Conversion of major soy isoflavone glucosides and aglycones in in vitro intestinal models

Authors


Abstract

Scope

This study compares conversion of three major soy isoflavone glucosides and their aglycones in a series of in vitro intestinal models.

Methods and results

In an in vitro human digestion model isoflavone glucosides were not deconjugated, whereas studies in a Caco-2 transwell model confirmed that deconjugation is essential to facilitate transport across the intestinal barrier. Deconjugation was shown upon incubation of the isoflavone glucosides with rat as well as human intestinal S9. In incubations with rat intestinal S9 lactase phlorizin hydrolase, glucocerebrosidase, and cytosolic broad-specific β-glucosidase all contribute significantly to deconjugation, whereas in incubations with human intestinal S9 deconjugation appeared to occur mainly through the activity of broad-specific β-glucosidase. Species differences in glucuronidation and sulfation were limited and generally within an order of magnitude with 7-O-glucuronides being the major metabolites for all three isoflavone aglycones and the glucuronidation during first pass metabolism being more efficient in rats than in humans. Comparison of the catalytic efficiencies reveals that deconjugation is less efficient than conjugation confirming that aglycones are unlikely to enter the systemic circulation.

Conclusion

Altogether, the data point at possible differences in the characteristics for intestinal conversion of the major soy isoflavones between rat and human, especially with respect to their deconjugation.

Abbreviations
BSβG

broad-specific β-glucosidase

DMEM

Dulbecco's modified Eagle's medium

LPH

lactase phlorizin hydrolase

NB-DGJ

N-(n-Butyl) deoxygalactonojirimycin

TEER

transepithelial electrical resistance

UPLC

ultra performance liquid chromatography

1 Introduction

Isoflavones are naturally occurring ingredients predominantly found in soybean and food products. They can be present as esterified forms (i.e. malonyl glucoside, acetyl glucoside, or glucoside) [1-3]. In whole soybeans and other soy protein products, 97–98% of the isoflavones are present as their esterified conjugated forms predominantly as glucosides, whereas fermented products mostly contain aglycones [1, 4]. The chemical structures of the major isoflavone glucosides and aglycones are shown in Fig. 1. Because of several proposed beneficial health effects including lower incidences for a number of chronic diseases such as mammary, prostate, and colon cancer, osteoporosis, and cardiovascular diseases [5-7], a variety of food supplements containing soy isoflavones are commercially available and the consumption of these food supplements in Western countries increased markedly in the last decades [8]. Being derived from a natural source, consumers often consider soy supplements a safe alternative for “hormone replacement therapy” for menopausal women [9-11] to supplement the decreasing E2 levels and to prevent adverse effects in the body during that period. However, likely adverse effects related to possible promotion of different types of cancer such as breast, endometrial, cervical, and ovarian cancer have also been reported [12-14] and may be related to their estrogen receptor agonist activity [8, 13]. Exposure to isoflavones may also result from the use of isoflavone-containing products as part of the regular diet, such as the use of fermented soy products including tempeh, miso, or natto, which are part of the traditional diet in many Asian countries and now popular in Western societies especially for vegetarians [15]. The health benefits or risks of isoflavones are still controversial [8, 13, 16] and in order to perform an adequate evaluation of their health effects, it is crucial to have knowledge of the biological fate of dietary isoflavones, in particular of their absorption, distribution, metabolism, and excretion characteristics. These absorption, distribution, metabolism, and excretion characteristics will determine the ultimate physiological concentrations of the isoflavones and their metabolites upon intake at dietary relevant levels, but also the health effects upon high-dose exposure in rodent models or upon human intake of highly dosed supplements. The glucoside forms of the soy isoflavones are generally too polar to cross cellular membranes by diffusion [17], and this hampers their cellular uptake and bioavailability [18, 19]. Hence, after intake of soy isoflavone glucosides, hydrolysis is essential to release the biologically active aglycones.

Figure 1.

Chemical structures of soy isoflavone glucosides and their corresponding aglycones.

Hydrolysis of isoflavones to their aglycones has previously been assumed to be predominantly catalyzed by microflora in the colon [17, 20], but recent in vivo human and animal data suggest that it may already occur in the duodenum and proximal jejunum by different intestinal β-glucosidase enzymes [19, 21-24]. Using saliva from human volunteers, Allred et al. [25], reported that genistin can deconjugate to its aglycone genistein already in the mouth. In addition, it is also interesting to know whether the low pH in the stomach can chemically deconjugate the isoflavone glucosides since a considerable number of analytical procedures use acid hydrolysis of these glucosides [2, 26, 27]. Following deconjugation, the aglycones may become conjugated again by first pass metabolism along their transport over the intestinal wall and/or in the liver, before entering the systemic circulation. Clearly the outcome of these processes will influence the nature and amount of the soy isoflavones entering the systemic circulation and exerting a biological effect in vivo. In addition, possible species differences between rats and humans might determine to what extent the rat can be considered an adequate model to predict bioavailability of isoflavones in humans. Therefore, the aim of the present study was to charac-terize possible differences between rats and humans in the conversion of the three major soy isoflavones in the intestine and the liver using S9 fractions. To this end, conversion of the major soy isoflavones was quantified using well-designed in vitro model systems.

