Formation of Reactive Cyclopentenone Compounds in Vivo as Products of the Isoprostane Pathway*

Cyclopentenone prostaglandins A2 and J2 are reactive compounds that possess unique biological activities. However, the extent to which they are formed in vivo remains unclear. In this study, we explored whether D2/E2-isoprostanes undergo dehydration in vivo to form A2/J2-isoprostanes. Oxidation of arachidonic acid in vitro generated a series of compounds that were confirmed to be A2/J2-isoprostanes by mass spectrometric analyses. A2/J2-isoprostanes were detected in vivo esterified to lipids in livers from normal rats at a level of 5.1 ± 2.3 ng/g, and levels increased dramatically by a mean of 24-fold following administration of CCl4. An A2-isoprostane, 15-A2t-isoprostane, was obtained and found to readily undergo Michael addition with glutathione and to adduct covalently to protein. A2/J2-isoprostanes could not be detected in the circulation, even following CCl4administration, which we hypothesized might be explained by rapid formation of adducts. This was supported by finding that essentially all the radioactivity excreted into the urine following infusion of radiolabeled 15-A2t-isoprostane into a human volunteer was in the form of a polar conjugate(s). These data identify a new class of reactive compounds that are produced in vivo as products of the isoprostane pathway that can exert biological effects relevant to the pathobiology of oxidant injury.

tory responses (9 -11). Furthermore, CP-PGs exhibit antiviral activity (12). The common feature in these compounds is the presence of a reactive ␣,␤-unsaturated carbonyl group, which is very susceptible to nucleophilic addition reactions and seems to be essential for many of their biological effects (13)(14)(15).
Although the biological effects exerted by CP-PGs have been studied in some detail, the extent to which they are formed in vivo has been the subject of continuing controversy for over 2 decades (16 -18). Fueling this controversy has always been the uncertainty as to what extent dehydration of PGE 2 and PGD 2 ex vivo during sample processing contributes to the amount of PGA 2 and PGJ 2 detected. Recently, ⌬ 12 -PGJ 2 was definitively identified in human urine by Hayaishi and co-workers (19). However, the amounts in urine from males were Ͼ2-fold higher than the amounts in urine from females. This is difficult to reconcile with the evidence suggesting that there is no sexual difference in the amount of PGD 2 produced in vivo in humans (20). Convincing evidence was presented that the ⌬ 12 -PGJ 2 detected in urine unlikely arose as a result of dehydration of urinary PGD 2 ex vivo during sample processing. However, it is difficult to know to what extent PGD 2 may undergo dehydration in the genitourinary tract prior to voiding. This is of particular interest since the same authors recently reported that high levels of PGD synthase are present in human male reproductive organs and that seminal plasma greatly facilitates dehydration of PGD 2 (21). Furthermore, they also recently reported that the level of PGD synthase in male urine is approximately twice that found in female urine (22). Taken together, these findings suggest that at least some of the ⌬ 12 -PGJ 2 detected in urine may have arisen from dehydration of PGD 2 in the genitourinary tract and may explain the higher levels of ⌬ 12 -PGJ 2 in urine from males. This does not confute the occurrence of ⌬ 12 -PGJ 2 in human urine, but only raises the question of its origin, that being whether it arose from systemic sources or from local production in the genitourinary tract. Therefore, it still remains unclear whether CP-PGs are ubiquitously produced throughout the body.
Free radicals have been increasingly implicated in the pathogenesis of a wide spectrum of human disease processes (23,24). Previously, we reported the discovery of PG-like compounds, now termed isoprostanes (IsoPs), that are produced in vivo nonenzymatically as products of free radical-induced peroxidation of arachidonoyl lipids (25,26). Our initial report described the formation of PGF 2 -like compounds (F 2 -IsoPs). Subsequently, we reported the finding that PGD 2 -and PGE 2 -like compounds (D 2 /E 2 -IsoPs) are also produced in abundance in vivo (27). In the studies reported herein, we explored whether D 2 /E 2 -IsoPs undergo dehydration in vivo to produce CP-IsoPs (A 2 /J 2 -IsoPs). Our interest in this stems from the fact that such compounds may be identified as a new class of reactive compounds that are formed as products of free radical-induced lipid peroxidation that exert unique biological actions relevant to the pathobiology of oxidative stress.  (29,30), purified by normal phase HPLC, and analyzed by NMR for structural confirmation.
