CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

FEBS Letters 586 (2012) 3723–3730

journal homepage: www.FEBSLetters.org

Glutathione promotes prostaglandin H synthase ()-dependent

formation of malondialdehyde and 15(S)-8-iso-prostaglandin F2a ⇑ Dimitrios Tsikas , Maria-Theresia Suchy, Jonas Niemann, Paschalis Tossios, Yessica Schneider, Sabine Rothmann, Frank-Mathias Gutzki, Jürgen C. Frölich, Dirk O. Stichtenoth

Institute of Clinical Pharmacology, Hannover Medical School, Hannover, Germany

article info abstract

Article history: Prostaglandin (PG) H synthases (PGHS) or (COX) catalyse the peroxidation of arachi- Received 30 August 2012 donic acid (AA) to PGG2 and PGH2 which are further converted to a series of prostaglandins and Accepted 3 September 2012 . Here, we report that GSH promotes concomitant formation of the current oxida- Available online 13 September 2012 tive stress biomarkers malondialdehyde (MDA) and 15(S)-8-iso-prostaglandin F2a from AA via PGHS. This illustrates an uncommon interplay of enzymatic and chemical reactions to produce species that Edited by Barry Halliwell are considered to be exclusively produced by free-radical-catalysed reactions. We propose mecha-

nisms for the PGHS/AA/GSH-dependent formation of MDA, 15(S)-8-iso-prostaglandin F2a and other Keywords: F -isoprostanes. These mechanisms are supported by clinical observations. 2 Biomarker Ó 2012 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Cyclooxygenase Glutathione Prostaglandin H synthase

1. Introduction (NAC). On the other hand, conversion of endogenously produced substances, such as the arachidonic acid-derived epoxide leukotri-

Glutathione (GSH) is one of the most important and most ene A4 (LTA4) to the cysteinyl leukotriene C4 (LTC4), requires only abundant intra-cellular endogenously produced antioxidants. The very low GSH amounts. This also applies to the prostaglandin chemical reaction of the sulfhydryl (SH) group of GSH with oxi- (PG) H synthase (PGHS)-catalysed oxidation of arachidonic acid

dants such as hydrogen peroxide (H2O2) yields glutathione disul- (AA) to a variety of PGs and thromboxane A2 (TxA2) which are pres- phide (GSSG) and water. GSSG is also formed by transition metal ent in pM-concentrations in biological fluids and tissues [5]. ions-catalysed oxidation of GSH. In well-functioning cells, GSSG The PGHS isoforms, PGHS-1 and PGHS-2, are commonly is effectively reduced to GSH by glutathione reductase, so that referred to as cyclooxygenase-1 (COX-1) and cyclooxygenase-2 the molar ratio of GSH to GSSG (GSH/GSSG) is in great favour of (COX-2), respectively. It has been demonstrated that GSH is oxi- GSH, e.g., about 750:1 in human blood [1] and 200:1 in human dised to its thiyl free radical metabolite by PGHS [6]. Also, GSH is erythrocytes [2]. The intra-cellular GSH/GSSG molar ratio is there- involved in the metabolism of PGs and can inhibit PGHS activity

fore a widely used quantitative parameter of oxidative stress in [5,7]. Especially in human , formation of TxA2 has been health and disease [3,4]. reported to be accompanied by temporary formation of GSSG, Yet, GSH is not merely an antioxidant but it also serves as a malondialdehyde (MDA) and the complementary 12-hydroxy- cofactor/substrate for various enzymes, for instance for GSH 5,8,10-heptadecatrienoic acid (HHTrE) in considerable amounts S-transferases. In most of these cases GSH is consumed without [8–11]. Yet, the mechanisms leading to GSSG formation during forming GSSG. In case of intoxication by certain xenobiotica, such PGHS-catalysed AA peroxidation are incompletely understood.

as paracetamol (acetaminophen) overdose, considerable amounts Despite increasingly use of F2-isoprostanes including 15(S)-8- of GSH are consumed. In such cases GSH must be substituted by iso-prostaglandin F2a (15(S)-8-iso-PGF2a) as biomarkers of oxida- intravenous administration of large amounts of N-acetylcysteine tive stress [12], MDA is still one of the most widely measured biomarkers of oxidative stress [3,4]. In accordance with the cur- rently prevailing theory of oxidative stress, the antioxidant GSH ⇑ Corresponding author. Address: Institute of Clinical Pharmacology, Hannover should prevent . Consequently, GSH should keep Medical School, Carl-Neuberg-Str. 1, 30625 Hannover, Germany. Fax: +49 511 532 low the concentration of oxidative stress biomarkers. The relation- 2750. E-mail address: [email protected] (D. Tsikas). ship between MDA, 15(S)-8-iso-PGF2a, PGHS and GSH, and its

0014-5793/$36.00 Ó 2012 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.febslet.2012.09.001 3724 D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730

