Transport Mechanisms for Soy Isoflavones and Microbial Metabolites Dihydrogenistein and Dihydrodaidzein Across Monolayers and Membranes

Transcription

1 Biosci. Biotechnol. Biochem., 77 (11), , 213 Transport Mechanisms for Soy Isoflavones and Microbial Metabolites Dihydrogenistein and Dihydrodaidzein Across Monolayers and Membranes Shoko KBAYASHI, 1;y Miki SHINHARA, 2 Toshitada NAGAI, 2 and Yutaka KNISHI 3 1 Research Center for Food Safety, Graduate School of Agricultural and Life Sciences, The University of Tokyo, Yayoi, Bunkyo-ku, Tokyo , Japan 2 Department of Food and Life-Science, Takasaki University of Health and Welfare, 37-1 Nakaorui, Takasaki, Gumma 37-33, Japan 3 Research Laboratories for Beverage Technologies, Kirin Company, Namamugi, Tsurumi-ku, Yokohama , Japan Received May 21, 213; Accepted August 7, 213; nline Publication, November 7, 213 [doi:1.1271/bbb.1344] Isoflavone data concerning the metabolism and permeability on intestinal epithelial cells are scarce, particularly for microbial isoflavone metabolites. This study evaluates the absorption mechanisms for the isoflavones, genistein and daidzein, and their microbial metabolites, dihydrogenistein () and dihydrodaidzein (DHD). The permeability characteristics of isoflavones were compared by using the Caco-2 human colon adenocarcinoma cell line for a parallel artificial membrane permeability assay, and comparing their physicochemical properties. The data suggest that genistein, and DHD were efficiently transported by passive diffusion according to the ph-partition hypothesis. was conjugated by phase II metabolizing enzymes and acted as a substrate of the breast cancer resistance protein (BCRP). Daidzein was not conjugated but did act as a substrate for BCRP, multidrug resistance-associated proteins, and P-glycoprotein. In contrast, and DHD were markedly more permeable than their parent isoflavones; they were therefore difficult to transport by the efflux effect, and glucuronidation/sulfation was limited by the flux time. Key words: dihydrogenistein; dihydrodaidzein; artificial membrane permeation assay; Caco-2 Isoflavones are polyphenols found in soybeans and soy food products. The consumption of dietary supplements and functional foods containing soy isoflavones is popular due to a correlation between the soy isoflavone intake in the Asian diet and the reduced incidence of such diseases as gland cancers. 1) High-concentration isoflavone is not part of the traditional Asian diet, however, and data from animal studies have raised concerns over possible endocrine disruption resulting from an excessive soy isoflavone intake. 2,3) Despite this, the intestinal permeability mechanism for isoflavones is poorly understood, and a safe daily allowance needs to be established. and daidzein are commonly present as glycosides (genistin and daidzin, respectively) in intact soybeans. rally administered glycoside forms are hydrolyzed by -glucosidases to aglycones in the jejunum and are then absorbed. 4) Fermented soybean foods are rich in aglycones. 5) The aglycone forms are either absorbed intact or further metabolized by intestinal microflora. Studies on isoflavones have suggested the important biological properties of parent compounds and their microbial metabolites; various metabolites of isoflavones have been detected in plasma and urine, 6) which are key to understanding the beneficial effects of isoflavones, because the metabolite rather than the parent compound is often absorbed. Colonic microflora play an important role in the metabolism of isoflavones. and daidzein are respectively first converted by gut microflora to dihydrogenistein () and dihydrodaidzein (DHD). These metabolites bind to estrogen receptors and exert antioxidative effects. 7,8) High concentrations of and DHD have been detected in human urine and plasma which may act as bioactive components of their parent isoflavones. However, data showing the intestinal permeability of and DHD are scarce. Isoflavones are absorbed in their aglycone forms by intestinal epithelial cells and are thus present at high concentrations in their glucuronized/sulfated forms in plasma and urine. Phase II metabolizing enzymes, particularly UDP-glucuronosyltransferases (UGT) and sulfotransferases (SULT), play important roles in the metabolism, elimination, and detoxification of phenolics. 9) Intestinal conjugation by phase II metabolites and intestinal excretion are key to the first-pass impact of isoflavone absorption and are critical factors for restricting bioavailability (BA). The limitations of isoflavone consumption involve intestinal absorption and metabolism. BA of isoflavones is low, and the plasma concentration is generally lower than the effective in vitro concentration. 1) Substantial study is still required y To whom correspondence should be addressed. Tel: ; Fax: ; ashoko@mail.ecc.u-tokyo.ac.jp Abbreviations: AUC, area under the blood concentration-time curve; BA, bioavailability; BCRP, breast cancer resistance protein; DHD, dihydrodaidzein;, dihydrogenistein; DMEM, Dulbecco s modified Eagle s medium; DMS, dimethyl sulfoxide; ECD, electrochemical detector; HBSS, Hanks balanced salt solution; MDR, multidrug-resistance; MRP, multidrug resistance-associated protein; P-gp, P-glycoprotein; PAMPA, parallel artificial membrane permeability assay; SULT, sulfotransferases; UGT, UDP-glucuronosyltransferases; TER, transepithelial electrical resistance

