ANALYTICAL BIOCHEMISTRY 152, lo?- 112 ( 1986)

Quantitation of Lipid Peroxidation Products by Gas Chromatography- Mass Spectrometry’

HELEN HUGHES, CHARLES V. SMITH, JANICE 0. TSOKOS-KUHN, AND JERRY R. MITCHELL

Department of Medicine and Institute for Lipid Research, Baylor College of Medicine, Houston, Texas 77030

Received May 2, 1985

A method for the quantitation of lipid peroxidation products in total hepatic lipid has been developed. Lipid extracts are reduced and subsequently transmethylated with sodium methoxide. The hydroxy fatty acid methyl esters are isolated by silicic acid chromatography and derivatized to their trimethylsilyl ethers prior to analysis by gas chromatography-mass spectrometry. Three isomers, 1 I -, 12-, and 15-hydroxyeicosateraenoic acid (HETE), are quantitated using selected ion monitoring techniques relative to the internal standard, methyl I S-hydroxyarachidate. In mice treated with carbon tetrachloride (2 ml/kg), the HETE levels in total hepatic lipid were 20-fold greater than those found in control animals. HETE levels were also elevated (5- to IO-fold) in hepatic lipid from rats treated with the same dose of carbon tetrachloride. Studies on subcellular fractions with this methodology show that these lipid peroxidation products are 5-to 6-fold higher in hepatic plasma membrane vesicles isolated from rats treated with carbon tetrachloride when compared with those isolated from control animals. 0 1986 Academic press. IIIC.

I&Y WORDS: gas chromatography-mass spectrometry: lipids; peroxidation; toxins.

The peroxidation of cellular lipids is one of a number of alterations to cellular macro- molecules that have been implicated in chem- ical-induced acute lethal injury. The peroxi- dation of membrane lipids associated with cell death has been of particular interest, and most thoroughly studied, following intoxication with halogenated hydrocarbons such as carbon tetrachloride ( l-3). Lipid peroxidation has also been proposed as having a causal role in the toxicity of reactive oxygen species and of compounds such as diquat and adriamycin that generate reactive oxygen products in vivo (4-6). The relative importance of lipid per- oxidation in the pathogenic processes, com- pared with other biochemical changes, such as covalent binding to lipids or protein, re- mains unclear and further studies in this area are required.

The measurement of the extent to which lipid peroxidation occurs in vivo is not

’ This work was supported by Grant GM-34 I20 from the National Institute for General Medical Science.

straightforward. Many of the nonspecific methods utilized successfully in the more widely studied in vitro systems are subject to interferences in vivo. Measurements of thio- barbituric acid-reactive substances, fluorescent products, conjugated dienes, chemilumines- cence, ethane, and pentane have all been em- ployed in studies on lipid peroxidation (7,8). In addition, we have recently reported a spe- cific method utilizing high-performance liquid chromatography for the analysis of hydrox- ylated fatty acids that are formed in liver phosphatidylcholines of carbon tetrachloride- treated animals (2).

Studies in our laboratory on hepatotoxins other than carbon tetrachloride, however, showed a discrepancy between findings where ethane and pentane expiration was used as an index of lipid peroxidation when compared with the specific method for determining hy- droxylated fatty acids in phospholipids (9). One possible explanation for this discrepancy was that the peroxidation was occurring in a class of lipids other than the phosphatidyl-

107 0003-2697186 $3.00 Copyright 0 1986 by Academic Press, Inc. All rights of reproduction in any form reserved

108 HUGHES ET AL.

cholines or alternatively in a small subcellular compartment where peroxidation could not be detected with the less sensitive HPLC methodology. We therefore required a method for determining lipid peroxidation occurring in all lipid classes, which would provide greater sensitivity while retaining the specificity of the HPLC method. To accomplish this, we have developed a method based on gas chromatog- raphy-mass spectrometry for the analysis of specific hydroxylated isomers of eicosatetra- enoic acid in total lipid extracts.

MATERIALS AND METHODS

Preparation of standards. Methyl 15hy- droxyarachidate (15-OH 20:0) was prepared from arachidonic acid by first preparing the unsaturated analog with soybean lipoxygenase ( 10). The resulting 15-HETE* was methylated with diazomethane, purified by HPLC, and hydrogenated over platinum oxide as previ- ously described (2).

For the preparation of standard curves, 1 1-, 12-, and 15HETEs were prepared from arachidonic acid using cupric oxide and hy- drogen peroxide as described by Boeynaems et al. (11). The isomers were methylated with diazomethane and purified by HPLC as de- scribed below with the exception that two Po- rasil columns were placed in series and the proportion of isopropanol was reduced to 0.3% when 12- and 15-HETE methyl esters were isolated.

