Biosci. Biotechnol. Biochem., 68 (5), 1051–1058, 2004

Isoflavones Found in Korean Paste as 3-Hydroxy-3-methylglutaryl Coenzyme A Reductase Inhibitors

y Jang Hoon SUNG,1 Seung Jun CHOI,1 Soon Won LEE,2 Kwan Hwa PARK,1 and Tae Wha MOON1;

1Department of Food Science and , School of Agricultural Biotechnology and Center for Agricultural Biomaterials, Seoul National University, Seoul 151-742, Korea 2Department of Chemistry, College of Natural Science, Sungkyunkwan University, Suwon 440-746, Korea

Received October 27, 2003; Accepted February 10, 2004

3-Hydroxy-3-methylglutaryl CoA (HMG-CoA) re- concentration and heart disease have been reported in ductase is the rate-limiting enzyme in the biosynthesis the literature.5–8) Korean soybean paste () is a of in mammals. Some microbial metabolites unique soy food, which is fermented by diverse micro- have been found to be HMG-CoA reductase inhibitors. organisms including fungi and bacilli during its manu- Korean soybean paste is a unique food fermented by facture.9,10) But not all fungi and bacilli related to the many microorganisms. The enzymatic method using the fermentation and the components of soybean paste have catalytic domain of Syrian hamster HMG-CoA reduc- been identified completely. tase was employed for the screening of HMG-CoA The objective of this study was, therefore, to screen reductase inhibitors. Soybean paste extract was fractio- for inhibitors of HMG-CoA reductase yet unidentified in nated by vacuum liquid chromatography. Fractions Korean soybean paste. showing relatively high HMG-CoA reductase inhibition were further purified through Sephadex LH-20 column Materials and Methods chromatography and C18 preparative HPLC, and the inhibitory compounds were identified as , Extraction of soybean paste. Korean soybean paste , and . was freeze-dried, crushed, sieved through a 35-mesh screen, and stored at 18 C until extraction. A Key words: Korean soybean paste; 3-hydroxy-3-methyl- powdered sample was extracted with a mixture of glutaryl coenzyme A (HMG-CoA) reduc- methanol and dichloroform (1:1, v/v) to achieve high tase; isoflavone; genistein; daidzein extraction potential. Sample (20 g) with (1:30, w/v) was homogenized at 10,000 rpm for 5 min with a Extensive epidemiological studies have shown that disperser (ULTRA-TURRAX T25, IKA Labortechnik, increased blood cholesterol level is a major cause of Janke & Kunkel GmbH & Co. KG, Staufen, Germany), coronary heart disease. Each cell of the higher animals and the mixture was filtered through a filter paper (No. 1, must balance external and internal sources of cholester- Whatman International Ltd., Maidstone, UK). The ols so as to sustain mevalonate synthesis while avoiding filtrate was then concentrated using a rotary evaporator overaccumulation. This balance is achieved with silica gel added to the filtrate, and the temperature through the feedback regulation of at least two sequen- was kept below 40 C. The resulting powder was tial enzymes in mevalonate synthesis, 3-hydroxy-3- subjected to vacuum liquid chromatography. methylglutaryl coenzyme A (HMG-CoA) synthase and HMG-CoA reductase, as well as low-density lipoprotein Vacuum liquid chromatography. The column was (LDL) receptors.1) It is well known that HMG-CoA prepared in a sintered glass funnel (Iwaki Glass, porosity reductase (EC 1.1.1.34) is the rate-limiting enzyme in 5–10 m, Tokyo, Japan) using TLC-grade packing the biosynthesis of cholesterol in mammals.2) Some (Silica gel 60H, Merck Ltd., Darmstadt, Germany). microbial metabolites and their analogs, which are The sample was applied uniformly at the top of the HMG-CoA reductase inhibitors, have been synthesized, support. A stepwise gradient with organic was including statins, reversible competitive inhibitors of used, and the column was allowed to run dry after the HMG-CoA reductase.3,4) collection of each fraction.11,12) Two vacuum liquid The relationship between diet and cardiovascular chromatography (VLC) columns of different sizes were disease was recognized early in the 20th century. Soy used for the crude purification: one of 90 mm and 50 mm food intake and its beneficial relation to blood lipid and the other of 65 mm and 40 mm in diameter and