A A

B B

C

C

Fig. 2. PGHS-1-dependent formation of GSSG upon incubation of GSH in the assay buffer under various conditions. (A) GSSG formation from GSH (100 lM) upon a 5- min pre-incubation with various PGHS-1 concentrations followed by addition of AA either at 0 lMor10lM and subsequent incubation for 30 min. (B) GSSG formation from GSH (100 lM) upon a 5-min pre-incubation with PGHS-1 (10 U), AA (10 lM) and subsequent incubation for 30 min. (C) GSSG formation from GSH (100 lM) upon a 5-min pre-incubation with PGHS-1 (10 U), acetylsalicylic acid (ASA, 1000 lM) or diclofenac (Diclo, 1 lM) followed by addition AA (10 lM) and subsequent incubation for 30 min. Data are shown as the mean from two Fig. 1. Formation of MDA (A) and 15(S)-8-iso-PGF2a (B) in incubation mixtures of AA (10 lM) and PGHS-1 (1.4 nM) or PGHS-2 (10.7 nM) at varying GSH concentra- independent experiments (A and B) and as mean ± S.D. from three independent tions (0–1000 lM) at 37 °C for 30 min under aerobic conditions. (C) Linear experiments (C). relationship between MDA and 15(S)-8-iso-PGF2a measured in this experiment. Data are shown as mean ± S.D. from three independent experiments (A) and as the mean from two independent experiments (B).

Paradoxically, we observed that GSH does not prevent but it pro-

motes PGHS-dependent MDA and 15(S)-8-iso-PGF2a formation significance in the context of oxidative stress are elusive. In the from AA in vitro. Based on these findings we propose mechanisms present study, we investigated a possible role of GSH in the forma- for the interaction between GSH and PGHS in the peroxidation of tion of PGHS-dependent formation of MDA and 15(S)-8-iso-PGF2a. AA to MDA, 15(S)-8-iso-PGF2a, TxA2 and other metabolites. D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730 3725

2. Materials and methods intravenously administered NAC prevented formation of reactive oxygen and reactive nitrogen species in transmural left ventricular 2.1. Materials biopsy specimens [16]. In that study patients had been treated with NAC (NAC group; prime 100 mg/kg into cardiopulmonary by- 2 1 1 Arachidonic acid, [5Z,6,8Z,9,11Z,12,14Z,15- H8]-arachidonic pass, followed by infusion 20 mg kg h ) or placebo (CON 2 acid (d8-AA; 99.5 atom% H), acetylsalicylic acid (ASA), diclofenac, group) immediately before (NAC0h, CON0h) and at the end of the ibuprofen and indomethacin were obtained from Sigma (Munich, surgery (3 h; NAC3h, CON3h), as well as 24 h after surgery Germany). Recombinant ovine PGHS-1 (http://www.caymanchem. (NAC24h, CON24h). In urine samples collected from this study, com/pdfs/60100.pdf) and recombinant ovine PGHS-2 (http://www. 15(S)-8-iso-PGF2a and PGE2 (i.e., 15(S)-PGE2) were quantified by 2 caymanchem.com/pdfs/60120.pdf), H4-15(S)-8-iso-PGF2a, other GC–MS/MS after IAC. Data from this study are shown as mean 2 2 H-labelled prostaglandins and H-labelled thromboxane B2, and ± S.E.M. Statistical significance was tested by t test. immunoaffinity column chromatography (IAC) cartridges with immobilized antibodies raised against 15(S)-8-iso-PGF2a (http:// 3. Results www.caymanchem.com/app/template/Product.vm/catalog/400058) or PGE2 (https://www.caymanchem.com/pdfs/500450.pdf) were 3.1. In vitro studies obtained from Cayman Chemicals (Ann Arbor, MI, USA). These materials were used as received and in accordance with the Fig. 1 shows that formation of MDA and 15(S)-8-iso-PGF2a from manufacturer’s recommendations. PGHS-catalysed peroxidation of AA is dependent upon GSH. In the PGHS-1 and PGHS-2 incubation mixtures, MDA concentrations

2.2. In vitro studies were 200–250 times higher than 15(S)-8-iso-PGF2a concentrations (Fig. 1C). There were highly linear relationships (r > 0.9) between

