Iso[7]LGD2−Protein Adducts Are Abundant in Vivo and

Apr 1, 2004 - free radical-induced oxidation of 1-palmitoyl-2-arachidonyl-sn-glycero-3-phosphocholine in vitro. We also show that iso[7]LGD2-protein a...
2 downloads 0 Views 225KB Size
Download PDF
Chem. Res. Toxicol. 2004, 17, 613-622

613

Iso[7]LGD2-Protein Adducts Are Abundant in Vivo and Free Radical-Induced Oxidation of an Arachidonyl Phospholipid Generates This D Series Isolevuglandin in Vitro Eugenia Poliakov, Susan Gillette Meer, Subhas C. Roy, Clementina Mesaros, and Robert G. Salomon* Department of Chemistry, Case Western Reserve University, Cleveland, Ohio 44106-7078 Received September 14, 2003

Isolevuglandins (isoLGs) are a family of γ-ketoaldehydes, aka isoketals or neuroketals, that are generated by free radical-induced oxidation of polyunsaturated fatty acid-containing lipids. Because of their high reactivity toward -amino groups of lysyl residues, isoLGs are found as protein adducts in vivo. Plasma levels of isoLG-derived protein modifications are orders of magnitude higher than levels of the corresponding isoprostane. This suggests that while isoprostanes are rapidly cleared from the circulation, isoLG-protein adducts accumulate over the lifetime of the protein, which can be weeks, and this may provide a dosimeter for oxidant stress. We now confirm the postulated formation of the first D series isoLG, iso[7]LGD2, by free radical-induced oxidation of 1-palmitoyl-2-arachidonyl-sn-glycero-3-phosphocholine in vitro. We also show that iso[7]LGD2-protein adduct levels in blood are the highest known for an isoLG-derived epitope. They average 30-fold higher than isoLGE2-protein and 3-fold higher than iso[4]LGE2-protein levels. Similarly, iso[7]LGD2-derived epitope levels in oxidized low density lipoprotein are 20 times higher than isoLGE2-protein and five times higher than iso[4]LGE2-protein levels. Previous studies showed that plasma levels of protein-bound E series isoLGs, i.e., isoLGE2 and iso[4]LGE2, are elevated in individuals with atherosclerosis as compared with age-matched controls. Plasma iso[7]LGD2-protein immunoreactivity in individuals with atherosclerosis averages 8.5 ( 3.1 nmol/mL, significantly higher (P ) 0.01) than the 3.5 ( 0.1 nmol/mL in healthy controls. Plasma levels of iso[7]LGD2-protein adducts are strongly correlated with iso[4]LGE2- (r ) 0.933) and isoLGE2-protein adducts (r ) 0.877). This supports the hypothesis that isoLGs are generated in vivo by parallel competing radicalinduced pathways.

Introduction To acquire a fundamental understanding of atherogenesis and other biological sequelae of oxidative injury, knowledge of the molecular structures and chemistry of lipid oxidation products is crucial. Guided by model studies, we found that the prostaglandin endoperoxide PGH2 readily undergoes a novel rearrangement forming levulinaldehyde derivatives with prostaglandin side chains (1) that we named LGs. Our discovery of free radicalinduced nonenzymatic formation of LGE2 isomers (2) provided the first evidence that a family of eight structurally isomeric LG diastereomers, isoLGs,1 is formed from arachidonate-containing phospholipids by analogous rearrangements of four structurally isomeric endoperoxide intermediates (Figure 1). The endoperoxide intermediates in isoLG biosynthesis are also putative precursors of isoPs, e.g., 8-epi-PGF2R (Figure 1). An important difference distinguishes prostaglandins and isoPs from LGs and isoLGs. Many PGs and isoPs are readily isolable from biological tissue and fluids. In contrast, free LGs or isoLGs have eluded detection in biological samples. A high proclivity for * To whom correspondence should be addressed. Tel: 216-368-2592. Fax: 216-368-3006. E-mail: [email protected].

covalent adduction results in rapid sequestration by proteins. LGE2 binds avidly (3) with the -amino group of protein lysyl residues forming Schiff bases (4) and pyrroles (5) that are readily oxidized (6) to hydroxylactam (HL) end products (Figure 1). Covalent protein modification by LGs or isoLGs may contribute to the pathological consequences of oxidant stress. Receptor recognition of LG or isoLG-derived epitopes can trigger immune responses and endocytosis. Autoantibodies recognizing iso[4]LGE2-protein adducts were detected in human blood (7). LGE2-modified LDL competes with oxLDL for receptor recognition and endocytosis by macrophage cells (8), pivotal steps in foam 1 Abbreviations: AA, arachidonic acid; apoB, apolipoprotein B; BCA, bicinchoninic acid; BHT, butylated hydroxytoluene; BSA, bovine serum albumin; CEO, chicken egg ovalbumin; CHP, carboxyheptylpyrrole; DHA, docosahexaenoic acid; DMAB, p-(N,N-dimethylaminobenzaldehyde); DPM, decompositions per minute; DTPA, diethylenetriaminepentaacetic acid; EDTA, ethylenediaminetetraacetate; EPA, eicosapentenoic acid; ESI, electrospray ionization; ETA, eicosatrienoic acid; FPLC, fast protein liquid chromatography; HL, hydroxylactam; HNE, (E)-4-hydroxy-2-nonenal; HSA, human serum albumin; IgG, immunoglobin G; isoLGs, isolevuglandins; isoPs, isoprostanes; KLH, keyhole limpet hemocyanin; LA, linoleic acid; LDL, low density lipoprotein; LG, levuglandin; LSC, liquid scintillation counting; MPO, myeloperoxidase; MRM, multiple reaction monitoring; ON, 4-oxononanal; oxLDL, oxidatively modified low density lipoprotein; PA-PC, 1-palmitoyl-2arachidonyl-sn-glycero-3-phosphocholine; PC, phosphatidylcholine.

10.1021/tx034185+ CCC: $27.50 © 2004 American Chemical Society Published on Web 04/01/2004

614

Chem. Res. Toxicol., Vol. 17, No. 5, 2004

Poliakov et al.

Figure 1. Free radical-induced oxidation of PA-PC is postulated to generate a family of eight structurally isomeric isoLG diastereomers through isoprostanoid endoperoxide (isoPGH2) intermediates. HL end products are formed by the reaction of isoLGs with protein lysyl residues.

cell formation and atherogenesis. Some LG and isoLGbased oxidative protein modifications can cement proteinprotein (9, 10) and protein-DNA cross-links (11). Mean plasma levels of protein-bound isoLGE2 are elevated in individuals with atherosclerosis (306 pmol/ mL) as compared with age-matched controls (121 pmol/ mL) (7). Similarly, mean plasma levels of the isoP, 8-epiPGF2R, are higher in individuals with atherosclerosis (1.1 pmol/mL) as compared with age-matched controls (0.35 pmol/mL) (12). In view of the common intermediacy of isoPGH2-PC in the biosynthesis of both isoLGE2 and 8-epi-PGF2R (Figure 1), it is noteworthy that plasma levels of the isoLG are orders of magnitude higher than those of the isoP. Two factors may underlie this difference. First, the half-life of isoPs in blood is limited by their rapid metabolism and excretion. For example, greatly increased plasma levels of 8-epi-PGF2R, caused by oxidative stress associated with myocardial reperfusion, drop precipitously within minutes (13). Second, because isoLGs rapidly bind to proteins, their half-lives are expected to correspond to those of the proteins, and many plasma proteins have half-lives of several weeks (14). Thus, isoP levels provide a snapshot of oxidant stress in vivo over a relatively brief period of time while isoLG-protein adducts can accumulate over days or weeks providing an integrated assessment of exposure. Ten years ago, we postulated the existence of an isoLG pathway (15) and later showed that isoLGE2-protein (2) and iso[4]LGE2-protein adducts (16) are generated upon oxidation of LDL in vitro. According to our isoLG hypothesis, hydrogen atom abstraction from the 13-, 10-, or 7-positions in the arachidonyl group in PA-PC leads

to isoLGE(D)2-PC, iso[4]LGE(D)2-PC, and iso[7]LGE(D)2-PC, respectively (Figure 1). We now confirm the postulated formation of an iso[7] and a D series LG isomer. Thus, iso[7]LGD2-PC is generated upon free radical-induced oxidation of PA-PC in vitro. Polyclonal rabbit antibodies were raised against iso[7]LGD2-derived protein epitopes to facilitate their detection in vivo. The notably low structural selectivity exhibited by these iso[7]LGD2-protein antibodies indicates a role for multiple functional group-immunoglobin interactions in the highly selective recognition of isoLGE2 and iso[4]LGE2 epitopes by the corresponding antibodies that we observed previously. The anti-iso[7]LGD2-protein antibodies cross-react 31% with isoLGE2- and 9% with PGA2-protein adducts. Nevertheless, because mean levels of iso[7]LGD2-derived protein immunoreactivity are higher than those of any isoLG examined previously, these antibodies provided good presumptive evidence for the presence of iso[7]LGD2-derived epitopes in human plasma and oxidized LDL. While free isoPs can be rapidly cleared form the circulation, cyclopentenone isoPs have the potential to form protein adducts that might accumulate. However, immunological evidence now shows that PGA2-protein adducts are unstable. Therefore, they are not likely to accumulate in vivo or to interfere with immunological detection of iso[7]LGD2-protein epitopes in biological samples.

