The Structure of an Inhibitor of Cholesterol Biosynthesis Isolated from Barley*

Purification of the oily, nonpolar fraction of high protein barley (Hordeum vulgare L.) flour by high pressure liquid chromatography yielded 10 major components, two (I, 11) of which were potent inhibitors of cholesterogenesis in vivo and in vitro. The addition of purified inhibitor I (2.5-20 ppm) to chick diets significantly decreased hepatic cholesterogenesis and serum total and low density lipoprotein cholesterol and con-comitantly increased lipogenic activity. The high res- olution mass spectrometric analysis and measurement of different peaks of inhibitor I gave a molecular ion at m/e 424 (CZsH4,O2) and main peaks at m/e 205,203, and 165 corresponding to C13H1702, C13H1602, and C,,H1302 moieties, respectively. which are characteristic of d-a-tocotrienol. This identification was con- firmed against synthetic samples. The tocotrienols are widely distributed in the plant kingdom and differ from tocopherols (vitamin E) only in three double bonds in the isoprenoid chain which appear to be es- sential for the inhibition of cholesterogenesis. studies show that of the cereal grains, barley is the most effective in lowering blood cholesterol levels of experimental animals (5-10). p-D-Glucans, the principal fiber component of barley endosperm might influence cholesterol

Purification of the oily, nonpolar fraction of high protein barley (Hordeum vulgare L.) flour by high pressure liquid chromatography yielded 10 major components, two (I, 11) of which were potent inhibitors of cholesterogenesis in vivo and in vitro. The addition of purified inhibitor I (2.5-20 ppm) to chick diets significantly decreased hepatic cholesterogenesis and serum total and low density lipoprotein cholesterol and concomitantly increased lipogenic activity. The high resolution mass spectrometric analysis and measurement of different peaks of inhibitor I gave a molecular ion at m/e 424 (CZsH4,O2) and main peaks at m/e 205,203, and 165 corresponding to C13H1702, C13H1602, and C,,H1302 moieties, respectively. which are characteristic of d-a-tocotrienol. This identification was confirmed against synthetic samples. The tocotrienols are widely distributed in the plant kingdom and differ from tocopherols (vitamin E) only in three double bonds in the isoprenoid chain which appear to be essential for the inhibition of cholesterogenesis. studies show that of the cereal grains, barley is the most effective in lowering blood cholesterol levels of experimental animals (5-10). p-D-Glucans, the principal fiber component of barley endosperm might influence cholesterol excretion (11). However, a major cholesterol-suppressive action of barley is at the level of HMG-CoA reductase (5-10). High protein barley flour (HPBF), the commercial pearling fraction consisting of the aleurone, subaleurone, and germ, is the richest source of the cholesterol-suppressive factors (12,13). Sequential extraction of HPBF with petroleum ether, ethyl acetate, methanol, and water remove all cholesterol-suppressive activity from HPBF; each of the solvent fractions contains HMG-CoA reductase-suppressive material (14). In this report, we describe the isolation of two cholesterol-suppressive metabolites, cholesterol inhibitors I and 11, from the petroleum ethersoluble fraction (PESF) of HPBF. The determination of the structure of cholesterol inhibitor I and the results of studies in vivo and in uitro confirming its cholesterol-suppressive action are reported.

EXPERIMENTAL PROCEDURES
It is well established that 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA)' reductase (EC 1.1.1.34) is the first ratelimiting enzyme in the biosynthetic pathways for cholesterol and other isoprenoids (1). There is strong evidence that cholesterol and mevalonic acid (or post-mevalonate, nonsterol products) independently exert feedback regulation on mevalonate biosynthesis (1 3-7 (1984). Mention of a trademark of proprietary product does not constitute a guarantee or warranty of the product by the United States Department of Agriculture and does not imply its approval to the exclusion of other products that may also be suitable. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom reprint requests should be sent.

105
Chemicaki-Sources of chemicals, substrates, labeled substrates, enzymes, and diagnostic kits were identified previously (12-16). Chemicals and solvents were of analytical grade. HPBF was contributed by the Minnesota Grain Pearling Co., East Grand Forks, MN, and d-a-tocotrienol by Dr. Shiro Urano, Tokyo Metropolitan Institute of Gerontology, Japan (17). The PESF of HPBF was prepared as previously described (14); 100 g of HPBF yields 3.5 g of PESF-HPBF.