All PGHS-catalysed reactions and respective controls were per- MDA and 15(S)-8-iso-PGF2a concentrations for both isozymes. Be- formed under aerobic conditions at 37 °C in 100 mM sodium/ cause MDA is a C3-compound (i.e., C3H4O) and 15(S)-8-iso-PGF2a is potassium phosphate buffer (pH 8.0) that contained 2 mM phenol, a C20-compound (i.e., C20H34O5), MDA and 15(S)-8-iso-PGF2a must 5 mM EDTA and 1 lM hematin. These conditions are recom- originate from two AA (a C20-compound, i.e., C20H32O2) molecules. mended by the manufacturer and were proven to be optimal in In agreement with previous reports [5,7,17], we found that GSH our experiments. The reaction was started by adding AA at a final concentration of 10 lM, which is close to the KM values towards PGHS-1 (10 lM) and PGHS-2 (2.5 lM) as observed in preliminary A experiments. Reaction mixtures were incubated for 30 min or as reported in the respective experiments. For the sake of clarity, experimental conditions are detailed in the figure legends. In our study, all PGHS inhibitors were used at pharmacologically relevant concentrations being close to their reported IC50 values (see Section 3). For MDA analysis in PGHS incubation mixtures, 100-lL aliquots were taken from the reaction mixtures and immediately treated with 400-lL aliquots of ice-cold acetone to stop enzymatic reac- 2 tions. Subsequently, the internal standard [1,3- H2]-MDA was added (1 lM) and derivatization and GC–MS/MS analysis were performed as described elsewhere [13]. For 15(S)-8-iso-PGF2a anal- ysis 100-lL aliquots were taken from the reaction mixture and immediately acidified to pH 4.5. Subsequently, the internal standard 2 H4-15(S)-8-iso-PGF2a was added (1.75 nM), the analytes were ex- tracted by IAC, derivatized and analysed by GC–MS/MS [13,14]. Other prostaglandins were also analysed by GC–MS/MS [14] after B solvent extraction with ethyl acetate from acidified (pH 4.5) samples using their respective 2H-labelled analogues (1 nM). GSSG was quantitated by HPLC with UV absorbance detection at 205 nm [15]. Data from other experiments are shown as the mean from two independent experiments or mean ± S.D. from three independent experiments.

2.3. In vivo studies

To examine the relationship between plasma concentrations of the oxidative stress biomarkers MDA and 4-hydroxy-2-nonenal (HNE), we analysed plasma samples from 15 males and five fe- males subjects (age 68 ± 8 years) suffering from peripheral arterial occlusive disease. These samples were obtained on 2 days with an interval of 90 days between. Plasma MDA, HNE and total

15(S)-8-iso-PGF2a (i.e., free + esterified fractions) were measured by GC–MS/MS [13,14]. The study was approved by the Ethics Com- Fig. 3. (A) PGHS-1 (1.4 nM) and (B) PGHS-2 (10.7 nM) dependent formation of MDA from AA (10 lM) upon incubation (30 min) of GSH at 0 lM (-GSH) or 100 lM in the mittee of the Hannover Medical School. absence (Control) or in the presence of the PGHS inhibitors acetylsalicylic acid (ASA; In a randomized, placebo-controlled clinical trial in patients 1000 lM), diclofenac (Diclo; 1 lM), ibuprofen (Ibu; 1 lM) or indomethacin (Indo; undergoing cardiac surgery study we showed previously that 1 lM). Data are shown as mean ± S.D. from three independent experiments. 3726 D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730

2 Fig. 4. GC–MS spectrum of the pentafluorobenzyl derivative of the reaction product obtained from the incubation of 10 lM [5,6,8,9,11,12,14,15- H8]-arachidonic acid (d8-AA) with PGHS-2 (10.7 nM) in the presence of GSH (100 lM) for 30 min at 37 °C. The insert on the left shows the GC–MS spectrum of the pentafluorobenzyl derivative of synthetic d2-MDA [13]. Both mass spectra were obtained from the gas chromatographic peak with the same GC retention time of 5.9 min. GSH mediated PGHS-2-catalysed formationof 2 two distinctly H-labelled MDA species, i.e., d1-MDA (m/z 252) and d2-MDA (m/z 253). d8-AA was used immediately after its receipt from the supplier. inhibited PGHS-1 and PGHS-2 activity up to 90% when measured as the COX inhibitors tested. Thus, PGHS-catalysed formation of

PGE2 formation rate (data not shown). MDA could not be completely blunted by the irreversible COX In GSH-containing PGHS incubation mixtures, we observed for- inhibitor ASA (1000 lM; IC50 0.75 and 1.25 mM for the recombi- mation of GSSG (Fig. 2). GSSG formation required presence of both, nant PGHS-1 and PGHS-2 isoforms), by the reversible COX inhibi-