2 to understand the absorption mechanism and metabolic pathway of isoflavones to develop viable strategies that overcome poor BA and to establish safe and effective doses. Tian et al. 11) have investigated the structure-permeability relationship of 36 flavonoids by using Caco-2 human colon adenocarcinoma cell line monolayers and have reported that isoflavones and flavanones had high permeability. The absorption characteristics of the flavanones, hesperetin, naringenin and eriodictyol, were elucidated 12,13) by evaluating the passive and active transport mechanisms, using an artificial membrane permeation assay (PAMPA) and Caco-2 cells. 14) These flavanones were actively and efficiently transported by passive diffusion according to the ph-partition hypothesis, depending on ionization of the H group present on C7. The pka and ionization curves obtained from the Marvin Sketch program should reflect ionization of the 7-H group for these flavanones. 14) Under in vivo conditions, the sodium/proton anti-porter at the brushborder membrane can generate a proton gradient sufficient to facilitate the transport of certain organic acids. 15,16), daidzein, and their microbial metabolites, and DHD, possess substituted H groups on C7 and are highly lipophilic. Hence, these isoflavones may also be efficiently transported by passive diffusion according to the ph-partition hypothesis. However, the transport mechanism for isoflavones and their metabolites on the intestinal epithelial cells has not been investigated. We therefore compared data from a Caco-2 cell-based assay and PAMPA to provide further insight into the mechanism governing the intestinal absorption of isoflavones. PAMPA is a quick, simple test to measure passive permeability 17) and has been used to predict oral absorption. 18) Caco-2 cell monolayers have many features of intestinal epithelial cells and are a widely accepted in vitro system for assessing the permeability and metabolism in the human intestines. 19) We evaluated in this study the impact of bidirectional metabolite transport by examining the intestinal absorption of genistein, daidzein,, and DHD. The results of this study provide insight into the health benefits and safety of these compounds in vivo. Materials and Methods Transport Mechanisms for Soy Isoflavones and Their Microbial Metabolites 2211 Chemicals. The structures of genistein, daidzein,, and DHD are presented in Fig. 1. These compounds were purchased from Toronto Research Chemicals (Brisbane, North York, Canada). All other chemicals used in this study were of analytical grade. Calculation of the physicochemical properties. All pka values were obtained by using SciFinder. The Marvin Sketch program (http: // was used to evaluate the ionization profiles, log Ps, log Ds, ionization curves, and ionizable H groups present in each molecule. Theoretical partition ratio according to the ph-partition hypothesis. The ratio of the ionic and nonionic forms of compounds in the gut lumen/plasma (C bl /C ap ) was calculated by using the formula for weak acids: C bl =C ap ¼ð1þ1 ph2 pka Þ=ð1 þ 1 ph1 pka Þ ð1þ Artificial membrane permeability studies. Experiments using PAMPA were performed according to the Millipore Protocol Note (lit. no. PC1545EN). This model consists of a 96-well microtiter plate that serves as the donor chamber and a MultiScreen-IP 96-well filter plate (Millipore, Bedford, MA, USA) that serves as the acceptor compartment. The acceptor plate, placed directly on the donor plate, is fitted with a hydrophobic microfilter disc (.45-mm Immobilion-P membrane) and impregnated with a phospholipid solution. In these tests, 5 ml of a 1% (w/v) solution of soy lecithin in dodecane was added to each well with a phospholipid membrane. Buffers of ph 6 or 7.4 were prepared from a phosphate buffer (28 mm KH 2 P 4 and 41 mm Na 2 HP 4 at ph 7) which was adjusted to the required ph value with NaH (ph 7.4) or H 3 P 4 (ph 6.). Assays for each sample were performed with or without a ph gradient between the acceptor and donor reservoirs. The concentration of the samples in both assays was 5 mm. The solutions were prepared by diluting stock solutions, all containing 5% dimethyl sulfoxide (DMS). The two plates were separated after 16 h, and the acceptor solutions were analyzed by using an HPLC-electrochemical detector (ECD) with a coulometric detection system (ESA, Boston, MA, USA). The permeability coefficients (log P app-pampa ) for the artificial membrane were calculated by using eq. (2): log P app-pampa ¼ log fc ln ð1 ½drugŠ acceptor =½drugŠ equilibrium Þg where C ¼ðV D V A =ðv D þ V A Þðarea timeþ. In this equation, V D is the donor volume (.15 cm 3 ), V A is the volume of