Chem. Res. Toxicol. 1989,2, 295-300
295
A Sensitive Electrochemical Method for Quantitative Hydroperoxide Determination' Coleen Young O'Gara, Krishna Rao Maddipati, and Lawrence J. Marnett* Department of Chemistry, Wayne State University, Detroit, Michigan 48202 Received March 6, 1989
We report a general assay for hydroperoxides that is simple, selective, and sensitive. The assay is based on the reduction of hydroperoxides by glutathione (GSH) catalyzed by GSH peroxidase. Stoichiometric amounts of oxidized glutathione (GSSG) are produced that are separated from GSH by HPLC. GSSG eluting from the column is quantitated with a coulometric detector operating in the oxidizing mode (E = 0.82 V vs Pd). Picomole amounts of GSSG can be measured and related to the hydroperoxide concentration in the incubation mixture. GSH peroxidase has broad substrate specificity to many different hydroperoxides. Therefore, this method allows the determination of the total hydroperoxide concentration in the reaction mixture. For analysis of peroxidized phospholipids, phospholipase A2 is included in the reaction to release fatty acid hydroperoxides from the 2-position of the glycerol moiety. The presence of hydroperoxide is verified by addition of sodium borohydride or stannous chloride to sample extracts of biological fluids before analysis. T h e applicability of this method was tested by examination of human plasma from normal individuals for hydroperoxide levels.
Introduction Lipid peroxides exert deleterious effects at the cellular and molecular level. They are believed to contribute to the genesis of inflammation, thrombosis, and tumor metastasis (1). Evaluation of the role of hydroperoxides in human disease requires sensitive and specific methods for their determination suitable for analysis of a variety of matrices. Several indirect methods are available for quantitation of lipid peroxides and lipid peroxidation (2, 3). The thiobarbituric acid assay estimates the amount of malondialdehyde produced during the decomposition of polyunsaturated fatty acid hydroperoxides. Although this method is very sensitive for the determination of malondialdehyde, it is notoriously nonspecific. A noninvasive technique is the measurement of exhaled alkanes (ethane and pentane) produced during the decomposition of fatty acid hydroperoxides. The thiobarbituric acid and alkane assays suffer from the limitation that they measure byproducts of hydroperoxide decomposition that are produced in low and variable yield from the hydroperoxides. Quantitation of lipid hydroperoxides by measurement of conjugated diene absorbance at 235 nm is sensitive but does not discriminate hydroperoxides from their alcohol reduction products (3). Recent advances in methodology for hydroperoxide analysis utilize detection techniques that are specific for the hydroperoxide moiety ( 2 , 3 ) . The postcolumn chemiluminescence assay using microperoxidase and isoluminol is sensitive and selective ( 4 , 5 ) . The hydroperoxides are separated by HPLC before quantitation, so knowledge of the physical properties of the hydroperoxides is important to ensure complete hydroperoxide recovery. Another method of hydroperoxide detection is based on the ability of fatty acid hydroperoxides to activate cyanide-inhibited prostaglandin synthase (6). This method has a narrow-
* Author to whom correspondence should be addressed at the Department of Biochemistry,Vanderbilt University School of Medicine, Nashville, TN 37232. 0893-228x/89/2702-0295$01.50/0
dynamic range and differential sensitivity to organic hydroperoxides and H20z. GSH peroxidase, which reduces both hydrogen peroxide and organic hydroperoxides, has been used to estimate hydroperoxides by using the classical assay of Paglia and Valentine (7,8). In this method, the stoichiometric amount of oxidized glutathione (GSSG)2produced during hydroperoxide reduction is estimated by coupling it to NADPH reduction with GSSG reductase: ROOH
+ 2GSH
GSH peroxidase
ROH
+ GSSG + HzO (1)
GSSG reductase
NADPH + GSSG NADP +2GSH (2) This method is highly specific for GSSG, but it sensitivity is limited by the molar absorptivity of NADPH (t = 6200). We have attempted to increase the sensitivity of this method by utilizing electochemical techniques for determination of GSSG. Extracts of biological fluids are treated with GSH/GSH peroxidase to generate amounts of GSSG stoichiometric with hydroperoxide. The GSSG is separated from GSH by HPLC and detected by using a coulometric detector. We find that this method allows the determination of picomole amounts of hydroperoxide.