y To whom correspondence should be addressed. Tel: +82-2-880-4854; Fax: +82-2-873-5095; E-mail: [email protected] Abbreviation: HMG-CoA, 3-Hydroxy-3-methyglutaryl coenzyme A 1052 J. H. SUNG et al. height respectively. CoA reductase is defined as the amount of enzyme that For the first chromatography with VLC, the sample catalyzes the oxidation of 1 mol of NADPH per min.13) (1.7 g) was fractionated using solvents in the order hexane, ethyl acetate, and methanol. The collection Lipophilic Sephadex LH-20 chromatography. Previ- volume was 200 ml for each step. Each fraction was ously prepared residue (0.12 g) dissolved in methanol monitored on a silica gel plate (Silica gel 60F254, Merck was subjected to Sephadex LH-20 (20 400 mm, Sigma Ltd.) developed with chloroform-methanol-acetic Chemical Co., St. Louis, MO, USA) column chroma- (85:10:5). For visualization of spots, two detection tography with methanol as the eluent at a flow rate of modes, irradiation with UV at 254 nm with a fluorescent 1 ml/min. The fractions were monitored with TLC. Each indicator, and iodine vapor adsorption, were used. Based fraction was grouped into five subfractions. The 4th (L4 on the assay results and UV observation on TLC, ethyl group) and 5th (L5 group) subfractions, which showed acetate and methanol eluates with relatively high relatively high inhibition activities, were separately inhibitory activity were pooled and evaporated. In collected and evaporated. addition, the residues containing hydrophilic compounds were separated with water and ethyl acetate. Water- HPLC analysis and purification. For HPLC analysis, soluble materials were discarded, and the ethyl acetate HPLC-grade acetonitrile (Fisher Scientific, Fair Lawn, layers were collected and evaporated for the next NJ, USA) and deionized water including 0.1% trifluoro- purification step. acetic acid (TFA) (Sigma Chemical Co.) were used as For the second VLC, the column was equilibrated the mobile phase. The mobile phase was filtered through with hexane. The ethyl acetate layer residue (0.73 g) was a 0.45-m pore-size filter (Millipore, Bedford, MA, then dissolved in a small volume (3–5 ml) of dichloro- USA). To determine the wavelength for analysis, each methane and loaded on the column, and chromato- sample was scanned from 200 nm to 400 nm using graphed using a stepwise gradient of n-hexane/CH2Cl2/ HPLC (Agilent 1100, Agilent , Palo Alto, EtOAc/MeOH. The collection volume was 100 ml for CA, USA) equipped with a diode array detector. The each fraction. The solvents used were hexane:CH2Cl2 sample was injected into the HPLC column (150 mm (90:10), hexane:CH2Cl2 (50:50), CH2Cl2 (100), 3.9 mm I.D. Symmetry column, thermostated at 40 C, CH2Cl2:EtOAc (98:2), CH2Cl2:EtOAc (90:10), packed with a 5-m particle-size RP18 from Waters CH2Cl2:EtOAc (50:50), EtOAc (100), EtOAc:MeOH Corporation, Milford, MA, USA, injection volume (95:5), EtOAc:MeOH (50:50), and MeOH. Among the 20 l). A SymmetryPrep column (300 mm 7.8 mm) ten fractions, based on the assay results and the similar packed with 7-m particle-size RP18 (Waters Corpo- appearance of spots after TLC analyses, the 6th, 7th, and ration) was used for the preparative HPLC, and the 8th eluents were pooled and evaporated. injection volume was 100 l. The flow rates were 0.7 and 5.6 ml/min for analytical and preparative chroma- Spectrophotometric assay using Syrian hamster tographies respectively. Alternation of the eluent from HMG-CoA reductase. Plasmid pKFT7-21 encoding 70% to 30% acetonitrile was carried out on L5 HMG-CoA reductase was a generous gift from Professor components to separate the active compound, L5-2. V.W. Rodwell of Purdue University. The catalytic Additionally, L4 components were separated into two domain of Syrian hamster HMG-CoA reductase was active components, L4-1 and L4-2, using 20% acetoni- purified as described by Frimpong et al.13) HMG-CoA- trile as the eluent. dependent oxidation of NADPH was monitored at 340 nm with a diode array spectrophotometer (Ultraspec Structural analysis of inhibitors. To measure the 4000, Amersham Biosciences Ltd., Piscataway, New molecular weight of the inhibitors, LC/MS was per- Jersey, USA) equipped with a cell holder maintained at formed using an Agilent 1100 HPLC system and a JMS- 37 C. Standard assay mixture contained 200 M HMG- LCmate single quadrupole mass spectrometer (JEOL CoA, 200 M NADPH, 100 mM NaCl, 1.0 mM EDTA, Ltd., Tokyo, Japan) equipped with an atmospheric 10 mM dithiothreitol, and 100 mM NaxPO4 (pH 6.8) at a pressure chemical ionization (APCI) interface at an ion final volume of 150 l. Reaction mixtures containing the source temperature of 500 C. The flow rate for all enzyme (5.3 g/150 l) and all components except analyses was 0.7 ml/min, and the entire effluent was HMG-CoA were first monitored to detect any HMG- introduced into the APCI probe. CoA-independent oxidation of NADPH. To check the To elucidate the structure of L5-2, NMR experiments, inhibitory activities of the fractions from each purifica- elemental analyses for C, H, and N (EA1110 elemental tion step on HMG-CoA reductase, each fraction, analyzer, CE Instrument, Milan, Italy), Fourier trans- evaporated by a SpeedVac (Savant Instruments, Hol- form-infrared spectroscopy (ABB Bomem Inc., Quebec, brook, New York, USA), was dissolved in DMSO Canada) with KBr pellet and X-ray structure determi- (100 l), and 0.1 N NaOH (900 l) was added. After nation were performed. 1H NMR (operating at heating at 50 C for 2 h, 15 l of the sample was taken 399.65 MHz), 13C NMR (operating at 100.40 MHz), and the reaction mixtures were added. The reaction was 1H–1H correlation spectroscopy (COSY), 13C–1H then initiated by adding HMG-CoA. One unit of HMG- COSY, Distortionless Enhancement by Polarization Isoflavones as HMG-CoA Reductase Inhibitors 1053 Transfer (DEPT), Heteronuclear Multiple Quantum Table 1. Relative Inhibition of HMG-CoA Reductase by VLC Correlation spectroscopy (HMQC), and Heteronuclear Fractions Multiple Bond Correlation spectroscopy (HMBC) spec- Concentration Relative activity of Fractions tra were obtained with a JEOL JNM-LA400 spectrom- (g/150 l) HMG-CoA reductasea (%) eter (JEOL Ltd.). All NMR spectra were obtained in 1VFb 30.0 113:4 15:7 DMSO-d6, with chemical shifts expressed as parts per 2VF 30.0 12:2 0:8 million () and coupling constants (J) in Hertz. The 3VF 36.0 6:9 3:3 spectra of L4-1 and L4-2 were compared with those of 4VF 30.8 68:6 1:1 the authentic standards, daidzein and glycitein (Sigma 2VFc 28.5 91:9 10:0 Chemical Co.), respectively. 3VF 24.0 78:2 6:0 Crystals of L5-2 were made as follows. The sample 4VF 63.0 98:6 10:3 5VF 34.5 105:3 4:9 was dissolved in the hot solvent, a mixture of methanol 6VF 9.8 32:6 9:1 and ethyl acetate (1:1). A small portion of the solvent 7VF 11.3 7:4 1:6 was added to the sample mixture for slow crystalliza- 8VF 15.8 30:0 2:8 tion, and the mixture was then evaporated slowly at 9VF 36.0 42:3 8:7 room temperature and the structure solved directly. All 10VF 36.0 98:9 5:9 hydrogen atoms were located and refined isotropically. a Each value represents mean standard deviation of three replicates. b First vacuum liquid chromatography fraction number. c Results and Discussion Second vacuum liquid chromatography fraction number.