Experimental Procedures General Methods. Absorbance values were measured on a Bio-Rad microplate reader using dual wavelengths (405 nm to

Iso[7]LGD2-Protein Adducts Are Abundant in Vivo read the plate and 650 nm as a reference). For all ELISAs, unless otherwise noted, duplicates of each sample were run on the same plate. Concentrations of binding inhibitor are expressed in units of pmol/well. Because 50 µL of sample or standard solutions was loaded into each well of the ELISA plates, this corresponds to pmol/50 µL. The concentration of protein-bound iso[7]LGD2-derived pyrrole was determined using Ehrlich’s reagent DMAB. The concentration of pyrrole adduct in iso[7]LGD2-protein adducts was estimated as described previously for protein-bound iso[4]LGE2-derived pyrrole using the equation [pyrrole (µmol)] ) 2.22 (absorbance at 586 nm) (16). Materials. Spectrapor membrane tubing (MW cutoff 14 000 No. 2) for dialysis was obtained from Fisher Scientific Co. The following commercially available materials were used as received: AA, DHA, EPA, ETA, LA, CEO (grade V, 99%), BSA (fraction V, 96-99%), HSA (fraction V), and disodium pnitrophenyl phosphate (Sigma, St. Louis, MO); PA-PC (Avanti Polar Lipids, Inc., Alabaster, AL); KLH (ICN Biochemicals); goat anti-rabbit IgG-alkaline phosphatase (Boehringer-Mannheim); MPO (Calbiochem, La Jolla, CA); and DMAB and GSH (Aldrich, WI). Synthetic PGA2 and PGE2 were purchased from Cayman Chemical Company (Ann Arbor, MI). [5,6,8,11,12,14,15-3H]PGE2 was purchased from NEN Life Science Products (Boston, MA); methoxy-d3-amine HCl was purchased from CDN Isotopes (Pointe-Claire, Quebec, Canada); CDCl3 was purchased from Norell Inc. (NJ). A kit for antibody purification by protein G column chromatography was purchased from Pierce. LDL was isolated from human plasma as described previously (17). LGE2-KLH (18) and iso[4]LGE2-KLH (16) antibodies as well as iso[4]LGE2-KLH, iso[4]LGE2-BSA, iso[4]LGE2-HSA, iso[7]LGD2, iso[7]LGD2 bismethoxime, i.e., 8-(2-aza-2-methoxyvinyl)-9-(2-aza-1-methyl-2-methoxyvinyl)-5-hydroxyheptadeca6,11-dienoic acid, and iso[7]LGD2-PC (19) were prepared as described previously. Iso[7]LGD2 bis-d3-methoxime was prepared analogously by treatment of iso[7]LGD2 with methoxyd3-amine HCl in pyridine. In Vitro Oxidation and Derivatization of PA-PC in Unilamellar Vesicles. Small unilamellar PA-PC vesicles (2 mg/ mL, 1 mL) were prepared in argon-sparged sodium phosphate buffer (50 mM, pH 7.4) containing 200 mM DTPA, as described previously (20). Vesicles (1 mL), diluted to a final concentration of 0.2 mg of lipid/mL, were oxidized by incubation at 37 °C for 7 h in the presence of 30 nM MPO, 100 mg/mL glucose, 100 ng/mL glucose oxidase, and 50 mM NaNO2 (21). Methoxylamine-hydrochloride (30% aqueous solution, v/v) was then added to a final concentration of 3% methoxylamine-HCl. The mixture was incubated for 45 min at room temperature, then treated with an equal volume of a 15% KOH in methanol solution for 30 min at 37 °C, and finally acidified to pH 4 with 1 N HCl. Derivatized fatty acids were extracted from the resulting solution (1 mL) by addition of NaCl (20 mg) and then 2.5 mL of 2-propanol:heptane:2 M acetic acid (40:10:1, v/v/v). The mixture was vortexed, and then heptane (2.5 mL) was added, and the mixture was again vortexed. Samples were centrifuged at 4 °C for 5-10 min at 3000 rpm, and the top (heptane) layer was collected. The bottom layer was reextracted with heptane (2.5 mL) and then a third time with hexane (2.5 mL). All heptane and hexane layers were combined and evaporated under nitrogen. The residue was resuspended in methanol (HPLC grade, 0.5 mL) and stored in an amber vial under argon at -20 °C until analysis. LC/ESI/MS/MS Analysis. Mass spectrometric analyses were performed on a Quatro II triple quadrupole mass spectrometer (Micromass, Inc.) interfaced with an HP 1100 HPLC (HewlettPackard). Derivatized oxidized lipids were resolved on an Ultrasphere ODS C18 column (2 mm × 150 mm, 5 mm, Beckman Instruments) by isocratic elution at 0.2 mL/min using acetonitrile/water (90:10, v/v) with 0.1% formic acid as the solvent. HPLC with on-line ESI tandem mass spectrometry (LC/ ESI/MS/MS) detection of iso[7]LGD2 methoxime derivatives was achieved in the positive ion mode with MRM by showing cochromatography of multiple characteristic parent f daughter