PGHS and AA, and was dependent upon PGHS concentration and tors diclofenac (1 lM; IC50 60–220 nM), ibuprofen (1 lM; IC50 incubation time (Fig. 2). Substitution of GSH by cysteine (Cys) re- 9 lM) or indomethacin (1 lM; IC50 0.1–6 lM) (Fig. 3). ASA and vealed formation of cystine, i.e., the Cys disulphide, at concentra- indomethacin inhibited PGHS-2-catalysed formation of MDA with tions similar to those of GSSG (i.e., up to 4 lM; data not shown). no effect on PGHS-1. By contrast, ibuprofen inhibited PGHS-1- These observations suggest that the effects of GSH and Cys are catalysed formation of MDA but was inactive towards PGHS-2. chemical in nature. In the presence of the irreversible PGHS inhib- COX inhibitors exert their action on different sites of the PGHS itor acetylsalicylic acid (ASA, 1000 lM), GSSG was formed at lower enzymes and inhibit their activity by different mechanisms. The concentrations (Fig. 2C). In contrast, in the presence of the revers- distinct effects of the COX inhibitors tested in this study may ible PGHS inhibitor diclofenac (1 lM), GSSG formation occurred indicate that GSH intervenes at different stages of the multi-step and was comparable to control, i.e., in the absence of any PGHS and complex process of PGHS-catalysed peroxidation of AA (see inhibitor (Fig. 2C). Thus, GSH oxidation to GSSG occurs during Section 4 and Supplement). PGHS-catalysed oxidation of AA, presumably during the peroxidase To gain mechanistic insight into the PGHS-dependent, GSH- reaction of PGHS [11], and not simply by GSH autoxidation. mediated formation of MDA from AA, we investigated the In vitro, PGHS-catalysed and GSH-promoted formation of MDA PGHS-2-catalysed oxidation of the deuterium-labelled [5,6,8, 2 and 15(S)-8-iso-PGF2a (not shown) was influenced differently by 9,11,12, 14,15- H8]-arachidonic acid (d8-AA). By means of a D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730 3727

A intense ions at m/z 252 and m/z 253 (Fig. 4), implies rearrangement of the double bond system of AA during its oxidation by PGHS-2 and/or during the subsequent steps. This is supported by the

observation of PGE2, AA and HHTrE labelled with 8, 7 and 6 deuterium atoms, and by the formation of deuterium-labelled trans-arachidonic acid (Figs. S1–S3). trans-Arachidonic acid iso- mers have been shown to occur physiologically in human plasma [18] and to be formed from arachidonic acid added to human platelets [19].

3.2. In vivo studies

Like in vitro with the recombinant PGHS isoforms (Fig. 1C), in vivo in plasma samples of humans we observed a close correla-

tion between MDA and 15(S)-8-iso-PGF2a concentrations, as well as between HNE and MDA concentrations (Fig. 5). That MDA and HNE occur in human plasma at concentrations of the same order of magnitude may suggest that MDA and HNE are formed from the same AA molecule by a common PGHS/AA/GSH-involving B mechanism (see Section 4). Many in vivo and in vitro studies reported that the F2-isopros- tane 8-iso-PGF2a is produced by PGHS [19–28], while PGF2a, which is generally accepted to be exclusively produced by PGHS, is pro- duced independently of PGHS [29]. In urine samples collected from a previously performed study [30] on 15 healthy young volunteers who had received celecoxib, a PGHS-2 selective inhibitor, indo- methacin, a non-selective PGHS inhibitor, or placebo, we measured

retrospectively 15(S)-8-iso-PGF2a. We found that excretion of 15(S)-8-iso-PGF2a was reduced tendentiously both by celecoxib (32.7 ± 3.17 nmol/mol creatinine; P = 0.097) and indomethacin (34.9 ± 3.74 nmol/mol creatinine; P = 0.162) compared to placebo (45.5 ± 7.25 nmol/mol creatinine). This observation collaborates with previous findings [19–28] and argues for PGHS-catalysed for-

mation of 15(S)-8-iso-PGF2a. NAC, a precursor of GSH, and ascorbic acid (vitamin C) have been widely used as antioxidants with contradictory outcomes. For instance, NAC (10 mg/kg body weight) and ascorbic acid Fig. 5. Relationship between concentrations of (A) MDA and total 15(S)-8-iso-PGF2a (12.5 mg/kg body weight) increased the plasma concentration of (i.e., free and esterified to lipids) and of (B) HNE and MDA in plasma of 15 males and 15(S)-8-iso-PGF2a, suggesting enhancement of oxidative stress in five females subjects (age 68 ± 8 years) suffering from peripheral arterial occlusive disease. Note the different concentration units for MDA and HNE (i.e., lM), and for exercising humans [29]. An alternative explanation for the increase

15(S)-8-iso-PGF2a (i.e., nM). Plasma was analysed on 2 days with an interval of in plasma of 15(S)-8-iso-PGF2a concentration seen in the study by 90 days between. Insets are the regression equations, r and P values from linear Childs et al. [29] may be GSH- and/or ascorbic acid-promoted and regression analysis of (A) MDA (y) vs. 15(S)-8-iso-PGF2a (x) and of (B) HNE (y) vs. PGHS-mediated formation of 15(S)-8-iso-PGF2a. It is worth men- MDA (x). tioning that ascorbic acid may mimic the action of thiols on PGHS activity [31]. In patients undergoing cardiac surgery we showed that NAC specific GC–MS method [13] we identified two different deute- administration reduced tissue 3-nitrotyrosine content suggesting 2 rium-labelled MDA species: [1,3- H2]MDA (d2-MDA) and reduction of nitrative stress by NAC and/or its metabolites includ- 2 [1- H]MDA (d1-MDA) (Fig. 4). Formation from d8-AA of MDA with ing GSH [16]. In urine samples collected in that study were mea- two deuterium atoms (i.e., d2-MDA) and one deuterium atom (i.e., sured retrospectively 15(S)-8-iso-PGF2a and PGE2. Table 1 shows d1-MDA) at comparable concentration, as indicated by the equally that both NAC and placebo increased temporarily excretion of