To confirm the enzymatic method using HMG-CoA reductase of a Syrian hamster, the concentration re- 140 quired for 50% inhibition (IC50) was measured with lovastatin. IC50 for lovastatin was calculated to be 3) 0.018 M, whereas the reported value is 0.023 M. 120 Hence we concluded that the spectrometric assay method using Syrian hamster HMG-CoA reductase was proper for screening of the inhibitors. 100 Oily residue of the solvent extract of soybean paste was subjected to VLC to separate the active compo- 80 nents. Spots of the active fractions observed through TLC revealed more than one active component. Fur- mAU thermore, changing the solvent gradient did not result in 60 a higher resolution than that obtained from the second VLC. Through the VLC steps, non-active or relatively inactive fractions, which have relatively high hydro- 40 phobic and hydrophilic components, were discarded. The relative inhibition activities of the fractions are 20 shown in Table 1. 2VF and 3VF fractions that exhibited relatively high inhibition activities were pooled for a second fractionation. Among the second VLC eluates, 0 the 6th to 8th fractions showed relatively high inhibition activities with similar Rf values detected on TLC. The 200 300 400 500 600 resulting active fractions of the second VLC were nm pooled and loaded onto a Sephadex LH-20 column, and more inactive compounds were excluded. The inhibition Fig. 1. UV Absorbance Spectrum of L5-2 Recorded with a Photo- diode Array Detector. activity levels of the L4 and L5 groups of LH-20 eluate were 63:6 9:5 and 49:9 12:5% at 4.5 and 3.8 g/ 150 l respectively. Thus L5 was selected for purifica- identified as genistein.14) From the spectrum of L5-2, the tion, and its structure was determined through LC/MS, analysis wavelength was determined to be 260 nm; thus, NMR, FT-IR, elemental analysis, and X-ray diffraction. all HPLC detections were carried out at 260 nm. HPLC Based on the results of L5, the structures of the L4 group chromatograms of the L5 and L4 groups based on components were inferred from the LC/MS data and 260 nm absorption are displayed in Fig. 2. The main identified by comparing them with the NMR spectra of peak of L5-2 appeared at a retention time of 6.16 min the authentic standards. when L5 was eluted with 30% acetonitrile with 0.1% The UV absorption spectrum of L5-2 (major peak of TFA. The L4 group was found to have two main L5) is shown in Fig. 1. The purified L5-2 had a pale compounds. The active components purified through yellow appearance. This UV absorption spectrum was preparative HPLC were pale yellow, nearly white, 1054 J. H. SUNG et al.