Chem. Res. Toxicol., Vol. 17, No. 5, 2004 615 ion transitions with an authentic standard sample of iso[7]LGD2 bismethoxime [mass-to-charge ratio (m/z) 393 f 331, 393 f 316, 393 f 222]. Iso[7]LGD2-KLH Antibodies and Immunoassays. 1. Iso[7]LGD2-KLH Antigen. A solution of iso[7]LGD2 (6 mg) in 1.2 mL of 10 mM, pH 7.4, PBS and 1.2 mL of KLH (10.2 mg/ mL) was incubated at room temperature for 1 h followed by two successive 24 h dialyses against 1 L of 10 mM PBS (pH 7.4). The final concentration of protein was 3.5 mg/mL, and the pyrrole concentration was 6.46 mM. 2. Iso[7]LGD2-BSA Coating Agent. A solution of iso[7]LGD2 (8.4 mg) in 1 mL of 10 mM PBS (pH 7.4) was incubated with BSA (35 mg) at 37 °C for 12 h. The resulting mixture was dialyzed (MW cutoff 10 000) against 10 mM PBS (2 × 1 L) for 48 h. The concentration of pyrrole was 21.3 mM. The final concentration used for ELISA was 2 µg/mL of protein (1:1000). 3. Iso[7]LGD2-HSA Standard. A solution of iso[7]LGD2 (4.2 mg) in 1 mL of 10 mM PBS (pH 7.4) was incubated with HSA (35 mg) at 37 °C for 16 h. The resulting mixture was dialyzed (MW cutoff 14 000) against 10 mM PBS (2 × 1 L) for 48 h. The concentration of pyrrole was 12.5 mM. The concentration of protein was 30.2 mg/mL. Immunization. The immunogen, an iso[7]LGD2-KLH adduct (0.7 mg) containing 1.85 µmol of iso[7]LGD2/mg of KLH in 10 mM PBS (200 µL), was emulsified with Freund’s complete adjuvant (200 µL). A Pasturella free, New Zealand white rabbit (Hazelton) was inoculated intradermally into several sites on the back (125 µL) and rear leg (125 µL). Booster injections of iso[7]LGD2-KLH (200 µL) in Freund’s incomplete adjuvant (200 µL) were given every 21 days. Antibody titer was monitored 10 days after each inoculation by ELISA as described before (18). A strong immunogenic response was seen within a month. Booster injections and blood collections were continued until the antibody titer remained fairly constant. Serum was assayed (1:10 000), and after 90 days, all blood was collected. Antibody Purification. The crude antibody serum from the bleeding with the highest titer (70 day bleeding) was purified on a protein G column. A disposable (5 mL) polypropylene column was charged with immunopure immobilized protein G gel (2 mL) and was equilibrated with 5 column volumes of immunopure IgG binding buffer (pH 5). Binding buffer (2 mL) was added to the crude serum (2 mL), and the resulting mixture was vortexed for 1 min. The solution then was applied to the equilibrated column and allowed to flow completely into the gel. It was then washed with binding buffer until the absorbance of the eluant at 280 nm was less then 0.05. The IgG protein was then eluted with IgG elution buffer (pH 2.6). All eluates were collected into fractions (1 mL) containing 1.0 M Tris-HCl buffer (pH 8.8, 200 µL). The fractions containing IgG proteins were pooled and dialyzed against 10 mM PBS (pH 7.4, 2 × 500 mL) for 24 h at 5 °C. The resulting protein concentration was determined by the micro BCA method (Pierce, IL) (22). The final concentration of antibody for ELISA was e1 µg/mL. Antibody Titers. For determination of antibody levels in rabbit blood serum, iso[7]LGD2-BSA containing 41 mol pyrrole per mol protein was used as coating agent. The iso[7]LGD2BSA conjugate (100 µL of a solution containing 2 µg/mL in pH 7.4 PBS) was added to each well of a sterilized Baxter ELISA plate. The plate was then incubated at 37 °C for 1 h in a moist chamber. After the coating solution was discarded, each well was washed with PBS (3 × 300 µL), then filled with 1.0% CEO in PBS (300 µL), and incubated at 37 °C for 1 h to block remaining active sites on the plastic phase. Each well was washed with 0.1% CEO in PBS (300 µL) and then 100 µL of rabbit serum from each bleeding diluted 1:10 000 with 0.2% CEO in PBS or 0.2% CEO in PBS without serum for a blank was dispensed into the sample wells. Normal rabbit serum, i.e., prior to inoculation with antigen, diluted as above, was employed as a negative response control. The ELISA was completed as described previously (18). The antibody titer rose abruptly after 3 weeks, reaching a maximum within about 70 days.

616

Chem. Res. Toxicol., Vol. 17, No. 5, 2004

Competitive Antibody Binding Inhibition Studies. For iso[7]LGD2 antibodies, all binding inhibition studies were done as previously described with small modifications (18). Iso[7]LGD2-HSA was used as standard, and iso[7]LGD2-BSA (100 µL, 21 µM iso[7]LGD2 bound to BSA) was used as a coating agent (unless otherwise stated). Tween 20 (0.01%) was added to the blocking (1% CEO) and to the washing agent (0.1% CEO). A solution of antibody was prepared by dissolving the required amount of antibody in 0.2% CEO containing 0.01% Tween 20. The final concentration of antibody for ELISA was 1 µg/mL. For LGE2-KLH and iso[4]LGE2-KLH, ELISAs were done as described before (16, 18). Iso[7]LGD2-KLH Antibody Binding Study. 1. Iso[7]LGD2-HSA. A stock solution of iso[7]LGD2-HSA standard in 12.5 mM protein-bound iso[7]LGD2 was diluted 1:100 or 1:1000 with PBS to 6250 or 625 pmol/well and was used as a standard in an ELISA binding study for iso[7]LGD2-KLH antibodies. The solution of iso[4]LGE2-HSA in PBS was also used for an ELISA binding cross-reactivity study for LGE2-KLH, iso[4]LGE2KLH, and PGA2-KLH antibodies. To standardize the IC50 values between different plates, all IC50 values were scaled to set this standard value for iso[7]LGD2-HSA at 24.5 pmol/well for iso[7]LGD2-KLH antibodies for all cross-reactivity studies with 0.001% Tween 20. 2. LGE2-HSA, Iso[4]LGE2-HSA, and PGA2-HSA Adducts. A stock solution of LGE2-HSA in PBS (0.34 µmol protein-bound LGE2/mL PBS) was diluted with PBS to an appropriate concentration and was used as a standard for an ELISA binding study for LGE2-KLH antibodies. A stock solution of iso[4]LGE2-HSA standard in PBS (0.41 µmol protein-bound iso[4]LGE2/mL PBS) was diluted with PBS to an appropriate concentration and was used as a standard in an ELISA binding study for iso[4]LGE2-KLH antibodies. The same solutions of iso[4]LGE2-HSA and LGE2-HSA in PBS were also used in an ELISA binding cross-reactivity study for iso[7]LGD2KLH antibodies. A stock solution of PGA2-HSA standard in PBS (0.159 µmol protein-bound PGA2/mL PBS) prepared as described below was diluted to the appropriate concentration and used for cross-reactivity studies with iso[7]LGD2-KLH antibodies. Autoxidation of Polyunsaturated Fatty Acids (PUFAs) in the Presence of HSA. Fatty acid (i.e., AA, LA, EPA, ETA, 2 mg) and HSA (30 mg, 0.45 µmol) were dissolved in 0.1 M PBS (10 mL). Autoxidation was started by addition of 20 mM sodium ascorbate (510 µL) and 0.8 mM FeSO4‚7H2O (510 µL). The solutions were opened to air at room temperature for 30 min and then incubated at 37 °C for 24 h under air. After incubation, the reaction was quenched by adding 1 mM EDTA (200 µL) to each solution, which then was dialyzed against pH 7.4 PBS (2 × 2 L) for 40 h at room temperature. Samples of PUFAs that had been oxidized in the presence of HSA were analyzed for iso[7]LGD2-derived epitopes or cross-reacting epitopes by ELISAs using iso[7]LGD2-KLH antibodies in pH 7.4 PBS containing 0.001% Tween 20 and 0.2% CEO. Iso[7]LGD2-Derived Immunoreactivity in Oxidized LDL. LDL (0.5 mg/mL, 12 mL) was dialyzed for 5 h at 5 °C against pH 7.4 PBS (2 L), and then for 12 h against fresh buffer (2 L). The LDL was then dialyzed at 37 °C against 10 mM CuSO4 in PBS (1 L) (oxidation with simultaneous dialysis). Aliquots were removed periodically. The free radical-induced oxidation reaction in each aliquot was stopped by adding EDTA (2 mM, final concentration) and BHT (40 µM, final concentration). For determining the time dependence of appearance of particlebound pyrrole-derived immunoreactivity during the oxidation of LDL, the protein concentration in each aliquot of oxLDL was measured using the BCA assay (22). ELISA of the oxLDL with iso[7]LGD2-KLH antibodies was performed as for the competitive antibody binding inhibition studies described above. The starting concentration was the undiluted samples, and a dilution factor of 0.3 was employed. Iso[7]LGD2-Derived Immunoreactivity in Human Blood. Blood plasma was collected and stored as described previously