Animals. Male Swiss ICR mice (25-30 g) were fasted 18 h prior to receiving carbon tet- rachloride (2 ml/kg, ip) in mineral oil and were killed 30 min postdose. Studies on plasma membrane vesicles were carried out on male Sprague-Dawley rats (200-300 g) that were pretreated with phenobarbital (80 mg/kg) for 3 days prior to receiving carbon tetrachloride (2 ml/kg, ip). For studies on whole liver, male Fischer rats were pretreated with Arochlor

’ Abbreviations used: HETE, hydroxyeicosatetraenoic acid; 20:4, eicosatetraenoic; BSTFA, bistrimethylsilyl-(tri- fluoro)acetamide; TMS, trimethylsilyl.

1260 ( 100 mg/kg, po) 15 h prior to receiving carbon tetrachloride (2 ml/kg) in mineral oil. Control animals received vehicle alone. Ani- mals were killed 4 h postdose and exsangui- nated. The livers were removed and, for anal- yses on whole tissue, kept in ice-cold saline prior to lipid extraction. Plasma membrane vesicles were prepared as described by Kraus- Friedman et al. (12) and analyses carried out either on fresh samples or on those which had been stored below -20°C.

Isolation of lipid hydroxy acids. Lipid ex- tracts were prepared according to the method of Folch et al. (13). Fresh samples of liver (50- 150 mg) or plasma membrane vesicles (l-3 mg protein) were homogenized in 20 vol of chloroformmethanol (2: 1). Methyl 15-hy- droxyarachidate (0.3 nmol) and methyl non- adecanoate (1.8 pmol) were then added and the Folch extract was filtered and washed with saline (0.9%, 0.2 ~01).

After standing overnight at 4°C the lower organic phase was evaporated to dryness under nitrogen. Methanol (200 ~1) was added im- mediately and the samples were kept on ice. Triphenylphosphine (200 ~1, 1 mg/ml in ether) was added to reduce any hydroperoxides that were present to the more stable hydroxy de- rivatives. The samples were kept on ice for 1 h and then evaporated to dryness. Benzene (200 ~1) and sodium methoxide (500 ~1, 0.5 M in methanol) were added immediately and the resulting solution was left at room tem- perature. After 1 h, saline (5%, 1 ml) was added and the pH adjusted to 3 with glacial acetic acid. The transmethylated lipid sample was extracted with ether (10 ml) and evaporated to dryness. Diazomethane was then added to methylate the free fatty acids in the lipid ex- tract and the resulting methyl esters were re- dissolved in hexane. Extracts from plasma membrane vesicles were subjected to silicic acid chromatography in order to separate the hydroxylated fatty acid methyl esters from the great ,excess of unoxidized fatty acid esters. Samples dissolved in hexane (1 ml) were loaded on silicic acid (100-200 mesh) columns

LIPID PEROXIDATION PRODUCT CHROMATOGRAPHIC QUANTITATION 109

3 X 0.6 cm i.d. in Pasteur pipets. The columns were washed with 7% ether in hexane (4 ml) and the hydroxylated fatty acid methyl esters eluted with ether:hexane (50:50, 4 ml).

The extracts that had been isolated from whole liver were purified by HPLC. A Porasil column (30 cm X 3.9 mm, Waters Assoc.) was connected to a Glenco Model HPLPS-1 pumping system equipped with a Waters U6K loop injector. The mobile phase hexane:iso- propanol:acetic acid (995:4: 1) was pumped through the column at a rate of 2.4 ml/min. Arachidonic acid was quantitated in the first fraction eluting from 1 to 6 min. The second fraction eluting between 6 and 13 min, which contained methyl 15hydroxyarachidate and 1 1-, 12-, and 15-HETE methyl esters, was col- lected and analyzed by GC-MS.

For standard curves, mixtures of 1 l-, 12-, and 15-HETE methyl esters (15, 30, 60, and 150 pmol) were added to a chloroform/meth- anol extract of saline and carried through the procedure described above.

