On the Mechanism of Oxidation of Cholesterol at C-7 in a Lipoxygenase System*

Incubation of [7-2H2Jcholesterol with soybean lipoxy- genase and linoleic acid in the presence of oxygen gave a mixture of 5-cholestene-3B,7a-diol, 5-cholestene-38, 78-dio1, 3~-hydroxy-5-cholesten-7-one,5a,6a-epoxy-cholestan-38-01, and 5~,6/3-epoxycholestan-3~-01. The conversion into the 7-oxygenated products was associated with a very high intermolecular isotope effect (KH/KD = 15-17), suggesting that the rate-limiting step in the overall conversion is likely to be the ab- straction of hydrogen at C-7 in a radical reaction. Evidence that linoleic acid is to some extent directly involved was obtained with the use of [7-3H]choles-terol. Incubation of [7-3H]cholesterol resulted in a significant incorporation of 3H in the reisolated lino- leic acid fraction. The isotope associated with conversion of in the lipoxygenase the extraction of hydrogen is nonstereospecific. Incubation of with tadecadienoic the above 7-oxygenated products with relatively small


4).
It is concluded that the most important mechanism for oxidation of cholesterol at C-7 in the lipoxygenase system involves participation of radicals and that a carbon-centered linoleic acid radical can extract hydrogen directly from cholesterol. Fatty acid hydroperoxides and their secondary products seem to be less important as initiators in connection with oxidation of cholesterol.
It is well established that peroxidation of membranes and other structures containing cholesterol and unsaturated fatty acids yields a number of oxygenated cholesterol products in addition to fatty acid hydroperoxides and their degradation products (1, 2). It has been shown that NADPHdependent lipid peroxidation in liver microsomes gives a mixture of 5-cholestene-3P,7a-diol, 5-cholestene-3@,7P-diol, 3P-hydroxy-5-cholesten-7-one and that the formation of these products parallels the formation of TBA-reacting com-* This work was supported by Grant 03X-3141 from the Swedish Medical Research Council, Axel and Margareth Axelsson-Johnsons Foundation, Konung Gustaf V:s och Drottning Victorias Stiftelse, Svenska Nationalforeningen mot Hjart-och Lungsjukdomar, and Svenska sallskapet for Medicinsk Forskning. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Supported by grants from Tore Nilssons Fond for Medicinsk Forskning and Stiftelsen Lars Hiertas Minne.
To whom correspondence should be addressed.
pounds (3, 4). Incubation of cholesterol with linoleate and lipoxygenase is also known to give the above monooxygenated products of cholesterol in addition to hydroperoxides of cholesterol and fatty acids (4, 5 ) . Detailed and careful studies by Teng and Smith (4, 5) have shown that oxidation of cholesterol in the above two systems yields the 7a-and 7P-hydroperoxides of cholesterol as primary products. These hydroperoxides are then reduced to the corresponding diols or dehydrated to give 3P-hydroxy-5-cholesten-7-one. In the lipoxygenase system there was no evidence of a direct interaction between cholesterol and the enzyme, indicating that intermediates or products of the lipoxygenase reaction on the unsaturated fatty acids may be responsible for the oxidation of cholesterol. Since peroxidation of polyunsaturated fatty acids by lipoxygenase is known to involve participation of free radicals, Teng and Smith suggested that such species may be of importance for oxidation of cholesterol. They also pointed out that the transformations were similar to radiation-induced free radical oxidation of cholesterol by molecular oxygen. The exact nature of the species responsible for oxidation of cholesterol in the two systems could, however, not be defined.
In the present work we have studied the mechanism of oxidation of cholesterol in a system containing lipoxygenase + linoleate. To obtain information about the rate-limiting step in the reaction and a possible transfer of hydrogen from cholesterol to linoleate, we have used cholesterol specifically labeled with 'H or 3H in the 7a-and the 7P-positions.