the acceptor compartment (.3 cm 3 ), area is the accessible filter area (.24 cm 2 ), and time is the incubation time. The concentration of a compound in the acceptor compartment at completion of the assay is denoted by ½drugŠ acceptor, and the concentration of compound at theoretical equilibrium is denoted by ½drugŠ equilibrium. Transepithelial transport experiments across the Caco-2 monolayers. Caco-2 cells (passage 4 7) were cultivated as previously described ) Experiments investigating the transport of samples across Caco-2 cell monolayers were performed as previously described ) The apical-to-basolateral transport rate for each test compound (5 mmol/l) was measured across Caco-2 monolayers that had been grown for 2 3 weeks in Transwell inserts (Costar, Cambridge, MA, USA) coated with type-i collagen. The integrity of the cell layers was evaluated by measuring the trans-epithelial electrical resistance (TER) with Millicell-ERS equipment (Costar, Cambridge, MA, USA). nly those monolayers with TER > 3 cm 2, measured before and after each transport experiment, were used. Samples were prepared by diluting stock solutions, all solutions containing 1% DMS which did not affect transport. 2) The slope of the initial linear portion of the protein concentration curve (nmol/mg of protein) versus time (min) was calculated by a linear regression analysis and is defined as the permeation rate (nmol/min/mg of protein). The amount of a sample transported was determined by the aforementioned method for comparisons with the PAMPA data, using the permeability coefficient for Caco-2 cells of log P app-caco-2, as calculated by eq. (3): log P app-caco-2 ¼ log fc ð½drugš acceptor =½drugŠ initial,donor Þg where C ¼ðV A =area timeþ. V A in this equation is the volume of the acceptor compartment (.3 cm 3 ), area is the accessible filter area (1 cm 2 ), and time is the incubation time. The concentrations of a drug in the acceptor (basolateral) and donor (apical) compartments at the completion of the assay are respectively denoted by ½drugŠ acceptor and ½drugŠ initial,donor. Isoflavones and their metabolites were studied in the presence of the inhibitors, verapamil; the P-glycoprotein (P-gp) inhibitor, MK571; multidrug resistance-associated protein (MRP) inhibitor and estrone-3- sulfate; and breast cancer resistance protein (BCRP) substrate. Each reagent was added to both the apical and basolateral sides of Caco-2 cell monolayers. The isoflavone solution was added to the apical side after pre-incubating with the inhibitor solution for 1 h. Analytical methods. The isoflavone concentration was determined by HPLC ECD, using an HPLC ECD instrument fitted with a coulometric detection system as previously described ) Chromatographic separation was performed in an DS15 C18 column (MC Medical, Tokyo, Japan) with mobile phase A (solvent A) containing 5-mM sodium acetate and 5% methanol (ph 3.), and mobile phase B (solvent B) containing 5-mM sodium acetate, 4% acetonitrile, and ð2þ ð3þ

3 2212 S. KBAYASHI et al. Response (µa) [3 mv] 1. [2 mv] [ mv]. [6 mv] [5 mv] [4 mv] [8 mv] [7 mv] Retention time (min) Response (µa) [ mv]. Daidzein [3 mv] [2 mv] [5 mv] [4 mv] [6 mv] [7 mv] [8 mv] Daiazein Retention time (min) Response (µa) [2 mv] [ mv] [3 mv] [4 mv] [6 mv] [5 mv] [7 mv] [8 mv] Retention time (min) Response (µa) DHD [2 mv] [ mv] [3 mv] [4 mv] [6 mv] [5 mv] [7 mv] [8 mv] DHD Retention time (min) H H H H H H H H Daidzein DHD H H 2% methanol (ph 3.5). The elution profile (.6 ml/min) was as follows:.5 min, isocratic elution with 6% solvent A/4% solvent B; min, linear gradient from 6% solvent A/4% solvent B to % solvent A/% solvent B; min, isocratic elution with % solvent A/% solvent B; and min, isocratic elution with 6% solvent A/4% solvent B. Eight electrode detector potentials ( 7 mv, in increments of mv) were used to quantify the isoflavones. Enzymatic hydrolysis and determination of the isoflavone conjugates. After the transport experiments, Hanks balanced salt solutions (HBSS, 5 ml) from the apical and basolateral chambers were mixed with 5 ml of a solution containing sulfatase type H-5 (Sigma-Aldrich, St. Louis, M, USA), a.1-mol/l acetate buffer (ph 5.), 12.5 units of sulfatase, and approximately 125 units of -glucuronidase. The mixture was incubated at 37 C for 45 min. It is considered that differences between the genistein, daidzein,, and DHD contents before and after the sulfatase treatment were caused by the various concentrations of sulfate and glucuronide conjugates in the sample. Data analysis. Results are expressed as the mean SE. Two-group comparisons were made by using the unpaired two-tailed Student s t-test (Table 3). Multiple comparisons (Table 2 and Fig. 4) were respectively performed by repeated-measure 1-way ANVA followed by using Fisher s or Dunnett s multiple-comparison test. p values of <:5 were considered significant for all comparisons. Results Fig. 1. Chromatograms and Chemical Structures of, Daidzein,, and DHD., dihydrogenistein; DHD, dihydrodaidzein. Analysis of genistein, daidzein, and their