Gas chromatography-mass spectrometry. GC-MS analyses were carried out on a Ri- bermag R lo- 1 OC quadrupole mass spectrom- eter equipped with a Ribermag “SADR” data system. The instrument was operated in the electron impact mode (40 eV) with a source temperature of 200°C. Samples were deriva- tized with BSTFA:pyridine (5: 1, 12 ~1) for 1 h at 70°C and introduced into the mass spec- trometer via the GC-inlet. A glass column (3 ft X 2 mm id.) packed with 3% SP 2100 on Supelcoport ( lOO/ 120 mesh) was operated isothermally at 205°C. Selected ions were monitored at m/z 225, 295, and 335 for the TMS-ethers of 1 1-, 12-, and 15-HETE methyl esters, respectively, and m/z 343 for methyl 15-OTMS-arachidate. The small contribution to the m/z 225 ion current from the 15-hy- droxy derivative was computed from the m/z 335 ion peak area after analysis of the standard 15-HETE methyl ester TMS ether and sub- tracted from the m/z 225 peak area, to give the value for the 1 l-OTMS isomer. Quanti- tation from the biological samples was ac-

complished from standard curves on relative ion peak areas.

Gas chromatography. The less polar frac- tions from silicic acid chromatography, con- taining fatty acid methyl esters, were analyzed by gas chromatography on a Tracer-560 in- strument with a flame ionization detector. Ni- trogen (20 ml/min) was used as carrier gas and the samples were analyzed with injector and detector temperatures of 220 and 330°C re- spectively. Samples were dissolved in iso-oc- tane (2 ml) and l-2 ~1 of this solution was analyzed on a 12 ft X 2-mm i.d. column packed with 5% SE 30 on gas-chrom Q (80- 100 mesh). The column was heated at a rate of 2”C/min from an initial starting tempera- ture of 230°C. Methyl arachidonate eluted at 3.7 min and was quantitated relative to the standard methyl nonadecanoate (retention time 3.4 min) based on peak area calculations.

RESULTS AND DISCUSSION

A specific and sensitive method has been developed for the quantitation of lipid per- oxidation products formed in vivo. We have previously reported that during lipid peroxi- dation a large number of hydroxylated deriv- atives of linoleic, eicosatetraenoic, and doco- sahexaenoic acids are formed (2). In the pres- ent method we have chosen to quantitate three isomers, the 1 1-, 12-, and 15-HETEs, as rep- resentatives of this large group of hydroxylated unsaturated fatty acids. The method is based on gas chromatography-mass spectrometry for the analysis of the HETE isomers, which are determined following hydrolysis of total lipid extracts. More than one isomer was mea- sured to enable us to distinguish between an autoxidative process, where a multitude of isomers are formed ( 14) and a stimulation of a specific lipoxygenase, which would lead to elevation of only one hydroxylated derivative. Extracted standard curves for 1 l-, 12-, and 15- HETE methyl esters are illustrated in Fig. 1. The ions monitored for each isomer at m/z 225,295, and 335 for 1 I-, 12-, and 15-HETE derivatives, respectively, and m/z 343 for the

110 HUGHES ET AL.

15-oH 20:4 .

P mole

11-OH 20:4 12-OH 20:4 .

p mole p mole

FIG. 1. Standard curves for methyl 1 1-, 12-, and 15 hydroxyeicosatetraenoate analyzed by GC-MS as their trimethylsilyl ethers. Peak area ratios were measured for ions with m/z 225 (1 l-OH), 295 (12-OH), and 335 (15- OH) relative to m/z 343 for the internal standard, methyl 15-hydroxyarachidate. Correction was made for the con- tribution to the m/z 225 ion current from the 15-hydroxy isomer. See Materials and Methods.

internal standard represent major fragment ions in the electron impact mass spectrum of the TMS ethers. The ions are formed by cleav- age of the carbon-carbon bond adjacent to the TMS ether group and are therefore specific for the different isomers (2,12). Quantitation from biological samples was accomplished on peak area ratios from the standard curves. Analysis of nonextracted standard curves indicated that

the area ratios did not differ significantly from the extracted standards. Since the standards are methyl esters rather than hydroxy fatty ac- ids esterified to lipids, certain errors may occur if the transmethylation process was not com- plete. However, we observed no significant in- crease in HETE or arachidonic acid when the sodium methoxide transesterification reaction was continued for 2.5 h compared with the normal l-h period.

The validity of this method for measuring lipid peroxidation occurring in vivo was stud- ied in mice that had been treated with hepa- totoxic doses of carbon tetrachloride. Carbon tetrachloride is metabolized in the liver to the trichloromethyl radical ( 1,15), which is thought to initiate the peroxidative process by the abstraction of a bisallylic hydrogen atom from polyunsaturated fatty acids. The reaction of the resulting lipid radical with oxygen then leads to the hydroperoxy- and, on reduction, the hydroxy-fatty acid esters that we have identified in previous studies (2).