2 Materials and methods

2.1 Chemicals and materials

Genistin, daidzin, glycitin (all of >99% purity), and their corresponding aglycones, i.e. genistein, daidzein, and glycitein, were purchased from LC Laboratories (Woburn, MA, USA). Two glucuronide metabolites, namely genistein-7-O-glucuronide and daidzein-7-O-glucuronide were obtained from Extrasynthese (Genay, France). Analytical grade ethanol, NaCl, KSCN, NaH2PO4.H2O, Na2SO4, KCl, CaCl2, NaHCO3, KH2PO4, MgCl2.6H2O, NaOH, 37% HCl, HNO3, NH4Cl, urea, pepsin, d-glucose monohydrate, d-glucosamine hydrochloride, TFA, and ZnSO4 were purchased from VWR International (Darmstadt, Germany). N-(n-Butyl)deoxygalactonojirimycin (NB-DGJ) was obtained from EMD Chemicals (Darmstadt, Germany); uric acid, d-glucuronic acid, α-amylase, mucin, pancreatin (pig), lipase (pig), bile (bovine), BSA, ascorbic acid, and gluconolactone were obtained from Sigma-Aldrich (Steinheim, Germany). ACN (ULC/MS) and methanol (HPLC Supra-gradient) were purchased from Biosolve BV (Valkenswaard, the Netherlands). Conduritol B epoxide was from Santa Cruz Biotechnology (CA, USA) and DMSO from Acros Organics (NJ, USA). Fetal bovine serum was purchased from PAA Laboratories (Pasching, Austria). All cell culture reagents were purchased from Gibco (Paisley, UK). The pooled human intestinal S9 was ordered from Biopredic International (Rennies, France; batch FRA318008), whereas pooled Sprague Dawley (SD) male rat intestinal S9 was ordered from Xenotech (Kanses, USA; batch 1010434). Pooled human liver S9 and SD male rat liver S9 were obtained from BD Biosciences (MA, USA; batch 22877 and batch 88875, respectively). Fresh nanopure water was collected from an Elga ultra-high quality water purification system (High Wycombe, Buckinghamshire, UK).

2.2 Isoflavone content of commercial soy supplements

To select a suitable supplement for the in vitro human digestion model, isoflavone supplements were collected from the local market (Wageningen, the Netherlands) and analyzed for their isoflavone content. Extraction was done following a method described by Fiechter et al. [28]. In brief, 300 mg of the supplement were placed in a 30 mL glass extraction tube. To prevent possible hazy extracts, 150 mg ZnSO4 was added from a 0.3 g/mL stock solution in nanopure water. After that, 20 mL of 80% v/v methanol/water were added as extraction medium. The mixture was placed in an orbital shaker at 550 rpm for 1 h at room temperature followed by centrifugation at 2000 × g for 15 min at 10°C. After transferring the supernatant to a 100 mL volumetric flask, the remaining pellet was extracted once more using the same procedure with fresh 80% methanol, but without ZnSO4. The supernatants thus obtained were combined and the volume was adjusted to 100 mL with 80% methanol. The extract was subsequently stored at 4°C until analysis. Before ultra performance li-quid chromatography (UPLC) analysis, the extract was diluted ten times in nanopure water followed by centrifugation at 16 000 × g for 5 min and filtrated using a 0.2 μm cellulose acetate filter from VWR (West Chester, PA, USA). Identification of the different soy isoflavones and the corresponding aglycones was done by comparison of their retention times and UV spectra to those of commercially available standards. The compounds were quantified based on comparison of the area under the curve of their peaks in the UPLC chromatograms to those of calibration curves made using commercially available standards. The recovery factors for genistein, daidzein, and glycitein, of 94.5, 94.7, and 93.8%, respectively [28] were taken into account to calculate the total amount of isoflavones in the supplement. The amount of isoflavones claimed to be present in the supplements as well as the resulting estimated daily intake were established using the information provided by the supplier on the label or the relevant Internet site, usually recommending two capsules per day.

2.3 In vitro human digestion model

The in vitro human digestion model used was the one developed by the National Institute for Public Health and the Environment (RIVM, Bilthoven, the Netherlands). Figure 2 presents a schematic overview of this digestion model described in detail by Oomen et al. [29]. In accordance with this publication, different digestion juices (i.e. saliva, gastric, duodenal, and bile juice) were prepared 1 day before the experiment. Three hundred milligrams of a selected isoflavone supplement (containing genistin, daidzin, and glycitin at levels quantified (see Results)) without capsule were placed in a 30 mL glass tube and 6 mL of saliva juice (pH 6.0 ± 0.5) were added. After a quick shaking, a 0.1 mL sample was collected immediately. The mixture was then rotated with an upside-down rotor for 5 min at 40 rpm at 37°C. After this, another 0.1 mL sample was collected and subsequently 12 mL of gastric juice were added and the pH was adjusted to 1.5 ± 0.5 with 37% HCl. Then the mixture was placed in the rotor for 2 h at the same speed and temperature, followed by collection of again 0.1 mL sample. Finally, 12 mL of duodenal juice and 6 mL of bile juice were added to the mixture, the pH was adjusted to 6.0 ± 0.5 with 37% HCl and the sample was placed in the rotor for another 2 h. After that, the last sample of 0.1 mL was collected. Before starting the experiment, all the digestive juices were preincubated at 37 ± 2°C and the incubations were performed at the same temperature. Collected samples were stored at −80°C until analysis by UPLC.

Figure 2.

Schematic presentation of different steps of the human in vitro digestion model, indicating the collection of the samples to be analyzed after ingestion, passing the oral cavity, the stomach, and the intestinal compartment, respectively (amended from Omen et al. [29]).

2.4 Intestinal Caco-2 cell model

Caco-2 cells from ATCC (Manassas, VA, USA) were cultured in a humidified atmosphere of 5% CO2 at 37°C in Dulbecco's modified eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 1% minimum essential media nonessential amino acids, and 0.1% mg/mL gentamicin as described by Brand et al. [30]. For transport experiments, 1 × 105 cells/well were seeded from a stock of 2 × 105 cells/mL in a Corning Costar 12-well transwell plate with an insert membrane. This insert membrane has a pore size of 0.4 μm and growth area of 1.12 cm2. During formation of the monolayer, the medium was changed three times a week and 18 or 19 days postseeding the experiments were performed. The passage number of the cells used in the experiments was between 45 and 51. Before exposure, monolayers were washed with DMEM (without phenol red) and the integrity of the monolayers was checked by measuring transepithelial electrical resistance (TEER) values using a Millicell ERS volt/ohmmeter from Millipore (Bedford, MA, USA). Only monolayers demonstrating a TEER value between 500 and 1000 Ω·cm2 were used. Transport experiments were carried out with medium consisting of DMEM (without phenol red) supplemented with 1% v/v minimum essential media nonessential amino acids. To all media, 1 mM ascorbic acid (final concentration) was added to prevent auto-oxidation of the isoflavones.