metabolites after transport across the Caco-2 cell monolayers and PAMPA Representative chromatograms for transported genistein, daidzein,, and DHD in the basolateral media are presented in Fig. 1. The detection limit in the HPLC ECD analysis for all samples in the column was <:5 pmol. 21) The respective dominant oxidation potentials and retention times were as follows: genistein, 7 mv, 16.6 min;, 7 mv, 14.6 min; daidzein, 7 mv, 12.3 min; and DHD, 7 mv, 11.6 min. Determination of the physicochemical properties of isoflavones The physicochemical properties of the test compounds are summarized in Table 1. The compounds were of similar molecular weight. had the lowest pka value, although this property did not vary much among the test compounds. In contrast, the log P values varied between 2.38 and 3.8, with the highest log P value for genistein. and daidzein exist almost entirely in a non-ionized form at ph 6., with a one-sixth to one-seventh decrease in the non-ionized form at ph 7.4. Isoflavones and DHD exist almost entirely in non-ionized form at ph 6 and 7.4, although approximately 3% fewer molecules were in the nonionized form at ph 7.4 than at ph 6.. The ionization curves for these compounds obtained from the Marvin Sketch program may reflect ionization of the 7-H group following deprotonation. Theoretical partition ratios according to the ph-partition hypothesis In the presence of a proton gradient, the respective C bl /C ap ratios for genistein,, daidzein, and DHD

4 Table 1. Transport Mechanisms for Soy Isoflavones and Their Microbial Metabolites 2213 Physicochemical Properties of Test Compounds: Molecular Weight, Estimated pka, log P and log D and Presence of Non-Ionized Forms Isoflavone MW pka a log P b log D 6: b log D 7:4 b % of non-ionized forms ph 6. b Daidzein DHD a SciFinder Scholar. b Marvin Sketch ( dihydrogenistein; DHD, dihydrodaidzein; MW, molecular weight. ph 7.4 b Permeation (µl/mg of protein) Permeation (µl/mg of protein) Daidzein Time (min) Time (min) 5 1 DHD Time (min) Time (min) Permeation (µl/mg of protein) Permeation (µl/mg of protein) Fig. 2. Transport Characteristics of,, Daidzein, and DHD Across Caco-2 Cell Monolayers in the Presence ( ) and Absence ( ) of a Proton Gradient. The permeation of 5 mm genistein,, daidzein, and DHD from the apical side to the basolateral side was measured at 37 C in the presence ( ) and absence ( ) of a proton gradient (apical ph, 6. or 7.4; basolateral ph, 7.4). Each point represents the mean SE of three experiments., dihydrogenistein; DHD, dihydrodaidzein. were 8.76/1.31, 1.76/1.3, 3.45/1.1, and 1.49/1.2 according to eq. (1). In the absence of a proton gradient, the respective C bl /C ap ratios for genistein,, daidzein, and DHD were 8.76/8.76, 1.76/1.76, 3.45/ 3.45, and 1.49/1.49 according to eq. (1). Transport of genistein, daidzein, and their metabolites through PAMPA The permeability coefficients calculated from PAMPA experiments are listed in Table 2. The P app-pampa values for all compounds, except for daidzein, increased with increasing proton gradient. The P app-pampa value for genistein at ph 6/7.4 was the highest at 2:94 : cm/s and that for DHD at 7.4/7.4 was the lowest at :21 :1 1 6 cm/s. Transport of genistein, daidzein, and their metabolites through the Caco-2 cell monolayers The bi-directional apical-to-basolateral transport of genistein, daidzein, and their metabolites across the Caco-2 cell monolayers was examined in the presence or absence of an inwardly directed proton gradient (Fig. 2). In the presence of the proton gradient (apical at ph 6. and basolateral at ph 7.4), the respective permeation rates (J ap!bl ) for genistein,, diadzein, and DHD were :61 :29, 1:16 :17, :87 :17, and :96 :63 nmol/min/mg of protein. In all cases, these rates were greater than those in the absence of a proton gradient. With both apical and basolateral chambers at ph 7.4, the respective permeation rates (J ap!bl ) for genistein,, daidzein, and DHD were :36 :19, :6 :3, :51 :11, and :54 :4 nmol/min/mg of protein. Permeability coefficients through the Caco-2 cell monolayers were calculated according to eq. (3) (Table 2) to compare these data with the PAMPA data. Both the P app-caco-2 and J ap!bl values for the and DHD metabolites were greater than those for their precursors (Fig. 2 and Table 2). Both the P app-caco-2 and J ap!bl values for daidzein were higher than those for genistein, and the J ap!bl values were higher for than for DHD. The respective TER values in the presence of a proton gradient after genistein,, daidzein, and DHD transport were 618: 42:2, 584: 12:2, 741:7 13:2, and 42:3 53:3. The respective TER values in the absence of a proton gradient after genistein,, daidzein, and DHD transport were 562:7 15:8, 9:3 129:5, 777: 59:1, and 441:7 62:8. The TER values at the end of each transport experiment were maintained at >3 to ensure that neither TER

