24(S),25-Epoxycholesterol EVIDENCE CONSISTENT WITH A ROLE IN THE REGULATION OF HEPATIC CHOLESTEROGENESIS*

Herein we demonstrate that the nonsaponifiable extract from human liver tissue contains 24(S),25-epox-ycholesterol in an amount approximately relative to cholesterol. We show that 24(S),25-epoxycholes-terol, like many other oxygenated sterols, represses hydroxymethylglutaryl-CoA reductase activity in cultured cells and binds to the cytosolic oxysterol-binding protein. Furthermore, we show that this epoxide eluent, which gave cleaner separation of the peak thought to be the epoxide. Four repetitions of the latter process gave material which was homogeneous by HPLC. This material was used for mass spectroscopy and in the following experi- ments designed to establish the identity of the epoxide by demonstrat-ing: 1) its lability upon mild acid treatment, and 2) its conversion to 25-hydroxycholestero1 by treatment with lithium aluminum hydride. with two 5-ml washes of acetone, the solvent was evaporated under nitrogen, and the sterols were resuspended in 0.25 ml of 9O:lO CH30H:H20 for HPLC on a Waters instrument. The sterols were chromatographed in 90:lO CH3OHH20 at a flow rate of 1 ml/min on a Resolve 5-pm C18 column, and fractions were assayed for 3H. The radioactivity recovered after HPLC was 83% of that in the washed cells.

It was found by us (1,2 ) and by others (3,4) that inhibitors of oxidosqualene cyclase cause an accumulation of squalene 2,3(S),22(S),23-dioxide as well as squalene 2,3(S)-oxide during sterol biosynthesis. We then demonstrated that squalene 2,3(S),22(S),23-&0xide is converted by SI, rat liver homogenate to 24(S),25-epoxycholesterol via 24(S),25-oxidolanosterol, approximately as efficiently as squalene 2,3(S)-oxide is converted to cholesterol (5). We found further that incubation of acetate with rat liver homogenate in the absence of any cyclase inhibitor results in the formation of approximately 5% of 24(S),25-epoxycholesterol relative to cholesterol (6). These and other (7) observations suggest that sterol epoxides formed from squalene 2,3(S),22(S),23-dioxide may have biochemical significance, particularly since numerous cholesterol derivatives with oxygenated side chains are effective repres-*This research was supported by National Institutes of Health Grants HL 23083, CA 02758, and HL 05360. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
5 To whom correspondence should be addressed. sors of HMG'-CoA reductase, and, consequently, of cholesterol biosynthesis (8). Accordingly, it is important to determine whether production of such epoxy sterols occurs under physiological conditions, and whether these sterols play a role in cholesterol regulation.
In this study, we demonstrate that 24(S),25-epoxycholesterol is present in human liver. We also show that this epoxide, like sterols with side chain hydroxyl or keto groups, represses HMG-CoA reductase in cultured cells and binds to the cytosolic oxysterol-binding protein (9). Incubation of tritiated 24(S),25-epoxycholesterol with cultured fibroblasts leaves the epoxide largely unchanged, with no detectable conversion into 25-hydroxycholesterol, 24-hydroxycholesterol, desmosterol, or cholesterol. These results support a hypothesis that metabolic generation of 24(S),25-epoxycholesterol may be a mechanism for the regulation of cholesterol biosynthesis. Radiolabeled 24(S),25-epoxycholesterol was synthesized as follows. According to the method of Ikekawa and co-workers (lo), cholenic acid (Steraloids, Inc.) was converted to its tetrahydropyranyl ether, which was reduced with lithium aluminum hydride in tetrahydrofuran and then oxidized with pyridinium chlorochromate to afford the corresponding C-24 aldehyde as its tetrahydropyranyl ether. This material was labeled with 3H by triethylamine-catalyzed CY proton exchange according to the procedure of Barton et al. (ll), to afford 23-[3H]aldehyde which was treated with isopropylidene triphenylphosphorane, by a modification of a procedure used previously to prepare 25-['4C]desmosterol (12), to afford 23-[3H]desmosterol with specific activity = 94,000 dpm/pg. By the same procedure used with the unlabeled desmosterol, this material was converted to 23-[3H]24(S),25-epoxycholesterol. This labeled epoxide was purified before use by reverse-phase HPLC on a Resolve 5-pm CIS column (Waters) using 8 8 1 2 CH30H:H20 as solvent to give a single symmetrical radioactive peak with specific activity = 93,000 dpm/pg. 25-Hydroxycholesterol was purchased from Steraloids, Inc. 24-Hydroxycholesterol was prepared as a mixture of R and S epimers by NaBH4 reduction of 24-ketocholesterol (Analabs, Inc., North Haven, CT); this epimeric mixture, which was not resolved by the reverse-phase HPLC conditions employed, was used to test whether either 24(S)-hydroxycholesterol, the chemically plausible product of reduction of 24(S),25-epoxycholesterol, or 24(R)-hydroxycholesterol was being formed. 4-['4C]Chole~tero1 was purchased from Amersham Corp.