Materials and Methods Reagents. The following chemicals were purchased as analytical grade reagents and used without further purification: GSH, GSSG, L-a-phosphatidylcholine, L-a-diarachidonoylphosphatidylcholine, porcine pancreas PLA2, bovine erythrocyte GSH peroxidase (all purchased from Sigma Chemical Co.), and This work waa supported by research grants from the National Cancer Institute (CA 47479), the Wayne State University Institute of Chemical Toxicology, and the Gershenson Radiation Oncology Facility. C.Y.O. is the recipient of a postdoctoral fellowship from the American Heart Association of Michigan. Abbreviations: PPHP, 5-phenyl-4-pentenyl hydroperoxide; 1300H-182, 13-hydroperoxyoctadeca-9,ll-dienoicacid; PC-OOH, 1stearoyl-2-(hydroperoxylinoleoyloyI)phosphatidylcholine; PC-di-204-0OH, hydroperoxy-l,2-diarachidonoylphosphatidylcholine; GSH, reduced glutathione; GSSG, oxidized glutathione; PLA2,phospholipase A>
0 1989 American Chemical Society
296 Chem. Res. Toxicol., Vol. 2, No. 5, 1989 sodium phosphate monohydrate (EM Science). Hydroperoxide standards, PPHP, and 13-00H-182 were prepared by previously described methods (9, 10). Hydroperoxides of L-a-phosphatidylcholine and L-a-diarachidonoylphosphatidylcholinewere prepared by reaction with singlet oxygen. L-a-Phosphatidylcholine or L-a-diarachidonoylphosphatidylcholine (20 mg) was mixed with methylene blue (1mg) in 20 mL of chloroform and irradiated with a 1000-W sodium lamp (filtered to remove UV light) for 4 h at 4 "C with continuous oxygen bubbling. The solvent was evaporated and 200 pL of chloroform/methanol (2:l) added to the residue. The hydroperoxide was purified by using preparative TLC on silica (Si250F TLC plate, J. T. Baker Chemical Co.) with a mobile phase of chloroform/methanol(2:1). The hydroperoxide positive zone was removed from the plate, extracted three times with chloroform/methanol(21),and concentrated. The presence of hydroperoxide was verified by using a method described by Abraham et al. (11)and quantitated by the method of Mair et al. (12). Preparation of Reagents. Stock solutions were prepared in advance and stored. Tris-HC1 (0.15 M, pH 7.4) was prepared biweekly and stored at 4 "C. Bovine erythrocyte GSH peroxidase was made up in 50 mM sodium phosphate and 5 mM EDTA, pH 7.4, to a concentration of 100 units/mL, and 100-pL aliquots were stored in 1.5-mL vials at -80 "C. The enzyme is more stable when stored for long periods of time in concentrated form. When ready to be used, one vial of the enzyme was removed and diluted to 1 mL (10 units/mL) with the same buffer. Aqueous solutions of GSH (8 mM) were stored in 1.5-mL vials (1mL each) at -80 "C. Hydroperoxide Extraction. Method A. Samples (0.5-2 mL) were extracted three times with an equal volume of hexane/ethyl acetate (1:l). The organic phase was evaporated to dryness with a Speed Vac concentrator (Savant). Method B. Samples (1 mL) were diluted with water (10 mL) and passed through a 3-mL C18solid-phaseextraction column (J. T. Baker, Phillipsbury, NJ) conditioned with 3 mL of methanol and 15 mL of water. Samples were eluted with methanol (1 mL) and evaporated to dryness. Method C. A procedure based on a method described by Frei et al. (13)separates the extract into hexane and methanol phases. Two milliliters of methanol was added to 0.5 mL of sample. The solution was vortexed and further mixed with 10 mL of hexane. The hexane phase was collected and dried on a Speed Vac concentrator (Savant). The methanol phase was removed from the protein residue with a Pasteur pipet and then dried. Dried extracts could be stored a t -20 "C overnight with no alteration of hydroperoxide levels. Hydroperoxide Assay. A reaction mixture was prepared containing GSH (50 pL of stock) and GSH peroxidase (40 pL of stock) in 0.15 M Tris-HCI, pH 7.4 (1.5 mL). Extract residues were dissolved in 100 pL of methanol and added to the reaction mixture. For standard curves, hydroperoxide standards in 100 pM of methanol were added to reaction mixtures in the range of 2-250 pmol. After the samples were incubated at 37 "C for 2 min, the reaction was terminated with 6 N HCl(25 pL) and buffer (0.15 M Tris-HC1, pH 7.4) was added to make a total volume of 2 mL. The mixture was passed through a 1-mL C18solid-phase extraction column (conditioned with 1mL of methanol and 5 mL of water) to remove lipid. It should be noted that if the Baker solid-phase extraction columns are not properly end-capped by the manufacturer they can retain some GSSG. Therefore, all GSSG standards were run through the columns as well. After each sample was extracted, freshly prepared hydroquinone was added as an internal standard to give a final concentration of 5 pM. Samples were assayed immediately and run in triplicate to ensure reproducibility. They were kept at 4 "C until HPLC analysis. Two approaches were employed for the estimation of phospholipid hydroperoxide. In the first, 10 pL of PLA, (145 units) was added directly to the sample (1 mL) before extraction. In the second, 10 pL of PLA2 (145 units) was added to the reaction mixture, and the buffer was replaced with 0.15 M Tris-HC1, 10 mM CaC12,and 0.6 mM EDTA, pH 7.5. Standard curves for 13-00H-18:2 or PPHP were also run in this buffer. Standard curves for PC-OOH and PC-di-20:l-OOH were run with and without PLA,.