Quantitation of the specific products, 1 l-, 12-, and 15HETE in total hapatic lipid, by GC-MS showed that these isomers were ele- vated in carbon tetrachloride-treated mice to levels 20-fold greater than those found in con- trol animals (Table 1). This is in agreement with our earlier studies in which lipid perox- idation was monitored in mouse liver phos- phatidylcholines by HPLC (2).

One of our goals in developing this GC-MS assay was to achieve sufficient sensitivity that lipid peroxidation could be measured in sub-

TABLE 1

THE EFFECT OF CARBON TETRACHLORIDE (2 ml/kg) ON LIPID HYDROXY ACID LEVELS IN MOUSE LIVER 30 min AFTER L&SING

wml/g nmol/g

20~4 1 I-OH 20:4 12-OH 20~4 I5-OH 20:4

Control (n = 3) 17.9 * 0.3 0.25 + 0.05 0.22 r 0.02 0.80 f 0.31 CCL In = 3) 17.8 2 0.3 7.97 + 2.16 7.93 k 2.53 11.80 f 5.27

Note. Results are expressed as means k SE.

LIPID PEROXIDATION PRODUCT CHROMATOGRAPHIC QUANTITATION 111

cellular fractions where limited biological sample was available. It has been proposed that a common pathway in the development of ir- reversible acute lethal cell injury involves per- turbation of intracellular calcium (Ca”) levels [see (16) for review]. One of several mecha- nisms controlling calcium levels within the cell is the plasma membrane calcium pump. Re- cent studies in our laboratory have demon- strated that plasma membrane vesicles from carbon tetrachloride-treated rats have an im- paired ability to accumulate calcium ( 17.18). We have used the GC-MS assay described here to study lipid peroxidation in plasma mem- brane vesicles from drug-treated animals. In rats the levels of lipid hydroxy acids in whole liver at 4 h were lower than those found in mice at $ h (Table 2) being 5- to lo-fold that of control animals. Figure 2 illustrates the se- lected ion profiles for the HETEs in lipids iso- lated from plasma membrane vesicles that had been prepared from carbon tetrachloride- treated Sprague-Dawley rats. Levels of 1 l-, 12-, and 15-HETE were 47, 39, and 85 pmol/ mg protein (mean of two experiments) which were 5- to 6-fold higher than control values. Studies are underway to determine whether this observed peroxidation of plasma mem- branes may result in increased permeability of membrane vesicles and thus account for their inability to accumulate calcium.

Since polyunsaturated fatty acids are very susceptible to autoxidation, we were con- cerned that the values of HETEs in control animals were overestimated, due to autoxi-

dation occurring during the workup proce- dure. To determine to what extent autoxida- tion products were contributing to control values, the nonpolar lipid band from rat liver extracts, eluting between 1 and 6 min from HPLC and containing methyl arachidonate free of hydroxylated derivatives, was evapo- rated to dryness, dissolved in chloroform: methanol (2:1), and carried through the workup procedure. The lipid hydroxy acid levels in these extracts were 68 * 46,59 + 40, and 122 -t 83 pmol/g liver (mean -+ SD, n = 3) of 1 1-, 12-, and 15-HETE, respectively. The arachidonic acid content of these extracts was 21.1 pmol/g. The autoxidation during workup, although accounting for less than 0.002% of the arachidonic acid present, con- tributes to the intra-assay variability. The coef- ficient of variation in control rat extracts was 18, 19, and 22% (n = 4) for 1 l-, 12-, and 15- HETE, respectively. These HETE levels were not significantly decreased when the antioxi- dant butylated hydroxytoluene (ca. 0.05%) was added during the workup. The susceptibility of these lipids to autoxidation makes it difficult to put an exact figure on the levels of HETEs present in hepatic lipids in viva However, since all samples are treated identically, and the standard errors on values determined in control livers are low, relatively small increases in lipid peroxidation products caused by ad- ministration of hepatotoxins can be deter- mined with this methodology.