60 A (A) 100 11.311

40 80 1.071 L4-1 60 20 mAU L4-2 40 1.886 1.630 Relative intensity 0 20

0 -20 250 260 270 280 290 300 310 320 330 340 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 Mass (m /z ) Min

200 B 100

1.612 (B) 160 80

120

L5-2 60 80 mAU

L5-1 40 40 0.949 1.034 Relative intensity 3.209 1.716 1.445 0 20

-40 0 0246 8 250 260 270 280 290 300 310 320 Min Mass (m /z ) Fig. 2. HPLC Chromatograms of L4 (A) and L5 (B) Groups. Fig. 3. APCIþ Exact Mass Spectrum of L5-2. (A) Protonated molecular ions at m=z 271.0563 and 312.0876. (B) With methanol as the eluent. pellet. The purified components were identified by analytical HPLC and LC/MS analysis. Mass spectra were obtained from L5-2, L4-1, and L4- of carbon and hydrogen was calculated to be 66.7 and 2. When acetonitrile was used as the eluent, distinct 3.7%, and 65.2 and 3.9%, respectively. peaks in the mass spectrum of L5-2 were observed at Chemical shifts of the L5-2 protons were measured at þ þ m=z 271 ½M þ H , and 312 ½M:CH3CN þ H . The 12.95 (1H, s, HO-5), 10.86 (1H, s, HO-7), 9.58 (1H, s, exact mass spectrum of L5-2 exhibited a molecular ion HO-40), 8.29 (1H, s, H-2), 7.36 (2H, d, J ¼ 8:27 Hz, H- by APCI-MS at 271.0563 ½M þ Hþ, corresponding to 20, H-60), 6.81 (2H, d, J ¼ 8:27, Hz, H-30, H-50), 6.37 C15H11O5 (calculated m=z 271.0606) (Fig. 3A). The (1H, m, J ¼ 1:96, H-8), and 6.21 (1H, m, J ¼ 1:96,H- other molecular ion at 312.0876 was ambiguous when 6). The 13C-NMR spectrum was measured by 1H compared with the data obtained by NMR and elemental broadband decoupling. Chemical shifts of carbons were analysis. Therefore, the eluent was changed to methanol. measured at 180.21 (C-4), 164.26 (C-7), 161.99 (C-5), The resulting chromatogram showed molecular ions by 157.58 (C-9), 157.42 (C-40), 153.97 (C-2), 130.16 (C-20, þ þ 0 0 0 APCI-MS at 271 ½M þ H and 303 ½M:CH3OH þ H C-6 ), 122.27, 121.21 (C-3), 115.06 (C-3 , C-5 ), 104.46 (Fig. 3B). An MS chromatogram showed the protonated (C-10), 98.96 (C-6), and 93.66 (C-8). To assign the molecular ion of L5-2, determined to be 271, and 312 positions of carbons and hydrogens, 1H–1H COSY, m/z peak, determined to be an artefact, formed with 13C–1H COSY, DEPT, HMQC, and HMBC experiments acetonitrile. For L4-1 and L4-2 the protonated molecular were performed. The number of hydrogen atoms bonded weights were determined by APCI-MS with acetonitrile directly to a carbon nucleus was determined by as eluent. From the chromatogram, the protonated subjecting L5-2 to a DEPT experiment. The results molecular weights of L4-1 and L4-2 were determined showed that L5-2 has chemical shifts of five CH signals to be 255 ½M þ Hþ and 285 ½M þ Hþ, respectively. and eight quaternary carbon nuclei. The C–H COSY The expected molecular formula of L5-2 from the experiment revealed five peaks belonging to the carbon high-resolution mass spectroscopy, C15H10O5 (MW nuclei that have directly bonded protons. From the H–H 270), was confirmed by elemental analysis. Based on COSY and HMQC experiments, the resonance position the formula weight and elemental analysis, the content of the coupled nuclei was identified. To determine the Isoflavones as HMG-CoA Reductase Inhibitors 1055 Table 2. Hydrogen Bonding Parameters (nm, ) in L5-2

Bond D–H HA D–HADA Position of A O10–HO10O13 0.0928 0.1952 162.70 0.2847 x, y þ 1=2, z 1=2 O13–HO13O11 0.0846 0.1978 165.48 0.2800 x, y 1=2, z þ 1=2 O12–HO12O11 0.0886 0.1795 149.89 0.2626 intramolecular