Sterols-Unlabeled
Analysis of Human Liver for 24(S),25-Epoxycholesterol-Samples from two human livers were obtained from Dr. James Trudell of the Department of Anaesthesia, Stanford University School of Medicine. The samples were from donors in the Stanford Heart Transplant Program and were obtained as rapidly as possible following removal of the heart for cardiac transplantation. Donors were on cardiac bypass up to the time of excision of the heart. The liver samples were frozen at liquid nitrogen temperature immediately after excision and were kept frozen until they were saponified as described next. The ' The abbreviations used are: HMG, 3-hydroxy-3-methylglutaryl; HPLC, high performance liquid chromatography. protocol of Dr. Trudell was approved by the Stanford Human Subjects Experimentation Committee.
To a sample of liver (approximately 75 g, wet weight) which had been chopped into small pieces was added approximately 700 ml of ethanol and approximately 100 ml of 33% KOH in HzO, and the mixture was heated under reflux at 80 "C for 4 h. The cooled mixture was exhaustively extracted with petroleum ether. The extracts were washed with Hz0 and evaporated to afford approximately 1 g of nonsaponifiable extract. A 30-40-mg sample of this extract was dissolved in 3 ml of ethyl acetate and put through a Sep-Pak CIS cartridge (Waters). The eluted sample was then subjected to preparative TLC on Silica Gel 60 PFw4 +366 (EM Reagents, Cincinnati, OH) using three developments with 4 3 hexane:ethyl acetate. A broad band corresponding to cholesterol, which would contain any 24(S),25epoxycholesterol present, was scraped, eluted with ethyl acetate, and passed through a Sep-Pak Cl8 cartridge. Evaporation afforded material which was analyzed by reverse-phase HPLC on a Perkin-Elmer Series 3 instrument with a Perkin-Elmer LC-65T detector using UV light at 205 nm on an Ultrasphere ODS 5-pm, 4.6 X 250-mm column (Beckman Instruments) using 95:5 or 9O:lO CH30H:0.1% KzC03 in H20 at 30 "C with a flow rate 1.0 ml/min for 20 min and then 2.0 ml/ min. The 0.1% K&03 solution was used to prevent acid-catalyzed hydrolysis of the epoxide. Coinjection with authentic 24(S),25-epoxycholesterol was used to confirm identity of retention times.
The material with the retention time of 24(S),25-epoxycholesterol was collected from each of several injections with 95:5 CH30H:H20. These collected fractions were combined, reinjected, and recollected, with 9010 CH30H:HZO as eluent, which gave cleaner separation of the peak thought to be the epoxide. Four repetitions of the latter process gave material which was homogeneous by HPLC. This material was used for mass spectroscopy and in the following experiments designed to establish the identity of the epoxide by demonstrating: 1) its lability upon mild acid treatment, and 2) its conversion to 25-hydroxycholestero1 by treatment with lithium aluminum hydride.