5 2214 S. KBAYASHI et al. Table 2. Effective Permeability Values in the Presence (ph 6/7.4) and Absence (ph 7.4/7.4) of a Proton Gradient Across the Caco-2 Cell Monolayers and in the PAMPA Model A P app-pampa (1 6 ) (cm/s) B P app-caco-2 (1 6 ) (cm/s) Compound 6/ /7.4 6/ /7.4 2:94 :49 a 1:5 :18 b 52:86 3:77 a 37:75 1:15 b :74 :18 c :39 :3 d 99:44 :48 c 55:47 2:7 a Daidzein :35 :4 d :45 :4 d 67:97 2:76 d 4:64 :91 e DHD :48 :6 c :21.1 c 99:72 3:76 c 52:78 :64 a Data are presented as the mean SD. Statistical analyses were performed by repeated-measure 1-way ANVA followed by using Fisher s multiple-comparison test. A(P app-pampa ) and B (P app-caco-2 ) were compared independently. Means without a common letter differ, p < :5., dihydrogenistein; DHD, dihydrodaidzein; PAMPA, artificial membrane permeation assay. Papp, Caco-2 (1-6 cm/s) y = x Papp, PAMPA(1-6 cm/s) Fig. 3. Correlation between P app-caco-2 and P app-pampa. All data were plotted in the presence and absence of a proton gradient (apical or donor ph, 6. or 7.4; basolateral or accepter ph, 7.4). in the presence ( ) and absence ( ) of a proton gradient, in the presence ( ) and absence ( ) of a proton gradient, daidzein in the presence ( ) and absence ( ) of a proton gradient, DHD in the presence ( ) and absence ( ) of a proton gradient. The dashed circle includes genistein at ph 6/7.4 ( ) and ph 7.4/7.4 ( ). The solid circle includes DHD ( ) and at ph 6/7.4 ( )., dihydrogenistein; DHD, dihydrodaidzein; PAMPA, artificial membrane permeation assay. nor isoflavones permeation varied during the incubation process. Correlation between P app Caco 2 and P app PAMPA While P app-pampa is an estimate of passive transport, P app-caco-2 estimates both passive and active transport, which include secretory and absorptive transport, and first-pass metabolism. A comparison of data from permeability profiling using PAMPA and Caco-2 cellbased assays therefore indicates the permeability mechanism. 22,23) Figure 3 shows a graphical plot of P app-caco-2 vs. P app-pampa between ph 6. and 7.4. The characteristics of this plot can be explained by the much lower permeability of genistein and much higher permeability of DHD and with a proton gradient across the Caco-2 cell monolayers. The dashed circle indicates that genistein at 6/7.4 and 7.4/7.4 had a much lower P app-caco-2 than P app-pampa values. The continuous circle indicates that DHD and at 6/7.4 had higher P app-caco-2 than P app-pampa values. Effects of transport inhibitors on the efflux of genistein, daidzein, and their metabolites The effects of verapamil (a P-gp inhibitor), MK571 (MRP1 and 2 inhibitors), and estrone-3-sulfate (a BCRP substrate) on apical-to-basolateral transport are presented in Fig. 4. The permeability of genistein, and DHD was increased in the presence of estrone-3- sulfate, and the permeability of daidzein was increased in the presence of verapamil, MK571, and estrone-3- sulfate. The permeability of daidzein was increased almost two-fold in the presence of estrone-3-sulfate. Conjugation coefficients of genistein, daidzein, and their metabolites When transported samples on both the apical and basolateral sides were treated with a solution containing deconjugating enzyme sulfatase H-5, which contained both sulfatase and glucuronidase, only the apical genistein concentration increased over 4 min (Table 3). Discussion P app-caco-2 estimated high permeability and marked protone-coupled transport (Fig. 2 and Table 2) for all the compounds tested in this study. P app-caco-2 is an estimate of both passive and active transport, including influx and efflux transport, and first-pass metabolism. The compounds in this study were weakly acidic; therefore, they may have diffused passively through membranes according to the ph-partition hypothesis. 24) Compounds with high log P (>2) empirically have high passive permeability. All the isoflavones tested in this study had values for log P > 2 (Table 1), indicating high passive permeability. We used PAMPA to study the mechanisms underlying ph-dependent passive transport, because P app-pampa provides an estimate of passive diffusion permeability. The non-ionized form of these test compounds predominated at ph 6, allowing rapid passage through lipid membranes. Isoflavone dissociation subsequently increased at ph 7.4 (Table 1) after transport to the acceptor side. In this manner, a concentration gradient of non-ionized isoflavones across PAMPA was maintained until equilibrium of the nonionized form across the donor and acceptor sides was achieved. The coefficients, P app-pampa and P app-caco-2, for genistein,, and DHD were positively correlated with C bl /C ap, according to eq. 1, but not P app-pampa for daidzein. These results indicate that the high permeability of genistein,, and DHD in this study depended on passive permeability according to the ph-partition hypothesis. Compounds that are permeable via passive diffusion give strong correlation between PAMPA and Caco-2 cell monolayer permeation data. 23) Figure 3 shows that P app-caco-2 and P app-pampa were not correlated between ph 6. and 7.4, with high P app-caco-2 values for DHD and compared to P app-pampa. Compounds that are actively transported or diffuse passively as acids under a