7.39 (2H, d, J ¼ 8:55,H-20, H-60), 6.94 (1H, dd, J ¼ 8:89, 2.06, H-6), 6.86 (1H, m, J ¼ 1:96, H-8), and 6.81 (2H, d, J ¼ 8:55, H-30,H-50), and were compared with the data in the literature.20) Chemical shifts of carbons L4-1 were measured at 174.66 (C-4), 162.46 (C-7), 157.40 (C-9), 157.12 (C-40), 152.78 (C-2), 130.04 (C-20, C-60), 127.26 (C-5), 123.46 (C-10), 122.52 (C-3), 116.61 (C-10), 115.01 (C-30, C-50), 114.90 (C-6), and 102.06 (C-8).15) Chemical shifts of L4-2 protons were measured at 10.51 (1H, br s, HO-7), 9.55 (1H, br s, 0 ¼ Fig. 4. ORTEP Drawing of L5-2. HO-4 ), 8.27 (1H, m, J 0:48, H-2), 7.43 (1H, s, H-5), 0 0 The atom-labelling scheme and 50% probability thermal ellip- 7.39 (2H, d, J ¼ 8:31, H-2 , H-6 ), 6.93 (1H, s, H-8), soids are presented. 6.81 (2H, d, J ¼ 8:31, H-30, H-50), and 3.88 (3H, s), and were compared with data in the literature.15) Chemical shifts of L4-2 carbons were measured at 174.26 (C-4), long-range H–C connectivity, the HMBC experiment 157.05 (C40), 153.01(C-7), 152.41 (C-2), 151.73 (C-9), was performed. In addition to the NMR experiments and 146.95 (C-6), 130.01 (C-20,C-60), 122.92 (C-3), 122.77 LC/MS spectra, the structure of L5-2 was determined to (C-10), 116.09 (C-10), 114.89 (C-30, C-50), 104.68 (C-5), be genistein by X-ray diffraction analysis. After the 102.75 (C-8), and 55.79 (OMe).15,21) structural determination, NMR spectra of L5-2 were Soy isoflavones have been reported to show various compared with those of the authentic compound, biological activities including estrogenic,22) antioxida- genistein, in the literature. All protons and carbons of tive,23) anti-osteoporotic,24) and anticarcinogenic25) ac- L5-2 were assigned based on the results of NMR.15,16) tivities. Furthermore, recent studies have proposed that The molecular structure of L5-2 with the atom- isoflavones may be the main active ingredient respon- numbering scheme is shown in Fig. 4. The planes sible for the cholesterol-lowering properties of some soy defined by C10–C60 are almost planar, with an average foods.26,27) Our study of HMG-CoA reductase inhibitors atomic displacement of 0.0002 nm. The naphthalene- in soybean paste identified the active components to be 0 type ring defined by 10 atoms (C2–C10 and O1) is also genistein (4 ,5,7-trihydroxyisofalvone, C15H10O4), daid- 0 nearly planar, with an average atomic displacement of zein (4 ,7-dihydroxyisoflavone, C15H10O4), and glyci- 0 0.0019 nm. The dihedral angle between these two planes tein (4 ,7-dihydroxy-6-methoxyisoflavone, C16H12O5), is 53.1(1), and the torsion angle of C8–C9–O1–C2 is aglycone forms of the isoflavones, which are not statins very close to 180 (178.9(3)). Furthermore, all the or related compounds known to be inhibitors of HMG- hydrogen atoms of the hydroxyl group are involved in CoA reductase. The discovery of isoflavones as HMG- the formation of hydrogen bonds of the type O–HO, CoA reductase inhibitors, not yet reported elsewhere, in including one intramolecular and two intermolecular soy foods is intriguing. No structural similarity exists hydrogen bonds (Table 2). The data on the crystal between statins and isoflavones. But a kinetic study of structure are in accord with the report of Brenton et al.17) isoflavones revealed that they are competitive inhibitors The FT-IR spectrum (KBr) displayed an absorption at of HMG-CoA,28) like statins. Additionally, a comparison 3411 (intermolecular hydroxyl), 3106 (hydroxyl), 1649 of the inhibition activity of glycone, the non-fermented (; -unsaturated carbonyl or ketone), and 1616 cm1, form, and aglycone, the fermented form, of isoflavone which supports the structure of L5-2, and also is in revealed the aglycone form to be more active.28) agreement with the results of several previous In soy foods, each the three isoflavones, daidzein, reports.18,19) glycitein, and genistein, exists in four forms, and in L4-1 and L4-2 were deduced to be daidzein and nature 600-O-malonyl--glucoside is the predominant glycitein respectively, based on the LC-MS data and form.20) 600-O-malonylgenistin and 600-O-malonyldaidzin those obtained for the identification of L5-2, and were are the most abundant isoflavones of