Cells were exposed to 50 μM of the isoflavone glucosides and/or their corresponding aglycones at the apical side and 100 μL sample was collected at eight time points (0, 10, 20, 40, 60, 80, 100, and 120 min) both from the apical and the basolateral compartments. A complementary volume (i.e. 100 μL growth medium containing 50 μM soy isoflavones) was added every time only in the apical compartment just after sample collection. In all experiments, the concentration of DMSO was kept < 0.5%. Collected samples were kept on ice during the experiment and subsequently stored at −80°C until analysis by UPLC. At the end of the experiment, the TEER values were checked again to confirm the integrity of the monolayers.

2.5 Enzymatic deconjugation of isoflavone glucosides by rat and human intestinal S9

To obtain the kinetic constants for deconjugation of the isoflavone glucosides, incubation mixtures (total volume 100 μL) contained 100 mM potassium phosphate (pH 7.0), increasing concentrations of substrate (1 to 200 μM for genistin and daidzin and 1 to 500 μM for glycitin, respectively) all added from 200 times concentrated stock solutions in DMSO, and 0.15, 0.40, or 0.10 mg protein/mL rat intestinal S9 or 0.13, 0.40 or 0.10 mg protein/mL human intestinal S9 for genistin, daidzin, and glycitin, respectively. Incubations of genistin, daidzin, and glycitin with rat intestinal S9 were performed for 15, 10, and 40 min, respectively and with human intestinal S9 for 10, 10, and 15 min, respectively. The reactions were terminated by adding equal volumes of ice-cold methanol containing 0.8 mM ascorbic acid to stabilize the sample as described by Day et al. [24]. To identify the role of different glucosidases present in intestinal S9 fractions, incubations were also performed in the presence of specific inhibitors including 250 μM NB-DGJ, 5 mM conduritol B epoxide, and 10 mM δ-gluconolactone that were all used at concentrations selected based on literature information [24, 31, 32]. Table 1 presents an overview of the different enzyme inhibitors used and their specificity for the intestinal enzymes at the concentrations generally used in literature. Kinetic parameters were measured in duplicate.

Table 1. Different enzyme inhibitors with their specific inhibition potency for various intestinal enzymes
InhibitorExposure conc. (mM)Inhibited enzyme(s)Percentage of inhibitionReference
NB-DGJ0.25LPH60[31]
 10Glucocerebrosidase51[32]
Conduritol B epoxide5Glucocerebrosidase100[24]
 5LPH87[31]
 6.20Glucocerebrosidase100[32]
Gluconolactone10BSβG93[24]
  Glucocerebrosidase51[24]

2.6 Enzymatic conjugation of isoflavone aglycones by rat and human intestinal and liver S9

To study glucuronidation of aglycones, incubation mixtures (total volume 99.5 μL) were prepared containing (final concentrations) 10 mM MgCl2, 25 μg/mL alamethicin added from a 200 times concentrated stock solution in methanol and 10 mM uridine 5′-diphosphoglucuronic acid trisodium salt in 50 mM Tris-HCl (pH 7.4) [33]. The reactions were started by adding 0.5 μL of the substrate genistein, daidzein, or glycitein (final concentration ranges were between 0.5 to 400 μM added from 200 times concentrated stock solutions in DMSO) to incubation mixtures containing 0.05, 0.25, or 0.25 mg protein/mL rat intestinal S9, respectively, and incubated for 15, 10, or 10 min, respectively. Incubation with rat liver S9 were carried out using 0.5, 0.5, or 0.02 mg protein/mL for 5, 5, or 3 min, respectively. Incubations of genistein, daidzein, and glycitein were carried out with 0.05, 0.25, or 0.25 mg protein/mL human intestinal S9 for 10, 10, or 10 min, respectively and with 0.5, 0.5, or 0.5 mg protein/mL human liver S9 for 10, 10, or 10 min, respectively. The reactions were terminated by adding 25 μL ice-cold ACN. Under these conditions, metabolite formation was linear with time and with the amount of protein added (data not shown). All incubations were conducted at 37°C and activities were expressed in nmol min−1 mg protein−1. Samples were stored at −80°C until analysis by UPLC. Kinetic parameters were measured in duplicate.

To study sulfation of the aglycones, incubation mixtures (total volume 99.5 μL) contained (final concentrations) 100 μM 3′-phosphoadenosine 5′-phosphosulfate in 50 mM potassium phosphate (pH 7.4) containing 5 mM MgCl2 and S9 fractions. The reactions were started by addition of 0.5 μL genistein, daidzein, or glycitein (final concentrations ranging from 0.25 to 400 μM added from 200-fold concentrated stock solutions in DMSO) to incubation mixtures containing 0.5 mg protein/mL rat intestinal S9 and incubated for 90, 120, 30 min, respectively. Incubations with rat liver S9 were carried out using 0.2, 0.2, 0.1 mg protein/mL for 60, 15, 5 min, respectively. Incubation of genistein, daidzein, and glycitein was also carried out with 0.2, 0.2, and 0.5 mg protein/mL human intestinal S9 for 30, 30, and 10 min, respectively and with 0.2, 0.2, and 0.2 mg protein/mL human liver S9 for 10, 30, and 10 min, respectively. These reactions were terminated by adding 25 μL ice-cold ACN. Under these conditions, metabolite formation was linear with time and with the amount of protein added (data not shown). Samples were stored at −80°C until analysis by UPLC.