Table 1 a b Creatinine-corrected urinary excretion of 15(S)-8-iso-PGF2a and 15(S)-PGE2 and their molar ratio in two groups of patients undergoing cardiac surgery . ‘‘NAC’’ group received NAC, while ‘‘CON’’ group receive placebo.

c Group 15(S)-8-iso-PGF2a (nmol/mol creatinine) 15(S)-PGE2 (nmol/mol creatinine) 15(S)-8-iso-PGF2a to 15(S)-PGE2 ratio P value for the ratio NAC0h 40.5 ± 10.7 (19) 103.1 ± 37.6 (18) 0.74 ± 0.16 (18) 0.837 (0 h vs. 3 h) NAC3h 76.3 ± 10.2 (19) 166.2 ± 30.2 (19) 0.71 ± 0.12 (19) 0.0015 (0 h vs. 24 h) NAC24h 46.1 ± 7.18 (19) 24.9 ± 5.24 (19) 4.65 ± 1.28 (18) 0.0018 (3 h vs. 24 h) CON0h 37.8 ± 8.02 (18) 54.5 ± 9.88 (17) 0.958 ± 0.192 (15) 0.3079 (0 h vs. 3 h) CON3h 57.2 ± 10.3 (18) 133.8 ± 30.7 (17) 0.712 ± 0.136 (17) 0.0052 (0 h vs. 24 h) CON24h 32.8 ± 3.70 (17) 18.8 ± 2.85 (18) 3.549 ± 1.376 (17) 0.0003 (3 h vs. 24 h)

a Data are shown as mean ± S.E.M. In parentheses, the number of considered patients is given. b Urine samples were collected at three different time points (0 h, 3 h and 24 h). c There was no significant difference between the groups at each time point. 3728 D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730

A

B

Fig. 6. Proposed mechanisms for the GSH-induced formation of MDA and HNE (A), and of PGF2a, 15(S)-8-iso-PGF2a and other F2-isoprostanes (B) during PGHS-catalysed peroxidation of AA. (A) GSH attacks the electropositive O atom at C-11 of the 9,11-peroxido group of an intermediate which, without the action of GSH, would provide PGG2.A second GSH molecule reduces the S atom of the intermediate G–S(±0)–O–RÅ (RÅ, is a radical residue) to yield GSSG, while RÅ rearranges to finally produce HNE. (B) In total six GSH molecules are involved in the attack on electropositive (d+) O atoms of the 9,11-peroxido and 15-peroxido groups of the same intermediate as in (A) and the subsequent reactions to produce three GSSG molecules and a AA-trihydroxy-diradical that rearranges to isomeric PGF2a, 15(S)-8-iso-PGF2a and other F2-isoprostanes.

15(S)-8-iso-PGF2a and PGE2 in the urine. It is remarkable that 4. Discussion 15(S)-8-iso-PGF2a values returned to baseline 24 h after surgery.

By contrast, PGE2 values decreased below baseline 24 h after The mechanisms by which 15(S)-8-iso-PGF2a, MDA, prostaglan- surgery. The effects were stronger in the NAC group (reduction dins including PGG2, PGH2, PGE2, PGD2 and PGF2a, and TxA2 are by 76%) than in the placebo group (reduction by 66%). As the pa- formed during the PGHS-catalysed peroxidation of AA are still tients of both groups had received several other drugs, it is possible incompletely understood. For TxA2 and MDA, Ullrich et al. pro- that the effects of NAC on PGHS-catalysed formation of 15(S)-8-iso- posed an oxene transfer mechanism analogous to that catalysed PGF2a and inhibition of PGE2 have been superposed by actions of by cytochrome P450 [32]. Pace-Asciak and Smith suggested a the drugs. transient formation of an electropositive O atom at the C-11 and D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730 3729 subsequent cleavage of the 9,11-peroxido group (see Ref. [11]). Be- formation of MDA and 15(S)-8-iso-PGF2a from AA challenges the cause MDA (C3H4O), 15(S)-8-iso-PGF2a (C20H34O5) and TxA2 utility of MDA and 15(S)-8-iso-PGF2a as reliable biomarkers of (C20H32O5) cannot be formed from a single AA molecule at the oxidative stress. Assessment of circulating and/or excretory MDA same time, several different mechanisms and intermediates are and 15(S)-8-iso-PGF2a, and possibly of HNE and HHTrE, may not likely to be involved and might better explain differences in be proper biomarkers of oxidative stress in certain conditions.