O'Gara et al. Reduction of Hydroperoxides. Reduction of hydroperoxides was accomplished with NaBH4or SnCl, according to a procedure described by Frei et al. (13). Methanol/butanol (1:l) (0.5 mL) was added to the dried hexane/ethyl acetate or methanol extract of plamsa. In some experiments this solution was spiked with either 13-00H or PC-OOH. Freshly prepared NaBH, (10 mg/mL of MeOH) or SnC1, (10 mg/mL of MeOH) (0.5 mL of either) was added to the mixture. This solution was incubated for 1 h in the dark at 4 "C. One milliliter of methanol and 0.5 mL of water were added to the solution. Finally, 10 mL of hexane/ethyl acetate (1:l)was added and vortexed and the organic layer collected and dried. In the case of the extracted hexane phase, the same procedure was followed except 10 mL of hexane replaced hexane/ethyl acetate in the fiial extraction. Methanol (100 pL) was added to the dried extract and the solution transferred to the hydroperoxide assay reaction mixture. All organic solvents contained 100 pM BHA as an antioxidant. Control samples were run without NaBH, or SnClz. Plasma Preparation. Human blood was collected in Vacutainer tubes (125 X 16 mm, Becton Dickinson Co.) containing 1 mL of 50 mM Tris-HC1 and 100 mM EDTA, pH 7.4, as an anticoagulant. Plasma was separated by spinning the blood at 500 rpm on a tabletop centrifuge (Dynac) for 10 min. The supernatant was removed and centrifuged at 2000 rpm for 7 min. The final supernatant (plasma) was collected and divided into aliquots for immediate analysis. Equipment. The reaction mixtures were subjected to HPLC analysis on a 5-pm C18Ultrasphere ODS reverse-phase column (4.6 mm X 25 cm) (Beckman) using a Varian Vista 5500 liquid chromatograph. The mobile phase (50 mM sodium phosphate, pH 3.0) was run a t a flow rate of 1mL/min. The effluent was monitored by an ESA coulometric detector (Bedford, MA) with a dual-electrode analytical cell (ESA Model 5020) interfaced to a Varian DS 654 computer for data storage and analysis. In preliminary studies, a W detector was connected in series before the electrochemical detector (Varian 2050). Although not as sensitive as an electrochemical detector, a W detector at 230 nm can be used to detect GSSG amounts greater than 5 nmol(O.01 AU). EC-HPLC. Cyclic voltammetric experiments indicated that GSSG was oxidized at a maximum potential of 1.2 V (vs a Pd reference) at pH 3.0 in phosphate using a carbon electrode. With the use of a porous graphite electrode (ESA) one can monitor the oxidation of GSSG at 0.82 V using an isocratic mobile phase of 50 mM sodium phosphate, pH 3.0. For greater sensitivity higher oxidation potentials can be used. The coulometric detector system consists of two cells, an analytical and a guard cell. The analytical cell contains two porous graphite electrodes (Tl and Tz)that run in series. The upstream electrode (Tl) is used as a screening electrode to minimize the interference caused by compounds with lower oxidation potentials than the compound of interest as they elute from the HPLC column. The screening electrode is set at a potential that will not oxidize the compound of interest (Tl < T2).The downstream electrode (T2)detects the compound of interest and is set at or above the oxidation potential of the compound of interest. The guard cell is connected on line with the mobile phase before the HPLC injector and removes oxidizable impurities in the HPLC mobile phase. The potential of the guard cell is set above Tz. The EC-HPLC systerm was equilibrated for approximately 3 h at potentials of 0.82 V (T,)and 0.85 V (guard cell) with a mobile phase of sodium phosphate, pH 3, at a flow rate of 1 mL/min. The mobile phase was prepared by dissolving sodium phosphate monohydrate in distilled deionized water. The pH was adjusted to 3 by using 85% phosphoric acid (Fisher Scientific). The stock solution of mobile phase was stored at 4 "C and filtered through a Nylon 0.45-pm filter (Millipore Co.) prior to use. When necessary, a screening electrode was used at a potential of 0.60 V (TI). Base-line drift was minimized by employing longer equilibration times. The mobile phase can be pumped (not recirculated due to possible mold growth that can clog cells) at 0.5 mL/min overnight with the detector set at lowered potentials, 0.32 V ( T I and/or T2)and 0.35 V (guard cell). When the system was not in use for more than 2 days, the column and cells were washed with 1 L of water and then 100 mL of methanol and stored until use.