Synthesis of [(6),7,7-2H2,3 JCholesterol-500 mg of 3P-hydroxy-5a-cholestan-6-one and 250 mg of sodium methoxide were suspended in a mixture of 4.5 ml of ['H'lwater, 3 ml of tetrahydrofuran, and 3 ml of CH,O-['HI. The suspension was kept at 40 "C with continuous stirring for 15 h. After slight acidification with deuter-' The abbreviations used are: GC, gas chromatography; MS, mass spectrometry; HPLC, high performance liquid chromatography. ochloric acid, followed by dilution with water, the material was extracted with diethyl ether, washed with water, and dried with MgSO,. The solvent was then removed under reduced pressure. The product, [5,7,7-2H3]-3P-hydroxy-5a-cholestan-6-one, was acetylated in 3 ml of pyridine/acetic anhydride (2:1, room temperature, stirring overnight, work-up with diethyl ether and water). The acetate was then converted into cholesterol as described by Corey and Gregoriou (6), except that the dehydration with phosphorous oxychloride was carried out for 20 min and that sodium borodeuteride was replaced for sodium borohydride when [6,7,cholesterol was the desired end product. Typical overall yield was 25%. T h e isotopic purity was >98% when calculated according to Biemann (7).
Synthesis of 13-Hydroperoxy-9,ll -octadecadienoic acid-The synthesis was performed as described by Hamberg (8). Linoleic acid, 100 mg in 0.5 ml of ethanol, was dissolved in 100 ml of 0.1 M sodium borate buffer, p H 9.0, containing 0.745 g of KCl. Three portions (100,000 units each) of soybean lipoxygenase were added a t 10-min intervals. The suspension was kept at 0 "C with continuous stirring. T h e reaction was terminated with the addition of 200 ml of methanol after 30 min, followed by acidification with hydrochloric acid and extraction with 600 ml of diethyl ether. The ether phase was washed twice with water, dried with MgSO,, and evaporated in vacuo. The residue was purified on a 5-g SilicAR CC-4 silica column (Mallinckrodt). Linoleic acid was eluted with 150 ml of hexane/diethyl ether (90:10, v/v); 13-hydroperoxy-9,11-octadecadienoic acid was eluted with 150 ml of hexane-diethyl ether (70:30, v/v). The hydroperoxide was pure (>95%) as determined by HPLC with UV detection a t 234 n m (Nova-Pak C-18 cartridge, 4 pm, 8 X 100 mm, 1 ml/min of methanol/water/acetic acid 80:20:0.1, v/v/v) and by GC-MS of the compound reduced with triphenylphosphine and trimethylsilylated. Typical yield was about 15%.
Gas Chromatography-Gas chromatography of the trimethylsilylated compounds was performed using a n H P 5890 chromatograph equipped with a 15 m, 0.32 mm inner diameter, 0.25-pm phase thickness DB-5 (J&W Scientific, Folsom, CA) column. Temperature program: 220 "C, 1 min, 10 "C/min to 260 "C, 3-290 "C where the oven was kept for 10 min. All oxysterols studied were base line separated by the GC-MS as well as the GC chromatography system.
Thin Layer Chromatography-Thin layer chromatography of hydroperoxycholesterol was carried out on Silica Gel-60 plates (E. Merck, Darmstadt) in toluene/ethyl acetate 3:7 (v/v). The bands were visualized with sulfuric acid. In the case of radioactive cholesterol, the plates were scanned with a Berthold Tracemaster 20 thin layer chromatography scanner. Incubation of Cholesterol with Linoleic Acid and Soybean Lipoxygenase-The standard incubation was performed in the same way as the synthesis of 13-hydroperoxy-9,ll-octadecadienoic acid using half of the amounts. Incubation time was 30 min. Oxygen gas was blown over the solution, which was kept a t 37 "C with vigorous shaking. The reaction was carried out in an argon atmosphere as well. Cholesterol, 300 pg in 300 pl of acetone, was added immediately before the first addition of lipoxygenase. The extraction was performed without prior acidification, and the ether extract was, after evaporation in vacuo, purified with solid-phase cartridges (cf. below). For reasons of comparison, the incubations were also performed with ethyl linoleate as described by Teng and Smith (4).