MDA, GSSG and TxA2 (actually its stable analogue TxB2) concentra- These cases may include diseases associated with haemolysis and tions measured in vitro and in vivo. As the mechanism proposed by platelets activation or inactivation, and administration of drugs Pace-Asciak and Smith are more compatible with the results re- that inhibit PGHS activity such as aspirin [8], or drugs that may ported in the present article, we extended this mechanism by affect GSH homoeostasis such as NAC, a GSH-prodrug, as shown including GSH as a chemical partner in the PGHS-catalysed perox- in the present study (Table 1). idation of AA. As demonstrated by us here for MDA and 15(S)-8-iso-PGF2a and In accordance with this mechanism (Fig. 6), nucleophilic attack by others elsewhere for MDA, 15(S)-8-iso-PGF2a and other of the GSH thiolate on the electropositive O atom at C-11 of the F2-isoprostanes and E2/D2-isoprostanes [33], many of the current 9,11-peroxido group of an AA intermediate produces concomi- oxidative stress biomarkers are formed, at least in part, in the tantly MDA and HNE in stoichiometric amounts (Fig. 6A). The per- course of enzyme-catalysed reactions, without being necessarily oxidised, not yet cyclised AA triradical intermediate is different associated with oxidative stress, and their production is promoted from PGG2, because GSH and other thiols do not convert PGG2 into by antioxidants including GSH. This is only rarely considered in this MDA (not shown), PGE2 (Fig. S5) and PGF2a (Fig. S5). On the other area. The duality of oxidative stress biomarkers, i.e., their chemical hand, formation of PGF2a, 15(S)-8-iso-PGF2a and other F2-isopros- and enzymatic origin, seems to be a general phenomenon. It tanes requires simultaneous nucleophilic attack of the SH groups represents a considerable methodological challenge and requires of two GSH molecules on both O atoms of the 9,11-peroxido group proper consideration of potentially involved pathways and their of the peroxidised not yet cyclised AA triradical intermediate ramifications. (Fig. 6B). This mechanism requires the combination of the two rad- ical C atoms at positions 8 and 12 (Fig. 6B). Such a ring closure is Appendix A. Supplementary data likely to produce several isomeric compounds such as 15(S)-8- iso-PGF2a and PGF2a, and possibly other F2-isoprostanes of the type Supplementary data associated with this article can be found, in 15-F2t-IsoP [12] after rearrangement of the AA-trihydroxy- the online version, at http://dx.doi.org/10.1016/j.febslet.2012. diradical. A similar mechanism may also apply to the other three 09.001. types of F2-isoprostanes, i.e., 5-F2t-IsoP, 8-F2t-IsoP and 12-F2t-IsoP. Finally, formation of TxA requires nucleophilic attack of the SH 2 References group of a GSH molecule on the electropositive O atom at C-9 of the 9,11-peroxido group of PGH2 to produce TxA2 (Fig. S6). Such [1] Rossi, R., Milzani, A., Dalle-Donne, I., Giustarini, D., Lusini, L., Colombo, R. and a GSH-promoted PGHS-catalysed formation of TxA2 is in line with Di Simplicio, P. (2002) Blood glutathione disulfide: in vivo factor or in vitro artifact? Clin. Chem. 48, 742–753. recent findings showing that extra- GSH may enhance TxA2 synthesis in platelet-rich plasma [17]. It is worth mentioning that [2] Michaelsen, J.T., Dehnert, S., Giustarini, D., Beckmann, B. and Tsikas, D. (2009) HPLC analysis of human erythrocytic glutathione forms using OPA and N- the stereochemistry of TxA2 is prescribed in PGH2 [11] but not in acetyl-cysteine ethyl ester: evidence for nitrite-induced GSH oxidation to the opened cyclopentane-ring structure proposed by Ullrich et al. GSSG. J. Chromatogr. B 877, 3405–3417. [32]. [3] Dalle-Donne, I., Rossi, R., Colombo, R., Giustarini, D. and Milzani, A. (2006) Biomarkers of oxidative damage in human disease. Clin. Chem. 42, 601–623. The proposed reactions and their stoichiometry, in