6 Relative permeation (% of control) Control Transport Mechanisms for Soy Isoflavones and Their Microbial Metabolites 2215 Verapamil MK571 Estrone-3-sulfate Relative permeation (% of control) 2 Daidzein Control Verapamil MK571 Estrone-3-sulfate Relative permeation (% of control) Control Verapamil MK571 Estrone-3-sulfate Relative permeation (% of control) 14 DHD Control Verapamil MK571 Estrone-3-sulfate Fig. 4. Effects of Various Compounds on Isoflavone Efflux Across Caco-2 Cell Monolayers in the Presence of a Proton Gradient. Each value represents the mean SE of three experiments. Significant differences from the control value were identified by statistical analyses performed by repeated-measure 1-way ANVA and subsequent Dunnett multiple-comparison test ( p < :5; p < :1)., dihydrogenistein; DHD, dihydrodaidzein. Table 3. Sulfatase Treatment of Transported Isoflavones and Their Metabolites after Transepitherial Transport Experiments in the Presence of Proton Gradient Amount of compounds after the Compound sulfatase treatment (% of control) Apical side Basolateral side 136:84 12:8 12:95 6:8 94:54 3:97 14:23 4:67 Daidzein 13:7 1:7 15:7 4:26 DHD 93:29 1:4 91:74 5:91 Data are presented as the mean SD. p < :5 versus without treatment with sulfatase., dihydrogenistein; DHD, dihydrodaidzein. ph gradient produce relatively higher Caco-2 permeation. 23) In this study, the P app-caco-2 values for all isoflavones except genistein were higher than the respective P app-pampa values, particularly for DHD and in the presence of a proton gradient. DHD and were potentially actively transported via an unidentified transporter. Although P app-pampa was higher for genistein than for the other isoflavones, P app-caco-2 for genistein was the lowest (Table 2). The dashed circle indicates that genistein at ph 6/7.4 and 7.4/7.4 had a much lower P app-caco-2 values than P app-pampa (Fig. 3). These results indicate the conjugation of genistein by phase II metabolic enzymes in Caco-2 cell monolayers (Table 3). Conjugation is reportedly a major barrier to oral BA of genistein in support of this. 25) Among several isoflavones, genistein has been easily conjugated by phase II metabolic enzymes in the intestines. 26) The conjugation by phase II metabolic enzymes is therefore thought to involve limited genistein permeability. It is known that isoflavones are absorbed by intestinal epithelial cells as aglycones. These compounds subsequently predominate in their glucuronized/sulfated forms in plasma and urine. 27) However, the aglycone forms of isoflavones predominated in the present study and were transported from the apical to basolateral side for 4 min after their addition (Tables 2 and 3). These data suggest that the isoflavones were so quickly absorbed that glucuronidation/sulfation was limited by the flux time. Prolonged incubation of Caco-2 cell monolayers with flavonoids (for over one hour) leads to the formation of conjugates and to increased efflux. However, this evaluation of the initial permeability rate indicates that isoflavones were rapidly absorbed by the intestines. rgans other than the intestines (such as the liver) may therefore be the main sites for the phase II metabolism of isoflavones. To the best of our knowledge, this is first investigation of the intestinal absorption characteristics of and DHD. Microbial metabolites of daidzein in human fecal microbiota comprised 73% of total isoflavones in a previous study, with DHD being the most prevalent and accounting for >31%. 7) The metabolites, and DHD, have been detected at high concentrations in human urine and plasma and may play a role as bioactive components of their parent isoflavones. 7,8) Considering the higher yield and biological properties of microbial isoflavone metabolites in vivo, these bioactive compounds may be responsible for the health effects of dietary isoflavones. However, their absorption and distribution in the body are too poorly characterized to draw any conclusions about the health effects of these microbial metabolites. and DHD in the present study were more permeable than their parent isoflavones (Table 2 and Fig. 2), despite similar physicochemical properties. Moreover, their ionization ratios did not vary between ph 6 and 7.4. Passive diffusion is governed by such underlying physicochemical properties as lipophilicity (log P), pka, molecular weight, and hydrogen bonding 28) which may influence the permeability of isoflavones in a very complex manner. and daidzein (Fig. 1), with unsaturated C2 C3 bonds, form