Analysis of A 2 /J 2 -IsoPs-A 2 /J 2 -IsoPs were analyzed by GC/NICI/MS using a modification of methods described for the analysis of other IsoPs (31). Briefly, 10 ng of [ 2 H 4 ]PGA 2 was added to samples as an internal standard, after which compounds were partially purified using C 18 and Silica Sep-Pak cartridges (Waters), methoximated, and converted to pentafluorobenzyl (PFB) esters. Compounds were then purified by chromatography on silica 60ALK6D TLC plates (Whatman) using a solvent system of hexane/acetone (70:30, v/v). Compounds migrating in the region from 0.5 cm below the PFB ester of PGJ 2 to 1 cm above the PFB ester of PGA 2 were scraped, extracted with ethyl acetate, converted to a trimethylsilyl (TMS) ether derivative, and then analyzed by GC/NICI/ MS. The major ions generated in the NICI mass spectra of the PFB ester, O-methyloxime, TMS ether derivative of A 2 /J 2 -IsoPs and [ 2 H 4 ]PGA 2 are the carboxylate anions at m/z 434 and m/z 438, respectively. Quantitation was performed by integration of the peak areas.
Analysis of A 2 /J 2 -IsoPs as a Piperidyl-enol-trimethylsilyl Ether Derivative-Treatment of PGA with N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) and piperidine has been shown to convert it to a piperidyl-enol-TMS ether derivative, which is specific for A-ring prostanoids (32). Thus, we analyzed for the formation of this derivative with CP-IsoPs. PFB esters of putative CP-IsoPs were treated with a 1:1 mixture of BSTFA/piperidine for 1 h at 60°C and analyzed by GC/ NICI/MS monitoring of the carboxylate anions at m/z 562 for the IsoPs and m/z 566 for [ 2 H 4 ]PGA 2 .
Analysis of A 2 /J 2 -IsoPs by GC/Electron Ionization/MS-Putative A 2 /J 2 -IsoPs generated during oxidation of arachidonic acid in vitro were subjected to purification using two HPLC systems: (a) normal-phase HPLC on a 5-m Econosil SI column using an isocratic solvent system of hexane/isopropyl alcohol/acetic acid (97:3:0.1, v/v/v) and (b) reversedphase HPLC on a 5-m Econosil C 18 column using an isocratic solvent system of acetonitrile/water/acetic acid (38:62:  (27). The amounts of D 2 / E 2 -IsoPs reported in this study differ from those in our previous reports (27). In our previous reports, quantification was based on the intensity of a single prominent peak, whereas in this study, the integrated area under all peaks was used for quantification.
Oxidation of Arachidonic Acid in Vitro-Non-esterified arachidonic acid was dissolved in ethanol and then oxidized in 50 mM Tris buffer for 45 min using an iron/ADP/ascorbate oxidizing system as described (33). 1-Palmitoyl-2-arachidonoyl-phosphatidylcholine was prepared and oxidized in an identical fashion as non-esterified arachidonic acid at a concentration equivalent to 1 mg/ml arachidonic acid. After oxidation, the phospholipid was extracted and hydrolyzed as described below for liver phospholipids.
In Vivo Model of Free Radical-induced Liver Injury-Free radicalinduced lipid peroxidation in liver was induced in rats by administration of CCl 4 (25). Two hours after treatment, animals were killed. Livers were then removed, snap-frozen in liquid N 2 , and either processed immediately or stored at -70°C.
Extraction, Purification, and Hydrolysis of Phospholipids-Lipids from livers of CCl 4 -treated and control rats were extracted as described (31). The lipid extracts were then hydrolyzed enzymatically using Apis mellifera bee venom phospholipase A 2 (27).
Conjugation of 15-A 2t -IsoP and PGA 2 with GSH in Vitro-Radiolabeled 15-A 2t -IsoP and PGA 2 were incubated at 37°C with GSH in phosphate buffer (pH 6.5) in the presence of 19 units/ml glutathione S-transferase (Sigma). The molar ratio of the prostanoids to GSH was 1:10. At the designated time points, aliquots were removed, acidified immediately to pH 3, and extracted twice with 2 volumes of methylene chloride. The formation of a GSH conjugate was assessed by determining the percent of radioactivity that did not extract into organic solvent (14). We predetermined that 93 Ϯ 4% (mean Ϯ S.E.) of PGA extracts into methylene chloride from buffer solutions at pH 3. Control incubations were carried out in the absence of GSH and enzyme.