Chem. Res. Toxicol., Vol. 2, No. 5, 1989 297
Hydroperoxide Analysis
,
1000 a
+
800
I
-
CI
2
C
w
600-
Y
400-
U 200
-
B
0
I
I
100
200
300
[13-00H-(18:2(pmol) )3
7
(Ong Control)
Figure 2. Coulometric response as a function of 13-00H-18:2 added to the assay mixture. Current response to 2-250 pmol of 13-00H-182standard reacted in the hydroperoxide assay without PLA2 Analysis by HPLC is described in the text (14). 1000
,
1
11111111111 0
5
1 0 1 5 2 0
25
F.ETEh7OS TIME (mid
Figure 1. Comparison of the amount of GSSG formed from 100 pmol of 13-00H-182 to authentic GSSG. EC-HPLC profiles of 100 pmol of oxidized glutathione (a) and assay of 100 pmol of 13-00H1-18:2(b). (A) Reduced glutathione; (B)hydroquinone (internal standard);(C) oxidized glutathione. When 0 ng control (b, lower profile) is subtracted from the reaction assay, standard GSSG to GSSG formed from 13-00H-182standard in assay has a ratio of 1:l.
0
100
200
300
Results
PC-OOH @mol) Figure 3. Coulometric response as a function of PC-OOH added to the assay mixture. Current response to 2-250 pmol of PC-OOH standard reacted in the hydroperoxide assay with PLA,. Analyais by HPLC is described in the text (14).
Our approach to hydroperoxide analysis requires that GSH peroxidase catalyze quantitative oxidation of GSH to GSSG by hydroperoxides. We first carried out a series of experiments to maximize GSH oxidation. A GSH concentration of 0.2 mM or greater was necessary to effect complete hydroperoxide reduction. A glutathione peroxidase concentration of 0.2 IU catalyzed quantitative GSH oxidation at hydroperoxide concentrations of 2-125 nM. These studies helped to establish standard assay conditions: GSH (0.2 mM), GSH peroxidase (0.2 IU), and hydroperoxide sample (2-250 pmol) in Tris buffer, pH 7.4 (2.0 mL). The reaction was incubated at 37 OC for 2 min and terminated with HC1 (0.075 N). Samples were cleaned up by using a solid-phase extractor and immediately injected into the HPLC. The assay was run with hydroperoxide standards of H202,PPHP, 13-00H-18:2, PC-OOH, and PC-di-20:4OOH. Reaction of 100 pmol of 13-00H-18:2 in the standard incubation mixture produced GSSG. Figure l a shows separation of the GSSG from GSH and the internal standard, HQ. In the absence of peroxide, a small peak was observed due to the GSSG contamination of commercial GSH (Figure la, lower). Chromatography of 100 pmol of GSSG produces the same peak height and area as GSSG generated in the reaction with 100 pmol of 1300H-18:2 (Figure lb). When GSSG reductase and NADPH were included in the assay with peroxide, no GSSG was detected due to the reduction of GSSG to GSH. This experiment further confirms the identity of the peak as GSSG. The standard curve for analysis of 13-OOH-1832 is displayed in Figure 2; linearity was obtained from 2 to 250 pmol. The zero hydroperoxide control was subtracted from each reaction, and each hydroperoxide standard was as-
sayed and chromatographed in triplicate. GSSG areas and peak heights were plotted with a relative standard deviation between 1% and 20% for a concentration range of 250-5 pmol. The GSSG produced in the peroxide assay compared to GSSG standards in a 1:l relationship (Figure 1). Peroxides generated in cells as components of phospholipids are relatively poor substrates for GSH peroxidase. Free fatty acid peroxides can be released by treatment of the phospholipids with PLAP. PC-OOH was prepared by photooxidation of l-stearoyl-2-linoleoyl-PC, and its peroxide content was quantitated by iodometric titration. Reaction of PC-OOH with GSH/GSH peroxidase produced only 20% of the theoretical amount of GSSG. Inclusion of PLA, (145 units) in the GSH/GSH peroxidase reaction mixture produced GSSG stoichiometric with the peroxide content of the PC-OOH added. The standard curve under modified conditions was virtually superimposable on the 13-00H-182 curve (Figure 3). For analysis of biological fluids, procedures are required for efficient extraction of hydroperoxides. The three extraction procedures used in this study are referred to as methods A-C. In method A, samples (0.5-2.0 mL) were extracted three times by using equal volumes of hexane/ ethyl acetate (1:l). The organic phase was extracted and evaporated to dryness. For method B, samples (1.0 mL) were diluted with water (10.0 mL) and passed through a 3-mL Cu solid-phase extraction column. Samples were eluted with methanol (1.0 mL) and evaporated to dryness. Finally, method C separated the extract into hexane and methanol phases. Methanol (2.0 mL) was added to 0.5 mL of sample and mixed with hexane (10.0 mL). The hexane and methanol phases were collected and dried. Dried
O’Gara et al.