Incubations
Incubation of /7-'H]Cholesterol with Soybean Lipoxygenase-[7-'H]Cholesterol was incubated with soybean lipoxygenase and linoleic acid as above. Following the termination of the reaction with methanol, the suspension was made alkaline to pH 13 with 33% sodium hydroxide and extracted twice with diethyl ether, and the recombined ether phases (alkaline phase) were washed with water. The initial suspension was acidified with 6 M hydrochloric acid to p H 4 and extracted with diethyl ether twice more, and the recombined ether phases (acid phase) were washed with water. The alkaline phase was evaporated, dissolved in a minimal amount of methanol, and subjected to semipreparative reversed-phase HPLC. A fraction corresponding to linoleic acid was collected from 10.7 to 15 min of retention time. This fraction could also be expected to contain some 5-cholesten-3@,7a-diol and more polar cholesterol oxygenation products such as cholestan-3/3,5a,60-triol. This partially purified fraction was rechromatographed with radio-HPLC using the normal-phase system described below.
The acid-phase which contained most of the material, was chromatographed in 15 portions using the straight-phase system described below. Fractions corresponding to linoleic acid were collected from 12.7 to 15.7 min. The recombined phases were subjected to reversed-phase analytical HPLC using the fatty acid system. The eluent was collected between 19 and 22 min and assayed for radioactivity in an LKB-Wallac Rackbeta 1217 scintillation counter. The incubation was also performed according to Teng and Smith (4). The oxidized sample was hydrolyzed for 12 h in 1 M NaOH in 80% EtOH at room temperature under argon atmosphere, extracted, and subjected to semipreparative reversed-phase HPLC as above. The fraction corresponding to linoleic acid was rechromatographed with straight-phase radio HPLC (Fig. 3).
Incubation of Cholesterol with 13-Hydroperoxy-9,ll-octadecadienoic Acid-Typically, 300 pg of a cholesterol mixture was dissolved in 3 ml of ethanol together with 3 mg of 13-hydroperoxy-9,11octadecadienoic acid. The solution was kept in a closed container a t 100 "C for 4 h , or at 37 "C overnight.
The solvent was evaporated, and the steroids were purified with solid-phase cartridges (cf. below). The incubation was also performed at 100 "C in borate buffer containing linoleic acid as in the lipoxygenase system with and without 50 mM of ferrous sulfate. Unless otherwise stated, data refer to the conditions first mentioned.
Incubation of Cholesterol with Acetone Powder of Rat Liver Microsomes-Acetone powder of liver microsomes from male Sprague-Dawley rats (200-250 g) was prepared as described previously (IO), except that no EDTA was used in the homogenizing buffer. The incubations with cholesterol were performed as described previously (11).
Treatment of Cholesterol with Singlet Oxygen-800 pg of cholesterol were dissolved in 2 ml of pyridine containing 0.6 mg of hematoporphyrin. The solution was irradiated with a 254-nm UV source with continuous oxygen bubbling for 2 h. The temperature was kept at 5-10 "C. After evaporation of the solvent under reduced pressure, the residue was purified on thin layer chromatography. The 5a-hydroperoxy-6-cholesten-3@-01 thus obtained was allowed to stand overnight in chloroform in order to get 7a-hydroperoxy-5cholesten-3@-01 (12). Samples were drawn at different times and analyzed with GC-MS, in most cases upon reduction with lithium aluminium hydride.
Purification of Steroids with Solid-Phase Cartridges-The steroids were purified using Bond-Elut NH, cartridges (Analytichem Corp.) as described by Kaluczny et al. (13). Only the neutral lipid fraction was collected.