2.7 UPLC analysis for identification and quantification of isoflavones

All samples were analyzed using a Waters AcquityTM UPLC, (Milford, MA, USA) that consists of a binary solvent manager, sample manager, and photodiode array detector, equipped with a Waters Ethylene Bridged Hybrid (BEH) C18 1.7 μm 2.1 × 50 mm column (Waters, Ireland). Nanopure water with 0.1% TFA and 100% ACN were used as solvent A and B, respectively. After centrifugation at 16 000 × g for 4 min, samples were analyzed. In case of samples collected from supplement extraction and the in vitro human digestion model, an extra filtration step (0.2 μm filter, VWR, PA, USA) was performed before UPLC analysis. The injection volume for UPLC ana- lysis was 3.5 μL and the flow rate was 0.6 mL/min. Elution was started with 0% of solvent B followed by an increase in percentage of solvent B from 0 to 10, 15, 50, and 80% at 0.58, 2.85, 4.28, and 4.40 min, respectively. The 80% solvent B condition was kept until 4.52 min, and thereafter the percentage of solvent B was reduced to 0% at 4.63 min and maintained at that percentage until 5.80 min. Photodiode array spectra were detected between 200–360 nm and chromatograms acquired at 260 nm were used for quantification of the amount of isoflavone glucosides and aglycones using calibration curves of their commercially available reference compounds. Using the above-mentioned UPLC conditions, the retention times of the different test compounds were as follows; genistein 3.71 min (UVmax 260 nm); daidzein 3.36 min (UVmax 249 nm); glycitein 3.46 min (UVmax 257 and 319.8 nm); genistin 2.56 min (UVmax 259.4 nm); daidzin 1.76 min (UVmax 249 and 301.9 nm); and glycitin 1.92 min (UVmax 257.6 and 320.4 nm). The 7-O-glucuronides of genistein and daidzein are commercially available and the monoglucuronides were quantified based on calibration curves made with these available standards. The other monoglucuronide and monosulfate conjugates were quantified using the standard curve of their corresponding glucosides. This was possible because the 7-O-glucuronides of genistein and daidzein were shown to have similar UV spectra and calibration curves as their corresponding glucosides (data not shown).

2.8 Analysis of kinetic parameters

The maximum velocity (Vmax), expressed in nmol/min/mg S9 protein, and Michaelis–Menten constant (Km) expressed in micromoles, were determined by fitting the data to the Michaelis–Menten equation: v = Vmax * [S]/ (Km+ S), with [S] being the concentration of the substrate, using Graphpad Prism (version 5.02) from Graphpad Software (San Diego, CA, USA). Based on the kinetic constants obtained, the catalytic efficiency (Vmax/Km), expressed as mL/min/mg S9 protein or μL/min/mg S9 protein, for the different reactions was calculated.

3 Results

3.1 Isoflavone content in commercial soy supplements

Since we aimed at investigating the species differences between rat and human for conversion of isoflavones from soy supplements as a possible important source for current dietary exposure in the Western world, nine different supplements were randomly chosen from the local market. Figure 3A and B present the UPLC chromatograms of two analyzed commercial supplements. These chromatograms reveal that the isoflavone content and composition of different supplements vary considerably. Supplement number 1 (Fig. 3A) contains mainly the three major soy isoflavone glucosides daidzin, glycitin, and genistin, whereas supplement number 5 (Fig. 3B) contains only small amounts of daidzin and genistin and relatively larger amounts of their corresponding aglycones daidzein and genistein along with several minor unidentified ingredients. Figure 3C presents the overall isoflavone content determined for the different soy supplements compared to the amount claimed to be present on the labels or on the specified websites. For all supplements, except for supplement 1, the actual amounts detected appeared to be significantly lower than the amounts expected based on information provided by the supplier. Based on these results, supplement number 1 was selected as the supplement to be used in the subsequent in vitro human digestion model study since it appeared to contain all three model isoflavone glucosides in substantial amounts.

Figure 3.

Representative UPLC chromatograms of commercially available supplements containing especially isoflavone glucosides (A) or aglycones (B) and a comparison of the total isoflavone content (i.e. glucosides and aglycones) as detected in different commercial soy supplements and the amounts claimed to be present in the manufacturer's information (C). The values are given in milligram per 300 mg capsule content (each data point represents the average value of three independent experiments).

3.2 Deconjugation of soy isoflavone glucosides in an in vitro human digestion model

Table 2 shows the amount of total isoflavones (i.e. glucosides plus aglycones) and aglycones detected in different samples from the human digestion model. The results obtained reveal that incubation with different human model digestion juices does not convert the isoflavone glucosides to their respective aglycones to any significant extent, not even after 2-h incubation at low gastric pH (1.5 ± 0.5). No deconjugation was detected even when the experiment was repeated with modified gastric juices that did not contain BSA, mucin, and pepsin, thereby eliminating the possible protection against deconjugation by these proteins.

Table 2. Amount of isoflavones in milligram (total i.e. glucosides plus aglycones and aglycones detected in the human digestion model, (A) a digestion model with normal gastric juice and (B) a digestion model with modified gastric juice (without BSA, mucin, and pepsin). The values for total isoflavones and aglycones are presented in milligram per 300 mg of commercial supplement
Type of gastric juiceSample numberSituation representedIncubation timeTotal isoflavones (mg)Aglycones (mg)
A1Just after oral ingestion012.7 ± 1.30.35
 2After passing oral cavity5 min13.2 ± 3.50.26
 3After passing stomach2 h14.6 ± 3.50.35
 4After passing small intestine2 h16.1 ± 6.10.34
B1Just after oral ingestion013.6 ± 3.20.40
 2After passing oral cavity5 min12.8 ± 2.80.40
 3After passing stomach2 h14.9 ± 2.90.40
 4After passing small intestine2 h15.2 ± 2.90.50

3.3 Transport of isoflavones in a Caco-2 cell model

In a next step, it was investigated to what extent the soy isoflavone glucosides would be able to pass, at least to some extent, the intestinal barrier themselves. This was done by using a Caco-2 cell monolayer two-compartment transwell system, an in vitro model often applied to study transport across the intestinal barrier [30, 33, 34]. As can be seen in Fig. 4, no transport of the soy isoflavone glucosides across the Caco-2 monolayer was observed, whereas addition of the corresponding aglycones to the apical side of the monolayers resulted in a time-dependent increase in the amount of these isoflavone aglycones in the basolateral compartments. No metabolite formation was observed when Caco-2 cells were exposed to the aglycones (UPLC chromatograms not shown). The transport efficiency of three aglycones decreased in the order genistein > daidzein > glycitein, with the amount of genistein transported after 120 min being 2.3- and 4.1-fold higher than that of daidzein and glycitein, respectively.