Formation of 15-A 2t -IsoP and PGA 2 Covalent Adducts with Albumin-Radiolabeled PGA 2 and 15-A 2t -IsoP were incubated at 37°C with 20 mg/ml human serum albumin in phosphate buffer (pH 7.4). The molar ratio of prostanoids to albumin was 1:20. At the designated time points, aliquots were withdrawn, and 10 volumes of cold ethanol were immediately added to precipitate the albumin. The formation of covalent albumin adducts was assessed by determining the percent of radioactivity present in the albumin precipitate after centrifugation. Control incubations were carried out in the absence of albumin.
Formation of Polar Conjugates of 15-A 2t -IsoP in Humans-After informed consent was obtained, a small tracer quantity (1 Ci) of 15-A 2t -[ 3 H]IsoP (150 Ci/mmol) was dissolved in 200 l of ethanol, diluted to 10 ml with sterile saline, and infused over 10 min into the antecubital vein of a normal male volunteer. Individual voided urine specimens were collected over a period of 10 h and assayed for radioactivity. The formation of polar conjugates was assessed by determining the percent of radioactivity in a 1-ml aliquot of urine that did not extract into 2 volumes of methylene chloride at pH 3.

RESULTS
Initially, we explored whether A 2 /J 2 -IsoPs are formed during oxidation of arachidonic acid in vitro (Fig. 1). In the upper m/z 434 chromatogram are multiple peaks with a similar retention time as the m/z 438 peaks representing the syn-and anti-Omethyloxime isomers of [ 2 H 4 ]PGA 2 (lower chromatogram), consistent with the formation of CP-IsoPs. The amount of the compounds formed was 529 Ϯ 135 ng/mg of arachidonic acid (mean Ϯ S.E., n ϭ 4).
Additional analyses further supported the identity of these compounds as A 2 /J 2 -IsoPs. First, no m/z 433 peaks were present, indicating that the m/z 434 peaks were not natural isotope peaks of compounds generating an ion less than m/z 434. When analyzed as [ 2 H 9 ]TMS ether and O-[ 2 H 3 ]methyloxime derivatives, all of the original m/z 434 peaks disappeared, and an identical pattern of new peaks appeared 9 and 3 Da higher, respectively (data not shown), indicating that all of the compounds had one hydroxyl and one carbonyl group. When analyzed following catalytic hydrogenation, there was a disappearance of the m/z 434 peaks and the appearance of new intense peaks 6 Da higher at m/z 440 (Fig. 2). No peaks were detected at m/z 436, 438, or 442. This indicated that the compounds contained three double bonds. In addition, analysis following treatment with BSTFA/piperidine resulted in the formation of a piperidyl-enol-TMS ether derivative (Fig. 3). The amount of A 2 /J 2 -IsoPs analyzed as this derivative was calculated to be 118 Ϯ 32 ng/mg of arachidonic acid (mean Ϯ S.E., n ϭ 4), which is less than the amount formed when analyzed as an O-methyloxime, TMS ether derivative. This discrepancy can be explained by the fact that we have found that, whereas treatment of PGA 2 with BSTFA/piperidine efficiently converts it to the piperidyl-enol-TMS ether derivative, only trivial amounts of this derivative are formed with PGJ 2 . By analogy, therefore, only a portion of the mixture of A 2 /J 2 -IsoPs would be expected to form this derivative.
Further evidence for the identity of these compounds as CP-IsoPs was obtained by analyzing the compounds by electron ionization/MS to obtain complete mass spectra. Compounds were analyzed as a PFB ester, O-methyloxime, TMS ether derivative. One of the mass spectra obtained is shown in Fig. 4.  (307 ϩ 90)), m/z 199 ( ϩ CHϭCH-CH(Me 3 SiOH) (CH 2 ) 4 CH 3 ), m/z 181 (base) ( ϩ CH 2 C 6 F 5 ), and m/z 173 (Me 3 SiO ϩ ϭCH(CH 2 ) 4 CH 3 ). These ions or analogous ions and losses of other derivatives are also intense ions in the mass spectra of CP-PGs (34), 2 indicating that this was a mass spectrum of a 15-series CP-IsoP in which the side chain hydroxyl is located at C-15. Although the structure depicted in Fig. 4 is an A-ring IsoP, ions are not present that would allow a differentiation between an A-ring and a J-ring compound.