soybean hypocotyl directly compared with the authentic daidzein and and cotyledon, whereas 600-O-acetylgenistin and 600-O- glycitein samples. Chemical shifts of L4-1 protons were acetyldaidzin are the major isoflavones in texturized measured at 10.76 (1H, br s, HO-7), 9.52 (1H, br s, vegetable .29,30) The 600-O-acetyl--glucoside-, HO-40), 8.29 (1H, s, H-2), 7.97 (1H, d, J ¼ 8:75, H-5), -glucoside-, and aglycone forms are derived from 600- 1056 J. H. SUNG et al. O-malonyl--glucosides during soy food processing or Conclusions sample preparation and analysis.31,32) Depending on variety, crop year, and location, the isoflavone content of In this study, aglycone isoflavones in Korean soybean shows wide variations30,33) as supported by a paste have been identified as HMG-CoA reductase reported range of total isoflavone content of between inhibitors, even though there is no similarity in structure 1,161 g/g and 2,743 g/g observed for 210 soybean to statins. Isoflavones mainly exist in glycone forms in cultivars,34) a 2.5-fold variation from the lowest to the non-fermented soy foods, whereas, in fermented soy highest values. Approximately 68% of the isoflavones in foods such as soybean paste, they exist in aglycone soybeans are in the form of 600-O-malonyl glycoside, forms. Aglycone forms of isoflavones have higher which is highly labile to heat and decomposes into bioavailability and are more active in many cases than glycoside when heated to over 60 C in aqueous the conjugated forms. Our study suggests that the solvents.29) Less than 5% of the isoflavones in natural hypocholesterolemic effects of soy foods previously soybeans are aglycones, but they are the main forms reported in several epidemiological studies are due in present after fermentation.31) part to the inhibition of HMG-CoA reductase by and constitute 99% of all the isoflavones. Therefore, soybean paste is a good source isoflavones in dry soybeans.14) Isolated soy and of aglycone isoflavones which can prevent cardiovas- textured vegetable protein consist of a mixture of all cular diseases. three types of isoflavone conjugates.31) In unfermented soy foods such as soybean milk powder, aglycones are Acknowledgments present mainly in the glucoside forms and make up only 4–5% of the isoflavones.35) However, among fermented This study was supported by a grant from the National commercial soybean products, higher aglycone contents Research Laboratory Program funded by the Ministry of are observed.20,35,36) The liberation of aglycones from Science and Technology of the Republic of Korea. glucosides during fermentation occurs through the hydrolytic action of -glucosidase from fungi37) and References bacilli.38) These compositional differences may be important with regard to metabolism and bioavailability 1) Brown, M. S., and Goldstein, J. L., Multivalent feedback due to marked differences in the polarity of conjugated regulation of HMG-CoA reductase, a control mechanism and unconjugated isoflavones.39) coordinating isoprenoid synthesis and cell growth. J. Most daidzins and genistins were hydrolyzed after 4 Lipid Res., 21, 505–517 (1980). 2) Rodwell, V. W., Nordstorm, J. L., and Mitschelen, J. J., weeks of (Japanese soybean paste) fermentation.40) Regulation of HMG-CoA reductase. Adv. Lipid Res., 14, When the soybeans were soaked in water, the daidzein 1–74 (1976). and genistein contents increased, whereas those of 3) Endo, A., Kuroda, M., and Tanzawa, K., Competitive 41) daidzin and genistin decreased. Matsuura and Obata inhibition of 3-hydroxy-3-methylglutaryl coenzyme A partially purified three isoforms of -glucosidase from reductase by ML-236A and ML-236B fungal metabo- soybean cotyledons. But because these enzymes from lites, having hypocholesterolemic activity. FEBS Lett., soybeans were destroyed during steam-cooking, micro- 72, 323–326 (1976). organisms including A. oryzae probably produced new 4) Endo, A., HMG-CoA Reductase inhibitors. Nat. Prod. -glucosidase in the soybean paste fermentation. Riou Rep., 10, 541–550 (1993). et al.42) purified a glucose-tolerant -glucosidase from 5) Wong, W. W., Smith, E. O., Stuff, J. E., Hachey, D. L., A. oryzae found in wine and processings. Heird, W. C., and Pownell, H. J., Cholesterol-lowering effect of soy protein in normocholesterolemic and The chemical forms of isoflavones in food should be hypercholesterolemic men. Am. J. Clin. Nutr., 68, taken into consideration when evaluating their avail- 1380S–1389S (1998). 