Poliakov et al. (7). ELISAs of plasma samples were performed as for the competitive antibody binding inhibition studies described above using the same standards and coating agents. Statistical Analysis. For statistical analysis of data, P values were calculated by independent t-test assuming equal variances using Excel 97 for Macintosh. Correlation and regression analyses were done also using Excel 97. Studies with PGA2-KLH Antibodies. 1. PGA2-KLH Antigen and Antibodies. PGA2 (2 mg) was dissolved in 400 µL of 10 mM PBS. A solution of KLH in PBS (8 mg/mL, 400 µL) was added to the PGA2 solution. The resulting mixture was incubated for 48 h at 37 °C with periodic vortexing. Then, PBS (600 µL) was added and the resulting solution was dialyzed against 10 mM PBS (3 × 1L) for 36 h. For immunization, the antigen was emulsified in Freund’s complete or incomplete adjuvant (1:1) as described above for the iso[7]LGD2-KLH antigen. A white New Zealand rabbit was inoculated intradermally in several locations on its back as well as on the rear legs. Booster injections were given every 21 days, and blood was collected 10 days after each injection. A strong immunogenic response appeared within a month and rose steadily over the next 2 months. The rabbit was exsanguinated after 90 days. Crude serum was purified on a protein G column and used at a final concentration 0.5 µg/mL for ELISA. 2. PGA2-BSA Coating Agent. PGA2 (4 mg) was dissolved in 10 mM PBS (800 µL). BSA (8 mg, 20 mg/mL in PBS) was added to the PGA2 solution, and then, another aliquot of PBS (2.8 mL) was added. The resulting mixture was vortexed and incubated for 2 days at 37 °C. Then, the resulting solution was dialyzed against 10 mM PBS (2 × 1 L) for 24 h. 3. Radiolabeled PGA2-HSA Standard. An aliquot of CDCl3 (0.352 mL), containing PGE2 (2 mg, 5.7 µmol), was transferred into an NMR tube. The NMR spectrum was checked. Then, an ethanol-water solution (7:3) of radiolabeled PGE2 (11 µL, 1.1 µCi, 1 µCi/10 µL with specific activity 200 Ci/mmol) was added to the “cold” PGE2 in CDCl3. Cationic exchange resin, Dowex 50w-x8, (100 mg, 5.1 mequiv/dry g) (Bio-Rad, CA) was added to the mixture. The reaction mixture was shaken for 9 days. NMR analysis showed that conversion to PGA2 had proceeded almost quantitatively. Then, the solution was filtered through glass wool to remove the resin and then the solvent was evaporated. PGA2 was dissolved in 1 mL of 0.2 M PBS containing 10 mg of HSA. The final solution contained 0.67 mg of PGA2 (4280 dpm in 5 µL) and 10 mg of HSA in 1 mL of 0.2 M PBS. This solution was incubated for 3 days at 37 °C. Then, it was dialyzed against PBS 10 mM (2 × 500 mL) for 48 h. The dialyzed sample was transferred into new plastic tubes, and 100 µL of the sample was analyzed by LSC. The sample contained 204 µM or 10 200 pmol/well of bound PGA2 to protein, which corresponds to 1.4 mole per mole HSA (10.4% of the original PGA2 added was bound). Synthesis of PGA2-GSH Adduct. An aliquot of CDCl3 (0.9 mL), containing PGE2 (2 mg, 5.7 µmol), was transferred into an NMR tube. Then, an ethanol-water solution (7:3) of radiolabeled PGE2 (27.6 µL, 2.76 µCi, 1 µCi/10 µL with specific activity 200 Ci/mmol) was added to the “cold” PGE2 in CDCl3. An aliquot of 10 µL from the resulting solution was diluted with 1 mL of methanol, and 3 aliquots of 50 µL were counted by LS (average of 3376 dpm). Cationic exchange resin, Dowex 50wx8, (100 mg, 5.1 mequiv/dry g) (Bio-Rad) was added to the mixture. The reaction mixture was shaken for 24 h, and then, 50 mg more of resin was added and shaken for another 24 h. NMR analysis showed that conversion to PGA2 had proceeded almost quantitatively. Then, the solution was filtered through glass wool to remove the resin and then the solvent was evaporated. The PGA2 obtained was dissolved in fresh chloroform (0.9 mL), and LS counting showed 67 513 dpm in 10 µL, which corresponds to 2.76 µCi in the total PGA2 produced (1.9 mg, 5.7 µmol). PGA2 (1.58 mg, 4.72 µmol, 2.29 µCi) was combined with GSH (17.5 mg, 56.9 µmol) in 1.5 mL of 50 mM PBS (pH 7.4), and the mixture was incubated for 3 h at 37 °C. The 1:10 ratio of PGA2 to thiol is recommended in the literature (23).

Iso[7]LGD2-Protein Adducts Are Abundant in Vivo

Chem. Res. Toxicol., Vol. 17, No. 5, 2004 617

Figure 2. Positive ion electrospray (A) mass spectrum for iso[7]LGD2 bismethoxime and (B) tandem MS/MS fragmentation pattern for the pseudo parent ion, (MH-H2O)+, at m/z 393.

Then, the solution was acidified to pH 3 with 1 N HCl and then unreacted PGA2 (0.44 mg, 0.65 µCi, 1.35 µmol, 28.3%) was extracted with ethyl acetate. The aqueous layer was neutralized with 1 M NaOH. Then, the solvent was evaporated under high vacuum and the residue was dissolved in methanol. GSH is not very soluble in methanol (less than 5%); therefore, mostly PGA2-GSH conjugate is extracted into methanol with unreacted GSH remaining undisolved. The PGA2-GSH conjugate was further purified by TLC [Rf ) 0.3 with n-butanol-acetic acidwater (8:1:2)]. The 1H NMR (CD3OD) spectrum was consistent with that reported (24). The PGA2-GSH was dissolved in 10 mM PBS (2 mL) and used for ELISA. The final working concentration of radiolabeled PGA2-GSH was 0.32 mM (15.8 nmol/well). Antibody Binding Studies. 1. Antibody Titer. For determination of antibody levels in rabbit blood serum, PGA2-BSA containing 20 mg/mL of protein was used as a coating agent, and the amount of PGA2 bound to protein was estimated as 10% of the added amount of PGA2 in view of the results of binding radiolabeled PGA2 to HSA. The PGA2-BSA conjugate (100 µL of a solution containing 6.7 µg/mL of protein and presumably 0.33 µg/mL of PGA2-bound to protein in pH 7.4 PBS) was added to an ELISA plate. The ELISA was completed as described above for iso[7]LGD2 antibodies. The antibody titer rose abruptly after 3 weeks and continued to rise over 70 days and then reached a plateau. 2. Competitive Antibody Binding Inhibition Studies. For PGA2-KLH antibodies, all binding inhibition studies were done as previously described for LGE2-KLH antibodies. A stock solution of radiolabeled PGA2-HSA No. 1 containing 0.204 mM bound PGA2 to protein was diluted 10 times with PBS to give a solution containing 1020 pmol/well that was used as standard. A stock solution of PGA2-BSA was diluted 1:300 with PBS, and an aliquot (100 µL, ∼0.94 µM PGA2 bound to BSA) was applied as a coating agent to an ELISA plate. The final concentration of protein G purified antibody for ELISA was 0.5 µg/mL (1:1000 dilution from a stock solution). For LGE2-KLH and iso[4]LGE2-KLH antibodies, ELISAs were done as described before (2, 16, 18). Iso[4]LGE2-HSA and LGE2-HSA standards were prepared as described above for iso[7]LGD2-KLH competition assays. PGA2-Derived Immunoreactivity in Oxidized LDL. LDL (0.5 mg/mL, 12 mL) was dialyzed for 12 h at 5 °C against PBS (2 × 2 L). The LDL was then dialyzed at 37 °C against 10 mM CuSO4 in PBS (1 L) (oxidation with simultaneous dialysis). Aliquots were removed periodically. The free radical-induced oxidation reaction in each aliquot was stopped by adding EDTA (2 mM, final concentration) and BHT (40 µM, final concentration). The protein concentration in each aliquot of oxLDL was measured using the BCA assay (22). ELISA of oxLDL was performed the same as above for inhibition assays for PGA2KLH antibodies. The starting concentration was the undiluted samples, and a dilution factor of 0.5 was employed.