We then explored whether CP-IsoPs are also formed in vivo. Previously, we showed that IsoPs are initially formed in situ esterified in tissue phospholipids (26). Therefore, we assayed for the presence of A 2 /J 2 -IsoPs esterified in livers from rats that had been treated with CCl 4 to induce lipid peroxidation (Fig.   5A). Again, a series of m/z 434 peaks were present similar to what was found following oxidation of arachidonic acid in vitro, although the pattern of the peaks differed somewhat. This may be explained by our observation that there appears to be a steric influence of phospholipids on the formation of different IsoP isomers. In support of this notion, the pattern of peaks present following oxidation of arachidonoylphosphatidylcholine in vitro was found to be very similar to the pattern of peaks detected in esterified lipids in liver (Fig. 5B).
Further evidence for the identity of these compounds generated in vivo as A 2 /J 2 -IsoPs was obtained utilizing similar approaches used for the structural identification of the compounds formed in vitro. Analysis as a [ 2 H 9 ]TMS ether derivative, an O-[ 2 H 3 ]methyloxime derivative, and following catalytic hydrogenation indicated that all of the compounds had one hydroxyl group, one carbonyl, and three double bonds (data not shown). Furthermore, the compounds formed a piperidyl-enol-TMS ether derivative (Fig. 6), and the pattern of peaks was almost identical to that formed by the CP-IsoPs generated in vitro (Fig. 3), although the relative abundance of individual peaks differed somewhat.
We then compared the relative amounts of A 2 /J 2 -IsoPs and D 2 /E 2 -IsoPs present in livers from both normal and CCl 4treated rats to determine the extent to which D 2 /E 2 -IsoPs undergo dehydration in vivo (Fig. 7). Of interest was the finding that A 2 /J 2 -IsoPs could be detected esterified in the livers of normal rats at a level of 5.1 ng/g of liver. Relative to the amounts of D 2 /E 2 -IsoPs measured, A 2 /J 2 -IsoPs were present at levels indicating that the extent to which D 2 /E 2 -IsoPs undergo dehydration in vivo is not inconsequential. Following administration of CCl 4 , the levels of both A 2 /J 2 -IsoPs and D 2 /E 2 -IsoPs increased dramatically and to a similar extent by a mean of 23.9-and 21.2-fold, respectively.
We then carried out experiments to exclude the possibility that dehydration of D 2 /E 2 -IsoPs ex vivo during sample processing contributed significantly to the levels measured in liver. That this was not the case was supported by the finding that Ͻ1% of either PGE 2 or PGD 2 that was added to a lipid extract of liver and then processed and analyzed had undergone dehydration to form PGA 2 and PGJ 2, respectively.
Our major interest in the possibility that CP-IsoPs may be formed in vivo is because these compounds should be reactive molecules that are susceptible to nucleophilic addition reactions. Therefore, we compared the ability of one of the CP-IsoPs, 15-A 2t -IsoP, and PGA 2 to conjugate with GSH and adduct to protein, using albumin as a model. We recently showed that one of the E 2 -IsoPs that is produced in vivo is 15-E 2t -IsoP (25). Its dehydration product, 15-A 2t -IsoP should therefore also be one of the CP-IsoPs that is produced in vivo. We initially determined the time course of conjugation of GSH with radiolabeled 15-A 2t -IsoP and PGA 2 in the presence of glutathione S-transferase. Formation of GSH conjugates was assessed by determining the percent of radioactivity that did not extract into methylene chloride at pH 3. Approximately 70% of 15-A 2t -IsoP had conjugated with GSH within 2 min, and the conjugation was complete by 8 min (Fig. 8). The time course for the conjugation of PGA 2 with GSH was found to be essentially identical. In the absence of glutathione S-transferase, no appreciable conjugation occurred with either compound over the same time period (data not shown). Similarly, the time course for the formation of covalent adducts with albumin was essentially identical for both PGA 2 and 15-A 2t -IsoP (Fig. 9).
Following administration of CCl 4 to rats, plasma concentrations of non-esterified F 2 -IsoPs and D 2 /E 2 -IsoPs increase to very high levels (25,27). However, we unable to detect CP-IsoPs in the circulation of rats following administration of CCl 4 in the experiments described above. We hypothesized that this may be due to rapid formation of polar conjugates, e.g. with GSH, following the hydrolysis of the compounds from phospholipids. To test this hypothesis, we infused a tracer quantity of radiolabeled 15-A 2t -IsoP into a normal human volunteer and determined the percent of radioactivity excreted into the urine that did not extract into methylene chloride. Approximately 95% of the radioactivity recovered in the urine was excreted during the first 4 h following the infusion, representing ϳ30% of the total amount of radioactivity infused. Only ϳ5% of the radioactivity present in the urine was recovered in the organic extract, suggesting that all or almost all of the compounds were present in the form of a polar conjugate(s). DISCUSSION We report the finding that PGA 2 -and PGJ 2 -like compounds are formed in vivo as products of the IsoP pathway. Important and interesting was the finding that these compounds not only were detected in abundant quantities esterified in rat livers following induction of an oxidant injury, but were also present in readily detectable quantities in the livers of normal rats.