which GSH [4] Giustarini, D., Dalle-Donne, I., Tsikas, D. and Rossi, R. (2009) Oxidative stress participates either as a reductor (reactions (1)–(3)) or as a cata- and human diseases: origin, link, measurement, mechanisms and biomarkers. lyst (reaction (4); ‘‘[GSH]’’ indicates that GSH is not consumed), Crit. Rev. Clin. Lab. Sci. 46, 241–281. [5] Buckley, B.J., Kent, R.S. and Whorton, A.R. (1991) Regulation of endothelial cell are summarised below (see also Fig. S7). Considerable GSH oxi- prostaglandin synthesis by glutathione. J. Biol. Chem. 266, 16659–16666. dation to GSSG would be associated with formation of MDA, [6] Eling, T.E., Curtis, J.F., Harman, L.S. and Mason, R.P. (1986) Oxidation of glutathione to its thiyl free radical metabolite by prostaglandin H synthase. J. 15(S)-8-iso-PGF2a, HNE and HHTrE. As MDA (and HNE) formation Biol. Chem. 261, 5023–5028. rate is at least two orders of magnitude higher than that of [7] Feldman, R., Luncsford, A., Heinrikson, R.L., Westley, J. and Jarabak, J. (1981) 15(S)-8-iso-PGF2a and other F2-isoprostanes, the highest GSH Glutathione-related inhibition of prostaglandin metabolism. Arch. Biochem. consumption rate in PGHS-catalysed reactions would result from Biophys. 211, 375–381. [8] Szczeklik, A., Gryglewski, R.J., Grodzin´ ska, L., Musiał, J., Serwon´ ska, M. and MDA formation (reactions (1) and (3)). The low-extent formation Marcinkiewicz, E. (1979) Platelet aggregability, thromboxane A2 and of GSSG in the presence of the irreversible PGHS inhibitor ASA malondialdehyde formation following administration of aspirin to man. suggests that this drug acts in such a way that peroxidation of Thromb. Res. 15, 405–413. AA is not completely inhibited even at the high concentration [9] Tomita, T., Umegaki, K. and Hayashi, E. (1983) Basis aggregation properties of washed rat platelets: correlation between aggregation, phospholipid of 1000 lM. degradation, malondialdehyde, and thromboxane formation. J. Pharmacol. Methods 10, 31–44. AA-Triradical þ 2GSH ! MDA þ HNE þ GSSG ð1Þ [10] Burch, J.W. and Burch, P.T. (1990) Glutathione disulfide production during arachidonic acid oxygenation in human platelets. Prostaglandins 39, 123–134. AA-Triradical þ 6GSH ! 15ðSÞ-8-iso-PGF2a þ 3GSSG ð2Þ [11] Smith, W.L., Marnett, L.J. and DeWitt, D.L. (1991) Prostaglandin and AA-Monoradical þ 2GSH ! MDA þ HHTrE þ GSSG ð3Þ thromboxane biosynthesis. Pharmacol. Ther. 49, 153–179. [12] Schwedhelm, E., Benndorf, R.A., Böger, R.H. and Tsikas, D. (2007) Mass PGH2 þ½GSH!TxA2 þ½GSHð4Þ spectrometric analysis of F2-isoprostanes: markers and mediators in human disease. Curr. Pharmacol. Anal. 3, 39–51. In consideration of the antioxidant feature of GSH, the promo- [13] Dreissigacker, U., Suchy, M.T., Maassen, N. and Tsikas, D. (2010) Human tion of PGHS-dependent formation of MDA and 15(S)-8-iso-PGF2a plasma concentrations of malondialdehyde (MDA) and the F2-isoprostane from AA by GSH is paradoxical. This observation illustrates a 15(S)-8-iso-PGF(2alpha) may be markedly compromised by hemolysis: evidence by GC–MS/MS and potential analytical and biological ramifications. paradigm for an uncommon cooperation of enzymatic and chemi- Clin. Biochem. 43, 159–167. cal reactions to yield unexpected reaction products. MDA and [14] Tsikas, D., Schwedhelm, E., Suchy, M.T., Niemann, J., Gutzki, F.M., Erpenbeck, V.J., Hohlfeld, J.M., Surdacki, A. and Frölich, J.C. (2003) Divergence in urinary 8- 15(S)-8-iso-PGF2a are generally regarded as stable products formed iso-PGF(2alpha) (IPF(2alpha)-III, 15-F(2t)-IsoP) levels from gas by the reaction of free radicals and AA esterified to lipids [3,4]. That chromatography–tandem mass spectrometry quantification after thin-layer the antioxidant GSH promotes recombinant PGHS-dependent chromatography and immunoaffinity column chromatography reveals 3730 D. Tsikas et al. / FEBS Letters 586 (2012) 3723–3730