Response of Chick and Rat Hepatocytes to PESF-HPBF-Hepatocytes were isolated from livers of fasted (48 h) refed (72 h), 8-weekold white Leghorn pullets and 6-week-old Sprague Dawley male rats (15,16). The hepatocytes (30 mg of protein) suspended in Krebs-Henseleit buffer (pH 6.8) were incubated with PESF-HPBF (0-10 mg) and 10 pg of Tween 80 in a volume of 1 ml. After 15 min incubation at 37 (rat) or 42 "C (chicken), the incubation mixture was centrifuged at 5000 X g for 2 min at 4 "C. The sedimented cells were suspended in 0.4 ml of homogenizing buffer (0.1 M potassium phosphate, pH 7.4, 4 mM MgCl,, 1 mM EDTA, and 2 mM dithiothreitol) and homogenized with a Polytron homogenizer. Following preliminary centrifugation, the 100,000 X g supernatants (cytosolic fraction) and precipitates (microsomal fraction) were held at -20 "C prior to assay for enzymatic activities. Protein concentrations were determined by a modification of the biuret method using bovine serum albumin as the standard (la).
Fractionation of Cholesterol-suppressiue Factors by HPLC-The PESF-HPBF was fractionated by semipreparative HPLC. A 50-pl aliquot of PESF-HPBF (20 mg of PESF-HPBF in methanol: petroleum ether, 2:l) was eluted through a Cl* reverse phase column (30 cm X 10 mm inner diameter, 10 pm particle size) with HPLC grade methanol at a flow rate of 1 ml/min. Ten fractions (Fig. 1, detecting wavelength, 205 nm) were collected pooled fractions were taken to dryness under nitrogen at 5 "C; the dried fractions and Tween 80 were dissolved in sufficient Krebs-Henseleit buffer to provide 2 mg of HPLC fraction and 10 pg of Tween 80/ml. These was eluted through a Cls reverse phase column (30 cm X 10 mm inner diameter, 10-pm particle size) with HPLC grade methanol at a flow rate of 1 ml/min. Ten fractions (detecting wavelength, 205 nm) were collected. Superimposed are the UV spectra of the fractions which have HMG-CoA reductase-suppressive action. Each peak was scanned using the Perkin-Elmer LC-75 detector with an autocontrol variable wavelength detector.  fractions were tested for in vitro effects on HMG-CoA reductase and fatty acid synthetase activities.
Determination of the Inhibition Potency of HPLC Fractions of PESF-HPBF-Hepatocytes isolated from fasted refed 12-week-old White Leghorn pullets were incubated with 200 pg of each of the 10 fractions of PESF-HPBF separated by HPLC. The incubation was handled as described above except that it contained 1 pg of Tween 80. The results of this assay indicated that HPLC fractions 5 and 9 contained HMG-CoA reductase-suppressive metabolites. A short term feeding trial confirmed that these fractions carried the cholesterol-suppressive metabolites (results not shown).
Fractionation of PESF-HPBF by Column Chromatography-For large scale purification of the cholesterol inhibitors, 10 g of PESF-HPBF was dissolved in 10 ml of petroleum ether and applied to a 90 X 3.5-cm silicic acid (Bio-Si1 A 100-200 mesh, dried at 100 "C for 30 min) column. The sample was eluted with succcessive applications of petroleum etherdiethy1 ether mixtures. Preliminary trials led to the elution scheme described in Fig. 2a. Eleven fractions (A-K) were collected fractions B and H corresponded to HPLC fractions 9 and

5.
Response of Chicks to PESF-HPBF Chromatographic Fractions-PESF-HPBF (7 g/kg diet) and each of the chromatographic fractions    This grouping includes the control group and all treatment groups for which no treatment effects were noted. Chromatographic fractions were added 0.5 g/kg diet. The chicks were fed for 21 days, fasted for 2 days, and refed for 3 days x f S.D.
'PESF-HPBF was added 7 g/kg diet. e Final weight of column fraction E treatment group, 394 f 36 g.