298 Chem. Res. Toxicol., Vol. 2, No. 5, 1989 b
a
II
B
A
C
’;J
P.-
Y
Et:
a
r
C
0
i 0
5
10
15
20
25
RETENTION TJME ( m i d Figure 4. Coulometric-HPLC profile of assay of human plasma. EC-HPLC profile of human plasma (2 mL) extracted by method
B and analyzed by the hydroperoxide assay without PLA2. (A) GSH; (B)hydroquinone (internal standard); (C) GSSG.
extracts were dissolved in methanol (100 pL) and combined with the assay mixture. A set of standards was run for every experiment with each concentration run in triplicate. A comparison of the three extraction procedures using standard hydroperoxides showed free fatty acid hydroperoxides like 13-00H-182were recovered by greater than 90% with methods A or B. However, phospholipid hydroperoxides like PC-OOH were recovered by about 40% with method A. Recovery of PC-OOH by chloroform/ methanol (2:l) extraction was found to be no better than that obtained by method A. When method C was used, free fatty acids were recovered in the aqueous/methanol phase by greater than 50% while the hexane phase recovered 25%. Between 70% and 80% of the PC-OOH was recovered in the aqueous/methanol phase and 20-30% in the hexane phase. Therefore, routine extraction for free fatty acid hydroperoxides was done by using method A, and method C was used for phospholipid hydroperoxides. To test the feasibility of our method with a biological fluid, we analyzed the hydroperoxide content of human plasma. Plasma samples from normal human volunteers were analyzed for peroxide content by using the three different extraction procedures. The glutathione peroxidase reaction mixture of 1 mL of plasma extracted by method A gave an EC-HPLC profile as shown in Figure 4. The amount of GSSG was related to hydroperoxide concentrations by using standard curves generated with 13-OOH-182. There was always a small background peak of GSSG, so it was necessary to run the proper controls with all the reagents. Control values with no enzyme present were subtracted out. Formation of peak C (GSSG) required the presence of GSH and GSH peroxidase in the reaction mixture. Addition of GSSG reductase and NADPH eliminated peak C.
1 1 1 1 1 4
8
12
16
Zl
Retention Time (min)
24
r 0
4
8
12
16
10
24
Retention Time (min)
Figure 5. Comparison of hydroperoxide assay of human plasma with and without NaBH4. HPLC profile of human plasma (1mL each) extracted by method A and assayed with (a) and without (b) N&H4 present in extract before assaying for hydroperoxide. (A) GSH; (C) GSSG. (Internal standard HQ was oxidized out by screening electrode, TI= 0.60 V.)
Peroxide levels from human plasma extracted by method A or B without PLA2 varied from 0.0 to 1.7 pM with an average of 0.8 f 0.5 pM for 15 subjects. Method C gave slightly higher concentrations. The hexane phase showed an average hydroperoxide concentration of 1.07 f 1.43 pM, and the aqueous/methanol phase showed an average of 1.53 f 0.86 pM for four subjects. Levels varied from subject to subject and from day to day. Plasma samples spiked with 13-00H-18:2 or PPHP (100-120 pmol) and extracted immediately by methods A or B showed the anticipated GSSG levels (90-110 pmol). To probe the nature of the species in human plasma responsible for the formation of GSSG, the extracts were treated with NaBH, or SnC12to reduce the hydroperoxides before incubation with GSH and GSH peroxidase. Figure 5 shows an HPLC profile of the hydroperoxide analysis of human plasma with and without NaBH4 using extraction method A. No significant difference was observed for NaBH4-treated samples. Addition of 13-00H-18:2 (100 pmol) to human plasma followed by NaBH4 treatment of the extract indicated that NaBH4 quantitatively reduces fatty acid hydroperoxides in human plasma. These results strongly suggest the presence of materials in plasma that give rise to GSSG when incubated with GSH/GSH peroxidase but which are not hydroperoxides. These results indicate the necessity of running NaBH, reduction on extracts of biological fluids in order to verify the identity of the “peroxides” determined.