7 2216 S. KBAYASHI et al. planar structures. In contrast, the saturated C2 C3 bonds of and DHD (Fig. 1) lead to structural flexibility, and the multiple conformational and steric structures may explain the enhanced permeability of these metabolites. Equol is the most active microbiological metabolite of daidzein, and DHD is the intermediate for equol production. The level of DHD excreted in the urine of human subjects was higher than that of equol after an oral administration of the isoflavone, 29,3) while the plasma level of DHD was approximately equal to that of equol. 3) These data imply that DHD had, although being an intermediate product, some effect on the body mechanism. This finding and our results also suggest that DHD was quickly absorbed from the intestines, and could also be rapidly excreted in the urine. Waish et al. have reported that although DHD was present in the urine of all 6 subjects, equol was present in urine in only 3 out of the 6 subjects. 31) It can be considered that individual differences of microflora would affect the mechanism for DHD and equol. Further studies on the plasma and urine concentrations of DHD and equol should consider the effect of individual differences in the microflora. The permeability coefficients of all compounds tested in this study increased when Caco-2 cells had been treated with the BCRP inhibitor, estrone-3-sulfate (Fig. 4). Hence, these compounds may be substrates of BCRP. is a well-established MRP1 inhibitor, 32) although it may not be a substrate of MRP1. In contrast, Kato et al. 27) have shown that phase II metabolites of genistein were substrates of MRP2 and BCRP. Figure 4 demonstrates that BCRP-induced daidzein efflux was the most effective as compared with that of the other tested isoflavones. Moreover, the respective inhibition of P-gp or MRP with Verapamil or MK571 increased daidzein transport, suggesting that daidzein may be a substrate of P-gp and MRP. These data suggest that daidzein was easily recognized by efflux transporters, thereby reflecting its tendency to restrict intestinal absorption. Several studies have shown that BCRP played an important role in determining the disposition of genistein and its phase II metabolites in mice. 9,21) was present at moderately higher concentrations in the blood of Bcrp1 knockout mice than in wild-type mice. 33) and daidzein are known substrates of BCRP. 34) As a member of the ATP-binding cassette transporter family, BCRP is one of the most important xenobiotic efflux transporters. 34) BCRP is expressed not only in multidrug-resistant tumor cells but also in normal tissues, including the intestine, liver, kidney, brain endothelium, and placenta. Imai et al. 35) have shown that genistein effectively reversed BCRP-mediated drug resistance and thereby may be useful in overcoming this in tumor cells. The co-administration of genistein with BCRP substrate inhibitors may alter pharmacokinetics and increase the anti-tumor activity of these agents in cancer patients. 35) Tumor cells express such multidrug-resistance (MDR) transporters as P-gp, BCRP, and MRP which actively export various classes of chemotherapeutic compounds. Inhibiting MDR transporters has been associated with clinically relevant drug drug and food drug interactions, resulting in increased plasma exposure to the administered substrate drugs. The interaction of genistein or A Gut lumen Microbial metabolite B Microbial metabolite Daidzein DHD BCRP MRP BCRP BCRP MRP daidzein with various substrate drugs and MDR transporters has been increasingly studied. However, there are very few reports of the molecular interaction of microbial metabolites of isoflavones. The intestinal absorption mechanisms of and DHD shown in this study are therefore important for future food drug interaction studies. The results obtained in this study together with data from the literature 9,25) indicate a putative mechanism for the absorption of genistein, daidzein, and their metabolites, and DHD (Fig. 5)., and DHD are efficiently transported by passive diffusion according to the ph-partition hypothesis. All isoflavones are natural substrates of BCRP, and genistein was conjugated by phase II metabolic enzymes in this study. Daidzein was not conjugated by phase II metabolites, but was a substrate of several efflux transporters. and daidzein may be restricted in intestinal permeability by phase II enzymes or efflux transporters. However, and DHD were absorbed quickly, and glucuronidation/ sulfation was limited by the flux time. and DHD were more permeable than their parent isoflavones and they were actively transported via an unidentified transporter. Further studies to investigate the intestinal metabolism and permeability of other microbial isoflavone metabolites will help in assessing the health benefits and safety of isoflavone-containing foods. Acknowledgment P-gp Epithelial cells SULT or UGT Daidzein DHD Sul or Glc Plasma Fig. 5. Proposed Pathways for the Short-Term Absorption of Ingested and (A) and Daidzein and DHD (B). All tested compounds in the study were natural substrates of BCRP, and genistein was conjugated by phase II metabolic enzymes. Daidzein was not conjugated by phase II metabolites, but was a substrate of several transporters. Kato et al. have reported that the phase II metabolites of genistein were substrates of MRP2 and BCRP. BCRP, breast cancer resistance protein;, dihydrogenistein; DHD, dihydrodaidzein; MRP, multidrug resistance associated protein; P-gp, P-glycoprotein. This work was supported by grant-aid for young scientists B (227761).