Means within a column lacking a common superscript are different ( p < 0.01). and H were further purified by preparative HPLC and their purity checked by chromatography on a 0.5-mm Silica Gel G plate with a solvent system of 95% benzene, 5% ethanol (Fig. 26). After plate development and exposure to iodine vapor, examination under UV light revealed a single spot for each fraction.
The separation and testing of the isolated inhibitors were carried out in diffused light. Concentrated fractions were stored at -20 "C and when appropriate, the fractions were dissolved in olive oil.
Concentration-dependent Responses of Hepatocytes to Purified Cholesterol Inhibitor Z-Hepatocytes isolated from 10-week-old fasted refed White Leghorn pullets were incubated with 0-100 pg of purified cholesterol inhibitor I (column fraction H) under the conditions described above. Dose-response of Chicks to Dietary Cholesterol Inhibitor Z-Cholesterol inhibitor I (0-20 ppm) was fed to groups of nine 1-day-old broiler chicks for 21 days; the birds were fasted and refed prior to the collection of liver and blood. The remaining conditions of the experiment were identical for those described above except that serum LDL and HDL cholesterol levels were measured. Diets were prepared daily.  Dose-response of Chicks to Intraperitoneally Administered Cholesterol Inhibitor Z-White Leghorn cockerels, 4 weeks of age, were divided into six groups of six birds. The birds were fed the diet as described above except that proportions of corn and soybean meal were 61.5% and 30%, respectively. The birds were fasted for 48 h and then refed the diet for 72 h.  inhibitor. Injections were given also at 24 and 48 h. At 72 h, the birds were killed and blood and liver tissues collected for analysis. In this trial, hepatic microsomal cholesterol 7a-hydroxylase activity was determined.
Assays for Enzymatic Activities-Conditions for the radioassays of microsomal HMG-CoA reductase and cholesterol 7a-hydroxylase activities and for the spectrophotometric assay of cytosolic fatty acid synthetase activity have been described in detail (15).
Serum Lipid Assays-Serum cholesterol concentrations were determined using the Worthington "Cholesterol Reagent" set. Low density (LDL) and very low density lipoproteins were isolated from the serum (100 pl) by precipitation with a mixture of 10 pl each of 9.7 mM phosphotungstic acid and 0.4 M MgCl,. After standing for 5 min at room temperature, the mixtures were centrifuged at 2000 X g for 10 min, the supernatant removed and analyzed for high density lipoprotein cholesterol (HDL-cholesterol). The precipitate was dissolved in 0.1 M sodium citrate buffer (100 pl) and the concentration of cholesterol in very low density lipoprotein + low density lipoprotein was determined (15,19,20).
Expression of Data and Statistical Methods-Enzyme data are presented as specific activities (units/mg cytosolic or microsomal protein/min). Statistical comparisons were performed by analyses of variance; when the F test indicated a significant treatment effect, the differences between means were analyzed by a protected least significant difference test (21).
Identification of Cholesterol Inhibitor I-High resolution mass spectrometric analyses were performed using an MS-902 high resolution instrument (Associate Electrical Industries Ltd., Manchester, England). Samples of cholesterol inhibitor I, purified by rechromatography (analytical HPLC (three runs on a Cl, reverse phase column, 25 cm X 4.6 mm inner diameter, 5 pm particle size) were introduced directly into the ion source on a glass probe at temperature ranging from 90 to 180 "C. No spectral deviations were recorded thus indicating the purity of the sample. The potential of the ionizing electron beam was 70 eV. Mass spectra were also obtained of d-a-tocopherol, of acetylated cholesterol inhibitor I and of synthetic d-a-tocotrienol (17) which has a mass spectrum, ultraviolet absorption spectrum, and an HPLC elution pattern identical to those of cholesterol inhibitor I.

RESULTS
HMG-CoA reductase and fatty acid synthetase were suppressed by PESF-HPBF in a dose-dependent manner in both rat and chick hepatocytes, the response being slightly greater in the latter (Table I). These in vitro assays, the results of which are consistent with in uiuo responses to PESF-HPBF (14), provide a rapid means of monitoring the inhibitory potency of the HPLC fractions of PESF-HPBF. Preliminary studies had shown that PESF-HPBF failed to suppress the reductase activity when incubated with preparations of microsomes.