The advantage of the present methodology over existing methods for measuring lipid per-

TABLE 2

THE EFFECT OF CARBON TETRACHLORIDE (2 ml/kg) ON LIPID HYDROXY ACID LEVELS IN RAT LIVER 4 h AFTER DOSING

wnol/g nmol/g

20:4 1 I-OH 20:4 12-OH 20:4 1 S-OH 20:4

Control (n = 4) 23.6 f 0.4 0.24 + 0.02 0.24 f 0.02 0.39 f 0.06 CCI, (n = 3) 17.9 + 0.7 1.14 kO.15 2.07 + 0.14 2.65 k 0.19

Note. Results are expressed as means + SE,

112 HUGHES

5 m/t 335

mlr 295

ml.? 225

Retention time (mid

FIG. 2. Selected ion chromatographs of the trimethylsilyl ether derivatives of methyl 1 I-, I2-, and 15HETE (m/z 225, 295, and 335, respectively) that were isolated from total lipid extracts of plasma membrane vesicles from car- bon tetrachloride-treated rats. The m/z 343.ion profile corresponds to the internal standard 15hydroxyarachidate which elutes later than the unsaturated analogs. Samples were analyzed on 3% SP2 100 as described under Materials and Methods. In this figure ion profiles are normalized. Measurements were made on relative peak area ratios.

oxidation is primarily one of specificity. This specificity is achieved using selected ion mon- itoring techniques for GC-MS analyses, and the method is therefore not as subject to in- terferences from other endogenous or exoge- nous compounds as are diene conjugation and fluorescence measurements. Although ethane and pentane determinations by gas chroma- tography are specific, these compounds rep- resent minor breakdown products of lipid hy- droperoxides. We are concerned that the ob- served increases in hydrocarbon expiration may, in some instances, represent either an increased rate of breakdown of the hydroper- oxides, rather than enhanced formation of

ET AL.

lipid hydroperoxides, or a decreased metabo- lism of the endogenously formed alkanes ( 19).

I.

2.

3.

4.

5.

6.

I.

8.

9.

10.

II.

12.

13.

14.

15.

16.

17.

18.

19.

REFERENCES Recknagel, R. O., Glende, E. A., and Hruszkewycz,

A. M. (1977) in Free Radicals in Biology (Pryor, W. A., ed.), Vol. 3, pp. 97-132, Academic Press, New York.

Hughes, H., Smith, C. V., Homing, E. C., and Mitch- ell, J. R. (1983) Anal. Biochem. 130,43 l-436.

Rao, S. K., and Recknagel, R. 0. (1968) Exp. Mol. Pathol. 9, 21 I-218.

Mimnaugh, E. G., Trush, M. A., and Gram, T. E. (198 1) Biochem Pharmacol. 30,2791-2804.

Burk, R. F., Lawrence, R. A., and Lane, J. R. (1980) J. Clin. Invest. 65, 1024-1031.

Kellogg, E. W., and Fridovich, I. (1975) J. Biol. Chem. 250,8812-8817.

Tappel, A. L. (1980) in Free Radicals in Biology (Pryor, W. A., ed.), Vol. 4, pp. l-47, Academic Press, New York.

Logani, M. K., and Davies, R. E. (1980) Lipids 15, 485-495.

Mitchell, J. R., Smith, C. V., Lauterburg, B. H., Hughes, H., Corcoran, G. B., and Homing, E. C. (1984) in Drug Metabolism and Drug Toxicity (Mitchell, J. R., and Homing, M. G., eds.), pp. 301-3 19, Raven Press, New York.

Hamberg, M., and Samuelsson, B. (1967) J. Biol. Chem. 242, 5329-5335.

Boeynaems, J. M., Brash, A. R., Oates, J. A., and Hubbard, W. C. (1980) Anal. B&hem. 104,259- 261.

Kraus-Friedmann, N., Biber, T., Murer, H., and Car- afoli, E. (1982) Eur. J. Biochem. 129,7-12.

Folch, J., Lees, H., Sloan, E., and Standley, G. H. (1957) J. Biol. Chem. 226,497-509.

Porter, N. A., Wolf, R. A., Yarbro, E. M., and Weenen, H. (1979) Biochem. Biophys. Res. Commun. 89, 1058-1064.

Trudell, J. R., Bosterling, B., and Trevor, A. J. (1982) Proc. Natl. Acad. Sci. USA 19, 2679-2682.

Trump, B. F., and Berezesky, I. K. (1984) in Drug Metabolism and Drug Toxicity (Mitchell, J. R., and Horning, M. G., eds.), pp. 261-300, Raven Press, New York.

Tsokos-Kuhn, J. O., Todd, E. L., McMillin-Wood. J. B.. and Mitchell, J. R. (1983) Clin. Res. 31,869A.

Tsokos-Kuhn, J. O., Todd, E. L., McMillin-Wood, J. B., and Mitchell, J. R. (1985) Mol. Pharmacol., 28,56-61.

Frank, H., Hintze, T., Bimboes, D., and Remmer, H. (1980) Toxicol. Appl. Pharmacol. 56, 331-344.