Figure 4.

Time-dependent increase of the amount (nmol) of three soy glucosides and the corresponding aglycones in the basolateral compartment of Caco-2 transwell culture exposed to the isoflavones (50 μM) at the apical side. Each data point represents the average value from three independent experiments.

3.4 Enzymatic deconjugation of soy isoflavone glucosides by rat and human intestinal S9

Since naturally existing soy isoflavone glucosides were not deconjugated by the acidic stomach pH (see Section 'Deconjugation of soy isoflavone glucosides in an in vitro human digestion model') and also were not able to cross the Caco-2 monolayer barrier (see Section 'Transport of isoflavones in a Caco-2 cell model'), human and rat intestinal S9 fractions were tested for their soy isoflavone glucoside deconjugating ability. Table 3 depicts the kinetic parameters (Vmax and Km) for deconjugation by rat and human intestinal S9 as well as the catalytic efficiencies (Vmax/Km) derived from these values. Compared to the catalytic efficiencies for deconjugation by rat intestinal S9, catalytic efficiencies for deconjugation by human intestinal S9 were 3.9- and 6.5-fold higher for genistin and daidzin and almost similar for glycitin.

Table 3. Kinetic parameters for deconjugation of three soy isoflavone glucosides using human or rat intestine S9 fractions
Source of enzymeGlucosidesKmVmaxVmax/Km
  (μM)(nmol min−1 mg protein−1)(μL min−1 mg protein−1)
Rat intestine S9Genistin133.3 ± 7.84.8 ± 0.836.6
 Daidzin507.2 ± 77.22.0 ± 0.63.8
 Glycitin656.3 ± 611.2 ± 0.417.1
Human intestine S9Genistin28.7 ± 0.54.1 ± 0.7142.5
 Daidzin114.4 ± 21.02.8 ± 0.024.7
 Glycitin185.1± 79.53.1 ± 0.418.1

To investigate possible further differences between rat and human intestinal deconjugation of the soy isoflavone glucosides, inhibition of the reaction by selected glucosidase inhibitors (Table 1) was investigated. Figure 5 shows the inhibition of the dose-dependent deconjugation of the isoflavone glucosides in the absence and presence of the different inhibitors. The results presented point at a significant species dependent difference between rat and human with regard to the type of enzymes involved in deconjugation. Figure 5A and B show that deconjugation of genistin by rat intestinal S9 was inhibited by all inhibitors tested, whereas in incubations with human intestinal S9 especially gluconolactone inhibited the deconjugation. For the inhibition of the deconjugation of daidzin (Fig. 5C and D) and glycitin (Fig. 5E and F) similar species dependent differences were observed. These results suggest that in human intestinal S9 samples, especially broad-specific β-glucosidase (BSβG) is involved in the deconjugation of the isoflavone glucuronides, whereas in rat intestinal S9, there is an additional contribution from lactase phlorizin hydrolase (LPH) and/or glucocerebrosidase.

Figure 5.

Dose-dependent deconjugation of genistin (upper panel), daidzin (middle panel), and glycitin (lower panel) glucosides after incubation with rat intestinal S9 (left panel) and human intestinal S9 (right panel) in the absence (●), or presence of the following inhibitors: 250 μM NB-DGJ (▲) as a lactase domain selective inhibitor of LPH; 5 mM of Conduritol B epoxide (▪) as an inhibitor of both glucocerebrosidase and LPH; 10 mM gluconolactone (◻) as an inhibitor of cytosolic BSβG and glucocerebrosidase.

3.5 Evaluation of the use of intestinal S9 as a model for intestinal deconjugation

Given that part of the glucosidases present in intestinal S9 are intracellular enzymes with only LPH being located on the brush border of the enterocytes [31, 35], it was of interest to evaluate the possible overestimation of the level of deconjugation of isoflavone glucosides obtained when using the cell-free S9 samples. To obtain some insight in this possible overestimation, the in vitro catalytic efficiency for deconjugation of genistin by rat intestine S9 was compared to the catalytic efficiency calculated based on literature studies reporting intestinal perfusion data for the same compound. To allow comparison, the in vitro catalytic efficiency of 0.036 mL/min/mg S9 protein (Table 3) was first scaled to the whole organ by multiplying with an S9 protein yield of 11.4 mg S9/g intestine [36, 37]. At organ level, the catalytic efficiency for deconjugation is then 0.4 mL/min/g intestine. In three published perfusion studies with genistin [19, 38, 39], the clearance or catalytic efficiency of the intestine (CE expressed as mL/min) for deconjugation was calculated based on the reported inlet (Cin expressed as μmol/mL) and outlet concentrations (Cout expressed as μmol/mL) of genistin and the applied perfusion flow rate (Qp expressed as mL/min) with the following equation [40-42]:

display math

In this equation, the fraction of genistin that does not appear in the outlet ((Cin‒Cout)/Cin) is considered to completely correspond to the fraction of genistin that is deconjugated, since loss of genistin between the inlet and outlet due to possible uptake is also assumed to reflect deconjugation given that the aglycone is the preferential form for uptake [19, 23, 38]. In a study of Andlauer et al. [38], small intestines of male Sprague–Dawly rats were ex vivo perfused with a flow rate of 0.5 mL/min. Experiments were performed at three different inlet concentrations of 5.9, 12.0, and 23.8 μmol/L genistin. The reported outlet concentrations of genistin were 2.4, 7.0, and 10.2 μmol/L, respectively. Based on these data, deconjugation catalytic efficiencies were calculated to range from 0.2 to 0.3 mL/min for the three different inlet concentrations. In a study reported by Steensma et al. [39], the small intestines of male Wistar rats were in situ perfused with a flow rate of 1 mL/min at an inlet concentration of 55 μmol/L genistin. The average reported outlet concentration of genistin was 26 μmol/L, resulting in a deconjugation catalytic efficiency of 0.5 mL/min. In the study of Liu and Hu [19], different segments of the small intestines of male Sprague–Dawley rats were separately perfused in situ with a flow rate of 0.191 mL/min and the reported inlet concentration of genistin was 100 μmol/L and outlet concentrations of genistin were 43, 67, and 87 μmol/L for duodenum, jejunum, and ileum, respectively. Based on these values, the sum of the deconjugation efficiency by the different intestinal segments can be calculated to be 0.2 mL/min. Overall, the calculated deconjugation efficiencies for the in situ and ex vivo perfusion studies are comparable ranging from 0.2 to 0.5 mL/min for the whole small intestine corresponding to 0.06 to 0.14 mL/min/g intestine (calculated based on an average small intestine weight of 3.5 g for rats [43]). The scaled catalytic efficiency for deconjugation as determined with the incubations with rat intestine S9 fractions that amounted to 0.4 mL/min/g intestine, was three- to sevenfold higher than these ex vivo and in situ perfusion values.

Based on the incubations with the different inhibitors of glucosidase enzymes, an estimation can be made of the contribution of LPH to the catalytic efficiency for deconjugation by rat intestine S9. From Fig. 5A, it can be derived that the maximum deconjugation activity of genistin is about 60% lower in the presence of NB-DGJ, which inhibits LPH (Table 1). This suggests that 60% of the total S9 catalytic efficiency of 0.036 mL/min/mg S9 protein (Table 3) may be attributed to LPH activity, which corresponds to ∼0.022 mL/min/mg S9 protein or 0.25 mL/min/g intestine when scaled from S9 to the intestine. This latter value is quite comparable to the catalytic efficiencies derived from the data in literature on in situ and ex vivo perfusion studies ranging from 0.06 to 0.14 mL/min/g intestine. This confirms that the primary site for deconjugation is at the brush border in a reaction catalyzed by LPH. In incubations with human intestine S9 (Fig. 5B), the maximum deconjugation of genistin is only about 16% lower in the presence of NB-DGJ. Combining this result with the catalytic efficiency observed in incubations with human intestine S9 of 0.14 mL/min/mg S9 protein (Table 3), one could calculate that the deconjugation efficiency by LPH is ∼0.022 mL/min/mg S9 (i.e. 16% of the total deconjugation catalytic efficiency of 0.14 mL/min/mg S9) or 0.25 mL/min/g intestine when scaled from S9 to the intestine. This suggests that despite a relative lower contribution of LPH to deconjugation, the absolute deconjugation efficiency by LPH may still be comparable between humans and rats.

3.6 Enzymatic conjugation of isoflavone aglycones by rat and human intestinal and liver S9

Table 4 presents the different kinetic parameters of glucuronidation by small intestinal and liver S9 for both rat and human. Formation of 7-O- as well as 4′-O-glucuronide metabolites was observed, although for both rat and human intestine and liver catalytic efficiency for formation of the 7-O-glucuronides was generally one to two orders of magnitude higher than that for formation of 4′-O-glucuronides. This was due to lower Km in combination with higher Vmax values and indicates that formation of 7-O-glucuronides is preferred over formation of 4′-O-glucuronides in both rat and human. Furthermore, the catalytic efficiencies for the production of 7-O-glucuronides by rat intestinal S9 were 10.1-, 2.3-, and 2.4-fold higher than those for human intestinal S9 and those for rat liver S9 were 1.4-, 4.9-, and 11.7-fold higher than those for human liver S9 and genistein, daidzein, and glycitein, respectively. This suggests glucuronidation to be more efficient in rat than in human first pass metabolism, although the differences seem small and are generally less than one order of magnitude.

Table 4. Apparent Vmax and Km values of aglycones and catalytic efficiencies (Vmax/Km) derived from these values for glucuronidation using human and rat, small intestinal, and liver S9 fractions
Source of enzymeAglycones7-O-glucuronide4′-O-glucuronide
  KmVmaxVmax/ KmKmVmaxVmax/ Km
  (μM)(nmol min−1 mg(μL min−1 mg(μM)(nmol min−1 mg(μL min−1 mg
   protein−1)protein−1) protein−1)protein−1)
  1. ND, not detectable because of substrate inhibition at higher concentrations and too low conversion at lower concentrations.

Rat intestine S9Genistein2.8 ± 2.83.6 ± 0.1129026.4 ± 18.10.15 ± 0.016
 Daidzein7.1 ± 2.81.3 ± 0.119049.2 ± 22.30.15 ± 0.023
 Glycitein3.0 ± 1.43.2 ± 0.61070NDNDND
Rat liver S9Genistein51.6 ± 3.05.1 ± 1.310038.1 ± 2.80.3 ± 0.17
 Daidzein5.6 ± 3.71.4 ± 0.325077.7 ± 14.50.3 ± 0.13
 Glycitein1.8 ± 1.55.3 ± 4.32980NDNDND
Human intestine S9Genistein5.8 ± 6.70.7 ± 0.713018.9 ± 12.30.2 ± 0.111
 Daidzein2.7 ± 1.20.2 ± 0.1802.9 ± 1.10.1 ± 0.140
 Glycitein3.4 ± 1.51.5 ± 0.54504.2 ± 0.50.1 ± 0.0117
Human liver S9Genistein22.2 ± 15.11.5 ± 0.77031.9 ± 18.50.1 ± 0.013
 Daidzein18.9 ± 14.11.0 ± 0.05072.1 ± 61.10.2 ± 0.012
 Glycitein7.5 ± 5.61.9 ± 0.425064.3 ± 22.50.1 ± 0.021