In order to distinguish between the 3@-hydroxy-5-cholesten-7one formed from [7,7-'H2]cholesterol and that formed from unlabeled cholesterol, [26,26,26,27,27,27-'H6]cholesterol was used as a 7-unlabeled substrate instead of unlabeled cholesterol. Since the incubation mixture sometimes was contaminated to a very small degree with unlabeled cholesterol (in incubations with acetone powder, to a high degree, about lo%), it seemed advisable to have an additional deuterium label in the 7,7-'H2-labeled species in order to facilitate the mass spectrometric analysis. As for [6,7-'H2]5-cholestene-3@,7(a and @)-diol, part of the 6-as well as the 7-label is lost in the mass spectrometric fragmentation of the molecular ion to the M-TMSOH ion used for measurement. The reason for using this ion was its very high abundance compared to the molecular ion. By using synthetic unlabeled 6,7a-'H2-labeled 5-cholestene-3@,7@-diol and 6,7@-'H2-labeled 5-cholestene-36,7adiol as standards, it was possible to determine the exact isotope pattern of each species of interest. Using In the calculation of isotope effects, the mass spectrometric intensities from 'H6-labeled 5-cholestene-3(3,7-diol was divided by the mass spectrometric intensities recorded from *H,-labeled 5cholestene-3@,7-diol (after correction for presence of very small amounts of unlabeled compound in some experiments). This gives a crude measure of the KH/KD value. In case there was a small difference between the amount of [7-'H2]-and [7-2H2]cholesterol used as substrate, the KH/KD value was corrected for this. In the calculation of isotope effect, it must be assumed that the amount of product formed is small in relation to the amount of substrate. In all the experiments performed in which the isotope effect was calculated, the degree of conversion was always less than 10%. The isotope effect in the conversion into 3@-hydroxy-5-cholesten-7-one was calculated accordingly with the difference that the mass spectrometric intensities corresponding to 'H,-labeled product was recorded.

Extent of conversion (%) of [4-'4CJchoksterol into monooxygenated products under the different conditions
For experimental details, see "Materials and Methods." The extent of conversion was calculated after radio HPLC analysis as shown in Fig. 1, A-C. The results shown are those obtained from one typical experiment. Several experiments with each incubation condition gave essentiallv the same results.   be 15-17 in the conversion into the three 7-oxygenated products.

5-Cholestene-3&7a-diol+ 3p-Hydroxy-5-5a,Ga-Epoxy-5@,6@-Epoxy-
When the above experiment was repeated in the absence of oxygen, there was no significant conversion into oxygenated products. Addition of Fez+ to 50 PM did not further increase the degree of conversion (results not shown). Addition of EDTA to 50 PM did also not affect the conversion.
[4-'4C]Chole~ter~l as well as the above mixture with deuterated cholesterol species was also incubated with isolated 13-hydroperoxy-9,11-octadecadienoic acid under the conditions described under "Materials and Methods" (Fig. 1B). There was only a low yield of different oxysterols (Table I).
The intermolecular isotope effect measured in the conversion of [6,7,7,-2H3]cholesterol into the 7-oxygenated products was much lower than those obtained in the lipoxygenase system. As above, there was no significant intermolecular isotope effect in the conversion of [6,7,7-2H,]cholesterol into the two epoxides (Table 11).
For reasons of comparison, [4-'4C]cholesterol and the mixture of deuterated cholesterol species were also incubated with delipidated rat liver microsomes + NADPH (Table I, Fig. IC). The intermolecular isotope effects obtained here in the conversion into the 7-oxygenated products were lower than those obtained in the lipoxygenase system but higher than those obtained with the isolated 13-hydroperoxy-9,lloctadecadienoic acid (Table 11). When the above cholesterol species were treated with singlet oxygen as descibed under "Materials and Methods," there was a small conversion into 5-cholestene-3@,7a-diol and 3p-hydroxy-5-cholesten-7-one with little or no isotope effect.
In order to study stereospecificity aspects, a mixture of [*H6]cholesterol and [6,7a-*H2]cholesterol was incubated with the lipoxygenase system and linoleic acid ( Table 11). The isotope effect obtained in two different experiments varied between 2 and 3.