Results Free Radical-Induced Oxidation of PA-PC Generates Iso[7]LGD2-PC. The isoLG hypothesis postulates that a family of eight structurally isomeric LG diastereomers, isoLGs, is generated by rearrangements of four structurally isomeric isoprostanoid endoperoxides. The availability of an authentic sample iso[7]LGD2 by unambiguous total syntheses provided the opportunity to test whether this isoLG is generated upon free radicalinduced oxidation of PA-PC. Iso[7]LGD2 could be detected by collisionally induced positive ion tandem mass spectroscopy after conversion to a stable methoxime derivative and saponification of this derivative to the free acid. Iso[7]LGD2 bismethoxime exhibits a molecular cation with a mass-to-charge ratio (m/z) 411 (Figure 2). Daughter ions that are also found in the mass spectra of isoLGE2 and iso[4]LGE2 bismethoximes include 393, 361, 331, and 316. The m/z 393 ions represent dehydration of the parent ions, and all daughter ions result from further fragmentations of these pseudo parent (MH H2O)+ ions. Elimination of methanol from m/z 393 produces m/z 361. The bis-d3-methoxime of iso[4]LGE2 produces ions at m/z 417 (M + H)+, 399 (MH - H2O)+, and 331 (MH - H2O - 2OCD3)+. This indicates that the m/z 331 ion in the mass spectra of the bis-d0-methoximes corresponds to (MH - H2O - 2OCH3)+. An abundant daughter ion at m/z 222 is unique for iso[7]LGD2 bismethoximesit cannot be generated from any other isoLGsand corresponding ions at m/z 198 and 238 are unique for iso[4]LGE2 and isoLGE2 bismethoximes, respectively. Because the mass spectrum of the bis-d3methoxime of iso[4]LGE2 produces an ion at m/z 201 in place of the m/z 198 ion from the bis-d0-methoxime, the m/z 201 ion retains one OCD3 group and the m/z 198 ion must contain one OCH3 group. The loss of H2O, NOCH3, and the vinyl side chain from each isoLG isomer can account for the formation of these diagnostic isomer specific daughter ions (Figure 3). Using HPLC with on-line LC/ESI/MS/MS with MRM, we analyzed the oxidized lipids produced by exposure of small unilamillar PA-PC liposomes to air in the presence of human MPO, NO2-, and a constant flux of H2O2 generated by the glucose/glucose oxidase system (21). Any isoLG-containing phospholipids formed were converted into stable methoxime derivatives that were saponified to release free acid bismethoximes. The production of iso[7]LGD2-PC was detected by demonstrating cochromatography of the transitions between the bismethoxime pseudo parent at m/z 393 and the unique fragment at

618

Chem. Res. Toxicol., Vol. 17, No. 5, 2004

Figure 3. Isomer specific daughter ions produced in the positive ion electrospray mass spectra of isoLG bismethoximes.

Figure 4. Analysis of sample (oxPA-PC saponified methoximes) and standard (iso[7]LGD2 bis methoximes) by positive ion tandem LC/ESIMS/MS with MRM, monitoring transitions as indicated.

m/z 222 that cannot be generated from any other isoLG (Figure 3). Analysis of iso[7]LGD2 is complicated by a proclivity toward lactonization that we first encountered during a total synthesis (19). Previously, we isolated lactone-free iso[7]LGD2 by extraction into bicarbonate followed by careful acidification. Methoxime derivatives of the lactone (L) and free acid (A) are separable by HPLC, eluting with retention times of 3.0 and 4.3 min, respectively (Figure 4). The ratio of these peaks varies with sample handling, with lactonization being promoted by acid catalysis (data not shown). However, because both the free acid and the lactone produce the same (pseudo) parent ion with m/z 393 (Figure 3), both products are detected in the same selective reaction monitoring chromatogram, i.e., m/z 393

Poliakov et al.

f 222. The transitions from the pseudo parent to fragment ions at m/z 316 and 331, i.e., m/z 393 f 316 and 393 f 331, are common to all isoLG bimethoximes. The bismethoxime derivatives of other isoLGs have retention times (4-6 min) that overlap with that of the iso[7]LGD2-bismethoxime free acid, but the iso[7]LGD2bismithoxime lactone apparently elutes first in the selective reaction monitoring chromatograms of the oxidized lipid vesicle product mixture (Figure 4 bottom two chromatograms). Additional complexity that is resolved in the lactone peak of the oxidized lipid vesicle product mixture presumably is due to the presence of syn and anti methoxime isomers. Iso[7]LGD2-protein antibodies exhibit marginal structural selectivity. To facilitate detection and identification of iso[7]LGD2 in vivo, we prepared iso[7]LGD2-KLH rabbit polyclonal antibodies. The structural specificity of these rabbit polyclonal iso[7]LGD2-KLH antibodies was evaluated by testing the ability of various haptens to competitively inhibit binding of the antibodies to antigen using iso[7]LGD2-BSA as coating agent and iso[7]LGD2HSA as a standard. To prevent weak and nonspecific binding of antibody, assays were conducted with the presence of 0.01% of Tween 20. Under these conditions, the concentration at 50% inhibition of binding (IC50) to the iso[7]LGD2-HSA standard was 24.5 pmol/well for iso[7]LGD2-HSA. The iso[7]LGD2-KLH antibodies crossreact significantly with isoLGE2-HSA (31%) and PGA2HSA (8.9%) but very weakly with iso[4]LGE2-HSA (1.2%). For comparison, the cross-reactivity of iso[7]LGD2-HSA with isoLGE2-KLH, PGA2-KLH, and iso[4]LGE2-KLH antibodies relative to the corresponding standards was measured and found to be negligible (data not shown). PGA2-Protein Antibodies. A mechanism for the generation of bioactive prostaglandins independent of cyclooxygenase involves free radical-induced lipid oxidation that produces significant quantities of PGE2 and its enantiomer (25). These can readily undergo dehydration to give racemic PGA2. It was proposed that racemic PGE2 is produced by acid- or base-catalyzed epimerization of racemic 15-E2t-IsoP, aka 8-iso-PGE2, that is “formed in abundance in vivo” through the isoP pathway (25). To allow the measurement of PGA2-protein adducts in biological samples, we prepared rabbit polyclonal PGA2KLH antibodies. A standard with a known amount PGA2 bound to protein was required in order to develop an ELISA using PGA2-KLH antibodies. To quantify the amount of PGA2 bound to protein, we conjugated radiolabeled PGA2 to HSA. Because tritiated PGA2 is not commercially available, we developed a clean method for dehydration of PGE2 to PGA2. The dehydration can be catalyzed by alkali or acid, but under basic conditions, PGA2 rearranges to PGB2 (26, 27). Methods described previously for dehydration of PGE2 to PGA2 used acidic catalysts (26, 28, 29) but resulted in a mixture of products primarily because PGE2 has two reactive hydroxyl groups. However, we found that the use of the strongly acidic cationic exchange resin Dowex 50w-x8 (30) as catalyst in chloroform leads to clean conversion of PGE2 to PGA2. Progress of the reaction was followed by 1H NMR spectroscopy. Application of the same reaction to the conversion of 8-epi-PGE2 to 8-epi-PGA2 was complicated by isomerization of 8-epi-PGA2 to PGA2 under the reaction conditions as we detected by comparison of the 1H NMR spectra of the reaction product mixture with that

Iso[7]LGD2-Protein Adducts Are Abundant in Vivo

Figure 5. ELISA binding inhibition of PGA2-KLH antibodies by PGA2-HSA (b) and PGA2-GSH (2) to PGA2-BSA coating agent.

of PGA2. This comparison showed that the product consisted of a mixture of PGA2 and 8-epi-PGA2. Radiolabeled PGA2 was incubated with HSA, and then, unbound PGA2 was removed by dialysis against PBS. Quantitative radiochemical analysis showed that 10.4% of the PGA2 present in the reaction mixture became bound to protein, corresponding to 1.4 mole of PGA2 per mole of HSA. In a second preparation of PGA2-HSA, 12.7% of the PGA2 present in the reaction mixture became bound. That corresponds to 2.1 mole of PGA2 per mole of HSA. The concentration at 50% inhibition (IC50) for both standards was 0.6-1.5 pmol/well. PGA2 bound to BSA was used as a coating agent. To test the ability of the antibodies to recognize PGA2-thiol Michael adducts, we synthesized PGA2-GSH conjugate (23, 31). Binding inhibition by PGA2-GSH of PGA2-KLH antibodies exhibits the same slope as binding inhibition by PGA2-HSA. This suggests that the same epitope is recognized in both PGA2-GSH and PGA2-HSA by the antibodies (Figure 5). The IC50 for binding of PGA2-KLH antibodies was 0.71 pmol/well for the PGA2-HSA standard and 1.64 pmol/well for PGA2-GSH, which corresponds to 46% cross-reactivity. To evaluate the structural specificity of the PGA2KLH antibodies, various haptens were tested to competitively inhibit binding of the antibodies to antigen using PGA2-BSA as a coating agent and PGA2-HSA as a standard. Because protein adducts of LGE2 and PGA2 have the same prostanoid side chains, the ability of the antibodies to distinguish between these adducts would depend on their ability to discern stereochemical differences and/or to recognize the cyclopentanone structure vs HL rings. In fact, the cross-reactivity to either LGE2HSA (0.26%) or iso[4]LGE2-HSA (0.01%) was very low. Weak cross-reactivity of LGE2-KLH antibodies toward free PGA2 (0.47%) was reported previously (32). Selectivity of LGE2-KLH antibodies was also very high, and cross-reactivity toward PGA2-HSA was negligible (0.86%). Iso[7]LGD2-KLH Immunoreactive Epitopes Are Generated upon Oxidation of EPA but Not ETA or LA in the Presence of HSA. In vitro free radical oxidation of a variety of PUFAs promoted by iron and ascorbate were performed in the presence of HSA. As predicted, immunoreactive protein-bound epitopes were detected by ELISAs with iso[7]LGD2-KLH antibodies in the reaction product mixture from AA. Similar experiments with ETA, EPA, and LA revealed the generation of cross-reacting protein epitopes from EPA (16% of the level of immunoreactivity generated from AA) but not ETA or LA.