Upon incubation of the isoflavone aglycones with rat or human intestinal or liver S9 and 3′-phosphoadenosine 5′-phosphosulfate formation of only 7-O-sulfate metabolites was observed (Table 5) with human intestinal and liver S9 showing higher catalytic efficiencies than the corresponding rat S9 fractions. The catalytic efficiencies for sulfation by human intestinal S9 were 558- and 1.3-fold higher than those of rat intestinal S9 to conjugate daidzein and glycitein, while for human liver S9, they were 29-, 2.2-, and 18-fold higher than those of rat liver to conjugate genistein, daidzein, and glycitein, respectively. Due to substrate inhibition at higher concentrations and too low conversion at lower concentrations, the kinetic parameters of genistein sulfation by human intestinal S9 fraction could not be quantified.

Table 5. Apparent Vmax and Km values of aglycones and catalytic efficiencies (Vmax/Km) derived from these values for sulfation using human and rat, small intestinal and liver S9 fractions (based on results from single experiments)
S9 fractionsAglyconesKm μMVmaxVmax/Km
   Nmol min−1 mg protein−1μL min−1 mg protein−1
  1. ND, not detectable because of substrate inhibition at higher concentrations and too low conversion at lower concentrations.

Rat intestine S9Genistein35.440.010.28
 Daidzein98.170.010.10
 Glycitein1.560.017
Rat liver S9Genistein33.040.134
 Daidzein1.820.0211
 Glycitein10.170.2020
Human intestine S9GenisteinNDNDND
 Daidzein0.350.0257
 Glycitein16.880.138
Human liver S9Genistein0.440.05114
 Daidzein0.820.0224
 Glycitein0.230.08348

4 Discussion

Knowledge on the conversion of soy isoflavones is crucial to understand and evaluate their possible health benefits and risks. Furthermore, quantification of possible species differences in kinetics of isoflavones is essential for adequate extrapolation of animal experimental data to the human situation. In the present study, different in vitro models were used to investigate the conversion of the major soy isoflavones and possible species differences between rat and human in this conversion. Apart from genistin and daidzin, an important soy isoflavone glycitin was included in the studies because information on the absorption and metabolism of this isoflavone is still lacking. Thus the present study provides a comparative investigation of metabolic characteristics in rat and human of the three major soy isoflavones and their corresponding aglycones in a series of in vitro models.

Before starting the metabolism studies, a series of commercially available soy supplements was analyzed for their isoflavone content to select a suitable supplement to be tested in the human digestion model and to obtain some insight in the nature of the isoflavones generally present in this possibly important dietary source of isoflavones in the Western diet. To this end, nine isoflavone supplements were collected from the local market (Wageningen, the Netherlands) and analyzed for their isoflavone content and composition. Results obtained reveal that eight out of nine supplements analyzed did not contain the amount of isoflavones as indicated on the labels or relevant websites. This illustrates that it is difficult to predict the exposure or intake by a target population based on the information presented by the suppliers. The calculated estimated daily intake of total isoflavones (glucosides plus aglycones) calculated based on the analytical results presented ranges from 17 to 88 mg per day which is in accordance with estimates presented by Eisenbrand et al. [44]. Several other groups also noticed such unexpected low yields compared to the labeled specifications for isoflavone supplements [2, 26, 28]. The results also indicated variation in the relative level of isoflavone glucosides or aglycones in the supplements, a factor that may affect bioavailability since plasma serum levels upon intake of genistein and daidzein have been shown to be much higher than those upon intake of the corresponding glucosides [45-47]. Based on the results obtained, the soy supplement with high levels of the major three soy isoflavone glucosides was selected to be tested in the human digestion model.

The in vitro human digestion model used was the model developed by the National Institute for Public Health and the Environment [29]. The results obtained clearly indicate that digestive juices applied are not able to deconjugate the isoflavone glucosides. This finding indicates that although concentrated acid hydrolysis at elevated temperature (≥80°C) is frequently used in analytical methods for quantification of isoflavones, physiologically relevant acidic pH values combined with a physiologically relevant temperature (37°C) are not sufficient to deconjugate the isoflavone glucosides. Our results also indicate that the 70% deconjugation of genistin after 90-min incubation with human saliva reported by Allred et al. [25] may be due to prolonged incubation time (90 min instead of the physiologically relevant 5-min incubation) in combination with the microbial activity present in the saliva juice applied in their in vitro digestion model.