Discussion Dietary fat has been implicated in a range of chronic human diseases including atherosclerosis, thrombosis, inflammation, and cancer. Oxidation of polyunsaturated fatty acids may be a key step in disease etiology because hydroperoxides react with transition metals to form potent oxidants such as metal-oxo complexes, alkoxy1 radicals, and peroxyl radicals. These oxidants are capable of inducing tissue damage, protein modification, and DNA damage. For example, selective inactivation of certain enzymes of prostacyclin and thromboxane biosynthesis may play a crucial role in thrombosis and metastasis (15).
Hydroperoxide Analysis
This is the basis for the concept of peroxide tone which suggests a tight regulation of fatty acid hydroperoxides in biological systems (16). Recent evidence also suggests that oxidized low-density lipoprotein particles are important for the development of atherosclerosis (17). Oxidation of low-density lipoprotein particles induces their uptake by macrophages to form foam cells. These cholesterol-laden cells (foam cells) accumulate and are trapped in the subendothelial space. The major obstacle in testing theories of hydroperoxide involvement in disease etiology has been the lack of a simple, selective, and sensitive method for determination of hydroperoxides in biological samples. We have developed an efficient general assay for hydroperoxides that is simple and provides an estimate of total hydroperoxide concentration. Commercially available bovine erythrocyte glutathione peroxidase is reactive to a wide variety of hydroperoxides including H202and most free fatty acid hydroperoxides. The enzyme will not utilize highly hydrophobic hydroperoxides like cholesterol hydroperoxide or phospholipid hydroperoxides as substrates. However, phospholipid hydroperoxides can become active substrates in the presence of PLA2, which will release the free fatty acid hydroperoxide. No matter what hydroperoxide is reduced by GSH/GSH peroxidase, the product is always GSSG. Therefore, the assay measures the total hydroperoxide content of whatever sample is analyzed. Several methods are available for the estimation of GSH and GSSG. These methods are excellent in sensitivity and selectivity for the estimation of GSH even in the presence of GSSG. However, estimation of small quantities of GSSG in the presence of a large excess of GSH is problematical. In the assay of hydroperoxides using GSH peroxidase, it is essential to use an excess of GSH to ensure the complete reduction of hydroperoxides of kinetic reasons. Therefore, it is important to utilize a method for the estimation of GSSG that does not interfere with the presence of excess GSH in the reaction mixture. The approach described here utilizing electrochemicaloxidation of GSSG should provide a general method for the estimation of hydroperoxides applicable to all classes of hydroperoxides in a variety of sample matrices. Application of GSH peroxidase chemistry to hydroperoxide estimation will give the method high specificity. Use of post-HPLC coulometric detection for the estimation of oxidized glutathione will enhance the sensitivity of the assay several orders of magnitude over any of the existing methods. The method should be a powerful tool to test theories that relate dietary fats and fatty acid hydroperoxides to the development of cardiovascular diseases and other pathogenic processes. Chromatographic separations of oxidized and reduced glutathiones using ion-exchange and reverse-phase columns (in the presence of ion-pairing agents) have been reported (18,19). Reverse-phase HPLC is a logical choice because of the hydrophilicity of the glutathiones. An isocratic solvent system is used because of its simplicity and easy adaptability to postcolumn detection methods. Postcolumn detection of glutathione by colorimetric, amperometric, and coulometric methods has been reported (18-20). The colorimetric method unlike the electrochemical methods can detect glutathione only in the high nanomolar range and is not sensitive enough for our purpose, which requires a dynamic range of detection to the picomolar level. The amperometric method utilizes gold-mercury electrodes and reductive electrochemistry. Reductive chemistry requires elaborate procedures to exclude oxygen from the HPLC solvents. The coulometric
Chem. Res. Toxicol., Vol. 2, No. 5, 1989 299