Chem. Res. Toxicol., Vol. 17, No. 5, 2004 619

Figure 6. Generation of iso[7]LGD2-protein adducts by oxidation of LDL (0.5 mg/mL) with Cu2+ (10 µM) detected by iso[7]LGD2-KLH (b) antibodies.

Stable Iso[7]LGD2-KLH Immunoreactive Epitopes Are Generated upon Oxidation of LDL. LDL was oxidized by dialysis of an aqueous solution of LDL in air against a buffer containing Cu2+. Oxidation was halted after various time periods by sequestration of Cu2+ with EDTA added to an aliquot of the reaction mixture. After an induction period, during which the endogenous antioxidants presumably were consumed, immunoreactivity toward iso[7]LGD2-KLH antibodies increased rapidly, reaching a plateau after 16 h (Figure 6). The levels of iso[7]LGD2-protein adducts in extensively oxidized LDL detected by iso[7]LGD2-KLH antibodies were almost five times higher than found previously for iso[4]LGE2protein adducts and almost 20 times higher than found for isoLGE2-protein adducts (2, 16). The immunoreactivity toward iso[7]LGD2-KLH antibodies in LDL oxidized for 9 h was found to be 2637 pmol/ mL. Levels of cross-reacting compounds, e.g., LGE2- (344 pmol/mL) and PGA2-protein adducts (26.0 pmol/mL) were also determined in LDL oxidized with Cu2+ for 9 h with the corresponding antibodies. Multiplied by the cross-reactivity factor, the levels of isoLGE2- and PGA2protein adducts detected in oxLDL account, respectively, for only 4.1 and 0.1% of the total immunoreactivity detected toward iso[7]LGD2-KLH antibodies. Weak crossreactivity (6.0%) toward carboxyheptylpyrrole (CHP) epitopes (33) was also observed with iso[7]LGD2-KLH antibodies. Therefore, the levels of CHP-protein adducts (44 pmol/mL) in oxLDL were also determined. This amount of CHP-protein adduct accounts for less than 0.1% of the total immunoreactivity in the oxLDL detected with iso[7]LGD2-KLH antibodies (Table 1). The total contribution of the cross-reacting protein adducts corresponds to less than 5% of the total immunoreactivity detected in oxLDL with iso[7]LGD2-KLH antibodies. Thus, detection of this immunoreactivity supports the presumption that iso[7]LGD2-protein adducts are generated during free radical-induced oxidation of LDL and suggests that iso[7]LGD2-KLH antibodies can be used to measure levels of iso[7]LGD2-derived protein adducts in biological samples. Unstable PGA2-KLH Immunoreactive Epitopes Are Generated upon Oxidation of LDL. Because we confirmed the high selectivity of PGA2-KLH antibodies toward PGA2-protein adducts, we could use PGA2-KLH antibodies to measure these adducts in LDL oxidized as previously described by dialysis of an aqueous solution in air against buffer containing Cu2+. Oxidation was stopped after various incubation times by adding BHT

620

Chem. Res. Toxicol., Vol. 17, No. 5, 2004

Poliakov et al.

Figure 7. Generation of PGA2-protein adducts by oxidation of LDL (0.5 mg/mL of protein) with Cu2+ (10 µM) detected by PGA2-KLH antibodies (b).

Figure 9. Cross-reacting epitopes contain seven-carbon carboxylic acid side chains.

correlated with iso[4]LGE2-protein adducts (r ) 0.933) and isoLGE2-protein adducts (r ) 0.877). The strong correlation observed between all isoLGs in plasma supports our hypothesis of parallel competing free radicalinduced generation of isoLGs in vivo.

Discussion

Figure 8. Levels of isoLG-protein adduct immunoreactivity in human plasma using polyclonal rabbit antibodies, isoLGE2KLH, iso[4]LGE2-KLH, and iso[7]LGD2-KLH, the corresponding adducts of HSA as standards, and the corresponding adducts of BSA as coating agent. Plasma from three healthy volunteers (normal) and six coronary artery bypass patients (atherosclerosis).

(40 µM) and EDTA (2 mM). Immunoreactivity toward PGA2-KLH antibodies increased rapidly in the first 3 h; however, subsequently, it decreased and almost disappeared by 24 h (Figure 7). Thus, although PGE2 is formed in significant quantities by free radical-induced lipid oxidation (25), protein adducts of its dehydration product, PGA2, do not accumulate. Plasma Iso[7]LGD2-Protein Adduct Levels Are Elevated in Atherosclerosis. A pilot clinical study compared iso[7]LGD2-protein adduct levels detected in plasma from six patients with atherosclerosis and three young control subjects (Figure 8). Levels in plasma of AS patients (8520 ( 3110 pmol/mL) were significantly elevated (P ) 0.01) as compared to healthy volunteers (3458 ( 82 pmol/mL). Interestingly, for all nine individuals, levels of iso[7]LGD2-protein adducts were strongly

Cross-Reactivity of Iso[7]LGD2-KLH Antibodies. The relatively low structural selectivity observed for the new iso[7]LGD2-KLH antibodies is an interesting exception to the outstanding selectivity that we generally find for antibodies that we raise against protein adducts of oxidized lipids. A possible explanation is that the high cross-reactivity is a consequence of the location of both the hydroxyl and the carboxyl functional groups in the same side chain in the iso[7]LGD2-KLH antigen so that the important antibody binding interactions are with the seven-carbon carboxylic acid side chain. All of the crossreacting epitopes, i.e., PGA2-protein, isoLGE2-HL, and CHP, have seven-carbon carboxylic acid side chains, while the epitope that does not cross-react, i.e., iso[4]LGE2-HL, has a four-carbon carboxylic acid side chain (Figure 9). Although PGA2-protein adducts incorporate the same prostanoid side chains as isoLGE2-protein adducts, their cross-reactivity with iso[7]LGD2-KLH antibodies is much lower than that of the isoLG epitopes. This selectivity presumably arises because the side chains are not coplanar in the PGA2-protein adduct but are coplanar in the isoLGE2-HL and the iso[7]LGD2HL (Figure 9). In contrast, the cross-reactivity of iso[7]LGD2-HSA with LGE2-KLH, PGA2-KLH, or iso[4]LGE2-KLH antibodies relative to the corresponding standards was negligible (data not shown). Thus, all antibodies raised previously against lipid-protein adducts, in which both isoLG side chains contain polar functional groups, can easily distinguish between the corresponding antigens and the protein adducts of any other isoLG including iso[7]LGD2. Presumably, antigens with functional groups

Table 1. Contributions of Cross-Reacting Epitopes to Immunoreactivity in oxLDL Detected by Iso[7]LGD2 Antibodies protein epitopes detected by iso[7]LGD2-KLH antibodies

epitope level in oxLDL (pmol/mL)

cross-reactivity factor

immunoreactivity detected by iso[7]LGD2 Ab (pmol/mL)

immunoreactivity detected by iso[7]LGD2 Ab (%)

isoLGE2-HSA iso[4]LGD2-HSA PGA2-HSA CHP-HSA total cross-reactivity iso[7]LGD2-HSA