Since isoflavone glucosides are thus likely to end up intact in the small intestine, their transport over an in vitro intestinal barrier model was investigated using the Caco-2 transwell model. Isoflavone glucosides were not able to cross the Caco-2 monolayer whereas the corresponding aglycones were translocated in a time-dependent manner. These results are in line with other studies reporting that initial hydrolysis is essential for absorption of related flavonoid glucosides [23, 34, 45]. The present study reveals differences in transport efficiency of the three major soy isoflavone aglycones, with transport decreasing in the order genistein > daidzein > glycitein. The transport of genistein is 2.3- and 4.1-fold faster than that of daidzein and glycitein, respectively. This difference between translocation of genistein and daidzein across the Caco-2 model layer is in line with the about 1.3- to 3.0-fold difference in rat plasma metabolite concentrations of genistein and daidzein upon dosing 7.9 μmol isoflavone aglycones per kilogram bw to rats [45] and with the about 1.3-fold faster transport of genistein compared to daidzein after 2 h incubation of Caco-2 cell monolayers with 50 μM isoflavone aglycones as reported by Steensma et al. [48]. The Caco-2 cells appeared unable to deconjugate the isoflavone glucosides and/or to metabolize the isoflavone aglycones to their corresponding glucuronide and sulfate metabolites, confirming the absence of LPH and phase II conjugating enzymes in these cells as reported before [19, 31]. Liu en Hu [19] and also Chen et al. [49] studied the transport of isoflavone glucosides or aglycones using cloned Caco-2 TC7 cells. Chen et al. [49] reported transport of especially isoflavone aglycones. Liu and Hu [19] reported limited transport of the glucoside genistin to the basolateral side, whereas in our studies this transport was not observed. This difference might be due to the fact that the studies of Liu and Hu [19] used the cloned Caco-2 TC7 cell line isolated from a late passage of the parental Caco-2 cells to provide a more homogeneous population, somewhat different pH values, and longer incubation times whereas they did not report the TEER values of their monolayer after the prolonged 4-h incubation, leaving the possibility that the limited transport observed may have been due to some loss in monolayer integrity. Results from other studies using parental Caco-2 cells are in line with our results reporting that glucosides are not transported across the Caco-2 monolayer to any significant extent [34, 48]. Our results are also in line with human data reporting that isoflavone glucosides are not absorbed as such across the enterocytes of healthy adults and that initial hydrolysis by intestinal enzymes is essential for their transport across the enterocytes [46].

Given that isoflavone glucosides will not be hydrolyzed before entering the small intestine deconjugation by rat and human intestinal samples was investigated. There is evidence in literature that some flavonoid glucosides can be hydrolyzed in the small intestine [23, 50]. They may be hydrolyzed by LPH, located in the brush border of the small intestine [31, 35], and/or by a BSβG and/or glucocerebrosidase upon entering the intestinal epithelial cells via specific transporters [24, 51, 52]. In subsequent experiments of the present study, incubations with intestinal S9 fractions from human and rat were used to study the deconjugation of the isoflavone glucosides in the absence or presence of specific inhibitors of LPH, glucocerebrosidase, and BSβG (Table 1). The results obtained point at a significant species-dependent difference between rat and human, since incubations with rat intestinal samples reveal inhibition by all inhibitors tested, whereas in incubations with human intestinal S9 predominantly gluconolactone inhibited the deconjugation almost completely. This indicates that in human intestinal S9 samples especially intracellular BSβG is involved in deconjugation, whereas in rat intestinal S9, there is a significant contribution from LPH and glucocerebrosidase. This finding together with the outcome of the Caco-2 study might lead to the conclusion that in human, deconjugation is largely carried out by gut microbiota as reported in the literature [17, 20]. However, Walsh et al. [23] reported that isoflavonoid glucosides are deconjugated and absorbed in the small intestine of human subjects with ileostomies, indicating that deconjugation can take place in the small intestine where the microbial contribution is very limited.

To further evaluate the appropriateness of the use of cell free S9 samples to study intestinal deconjugation, the catalytic efficiency for deconjugation calculated based on our in vitro data was compared to the in vivo cata-lytic efficiencies calculated based on data from intestine perfusion studies reported in literature. This comparison revealed that in vitro S9 clearance as determined by the intestinal S9 rat incubations results in three- to sevenfold overestimation compared to intestinal perfusion studies. The presence of all cytosolic and membrane-bound enzymes in intestinal S9 fractions (http://biofocus .com/offerings/adme-pk-laboratory/s9fraction) might be the cause of this overestimation in the in vitro model since in the perfusion studies glucocerebrosidase and BSβG may only be able to deconjugate glucosides to a limited extent since in this model they can only contribute after intracellular uptake of the isoflavone glucosides by specific transporters.

The results from subsequent conjugation studies revealed that glucuronidation of these isoflavone aglycones, especially at the 7-hydroxyl moiety, is the main metabolic route for glucuronidation. This efficient conjugation explains the low bioavailability of the free aglycones as reported in several studies [46, 53, 54] and indicates that 7-O-glucuronides are the major metabolites to be expected in the systemic circulation, with 4′-O-glucuronides being the least important isoflavone glucuronide expected in the systemic circulation. The additional contribution of sulfotransferases to the phase II metabolism of the isoflavone aglycones (see Table 5) will contribute to the conjugation of the isoflavone aglycones with relatively lower catalytic efficiencies than the glucuronidation [55, 56]. In this aspect, it is also of interest to notice that in vivo upon prolonged circulation of the isoflavones also combined metabolites are formed including 7-glucuronide 4-sulfate metabolites that have been reported to be found as major metabolites in human plasma after administration of a traditional Japanese roasted soy product called Kinako that contains considerable amount of soy isoflavone aglycones [53, 57]. This confirms that the 7-position is the preferred site for glucuronidation. Relatively modest species differences in the plasma levels of glucuronides and sulfates for rats and humans are in line with literature data reporting the proportion of isoflavone glucuronide and sulfates conjugates in serum of women and female rats after consumption of dietary containing soy protein isolate [58]. Finally, comparison of the catalytic efficiencies for deconjugation and conjugation reveals that deconjugation is less efficient than conjugation corroborating that isoflavones in their unconjugated form are unlikely to enter the systemic circulation. This is in line with the relatively low percentage of isoflavones in plasma present in the aglycone form [46, 53, 58]. Altogether the data point at possible differences in characteristics for intestinal conversion of the major soy isoflavones between rat and human, especially with respect to their deconjugation.

Acknowledgment

This work was commissioned (project number S/320001/11) and financed by the National Institute for Public Health and the Environment (RIVM), the Netherlands and co-financed by Food and Consumer Product Safety Authority (VWA), the Netherlands. The corresponding author is also indebted to Sher-e-Bangla Agricultural University, Dhaka, Bangladesh, for giving opportunity to conduct his PhD research at Wageningen UR, the Netherlands, and to Marelle G. Boersma (Division of Toxicology, Wageningen University) for her kind efforts in partially performing the kinetic studies.

The authors have declared no conflict of interest.

Ancillary