32) ability for absorption through the diet. Recently, 6) Nagata, C., Ecological study of the association between 43) Piskula et al. demonstrated that daidzein and genis- soy product intake and mortality from cancer and heart tein, but not their glucosides, are readily absorbed from disease in Japan. Int. J. Epidemiol., 29, 832–836 (2000). the stomach of rats. Similar results were observed in 7) Lichtenstein, A. H., Potential benefits of soy protein in humans.44) Isoflavone aglycone-rich products may be managing patients with hypercholesterolemia. Nutr. more effective than glucoside-rich products in prevent- Clin. Care, 58, 144–147 (2000). ing chronic diseases, such as coronary heart disease. Not 8) Choi, Y. S., Kim, J. H., and Yoon, D. H., Effects of soy only do fermented soybean foods at the least increase extract and sex on and liver lipid levels of the rat. the bioavailability of soy isoflavones, because glycosidic Food Sci. Biotechnol., 11, 10–13 (2002). 9) Park, K. J., Kim, Y. M., Lee, B. M., and Lee, B. K., forms are degraded into unconjugated forms before Fungal microflora on Korean home-made meju. Kor. J. absorption in the intestine, but they can also be Mycol., 5, 7–12 (1997). consumed in large quantities. Therefore, fermented soy 10) Lee, S. S., Meju fermentation for a raw material of foods such as soybean paste are good sources of Korean traditional soy products. Kor. J. Mycol., 23, aglycone isoflavones for the prevention of cardiovascu- 161–175 (1995). lar disease through inhibition of HMG-CoA reductase. 11) Coll, J. C., and Bowden, B. F., The application of Isoflavones as HMG-CoA Reductase Inhibitors 1057 vacuum liquid chromatography to the separation of mechanisms. Am. J. Clin. Nutr., 68 (suppl), 1390S– terpene mixtures. J. Nat. Prod., 49, 934–936 (1986). 1393S (1998). 12) Pieters, L. A. C., and Vlietinck, A. J., Vacuum liquid 27) Gardner, C. D., Newell, K. A., Cherin, G., and Haskell, chromatography and quantitative 1H NMR spectroscopy W. L., The effect of soy protein with or without of tumor-promoting diterpene esters. J. Nat. Prod., 52, isoflavones relative to milk protein on plasma lipids in 186–190 (1989). hypercholesterolemic postmenopausal women. Am. J. 13) Frimpong, K., Darnay, B. G., and Rodwell, V. W., Clin. Nutr., 73, 728–735 (2001). Syrian hamster 3-hydroy-3-methylglutaryl-coenzyme A 28) Sung, J. H., Lee, S. J., Park, K. H., and Moon, T. W., reductase expressed in Escherichia coli: Production of Isoflavones inhibit 3-hydroxy-3-methylglutaryl coen- homogeneous protein. Protein Expression Purif., 4, 337– zyme A reductase in vitro. Biosci. Biotechnol. Biochem., 344 (1993). 68, 428–432 (2004). 14) Fukutake, M., Takahashi, M., Ishida, K., Kawamura, H., 29) Kudou, S., Fleury, Y., Welti, D., Magnolato, D., Uchida, Sugimura, T., and Wakabayashi, K., Quantification of T., Kitamura, K., and Okubo, K., Malonyl isoflavone genistein and genistein in soybeans and soybean prod- glycosides in soybean seeds (Glycine max MRRILL). ucts. Food Chem. Toxicol., 34, 457–461 (1996). Agric. Biol. Chem., 55, 2227–2233 (1991). 15) Jha, H. C., Zilliken, F., and Bretmaier, E., Carbon-13 30) Wang, H. J., and Murphy, P. A., Isoflavone composition chemical shift assignments of chromones and isofla- of American and Japanese soybeans in Iowa: effects of vones. Can. J. Chem., 58, 1211–1218 (1980). variety, crop year, and location. J. Agric. Food Chem., 16) Hosny, M., and Rosazza, J. P. N., Microbial hydroxy- 42, 1674–1677 (1994) lation and methylation of genistein by Streptomycetes. J. 31) Barnes, S., Kirk, M., and Coward, L., Isoflavones and Nat. Prod., 62, 1609–1612 (1999). their conjugates in soy foods: extraction conditions and 17) Brenton, P. M., Precigoux, G., Courseille, C., and analysis by HPLC-Mass spectrometry. J. Agric. Food Hospital, M., Ge´nisteine. Acta Chrystallogr. Sec. B, 31, Chem., 42, 2466–2472 (1994). 921–923 (1975). 