344 1250 26.0 44.0

0.31 0.01 0.09 0.06

107 12.5 2.3 2.6 125 2637

4.1 0.5 0.1 0.1 4.8 100

1.00

Iso[7]LGD2-Protein Adducts Are Abundant in Vivo

on both side chains produce more specific antibodies because epitope recognition involves significant binding interactions with both side chains. Generation of Cross-Reacting Epitopes from EPA but not ETA or LA. According to the isoLG hypothesis, peroxy radical cyclization generates isoPGH endoperoxide intermediates that rearrange intramolecularly into isoLGs. This leads to the expectation that oxidation of LA in the presence of protein will not generate isoLG-protein epitopes. Thus, because the generation of isoLG endoperoxide intermediates requires three CdC bonds, the endoperoxide mechanism predicts that LA, with only two CdC bonds, cannot be a precursor for isoLGs. As expected, no iso[7]LGD2-KLH immunoreactive epitopes were generated upon oxidation of LA in the presence of HSA. Although ETA has three CdC bonds, oxidation of ETA in the presence of protein did not generate isoLG-protein adducts that cross-react with iso[7]LGD2-protein antibodies. This is because only two endoperoxide intermediates, isoPGH1 and iso[10]PGH1, are expected from ETA. The formation of a [7] series endoperoxide and isoLG from ETA is precluded by the absence of a 4,5-CdC bond. This selectivity of iso[7]LGD2-KLH antibodies is especially noteworthy in view of our previous demonstration that oxidation of ETA in the presence of HSA does generate epitopes that cross-react with isoLGE2-KLH antibodies (16). This is probably because isoLGE1 is generated from isoPGH1 and isoLGE1 cross-reacts strongly with isoLGE2-KLH antibodies.

In contrast, a [7] series isomer, iso[7]PGH3, can be generated from EPA. Strong cross-reactivity is expected from the derived iso[7]LGD3 since epitope recognition by iso[7]LGD2-KLH antibodies depends on binding interactions with an identical seven-carbon carboxylic acid side chain. The present study confirmed this prediction. Oxidation of LDL generates stable isoLG- and unstable PGA2-protein adducts. In contrast to the instability of PGA2-protein adducts generated during oxidation of LDL, isoLG-protein adducts were generated in oxLDL during the first several hours and then their levels reached a plateau but did not decrease (2, 16). We interpret the observed disappearance of immunoreactivity toward PGA2-KLH antibodies after prolonged oxidation of LDL as a consequence of reversibility of the initially formed thiol Michael adducts. Levels of PGA2protein adduct immunoreactivity reach only 40-60 pmol/ mL that is five times less than isoLGE2-protein adduct levels in oxidized LDL. PGs containing a cyclopentenone moiety (PGAs, PGJs) are inhibitors of cell cycle progression (34) and proliferation (35) at concentrations below the point at which they become toxic. As a result of blockage cell cycle progression, these prostaglandins also induce apoptosis (36). The mechanism of the antigrowth action of such prostaglandins is still not well-understood but covalent binding to

Chem. Res. Toxicol., Vol. 17, No. 5, 2004 621

proteins (29, 35) and GSH (24, 28, 29) is believed to be involved. The R,β-unsaturated keto group of the cyclopentenone is essential for their cytotoxicity because it allows covalent binding to thiols via a Michael addition (37, 38). Reversibility of PGA2 induced cell growth inhibition, by washing or by modulation of GSH levels, was reported previously (37, 39, 40). Presumably, such reversibility of the antigrowth effect could be explained by reversibility of protein Michael adduct formation under physiological conditions. Furthermore, reversibility of GSH conjugation of prostaglandins through the cyclopentenone moiety was reported previously (23). The present study provides presumptive evidence that PGA2 isomers generated through the isoP pathway during oxidation of LDL may form adducts with LDL protein, but if so, such adducts are formed reversibly. Although PGA2-KLH antibodies may be useful for in vivo monitoring of PGA2-protein adduct levels without interference of isoLG-protein adducts, the utility of monitoring PGA2-protein adduct levels as a measure of oxidative injury is questionable because of the reversibility of Michael adduct formation. Iso[7]LGD2 Immunoreactivity as a Cumulative Index of Oxidant Stress. As noted above, blood levels of iso[7]LGD2-protein immunoreactivity are strongly correlated with isoLGE2-derived and iso[4]LGE2-derived immunoreactivity. This is presumably because free radical-induced lipid oxidation produces all isoLG isomers through parallel pathways, and subsequent adduction of the resulting electrophilic oxidized lipids with nucleophilic functionality in proteins efficiently generates the corresponding immunoreactive epitopes. Because of relatively low structural specificity, some of the immunoreactivity detected with iso[7]LGD2-protein antibodies will be contributed by cross-reacting epitopes derived from other products of free radical-induced lipid oxidation. However, if levels of lipid-derived oxidative protein modifications are strongly correlated, then levels of immunoreactivity detected in blood with iso[7]LGD2protein antibodies will provide a reliable index of oxidant injury. In other words, contributions of cross-reacting epitopes to the immunoreactivity detected will enhance but not distort the signal resulting from free radicalinduced lipid oxidation.

Acknowledgment. We thank the NIH for support of this research through Grant GM12149. Mass spectrometry analyses were performed at the Cleveland Clinic Foundation Mass Spectrometry Core Facility, within the Center for Cardiovascular Diagnostics and Prevention, with the support of GCRC Grant No. M01 RR018390. Supporting Information Available: Negative ion ESIMS/MS spectra of 2, 2-d4, 3, 4, 8, 8-d3, 9, and HODA methoxime derivative. Positive ion ESI-MS/MS spectra of 16, 17, and CHD derivatives of HNE and benzaldehyde. 1H NMR spectrum of 13HPODE-PC and 9-HPODE-PC regioisomers. This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Salomon, R. G., Miller, D. B., Zagorski, M. G., and Coughlin, D. J. (1984) Solvent induced fragmentation of prostaglandin endoperoxides. New aldehyde products from PGH2 and a novel intramolecular 1,2-hydride shift during endoperoxide fragmentation in aqueous solution. J. Am. Chem. Soc. 106, 6049-6060. (2) Salomon, R. G., Subbanagounder, G., Singh, U., O’Neil, J., and Hoff, H. F. (1997) Oxidation of low-density lipoproteins produces levuglandin-protein adducts. Chem. Res. Toxicol. 10, 750-759.