heterogeneity of 8-iso-PGF(2alpha). Possible methodological, mechanistic and [24] Jourdan, K.B., Evans, T.W., Goldstraw, P. and Mitchell, J.A. (1999) Isoprostanes clinical implications. J. Chromatogr. B 794, 237–255. and PGE2 production in human isolated artery smooth muscle cells: [15] Tsikas, D., Denker, K. and Frölich, J.C. (2001) Artifactual-free analysis of S- concomitant and different release. FASEB J. 13, 1025–1030. nitrosoglutathione and S-nitroglutathione by neutral-pH, anion-pairing, high- [25] Sautebin, L., Ianaro, A., Rombolà, L., Ialenti, A., Sala, A. and Di Rosa, M. (1999) performance liquid chromatography. Study on peroxynitrite-mediated S- Cyclooxygenase-2-depedent generation of 8-epiprostaglandin F2alpha by nitration of glutathione to S-nitroglutathione under physiological conditions. lipopolysaccharide-activated J774 macrophages. Inflamm. Res. 48, 503–508. J. Chromatogr. A 915, 107–116. [26] Watkins, M.T., Patton, G.M., Soler, H.M., Albadawi, H., Humphries, D.E., Evans, [16] Tossios, P., Bloch, W., Huebner, A., Raji, M.R., Dodos, F., Klass, O., Suedkamp, M., J.E. and Kadowaki, H. (1999) Synthesis of 8-epi-prostaglandin F2a by human Kasper, S.M., Hellmich, M. and Mehlhorn, U. (2003) N-Acetylcysteine prevents endothelial cells: role of synthase. Biochem. J. 344, 747–754. reactive oxygen species-mediated myocardial stress in patients undergoing [27] McAdam, B.F., Mardini, I.A., Habib, A., Burke, A., Lawson, J.A., Kapoor, S. and cardiac surgery: results of a randomized, double blind, placebo-controlled FitzGerald, G.A. (2000) Effect of regulated expression of human clinical trial. J. Thorac. Cardiovasc. Surg. 126, 1513–1520. cyclooxygenase isoforms on eicosanoids and isoeicosanoids production in [17] Tsikas, D. and Niemann, J. (2012) Nitric oxide, peroxynitrite, S-nitrosothiols inflammation. J. Clin. Invest. 105, 1473–1482. and thiols are unlikely to exert their effects on recombinant cyclooxygenase-1 [28] Favreau, F., Petit-Paris, I., Hauet, T., Dutheil, D., Paret, Y., Mauco, G. and and cyclooxygenase-2 activity in vitro by modifying cysteine moieties. Nitric Tallineau, C. (2004) Cyclooxygenase 1-dependent production of F2- Oxide 26, 192–194. isoprostane and changes in redox status during warm renal ischemia– [18] Zghibeh, C.M., Gopal, V.R., Poff, C.D., Falck, J.R. and Balazy, M. (2004) reperfusion. Free Radic. Biol. Med. 36, 1034–1042. Determination of trans-arachidonic acid isomers in human blood plasma. [29] Childs, A., Jacobs, C., Kaminiski, T., Halliwell, B. and Leewenburgh, C. (2001) Anal. Biochem. 332, 137–144. Supplementation with vitamin C and N-acetyl-cysteine increases oxidative [19] Boulos, C., Jiang, H. and Balazy, M. (2000) Diffusion of peroxynitrite into the stress in humans after an acute muscle injury induced by eccentric exercise. human platelets inhibits cyclooxygenase via nitration of tyrosine residues. J. Free Radic. Biol. Med. 31, 745–753. Pharmacol. Exp. Ther. 293, 222–229. [30] Stichtenoth, D.O., Marhauer, V., Tsikas, D., Gutzki, F.M. and Frölich, J.C. (2005) [20] Pratico, D., Lawson, J.A. and FitzGerald, G.A. (1995) Cyclooxygenase- Effects of specific COX-2-inhibition on renin release and renal and systemic dependent formation of the isoprostanes, 8-epi prostaglandin F2alpha. J. prostanoid synthesis in healthy volunteers. Kidney Int. 68, 2197–2207. Biol. Chem. 270, 9800–9808. [31] Tsikas, D., Flentje, M., Niemann, J. and Modun, D. (2012) Antioxidants and [21] Bachi, A., Brambilla, R., Fanelli, R., Bianchi, R., Zucato, E. and Chiabrando, C. endothelial dysfunction in young and elderly people: is flow-mediated (1997) Reduction of urinary 8-epi-prostaglandin F2a during cyclooxygenase dilation useful to assess acute effects? Hypertension 60, e5. inhibition in rats but not in man. Br. J. Pharmacol. 121, 1770–1774. [32] Ullrich, V., Zou, M.H. and Bachschmid, M. (2001) New physiological and [22] Klein, T., Reutter, F., Schweer, H., Seyberth, H.W. and Nüsing, R.M. (1997) pathological aspects on the thromboxane A2–prostacyclin regulatory system. Generation of the isoprostane 8-epi-prostaglandin F2a in vitro and in vivo via Biochem. Biophys. Acta 1532, 1–14. the cyclooxygenases. J. Pharmacol Exp. Ther. 282, 1658–1665. [33] Morrow, J.D., Roberts, L.J., Daniel, V.C., Awad, J.A., Mirochnitchenko, O., Swift,

[23] Maclouf, J., Folco, G. and Patrono, C. (1998) Eicosanoids and iso-eicosanoids: L.L. and Burk, R.F. (1998) Comparison of formation D2/E2-isoprostanes and F2- constitutive, inducible and transcellular biosynthesis in vascular disease. isoprostane in vitro and in vivo – effects of oxygen tension and glutathione. Thromb. Haemost. 79, 691–705. Arch. Biochem. Biophys. 353, 160–171.