electrochemical system has the advantage of operating in the oxidative mode with an analytical cell containing two encapsulated porous graphite electrodes. Electrode 1 is often used as a screening electrode that can preoxidize (or reduce) any impurities. The second electrode is then set at a higher potential for analysis of the compound of interest. Higher potentials (electrode 2 > 0.82 V, guard cell > 0.85 V) can be used to increase the sensitivity, but high potentials decrease the lifetime of the cells. Alternatively, electrode 1can be used as a screening electrode (10.6 V), and electrode 2 can be utilized for analysis (20.82 V). However, it should be noted that the hydroquinone internal standard will be oxidized at the screening electrode. In this case, runs should be duplicated to ensure reproducibility. After approximately 150 sample injections the GSSG peak decreased by 2-fold. This could be corrected by cleaning the analytical cells with 6 N HC1 under the manufacturer’s guidelines. After approximately 500 sample injections slight changes occurred in retention times. This was corrected by regenerating the HPLC column under the manufacturer’s guidelines. Routine cleaning with 50% aqueous methanol also helped prevent these problems. It should be noted that the phosphate mobile phase should never be left immobile for any length of time due to the damaging effect high salt concentrations can have on the HPLC, column, and analytical cell. Hydroperoxide standards, H202, 13-OOH-1832, and PPHP all reacted with GSH/GSH peroxidase to produce GSSG, which was easily chromatographed and detected by coulometric detection. Lipid hydroperoxides, PC-OOH, and PC-di-20:40-00H were assayed in the presence of PLA2 and found to be fully reactive with glutathione peroxidase. Quantitative yields of GSSG were obtained under the conditions described. Analysis of human plasma indicated the presence of 0.8 f 0.5 pM “peroxide”. However, treatment of the initial plasma extract with agents that reduce hydroperoxides (e.g., NaBH, or SnC12)did not lower the level of hydroperoxide detected. This indicates that the material in the plasma extract was not actually a hydroperoxide but decomposed to one (or H202)sometime before or during the incubation with GSH/GSH peroxidase (no GSSG was formed if GSH peroxidase was omitted from the incubations). This result illustrates the importance of performing additional experiments to characterize the material that gives rise to GSSG in the assay. Estimations of the hydroperoxide concentration of human plasma have decreased in magnitude as the specificity of the analytical method increased. Initial estimates of 26-63 pM made using the TBA assay have been replaced by values of 0.1-1 pM obtained with the chemiluminescence or oxygenase assays (5,13,21). Warso and Lands reported a value of 0.5 pM using the oxygenase assay and demonstrated that inclusion of GSH/GSH peroxidase in the assay eliminated detectable hydroperoxide (22). Considering our finding that reduction of the plasma extract with NaBH, or SnC1, did not lower the amount of GSSG formed on subsequent incubation with GSH/GSH peroxidase, the control experiment in the Warso and Lands study does not eliminate the possibility that peroxide was generated during the assay. In fact, when GSH/GSH peroxidase is added to human plasma before extraction, no decrease in “hydroperoxide”levels is observed using our assay.3 This is consistent with the NaBH, reduction experiments and with experiments recently reported by C. O’Gara, unpublished result.
300 Chem. Res. Toxicol., Vol. 2, No. 5, 1989
Frei et al. (13). These investigators found that NaBH4 reduction of human plasma extracts did not decrease the intensity of a “hydroperoxide” peak detected by HPLC with chemiluminescent detection. They identified this peak as ubiquinol 10 which reduces O2 to H20z after eluting from the HPLC column. The H20zreacts with hematin and isoluminol to give a peak in the chemiluminescence assay. When allowance for this artifactual peak was made, Frei et al. were unable to detect any hydroperoxides in human plasma. Their results using a different hydroperoxide assay are very similar to our own which were obtained with an assay that measures total peroxide content. If one subtracts the amount of “peroxide” not reducible by NaBH4 from the amounts measured, one is unable to detect peroxides in human plasma at the limit of detection of our assay (2 nM with 2 mL of plasma).
Acknowledgment. We thank all the volunteers who donated blood for hydroperoxide analysis. We also thank Jan Crowley and Koni Stone for technical assistance, RBgine Lab6que for help in synthesizing hydroperoxy1,2-diarachidonoylphosphatidylcholine,and Max Funk for his expert advice in electrochemistry.