The PESF-HPBF was fractionated by semipreparative HPLC (Fig. 1). Ten crude fractions were collected and the HMG-CoA reductase-suppressive action of each tested in avian hepatocytes. Constituents of fractions 5 and 9 suppressed HMG-CoA reductase ( Table 11). Although components of both fractions 5 and 9 suppressed HMG-CoA reductase, components of the former enhanced and the latter suppressed fatty acid synthetase. The control activities for both enzymes (Table 11) are 3.5 times the activities shown in Table  I. Two factors may play roles; the concentration of Tween 80 required for the suspension of PESF-HPBF was 10 times that required for the suspension of the HPLC fractions. Alternatively, hepatocytes used in the second trial were from 12week-old pullets, whereas those used for the first test were from 8-week-old pullets. In a subsequent experiment, activities determined in hepatocytes from 10-week-old pullets in i FIG. 4. Mass spectral pattern and UV absorption spectrum of cholesterol inhibitor I. A, samples of the plant metabolite, purified by rechromatography (analytical HPLC, three runs on a CIS reverse phase column, 25 cm X 4.6 mm inner diameter, 5 pm particle elution with HPLC grade methanol) were introduced directly into the ion source of an MS-902 high resolution mass spectrometer on a glass probe at temperatures ranging from 90 to 180 "C. The potential of the ionizing electron beam was 70 eV. The actual scan shown was made using an ionization temperature of 100 "C. B, the UV spectrum of the HPLC fraction (see Fig. 3) scanned as described in the legend to Fig. 1. The chromatographic procedure developed for the isolation of these constituents in sufficient quantities to support a long term dietary study is shown in Fig. 2. All of the chromatographic fractions except A (less than 1.5% of the PESF-HPBF) were tested in the dietary study (Table 111). The treatment group fed the diet containing PESF-HPBF (7 g/kg diet) had significantly lower hepatic HMG-CoA reductase and fatty acid synthetase activities than the control group. HMG-CoA reductase was lower also in groups fed column fractions B and H (0.5 g/kg diet). Concomitantly, serum cholesterol levels were significantly lowered by the three treatments. Consistent with actions of the HPLC fractions, column fractions B (HPLC-9) suppressed and H (HPLC-5) enhanced fatty acid synthetase activity. No other fractions affected either of the enzyme activities or the serum cholesterol level. Weight gains of treatment groups B and H were equal to that of the control; the group fed PESF-HPBF had a greater weight gain. Column fraction E contained the component which significantly increased the gain. Calculations revealed that the PESF-HPBF group received per kg diet, 1.95 g of column fraction B, 0.7 g of fraction E, and 0.35 g of fraction H.
Subsequent studies addressed the isolation, characterization, and testing of cholesterol inhibitor I (HPLC fraction 5 or column fraction H). The characterizations of column fractions B and E will be reported elsewhere. Column fraction H was further purified by rechromatography on the C18 reverse phase column (Fig. 3). The leading shoulder of HPLC fraction 5 (Fig. 1) contains cholesterol inhibitor I. Purified cholesterol inhibitor I inhibited HMG-CoA reductase and enhanced fatty acid synthetase in a dose-dependent manner (Table IV). In vitro HMG-CoA reductase activity decreased linearly in response to 0-15 pg and fatty acid synthetase activity increased linearly in response to 0-25 pg of cholesterol inhibitor I.
The initial identification of cholesterol inhibitor I was based on a high resolution mass spectral analysis which gave a molecular ion at m/e 424 with main fragmentation peaks at m/e 205, 203, and 165 which correspond to molecular formula C2,H4,02 (424.3350), C13H1702 (205.1229), C13H15O2 (203.18821, and CloH& (165.0913) moieties, respectively (Fig. 4). The major peaks at m/e 205 and 203 are due to the loss of three isoprenoid units (m-219) and additionally, Hz (m-2). The loss of 40 mass units due to the cleavage of the chroman ring in the molecule gives rise to the peak at m/e 165 (Fig. 5). This pattern is characteristic of the fragmentation pattern of d-a-tocotrienol (17,22) and is identical with the fragmentation pattern for synthetic d-a-tocotrienol which was prepared by Urano et al. (17).