1) A sample sufficient to give a readily detectable HPLC signal of the material thought to be 24(S),25-epoxycholesterol (approximately 1 pg) was dissolved in 500 pl of tetrahydrofuran, and 2 drops of water and 100 p1 of 60% perchloric acid were added. The mixture was stirred at room temperature for 1 h. Ethyl acetate (10 ml) was added, and the mixture was washed with 1 ml of aqueous sodium bicarbonate and 1 ml of brine. Evaporation of the organic layer afforded a residue which was subjected to preparative TLC and HPLC analysis as described above. A control experiment was performed identically except that no perchloric acid was added.
2) Another sample of putative epoxide from liver was dissolved in 3 ml of freshly distilled dimethoxyethane at 0 "C under N,, and 20 mg of lithium aluminum hydride was added. The resulting suspension was heated at reflux for 72 h under N,, with care being taken to maintain the solvent level. The mixture was cooled and then treated with 3 ml of ethyl acetate, followed by 0.5 ml of HzO. The suspension was exhaustively extracted with ethyl acetate, and the organic extracts were evaporated to afford a residue which was subjected to preparative TLC and HPLC analysis as described above.
Normal-phase HPLC, used only to separate 24(S),25-epoxycholesterol and 24(R),25-epoxycholesterol, was performed under the same instrumental conditions used for reverse-phase analysis of liver samples on an Ultrasphere-Si 5-pm, 4.6 X 250-mm column (Beckman Instruments) with 98.81.2 hexane:isopropyl alcohol as eluent at a flow rate of 1.0 ml/min.
In order to determine the amount of material with the retention time of 24(S),25-epoxycholesterol relative to the amount of cholesterol, HPLC response curves for authentic samples of 24(S),25epoxycholesterol and cholesterol were determined. The intensity of the HPLC response at 205 nm was identical for the two compounds and was directly proportional to the amount of compound injected in both cases. A Hewlett-Packard H P 3380A integrator was then used to determine the ratio of epoxide to cholesterol in the liver samples at the earliest stage in the chromatographic purification described above at which an unobscured peak for the epoxide could be observed, which for both liver samples was usually, but not always, after initial passage of the nonsaponifiable extract through the Sep-Pak CIS cartridge.
Control experiments were run as follows to establish that the ratio of 24(S),25-epoxycholesterol to cholesterol thus determined was an accurate reflection of the ratio in intact liver. To the saponification mixture for a sample of liver were added known amounts of 23-[3H] 24(S),25-epoxycholesterol and 4-[14C]cholesterol. The nonsaponifia-ble extract was prepared exactly as usual, and an aliquot of the extract was analyzed for 3H and 14C. The per cent of the 3H and 14C having the TLC mobility of 24(S),25-epoxycholesterol and cholesterol was determined with another aliquot. The same analyses were repeated after the nonsaponifiable extract was passed through a Sep-Pak CI8 cartridge and after preparative TLC as described above.
Repression of HMG-CoA Reductase and Binding to the Oxysterolbinding Protein-Repression of HMG-CoA reductase activity in cultured L cells (mouse fibroblasts) by 24(S),25-epoxycholesterol was assayed as described (9). The relative binding affinity of the sterol for the oxysterol-binding protein was determined by measuring its ability to displace 25-[3H]hydroxycholesterol from specific sites. Conditions for preparation of the oxysterol-binding protein from L cells, for binding of the ligand at pH 5.5 or 7.4, and for assay of the complex by velocity sedimentation in a sucrose gradient were as described (9).