References (1) Hartland, H., Gilbert, J., and Brooks, C. (1973)Lipids of Human Atheroma: Oxidized Derivatives of Cholesteryl Linoleate. Biochim. Biophys. Acta 316,378-385. (2) Slater, T. F. (1984)Overview of Methods Used for Detecting Lipid Peroxidation. Methods Enzymol. 105,283-293. (3) Smith, C. V., and Anderson, R. E. (1987)Methods for Determination of Lipid Peroxidation in Biological Samples. Free Radical Biol. Med. 3(5), 341-344. (4) Yamamoto, Y.,Brodsky, M. H., Baker, J. C., and Ames, B. N. (1987)Detection and Characterization of Lipid Hydroperoxides at Picomole Levels by High-Performance Liquid Chromatography. Anal. Biochem. 160,7-13. (5) Miyazawa, T., Fujimoto, K., and Kaneda, T. (1987)Detection of Picomole Levels in Lipid Hydroperoxides by a Chemiluminescence Assay. Agric. Biol. Chem. 51,2569-2573. (6) Marshall, P. J., Warso, M. A., and Lands, W. E. M. (1985) Selective Microdetermination of Lipid Hydroperoxides. Anal. Biochem. 145,192-199. (7)van Kuijk, F. J. G. M., Handelman, G. J., and Dratz, E. A. (1985) Consecutive Action of Phospholipase A2 and Glutathione Peroxidase is Required for Reduction of Phospholipid Hydroperoxides and Provides a Convenient Method to Determine Peroxide Values in Membranes. J. Free Radicals Biol. Med. 1, 421-427. (8) van Kuijk, F. J. G. M., and Dratz, E. A. (1987)Detection of
O’Gara et al. Phospholipid peroxides in Biological Samples. Free Radical Biol. Med. 3(5), 349-354. (9) Weller, P.E., Markey, C. M., and Marnett, L. J. (1985)Enzymatic reduction of 5-phenyl-4-pentenylhydroperoxide: detection of peroxidases and identification of peroxidase reducing substrates. Arch. Biochem. Biophys. 243,633-€43. (10) Funk, M. O., Isaacs, R., and Porter, N. A. (1976)Preparation and Purification of Lipid Hydroperoxides from Arachidonic and y-Linolenic Acids. Lipids 11, 113-117. (11) Abraham, A., Davie, A. G., Llewellyn, D. R., and Thain, E. M. (1957) The Chromatographic Analysis of Organic Peroxides. Anal. Chim. Acta 17,499-503. (12) Mair, R. D., and Graupner, A. J. (1964)Determination of Organic Peroxides by Iodine Liberation Procedures. Anal. Chem. 36,194-204. (13) Frei, B., Yamamoto, Y.,Niclas, D., and Ames, B. (1988)Evaluation of an Isoluminol ChemiluminescenceAssay for the Detection of Hydroperoxidesin Human Blood Plasma. Anal. Biochem. 175,120-130. (14)Current and Coulomb calculations described in ESA Instruction Manual 87,revision 2, pp 5-24. (15) Moncada, S., and Vane, J. R. (1977) The Discovery of Prostacyclii-A Fresh Insight Into Arachidonic Acid Metabolism. In Biochemical Aspects of Prostaglandins and Thromboxanes (Kharasch,N., and Fried, J., Eds.) pp 155-177,Academic Press, New York. (16) Lands, W. E. M., Cook, H. W. and Rome, L. H. (1976)Prostaglandin Biosynthesis: Consequencesof Oxygenase Mechanism Upon In Vitro Assays pf Drug Effectiveness. In Biochemical Aspects of Prostaglandin and Thromboxane Research (Samuelsson, B., and Paoletti, R., Eds.) pp 7-17, Raven Press, New York. (17) Parthasarathy, S.,Wieland, E., and Steinberg, D. (1989)A Role for Endothelial Cell Lipoxygenase in the Oxidative Modification of Low Density Lipoprotein. Proc. Natl. Acad. Sci. U.S.A. 86, 1046-1050. (18) Keller, D. A., and Menzel, D. B. (1985)Picomole Analysis of Glutathione, Glutathione Disulfide, Glutathione S-Sulfonate, Cysteine S-Sulfonate By High Performane Liquid Chromatography. Anal. Biochem. 151,418-423, (19) Richie, J. P., and Lang, C. A. (1987)Determination of Glutathione, Cysteine and other thiols and Disulfides in Biological Samples Using High-Performance Liquid Chromatography with Dual Electrochemical Detection. Anal. Biochem. 163,9-15. (20) Carro-Ciampi, G., Hunt, P. G., Turner, C. J., and Wells, P. G. (1988)A High-Performance Liquid Chromatographic Assay for Reduced and Oxidized Glutathione in Embryonic, Neonatal, and Adult Tissues Using a Porous Graphite ElectrochemicalDetector. J. Pharmacol. Methods 19,75-83. (21) Warso, M. A., and Lands, W. E. M. (1984)Pathophysiologic Modulation of Arachidonate Metabolism. Clin. Physiol. Btochem. 2,70-77. (22) Warso, M. A. and Lands, W. E. M. (1985)Presence of Lipid Hydroperoxides in Human Plasma. J. Clin. Invest. 75,667-671.