In order to study the mechanism of extraction of hydrogen from cholesterol in the lipoxygenase system, [7-3H]cholesterol was incubated with lipoxygenase and linoleic acid. Linoleic acid and linoleic acid products were separated from cholesterol and its product as described under "Materials and Methods" (reversed-phase semipreparative HPLC followed by straight-phase HPLC or vice versa). 3H-Labeled material was eluted with the same retention volume as linoleic acid (Fig. 3). It was uncertain, however, whether all the 3H-containing fatty acid was linoleic acid or its conjugated diene, i.e. octadeca-9,ll-dienoic acid. Under the chromatogaphic conditions employed, these two fatty acids did not separate, and a compound with strong absorption at 234 nm eluted at a retention close to that of linoleic acid. The total recovery of 3H in the linoleic acid fraction was only about 10,000 cpm in an incubation with 250 Mcpm of [7-3H] cholesterol. Since the total conversion of the labeled cholesterol into 7-oxygenated products was about 5% and since only one of the two H atoms would be extracted in the conversion, about 5 Mcpm could be expected to be lost from [7-3H]cholesterol if there were no isotope effect at all in the conversion.
Since the tritium isotope effects could be expected to be at least 25-50 in the loss of hydrogen from cholesterol (14) this would correspond to an expected loss of H of about 100,000 cpm. There may, however, well be an isotope effect also in the addition of hydrogen from cholesterol to a linoleic acid radical, and thus it was not possible to calculate the exact degree of transfer of hydrogen from cholesterol to linoleic acid radicals. Teng and Smith (5) showed that oxidation of cholesterol by soybean lipoxygenase in incubations including ethyl linoleate as prime substrate gave the epimeric 7a-and 7phydroperoxides in the proportion 1:3 to 2:3. The 7-hydroperoxides were apparently the first formed oxidation products  of cholesterol with the epimeric 3@,7-diols and the 7-ketone as secondary products. We also found 5a76a-and 5~,6P-epoxycholestan-3P-o1 as products of cholesterol in the lipoxygenase reaction. The 5,g-epoxides formed from [7-2Hz]cholesterol had retained all the 2H. This as well as the absence of an isotope effect (cf. below) clearly excludes the possibility that the above hyperperoxides are intermediates in the reaction. On the basis of previous work, Smith (15) has suggested that cholesterol 5,6epoxides may be formed directly by attack of hydroperoxide on cholesterol.

DISCUSSION
The aim of the present work was to elucidate the initiation step in connection with the oxidative attack at C-7 in cholesterol in the lipoxygenase system. Since the 5a-saturated analogue of cholesterol, cholestanol, is not oxidized in the lipoxygenase system' it is evident that this attack requires allylic hydrogens a t C-7. Since the overall oxidation of cholesterol by the lipoxygenase system is dependent upon addition of linoleate and since the enzyme is not directly involved (4), it appeared possible that formation of 13hydroperoxy-9,ll-octadecadienoic acid is a prerequisite for the conversion. In accordance with this possibility, incubation of cholesterol with the above hydroperoxide gave all three 7-oxygenated products. The yield was, however, very low in relation to that obtained with linoleate and lipoxygenase. In addition the isotope effect in the hydroperoxidedependent oxidation of [7-2H2]cholesterol was much lower than that obtained with linoleate and lipoxygenase. It can thus be concluded that presence of 13-hydroperoxy-9,ll-'' E. Lund, U. Diczfalusy, and I. Bjorkhem, unpublished results.
octadecadienoic acid is not of major importance for the oxidation at C-7 of cholesterol in the lipoxygenase system.
A more likely mechanism is then that carbon-centered or oxygen-centered linoleic acid radicals may initiate the reaction by extracting one of the allylic hydrogens in cholesterol.