Metabolism of 24(S),25-Epoxycholesterol-Approximately 2 X IO7 cells from an L cell spinner culture (9) were pipeted into a 150-cm2 culture flask (Corning) and incubated for 2 h at 37 "C in a 5% CO, incubator. The cells attached to the flask's surface during this time, and the medium was then decanted and replaced with 10 ml of fresh medium containing 1 pg/ml vitamin E. The sterol solution was prepared by mixing 0.1 ml of an ethanol solution of 23-[3H]24(S),25epoxycholesterol with 1 ml of medium containing 5% bovine serum albumin (Sigma, essentially fatty acid-free). This mixture was then added to the culture medium to give a final concentration of epoxide of 0.37 pg (30,000 dpm)/ml. After 5 h of incubation at 37 "C, the cells were scraped into the medium, collected by centrifugation, and washed twice with 5 ml of 0.14 M NaCl. The sterols were extracted by resuspending the cells in 2 ml of 0.14 M NaCl and adding 5 ml of methanol, with mixing, followed by 2.5 ml of chloroform to form a single phase (13). After 30 min, the cell debris was removed by centrifugation, 2.5 ml each of chloroform and 0.14 M NaCl were added and mixed, and the phases were separated by centrifugation. The lower chloroform phase was dried under a stream of nitrogen, and the residue was dissolved in 1 ml of hexane:chloroform (19:l). This was applied to a Sep-Pak silica cartridge which was then washed with 2 5-ml volumes of hexane:chloroform. The sterols were eluted from the silica with two 5-ml washes of acetone, the solvent was evaporated under nitrogen, and the sterols were resuspended in 0.25 ml of 9O:lO CH30H:H20 for HPLC on a Waters instrument. The sterols were chromatographed in 90:lO CH3OHH20 at a flow rate of 1 ml/min on a Resolve 5-pm C18 column, and fractions were assayed for 3H. The radioactivity recovered after HPLC was 83% of that in the washed cells.

Analysis of Human Liver for 24(S),25-Epoxycholesterol-
Nonsaponifiable extracts from samples of two human livers were purified chromatographically as described under "Experimental Procedures" and were analyzed by reverse-phase HPLC. A peak was found which had a retention time identical with that of authentic 24(S),25-epoxycholesterol when either 95:5 or 9O:lO methano1:water was used as eluent. When 95:5 methano1:water was used, typical retention times were for 25hydroxycholesterol, 10.5 min; 24(S),25-epoxycholesterol, 13.5 min; desmosterol, 30.0 min; and cholesterol, 39.3 min. A representative HPLC trace of the "cholesterol fraction" from one liver is shown in Fig. 1.
Chemical evidence that the peak which co-migrated with 24(S),25-epoxycholesterol was indeed that substance was obtained as follows. The co-migrating peak was collected and purified until it was homogeneous by HPLC, as described under "Experimental Procedures." A sample of this purified material was treated with perchloric acid in tetrahydrofuran and, as would be expected if it were 24(S),25-epoxycholesterol (6,14), the product from this reaction showed no HPLC peak with the retention time of the S-oxide. This experiment was repeated twice with the same result, and a control reaction was carried out under identical conditions, except that no perchloric acid was added, in which case the S-oxide peak did not disappear.
Another sample of the putative purified 24(S),25-epoxycho- lesterol was treated with lithium aluminum hydride in an attempt to confirm its identity by showing that such treatment would convert it to 25-hydroxycholesterol, as we had demonstrated previously (5,6). HPLC analysis of the product from this reaction did not show an S-oxide peak but did show a peak which had the same retention time as authentic 25hydroxycholesterol. This experiment was repeated twice with the same results.
Confirmation that the substance isolated from liver was indeed 24(S),25-epoxycholesterol was obtained by mass spectroscopy. The purified liver-derived material showed the same molecular ion at m/z 400 and the same fragmentation pattern, including the diagnostic peaks at m/z 382 (M-H20) and at m/ z 367 (M-H,O-CH,), as authentic 24(S),25-epoxycholesterol.
However, none of the results described so far, including the mass spectroscopic data, distinguish between 24(S),25-epoxycholesterol and its epimer, 24(R),25-epoxycholesterol. Thus, it was considered necessary to establish that the substance isolated from liver was indeed the S-oxide. This was accomplished by use of normal phase HPLC under the conditions described under "Experimental Procedures" which effected resolution of the epimeric epoxides sufficient to permit demonstration by coinjection experiments that the material from liver was indeed 24(S),25-epoxycholesterol. Typical retention times were for 24(S),25-epoxycholesterol, 21.7 min; and 24(R),25-epoxycholesterol, 22.1 min.