The discovery of the formation of CP-IsoPs opens up numerous new avenues for scientific inquiry. The reactive nature of these compounds conferred by the ␣,␤-unsaturated carbonyl moiety that characterizes these molecules provides a basis for hypotheses regarding their potential role in the pathogenesis of oxidant injury. As mentioned previously, CP prostanoids exert unique biological effects. They inhibit cellular proliferation via their ability to modulate a variety of growth-related and stressinduced genes, induce apoptotic cell death at higher concentrations, and activate peroxisome proliferator-activated receptor-␥ (3)(4)(5)(6)(7)(8)(9)(10)(11). The reactive ␣,␤-unsaturated carbonyl seems essential for many of these biological actions (15). In this regard, one might anticipate that all of the CP-IsoPs might exert similar biological effects, lessening the potential importance of elucidating the precise structures of individual compounds in the mixture. As shown in Fig. 8, PGA 2 and 15-A 2t -IsoP covalently adduct to protein. Interestingly, however, in intact cells, PGA 2 seems to preferentially bind to particular proteins, the functions of which remain to be elucidated (35). As shown in Fig. 7, PGA 2 and 15-A 2t -IsoP also rapidly undergo glutathione S-transferase-catalyzed conjugation with GSH. We have previously shown that ⌬ 12 -PGJ 2 also rapidly conjugates with GSH, but does require catalysis by glutathione S-transferase (14). However, we (36) and others (37) have shown that conjugation with GSH acts to prevent the ability of CP-PGs to inhibit cellular proliferation and to induce apoptosis. Therefore, much remains to be known about the underlying molecular mechanisms that mediate the biological effects of CP prostanoids. Furthermore, the breadth of our understanding of the biological actions of CP prostanoids is probably far from complete. For example, virtually nothing is known about the biological consequences of the adduction of these molecules with DNA in regards to their potential mutagenicity.
Dehydration of PGE 2 and PGD 2 occurs in physiologic buffers, but has also been shown to be catalyzed by plasma and albumin and more recently also by human semen (21,28,38,39). Plasma components could be involved in promoting dehydration of E-ring and D-ring IsoPs following their release from tissue phospholipids into the circulation. However, this study, we identified the presence of CP-IsoPs esterified in tissue membrane lipids. Whether there are also factors in cellular membranes that promote the dehydration of D 2 /E 2 -IsoPs remains to be explored. Nonetheless, identification of factors that may promote this dehydration may reveal potential avenues for intervention to modulate the formation of CP-IsoPs in vivo.
Our observation that 15-A 2t -IsoP undergoes extensive conjugation in vivo in humans is consistent with our previous findings that ⌬ 12 -PGJ 2 also undergoes extensive conjugation in rats (14). These observations have potentially very important implications. This may explain the difficulty in the past in demonstrating the formation of CP-PGs in vivo because efforts were focused on detecting these compounds in free form. Even if detected in free form in small amounts, e.g. in urine as reported by Hirata et. al (19), this could potentially lead to the erroneous conclusion that the amounts formed in vivo may not be biologically relevant. However, quantification of the conjugated form(s) of CP-IsoPs in plasma and/or urine may provide an approach that more accurately assesses the magnitude of endogenous production of both CP-PGs and CP-IsoPs. In the case of CP-IsoPs, it is possible to assess their formation esterified in tissues of interest, but this is not amenable to human investigation. More important, as well, measurement of a conjugate of CP-PGs and CP-IsoPs would completely eliminate the confounding problem of artifactual formation ex vivo of what is being measured. This clearly provides the impetus for future studies aimed at elucidating the nature of these conjugates.
In summary, we have reported the discovery of a new class of reactive products of lipid peroxidation that are formed in vivo via the IsoP pathway. This provides a rational basis to explore in depth the biological activities of these novel molecules that may provide new insights into the biological consequences of their formation as it relates to the pathobiology of oxidant injury.