A significant transfer of 3H from [7-3Hz]cholesterol to linoleic acid (or its conjugated isomer) was found, suggesting at least some participation of a carbon-centered linoleic acid radical. It should be pointed out that transfer of 3H to a peroxyl radical or to an oxy1 radical yielding alcohol or hydroperoxide would lead to loss of 3H and go undetected.
Extraction of hydrogen from an unsaturated fatty acid in connection with a lipoxygenase reaction is known to be associated with a marked isotope effect, and KH/KO values of 8-9 have been reported (16,17). The intermolecular isotope effect observed here in connection with conversion of [7-*H2]cholesterol into the 7-oxygenated products was of the same magnitude, as could be expected if the extraction of hydrogen from cholesterol involves a radical reaction. When extraction of hydrogen is the rate-limiting step in a reaction, a marked isotope effect can be expected if this hydrogen is replaced with deuterium or tritium (14,18). If the substituted hydrogen is lost in a step which is not rate limiting, only small isotope effects can be expected. It has been stated that the lower limit of primary isotope effects of deuterium which may be interpreted as identifying the slowest or rate-limiting step in a multistep reaction is about 8 (19). The KH/Kn values obtained in the lipoxygenase reaction were, however, well above 8 in all experiments performed. It is thus tempting to suggest that extraction of one of the allylic hydrogens from C-7 in cholesterol is the ratelimiting step in the oxidation of cholesterol under the conditions studied here. It may be mentioned in this connection that monooxygenation of [7a-2H]cholesterol catalyzed by the liver microsomal cholesterol 7a-hydroxylase is not associated with an isotope effect, indicating a different type of reaction (20).
T h e very high degree of discrimination between unlabeled and 7-*H2-labeled cholesterol in the abstraction of one of the allylic hydrogens in the lipoxygenase system is probably dependent upon the specific properties of the attacking radical and the degree of hydrophobicity of the surroundings. We have failed to demonstrate such a high isotope effect when oxidizing [7-*H2]cholesterol in nonenzymatic model systems containing a synthetic radical initiator. In preliminary experiments we used 4,4'-azobiscyanovaleric acid as a radical generator and obtained isotope effects in the oxidation of [7-'Hz]cholesterol varying between 3 and 6, depend-ing on the solvent used and whether the free acid or the ethyl ester was used. The highest isotope effects were obtained under the most hydrophobic conditions with a lipophilic solvent and with an esterified reagent. Additionally, in the lipoxygenase system the interaction between cholesterol and the linoleic acid radical must occur in a lipophilic milieu since the cholesterol is solubilized in linoleate micells.
This situation is similar to that occurring in vivo where cholesterol is invariably associated with unsaturated fatty acids in membranes and lipoproteins and where a substantial part of the cholesterol may also be esterified to an unsaturated fatty acid.
In principle this may be due to a nonstereospecificity in the abstraction of hydrogen, to a nonstereospecificity in connection with the addition of oxygen, or to an epimerization of the initially formed 7-hydroperoxides. The fact that the isotope effect in the conversion of stereospecifically labeled [7a-2H]cholesterol into the two 7-diols was reduced to 2-3 shows that the initial extraction of one of the two allylic hydrogens must be nonstereospecific. The finding that the 7-diols were formed to the same extent suggests that the oxygen addition is nonstereospecific.
The possibility has been discussed that singlet molecular oxygen is involved in the soybean lipoxygenase-catalyzed oxidation of linoleate (for a discussion, see Ref. 5). After exposing a mixture of [7-ZH2]cholesterol and unlabeled cholesterol to singlet oxygen, there was a significant formation of 5-cholesten-3/3,7a-diol and 3P-hydroxy-5-cholesten-7one. These conversions occurred without significant isotope effect, indicating that the mechanism was different from that involved in the 7-oxygenation of cholesterol in the lipoxygenase system. This is in accord with previous results by Teng and Smith (5) and supports the contention that a singlet oxygen mechanism is of little or no importance in connection with oxidation of cholesterol in the lipoxygenase system.