The identification of cholesterol inhibitor I was further strengthened by mass spectral analysis of its acetate derivative (Fig. 6). The mass spectrum gives a molecular ion peak at M' 466 (C31H4603) which readily loses 42 mass units (CH2CO) giving rise to an intense peak at m/e 424, the parent compound. Peaks at m/e 247 and 245 represent the loss of three isoprenoid units (m-219) and additionally, Hz (m-2). Then, loss of CH2C0 gives rise to the peaks at m/e 205 and 203; splitting of the chroman ring yields a m/e of 165. The ultraviolet spectrum of cholesterol inhibitor I (Amax 292 nm, Fig. 4) is consistent with that of the synthetic d-a-tocotrienol and with the values reported for synthetic and natural tocotrienols (17,22) and with that of HPLC fraction 5 (Fig. 1).
Pure d-a-tocotrienol is light-and temperature-sensitive. As a constituent PESF of HPBF, the compound is stable for at least 2 weeks. The isolated compound, dissolved in olive oil, was stable for 2 days at room temperature.
The dose-dependent actions of d-a-tocotrienol isolated from the PESF-HPBF were confirmed by the results of a dietary trial (Table V). Broiler chicks were fed diets containing 0-20 ppm d-a-tocotrienol for a 3-week period. After a 2day fast, the chicks were refed for 3 days. Diet consumption by pen during the experimental period was 6.3-6.6 kg (-730 g/chick). The dose-dependent lowering of the HMG-CoA reductase activity and serum cholesterol by d-a-tocotrienol is shown in Table V. The fall in serum cholesterol is primarily in the LDL fraction; the decrease in HDL cholesterol is nonsignificant. Concomitant with the decrease in serum cholesterol is the decrease in cholesterol 7a-hydroxylase activity. As was recorded in the in vitro study (Table IV), d-a-tocotrienol effected a dose-dependent increase in induced fatty acid synthetase activity.
The protocol for the in vivo studies (Tables I11 and V) consisted of a 21-day ad lib feeding period, a %day period of fasting, and a 3-day period of refeeding the experimental diet to broiler cockerels. In the final study, White Leghorn cockerels which had been fasted for 2 days were administered da-tocotrienol by intraperitoneal injection at 24-h intervals during a 3-day period of refeeding. This protocol produced significant dose-dependent modifications of the hepatic enzyme activities (Table VI). Although the decrease in serum cholesterol was not significant, the level of LDL cholesterol was significantly lowered by this 3-day treatment.

DISCUSSION
The tocotrienols are widely distributed in the plant kingdom. The tocopherols are generally chloroplast components * pmol of 14C-labeled cholesterol into "C-labeled 7a-hydroxycholesterol per min/mg of microsomal protein.
e nmol of NADPH oxidized per min/mg of cytosolic protein. whereas the tocotrienols are components of cereal seeds (23). The biosynthetic paths are distinguished by the addition of geranylgeranyl pyrophosphate and phytyl pyrophosphate, respectively, to homogentistic acid in production of the tocotrienol and tocopherol metabolites (24). In animals, tocotrien-01s have 15-25% of the activity of vitamin E (25, 26). The function of the tocotrienols in plants is not known but one role may be as an antioxidant (23). Barley is rich in the various tocotrienols which differ from the tocopherols only in the double bonds of the isoprenoid chain (23). Ubiquinone-9 (27) and possibly squalene (1) suppress HMG-CoA reductase activity. The response of rats to 1.5 mg of ubiquinone-g/day (and ubiquinol-9 acetate) was similar to that recorded for 5 ppm of d-a-tocotrienol. Sterol synthesis was decreased by 40% and fatty acid synthesis increased 8% (27). Tocotrienols, ubiquinone-9, and squalene share in common the isoprenoid chain. This characteristic appears to be essential for the inhibition of cholesterogenesis.
Early in our studies of the feed value of cereal grains, we noted an effect of specific grains on plasma and liver cholesterol concentrations, a pattern of responses that was generally reflected in hepatic HMG-CoA reductase activities in chicks TABLE VI Concentration-dependent effects of d-a-tocotrienol administered intraperitoneally on cholesterol metabolism of fasted-refed White Leghorn cockerels Eight-week-old birds which had been fed for 4 weeks on a commercial diet were fasted for 48 h. During the 3day refeeding period, the birds were given d-a-tocotrienol intraperitoneally at 24-h intervals. Samples were taken 24 h following the final injection ( n = 6).