Although there is no reason to believe that the 24(S),25epoxycholesterol found in liver could have arisen by autoxidation, we attempted to exclude this possibility by leaving methanol solutions of cholesterol and desmosterol exposed to air and then analyzing them by HPLC after periods of 2, 8, and 18 weeks. No detectable amount of 24(S),25-epoxycholesterol was formed from either cholesterol or desmosterol.
Integration of their respective HPLC peaks, as described under "Experimental Procedures," was used to measure the relative amounts of 24(S),25-epoxycholesterol and cholesterol in the liver samples. After initial passage of the nonsaponifiable extract from one liver through a Sep-Pak C18 cartridge, the amount of 24(S),25-epoxycholesterol relative to the amount of cholesterol was 0.40 f 0.027% (three determinations). The analogously determined ratio in the extract from the second liver was 0.13 f 0.0078% (three determinations). Control experiments using radiolabeled materials, as de-scribed under "Experimental Procedures," showed that the recovery of both cholesterol and 24(S),25-epoxycholesterol was quantitative after preparation of the nonsaponifiable extract and passage through the Sep-Pak CIS cartridge, so these figures should be an accurate determination of the ratio of these substances in intact liver.
Confirmation of these results was provided by similar determinations of the ratio of 24(S),25-epoxycholesterol to cholesterol after subjection of the nonsaponifiable extracts further to normal-phase preparative TLC. The ratio of 24(S),25epoxycholesterol to cholesterol in the first liver sample was 0.35 * 0.072% (three determinations) and in the second liver sample was 0.091 f 0.016% (three determinations). The control experiments indicated that some selective loss of epoxide could occur upon preparative TLC, which may account for these latter ratios being slightly lower than those determined after passage through the Sep-Pak C18 cartridge.
Biochemical Assays and Metabolism Experiments-As shown in Fig. 2,24(S function at pH 5.5. The similar results obtained under both conditions suggest that this was not an important factor in the assay. The concentrations of the epoxide required to suppress HMG-CoA reductase by 50% (0.89 PM) and to displace 50% of the 25-[3H]hydr~xychole~ter~l from the binding protein (0.55 PM) can be compared with published values for a wide range of oxysterols (9). The activity of 24(S),25epoxycholesterol is intermediate between the least and most active sterols tested; it is approximately one-seventh that of 25-hydroxycholestero1.
To try to make certain that the repression of HMG-CoA reductase in L cells was caused by 24(S),25-epoxycholesterol itself and not by a metabolite, [3H]24(S),25-epoxycholesterol was incubated for 5 h with L cells, and the sterol fraction was extracted and analyzed by HPLC. The distribution of recovered radioactivity is shown in Fig. 3. About 88% of the counts were found as unaltered epoxide. The remaining radioactivity was scattered in the highly polar fractions; no significant amounts with the retention times of 25-hydroxycholesterol, 24-hydroxycholestero1, desmosterol, or cholesterol were detected.
These results suggest that the repression of HMG-CoA reductase is indeed caused by 24(S),25-epoxycholesterol and not by a product formed from it intracellularly. The fact that the activity measured in the binding assay is in good agreement with that determined by the reductase repression assay further supports the conclusion that the epoxide itself represses HMG-CoA reductase. By use of literature values indicating that the concentration of cholesterol in liver is approximately 2.5 mg/g, wet weight (15), and that water constitutes about 75% of the wet weight (16), the concentration of the epoxide can be estimated to be in the range of 10-30 PM. This range is well above the concentration of 0.89 PM required in a fibroblast cell culture to suppress HMG-CoA reductase activity by 50%. However, an unknown proportion of the epoxide present in liver may have been esterified and unavailable for a regulatory role. Furthermore, cultures of liver cells appear to be about 10 times more resistant than other cell cultures to the inhibitory effects of oxysterols (8). Nonetheless, despite the uncertainties involved in these estimates, the concentration of the epoxide in liver seems to be high enough to function in the regulation of HMG-CoA reductase. We have found (17) that similar levels of 24(S),25-epoxycholesterol, apparently adequate for regulation of the reductase, are synthesized from mevalonic acid in cultured fibroblasts, indicating that the epoxide could also play a regulatory role in non-hepatic cells.