622

Chem. Res. Toxicol., Vol. 17, No. 5, 2004

(3) Salomon, R. G., Jirousek, M. R., Ghosh, S., and Sharma, R. B. (1987) Prostaglandin Endoperoxides. 21. Covalent Binding of Levuglandin-E2 with Proteins. Prostaglandins 34, 643-656. (4) Boutaud, O., Brame, C. J., Salomon, R. G., Roberts, L. J., II, and Oates, J. A. (1999) Characterization of the lysyl adducts formed from prostaglandin H2 via the levuglandin pathway. Biochemistry 38, 9389-9396. (5) Iyer, R. S., Kobierski, M. E., and Salomon, R. G. (1994) Generation of Pyrroles in the Reaction of Levuglandin E(2) with Proteins. J. Org. Chem. 59, 6038-6043. (6) Brame, C. J., Salomon, R. G., Morrow, J. D., and Roberts, L. J. (1999) Identification of extremely reactive γ-ketoaldehydes (isolevuglandins) as products of the isoprostane pathway and characterization of their lysyl protein adducts. J. Biol. Chem. 274, 13139-13146. (7) Salomon, R. G., Batyreva, E., Kaur, K., Sprecher, D. L., Schreiber, M. J., Crabb, J. W., Penn, M. S., DiCorletoe, A. M., Hazen, S. L., and Podrez, E. A. (2000) Isolevuglandin-protein adducts in humans: products of free radical-induced lipid oxidation through the isoprostane pathway. Biochim. Biophys. Acta 1485, 225-235. (8) Hoppe, G., Subbanagounder, G., O’Neil, J., Salomon, R. G., and Hoff, H. F. (1997) Macrophage recognition of LDL modified by levuglandin E2, an oxidation product of arachidonic acid. Biochim. Biophys. Acta 1344, 1-5. (9) Iyer, R. S., Ghosh, S., and Salomon, R. G. (1989) LevuglandinE2 cross-links proteins. Prostaglandins 37, 471-480. (10) Jirousek, M. R., Murthi, K. K., and Salomon, R. G. (1990) Electrophilic levuglandin E2-protein adducts bind glycinesa model for protein cross-linking. Prostaglandins 40, 187-203. (11) Murthi, K. K., Friedman, L. R., Oleinick, N. L., and Salomon, R. G. (1993) Formation of DNA-protein cross-links in mammalian cells by levuglandin E2. Biochemistry 32, 4090-4097. (12) Vassalle, C., Botto, N., Andreassi, M. G., Berti, S., and Biagini, A. (2003) Evidence for enhanced 8-isoprostane plasma levels, as index of oxidative stress in vivo, in patients with coronary artery disease. Coron. Artery Dis. 14, 213-218. (13) Delanty, N., Reilly, M. P., Pratico, D., Lawson, J. A., McCarthy, J. F., Wood, A. E., Ohnishi, S. T., Fitzgerald, D. J., and Fitzgerald, G. A. (1997) 8-epi PGF2 alpha generation during coronary reperfusion. A potential quantitative marker of oxidant stress in vivo. Circulation 95, 2492-2499. (14) Spector, I. M. (1974) Animal longevity and protein turnover rate. Nature 249, 66. (15) Kobierski, M. E., Kim, S., Murthi, K. K., Iyer, R. S., and Salomon, R. G. (1994) Synthesis of a pyrazole isostere of pyrroles formed by the reaction of the -amino groups of protein lysyl residues with levuglandin E2. J. Org. Chem. 59, 6044-6050. (16) Salomon, R. G., Sha, W., Brame, C., Kaur, K., Subbanagounder, G., O’Neil, J., Hoff, H. F., and Roberts, L. J., II (1999) Protein adducts of iso[4]levuglandin E2, a product of the isoprostane pathway, in oxidized low-density lipoprotein. J. Biol. Chem. 274, 20271-20280. (17) Hatch, F. T. (1968) Practical methods for plasma lipoprotein analysis. Adv. Lipid Res. 6, 1-68. (18) Salomon, R. G., Subbanagounder, G., O’Neil, J., Kaur, K., Smith, M. A., Hoff, H. F., Perry, G., and Monnier, V. M. (1997) Levuglandin E2-protein adducts in human plasma and vasculature. Chem. Res. Toxicol. 10, 536-545. (19) Roy, S. C., Nagarajan, L., and Salomon, R. G. (1999) Total synthesis of iso[7]levuglandin D-2. J. Org. Chem. 64, 1218-1224. (20) Podrez, E. A., Poliakov, E., Shen, Z., Zhang, R., Deng, Y., Sun, M., Finton, P. J., Shan, L., Gugiu, B., Fox, P. L., Hoff, H. F., Salomon, R. G., and Hazen, S. L. (2002) Identification of a novel family of oxidized phospholipids that serve as ligands for the macrophage scavenger receptor CD36. J. Biol. Chem. 277, 3850338516. (21) Podrez, E. A., Schmitt, D., Hoff, H. F., and Hazen, S. L. (1999) Myeloperoxidase-generated reactive nitrogen species convert LDL into an atherogenic form in vitro. J. Clin. Invest. 103, 1547-1560. (22) Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner, F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N.

Poliakov et al.

(23)

(24)

(25)

(26) (27)

(28)

(29)

(30)

(31)

(32)

(33)

(34)

(35) (36)

(37)

(38)

(39)

(40)

M., Olson, B. J., and Klenk, D. C. (1985) Measurement of protein using bicinchoninic acid. Anal. Biochem. 150, 76-85. Atsmon, J., Sweetman, B. J., Baertschi, S. W., Harris, T. M., and Roberts, L. J., II (1990) Formation of thiol conjugates of 9-deoxyδ9,δ 12(E)-prostaglandin D2 and δ12(E)-prostaglandin D2. Biochemistry 29, 3760-3765. Bogaards, J. J., Venekamp, J. C., and van Bladeren, P. J. (1997) Stereoselective conjugation of prostaglandin A2 and prostaglandin J2 with glutathione, catalyzed by the human glutathione Stransferases A1-1, A2-2, M1a-1a, and P1-1. Chem. Res. Toxicol. 10, 310-317. Gao, L., Zackert, W. E., Hasford, J. J., Danekis, M. E., Milne, G. L., Remmert, C., Reese, J., Yin, H., Tai, H. H., Dey, S. K., Porter, N. A., and Morrow, J. D. (2003) Formation of prostaglandins E2 and D2 via the isoprostane pathway: a mechanism for the generation of bioactive prostoglandins independent of cyclooxygenase. J. Biol. Chem. 278, 28479-28489. Andersen, N. H. (1969) Dehydration of prostaglandins: study by spectroscopic method. J. Lipid Res. 10, 320-325. Monkhouse, D. C., Van Campen, L., and Aguiar, A. J. (1973) Kinetics of dehydration and isomerization of prostaglandins E 1 and E 2. J. Pharm. Sci. 62, 576-580. Parker, J., and Ankel, H. (1992) Formation of a prostaglandin A2-glutathione conjugate in L1210 mouse leukemia cells. Biochem. Pharmacol. 43, 1053-1060. Ohno, K., and Hirata, M. (1993) Characterization of the transport system of prostaglandin A2 in L-1210 murine leukemia cells. Biochem. Pharmacol. 46, 661-670. Beier, R., and Mundy, B. P. (1979) A facile removal of the tetrahydropyranyl protecting group from alcohol derivatives. Synth. Commun. 9, 271-273. Heasley, L. E., Watson, M. J., and Brunton, L. L. (1985) Putative inhibitor of cyclic AMP efflux: chromatography, amino acid composition, and identification as a prostaglandin A1-glutathione adduct. J. Biol. Chem. 260, 11520-11523. Subbanagounder, G. (1997) Part I. In vivo detection of levuglandins. Part II. Total synthesis of iso(4)-levuglandin E(2). Ph.D. Thesis, Case Western Reserve University; 361 pp. Kaur, K., Salomon, R. G., O’Neil, J., and Huff, H. (1997) (Carboxyalkyl)pyrroles in human plasma and oxidized low-density lipoproteins. Chem. Res. Toxicol. 10, 1387-1396. Hitomi, M., Shu, J., Strom, D., Hiebert, S. W., Harter, M. L., and Stacey, D. W. (1996) Prostaglandin A2 blocks the activation of G1 phase cyclin-dependent kinase without altering mitogenactivated protein kinase stimulation. J. Biol. Chem. 271, 93769383. Parker, J. (1995) Prostaglandin A2 protein interactions and inhibition of cellular proliferation. Prostaglandins 50, 359-375. Kim, I. K., Lee, J. H., Sohn, H. W., Kim, H. S., and Kim, S. H. (1993) Prostaglandin A2 and δ12-prostaglandin J2 induce apoptosis in L1210 cells. FEBS Lett. 321, 209-214. Honn, K. V., and Marnett, L. J. (1985) Requirement of a reactive alpha, beta-unsaturated carbonyl for inhibition of tumor growth and induction of differentiation by “A” series prostaglandins. Biochem. Biophys. Res. Commun. 129, 34-40. Chien, C. I., Kirollos, K. S., Linderman, R. J., and Dauterman, W. C. (1994) R,β-Unsaturated carbonyl compounds: inhibition of rat liver glutathione S-transferase isozymes and chemical reaction with reduced glutathione. Biochim. Biophys. Acta 1204, 175-180. Kim, H. S., Lee, J. H., and Kim, I. K. (1996) Intracellular glutathione level modulates the induction of apoptosis by δ12prostaglandin J2. Prostaglandins 51, 413-425. Narumiya, S., Ohno, K., Fukushima, M., and Fujiwara, M. (1987) Site and mechanism of growth inhibition by prostaglandins. III. Distribution and binding of prostaglandin A2 and δ12-prostaglandin J2 in nuclei. J. Pharmacol. Exp. Ther. 242, 306-311.

TX034185+