8 References Transport Mechanisms for Soy Isoflavones and Their Microbial Metabolites ) Andres S, Abraham K, Appel KE, and Lampen A, Crit. Rev. Toxicol., 41, (211). 2) Zhang WZ, Cui WM, Zhang X, Wang W, Jia XD, Zhang XP, and Li N, Biomed. Environ. Sci., 22, (29). 3) McVey MJ, Cooke GM, and Curran IH, Reprod. Toxicol., 18, (24). 4) Setchell KD, Brown NM, Zimmer-Nechemias L, Brashear WT, Wolfe BE, Kirschner AS, and Heubi JE, Am. J. Clin. Nutr., 76, (22). 5) Chen KI, Erh MH, Su NW, Liu WH, Chou CC, and Cheng KC, Appl. Microbiol. Biotechnol., 96, 9 22 (212). 6) Gardana C, Canzi E, and Simonetti P, J. Nutr. Biochem., 2, (29). 7) Liang XL, Wang XL, Li Z, Hao QH, and Wang SY, J. Agric. Food Chem., 58, (21). 8) Pfitscher A, Reiter E, and Jungbauer A, J. Steroid Biochem. Mol. Biol., 112, (28). 9) Yang Z, Zhu W, Gao S, Yin T, Jiang W, and Hu M, Drug Metab. Dispos., 4, (212). 1) Liu Y and Hu M, Drug Metab. Dispos., 3, (22). 11) Tian XJ, Yang XW, Yang X, and Wang K, Int. J. Pharm., 367, (29). 12) Kobayashi S and Konishi Y, Biochem. Biophys. Res. Commun., 368, (28). 13) Kobayashi S, Tanabe S, Sugiyama M, and Konishi Y, Biochim. Biophys. Acta, 1778, (28). 14) Kobayashi S, Nagai T, Konishi Y, Tanabe S, Morimoto K, and gihara T, J. Pharm. Pharmacol., 64, 52 6 (212). 15) Shiau YF, Fernandez P, Jackson MJ, and McMonagle S, Am. J. Physiol., 248, G (1985). 16) Takagi M, Taki Y, Sakane T, Nadai T, Sezaki H, ku N, and Yamashita S, J. Pharmacol. Exp. Ther., 285, (1998). 17) Kansy M, Senner F, and Gubernator K, J. Med. Chem., 26, 7 11 (1998). 18) Di L, Kerns EH, Fan K, McConnell J, and Carter GT, Eur. J. Med. Chem., 38, (23). 19) Dempe JS, Scheerle RK, Pfeiffer E, and Metzler M, Mol. Nutr. Food Res., 57, (213). 2) Da Violante G, Zerrouk N, Richard I, Provot G, Chaumeil JC, and Arnaud P, Biol. Pharm. Bull., 25, (22). 21) Guo C, Cao G, Sofic E, and Prior RL, J. Agric. Food Chem., 45, (1997). 22) Hallifax D, Turlizzi E, Zanelli U, and Houston JB, Eur. J. Pharm. Sci., 45, (212). 23) Kerns EH, Di L, Petusky S, Farris M, Ley R, and Jupp P, J. Pharm. Sci., 93, (24). 24) Neuhoff S, Ungell AL, Zamora I, and Artursson P, Eur. J. Pharm. Sci., 25, (25). 25) Zhu W, Xu H, Wang SW, and Hu M, AAPS J., 12, (21). 26) Chen J, Wang S, Jia X, Bajimaya S, Lin H, Tam VH, and Hu M, Drug Metab. Dispos., 33, (25). 27) Kato K, Kusuhara H, Kumagai Y, Ieiri I, Mori H, Ito S, Nakai Y, Maeda K, and Sugiyama Y, Pharmacogenet. Genomics, 22, (212). 28) Cruciani G, Milletti F, Storchi L, Sforna G, and Goracci L, Chem. Biodivers., 6, (29). 29) Franke AA, Halm BM, and Ashburn LA, Nutr. Cancer, 6, (28). 3) Franke AA, Hebshi SM, Pagano I, Kono N, Mack WJ, and Hodis HN, Cancer Epidemiol. Biomarkers Prev., 19, (21). 31) Walsh KR, Haak SJ, Bohn T, Tian Q, Schwartz SJ, and Failla ML, Am. J. Clin. Nutr., 85, (27). 32) Abdul-Ghani R, Serra V, Györffy B, Jürchott K, Solf A, Dietel M, and Schäfer R, ncogene, 25, (26). 33) Álvarez AI, Vallejo F, Barrera B, Merino G, Prieto JG, Tomás-Barberán F, and Espín JC, Drug Metab. Dispos., 39, (211). 34) Enokizono J, Kusuhara H, and Sugiyama Y, Mol. Pharmacol., 72, (27). 35) Imai Y, Tsukahara S, Asada S, and Sugimoto Y, Cancer Res., 64, (24).