32) Wang, H. J., and Murphy, P. A., Mass balance study of 18) Goto, J., Matsuda, Y., Asano, K., Kawamato, I., isoflavones during soybean processing. J. Agric. Food Yasuzawa, T., Shirahata, K., Sano, H., and Kase, H., Chem., 44, 2377–2383 (1996). K-254-I (genistein), a new inhibitor of Ca2þ and 33) Hoeck, J. A., Fehr, W. R., Murphy, P. A., and Welke, G. calmodulin-dependent cyclic nucleotide phosphodiester- A., Influence of genotype and environment on isoflavone ase from Streptosporangium vulgare. Agric. Biol. Chem., contents of soybean. Crop. Sci., 40, 48–51 (2000). 51, 3003–3009 (1987). 34) Wang, C., Sherrard, M., Pagadala, S., Wixon, R., and 19) Miyazawa, M., Sakano, K., Nakamura, S., and Kosaka, Scott, R. A., Isoflavone content among maturity group 0 H., Antimutagenic activity of isoflavones from soybean to II soybeans. J. Am. Chem. Soc., 77, 483–487 seeds (Glycine max merrill). J. Agric. Food Chem., 47, (2000). 1346–1349 (1999). 35) Wang, H. J., and Murphy, P. A., Isoflavone content in 20) Coward, L., Barnes, N. C., Setchell, K. D. R., and commercial soybean foods. J. Agric. Food Chem., 42, Barnes, S., Genistein, daidzein, and their -glycoside 1666–1673 (1994). conjugates: antitumor isoflavones in soybean foods from 36) Nguyenle, T., Wang, E., and Cheung, A. P., An American and Asian diets. J. Agric. Food Chem., 41, investigation on the extraction and concentration of 1961–1967 (1993). isoflavones in soy-based products. J. Pharm. Biomed. 21) Park, H. J., Park, J. H., Moon, J. O., Lee, K. T., Jung, W. Anal., 14, 221–232 (1995). T., Oh, S. R., and Lee, H. K., Isoflavone glycosides from 37) Murakami, H., Asakawa, T., Terao, J., and Matsushita, the flowers of Pueraria thunbergiana. Phytochemistry, S., Antioxidative stability of and liberation of 51, 147–151 (1999). isoflavones by fermentation. Agric. Biol. Chem., 48, 22) Messina, M., Soyfoods and soybean phyto-oestrogens 2971–2975 (1984). (isoflavones) as possible alternatives to hormone re- 38) Klus, K., and Barz, W., Formation of polyhydroxylated placement therapy (HRT). Eur. J. Cancer, 36, S71–S72 isoflavones from the soybean seed isoflavones daidzein (2000). and glycitein by bacteria isolated from tempe. Arch. 23) Kapiotis, S., Hermann, M., Held, I., Seelos, C., Ehringer, Microbiol., 164, 428–434 (1995). H., and Gmeiner, B. M., Genistein, the dietary-derived 39) Setchell, K. D. R., : The biochemistry, angiogenesis inhibitor, prevents LDL oxidation and physiology, and implications for human health of soy protects endothelial cells from damage by atherogenic isoflavones. Am. J. Clin. Nutr., 68 (suppl), 1333S–1346S LDL. Arterioscler. Thromb. Vasc. Biol., 17, 2868–2874 (1998). (1997). 40) Chiou, R. Y. Y., and Cheng, S. L., Isoflavone trans- 24) Anderson, J. J. B., and Garner, S. C., The effects of formation during soybean koji preparation and subse- phytoestrogens on bone. Nutr. Res., 17, 1617–1632 quent miso fermentation supplemented with and (1997). NaCl. J. Agric. Food Chem., 49, 3656–3660 (2001). 25) Herman, C., Adlercreutz, T., Goldin, B. R., Gorbach, S. 41) Matsuura, M., and Obata, A., -Glucosidases from L., Ho¨ckerstedt, K. A. V., Watanabe, S., Ha¨ma¨la¨inen, E. soybeans hydrolyze daidzin and genistin. J. Food Sci., K., Markkanen, M. H., Ma¨kela¨, T. H., and Wa¨ha¨la¨,K. 58, 144–147 (1993). T., Soybean intake and cancer risk. J. 42) Riou, C., Salmon, J. M., Vallier, M. J., Gu¨nnata, Z., and Nutr., 125, 757S–770S (1995). Barre, P., Purification, characterization, and substrate 26) Anthony, M. S., Clarkson, T. B., and Williams, J. K., specificity of a novel highly glucose-tolerant -glucosi- Effects of soy isoflavones on atherosclerosis: potential dase from Aspergillus oryzae. Appl. Environ. Microbiol., 1058 J. H. SUNG et al. 64, 3607–3614 (1998). M., Kataoka, S., Kubota, Y., Kikuchi, M., and Tobe, K., 43) Piskular, M. K., Yamakoshi, J., and Iwai, Y., Daidzein Soy isoflavone aglycones are absorbed faster and in and genistein but not their glycosides are absorbed from higher amounts than their glucosides in humans. J. Nutr., the rat stomach. FEBS Lett., 447, 287–291 (1999). 130, 1695–1699 (2000). 44) Izumi, T., Piskula, M. K., Osawa, S., Obata, A., Saito,