peroxidase (33 and references therein). A similar view of the enzyme origin of the prostaglandin synthase radical has been suggested (34). These enzyme states of horseradish peroxidase and cytochrome c peroxidase contain an organic free radical in one-to-one stoichiometry with the heme (35, 36), and the chemical identity of these free radicals is an area of very active research. The cytochrome c peroxidase radical cannot be power-saturated at liquid nitrogen temperature (33), and the horseradish peroxidase radical can be observed quantitatively only at liquid helium temperatures under conditions of rapid adiabatic passage (35), indicating that, in these cases, the organic free radical is closer to the paramagnetic iron of the heme than it is in the case of the ram seminal vesicle radical. The marked differences in linewidths of the free radicals seen with myoglobin, methemoglobin, and ram seminal vesicles may be the result of small differences in the radical-heme distance.
In addition to phenol, aminopyrine and aromatic amines are good one-electron donators in the horseradish peroxidase/ H2Oz, the metmyoglobin/H202 (37), and the prostaglandin synthase/PGG2 systems (11, 38). In the presence of these substrates, one electron is transferred from the substrate (AH2) to the free radical enzyme intermediate.

E*(X?, Fe'") + AH2 -+ E ( X , Fe'") + AH'
A one-electron transfer plus a proton transfer is equivalent to the loss of a hydrogen atom depicted above, and no preference for either possibility is implied. The formation of Fe'" in this sequence of reactions is widely accepted in the peroxidase literature (35 and references therein), but it is not an integral part of this scheme. The incorporation of 0'' from PGG2 into a substrate is clearly a reaction which our incomplete scheme cannot explain. Similar oxygen transfer reactions have been described as peroxygenase-catalyzed reactions, whereas peroxidases oxidize the substrate without a transfer of oxygen (39). Other examples of the incorporation of molecular oxygen into substrates may well be the result of substrate free radicals reacting with molecular oxygen (34, 40).
Further studies of the role of the ram seminal vesicle hemoprotein radical in the peroxidase activity of prostaglandin synthase are necessary. In particular, neither the stoichi-ometry of the heme and the free radical, nor the stoichiometry of the enzyme radical and substrate radical has been determined. In fact, with exactly the same kinds of evidence Shiga and Imaizumi (12) proposed that the hemoprotein radical of the methemoglobin/HzOz system was the result of the decomposition of X, which was not observable with ESR. Furthermore, the species X was proposed to possess chemical reactivity similar to that of compound I and/or compound 11 of horseradish peroxidase.
MetHb + HzOz & X + MetHb radical X + AH2 + MetHb (?) + AH' This scheme, also, is consistent with all of the available evidence. It is even possible that some hemoprotein other than prostaglandin synthase is a substrate and reacts with X to form the ram seminal vesicle hemoprotein radical. In this case the enzyme free radical would be merely a consequence of the peroxidase activity of ram seminal vesicles. This last possibility is supported by the recent finding that purified prostaglandin cyclo-oxygenase/hydroperoxidase does not show any ESR signal at all when arachidonic acid is oxygenated (41). The interpretation of this result is complicated by the fact that heme must be added to the purified enzyme, which is an apoprotein, to obtain an active preparation. The significance of the hemoprotein-derived free radical observed in ram seminal vesicles is clearly unknown.
These ESR investigations have shown that the free radical observed in ram seminal vesicles and long described as a small oxygen-centered free radical is, in fact, a hemoprotein free radical. It is related to, but distinct from, a free radical formed by the reaction of methemoglobin with Hz02. The metmyoglobin free radical may be an oxidation product of tyrosine (42). The cytochrome c peroxidase free radical has been proposed to be a cluster containing at least one methionine, in which the proximate amino acids share the charge created by the loss of one electron (33). After significant controversy, the free radical of horseradish peroxidase compound I has been found to be the porphyrin a-cation radical of the heme (43, 44). More elaborate techniques such as ENDOR (33, 44) or spin echo (45) will be needed to identify further the ram seminal vesicle free radical.