Red clover (Trifolium pratense L.) is used as a source for isoflavone (IF) dietary supplements. In this study, we focused on the red clover IF irilone (IRI), because of its reported comparatively high bioavailability. Because the conjugative metabolism plays a key role in the elimination of IF, we investigated the species-specific differences and glucuronidation kinetics of IRI using different liver microsomes as well as the recombinant UDP-glucuronosyltransferases (UGTs) 1A1, 1A7, 1A8, 1A9, 1A10, and 2B15. Both possible monoglucuronides, the IRI-O-4′-monoglucuronide (IRI-G4′) and the IRI-O-5-monoglucuronide (IRI-G5), were detected. Human liver microsomes (HLM) as well as rat liver microsomes predominantly formed IRI-G5, whereas for porcine liver microsomes, IRI-G4′ prevailed. HLM showed an apparent Vmax value of 0.43 nmol/min · mg and an apparent Km value of 9.8 μM for the formation of IRI-G5 and a Vmax of 0.35 nmol/min · mg and a Km of 64.7 μM in the case of IRI-G4′. Formation of both glucuronides was best fit using the substrate inhibition equation. The glucuronidation of IRI by UGTs led to values for the intrinsic clearance varying between 4 and 100 ml/min · mg, with UGT1A7 showing the lowest and UGT1A10 the highest IRI conversion rate. The results indicate that IRI undergoes an efficient glucuronidation, presumably in the intestine and liver, following atypical kinetic profiles.
Introduction
Isoflavones (IF) belong to the group of phytoestrogens displaying either weak estrogenic or antiestrogenic properties. Some, but not all, epidemiological and clinical studies showed beneficial effects of IF on human health, such as maintaining bone density or alleviating menopausal symptoms in postmenopausal women (Jacobs et al., 2009; Liu et al., 2009). As a consequence, a growing market for dietary supplements, based on soy and red clover as the major sources of IF, has developed worldwide. However, in recent years, there have been increasing concerns regarding the safety of such IF preparations, because animal studies provided evidence of potential adverse health effects that include an increased risk of breast cancer under certain conditions (Helferich et al., 2008).
Because the biotransformation seems to play a crucial role in the understanding and interpretation of the physiological effects of these compounds (Larkin et al., 2008; Lampe, 2009), a large number of studies investigated the metabolic pathways of IF in vitro and in vivo. To our knowledge, all IF studied so far undergo, at least in part, a transformation by the colonic microbiota (Kelly et al., 1993; Heinonen et al., 2004; Ruefer et al., 2007). An exception is 5,4′-dihydroxy-6,7-methylenedioxy-IF (irilone or IRI), which is almost resistant to microbial degradation (Braune et al., 2010). This behavior might be seen as the major reason for the high bioavailability of IRI in comparison with other IF, as very recently shown in a pilot intervention study (Maul and Kulling, 2010).
IRI possesses a methylenedioxy group attached to the aromatic A-ring of the IF skeleton, which might affect its biological behavior (Fig. 1). For example, other phytochemicals bearing a methylenedioxy group, such as safrol and piperin, are known to interfere with the metabolism of drugs (Velpandian et al., 2001) and are described as inhibitors of cytochrome P450 enzymes (Ueng et al., 2005).
The conjugation of IF with activated glucuronic acid catalyzed by UDP-glucuronosyltransferases (UGTs) represents the major metabolic pathway for IF because >60% of the IF is renally excreted from the body in the monoglucuronide form (Adlercreutz et al., 1995). The aim of the present study was to characterize the UGT isoform-specific glucuronidation of IRI in detail. Liver microsomes of three different species (human, rat, and pig) were chosen as a common screening model. We hypothesized that the glucuronidation of each hydroxyl group of IRI is catalyzed by different UGTs (see Fig. 1). Therefore, human recombinant UGTs were applied in addition to the liver microsomes to investigate site-specific glucuronidation. Furthermore, the different expression levels of UGTs in various tissues permitted us to estimate in which tissue and to which extent glucuronidation might take place. To date, not much information about UGTs' specificity for IF has been published, and the few studies carried out so far indicate that a single isoform may convert two different IF to a totally different extent (Pritchett et al., 2008). For this study, we selected those UGTs that so far have been described to glucuronidate (iso)flavonoids (Lepine et al., 2004). For characterization of the glucuronidation patterns of IRI by various enzyme sources, incubations were carried out over a wide concentration range.
Materials and Methods
Chemicals and Enzymes.
IRI (9-hydroxy-7-(4-hydroxyphenyl)-[1,3]dioxolo[4,5-g]chromen-8-one, purity 98% according to high-performance liquid chromatography-diode array detector (HPLC-DAD) analysis) was purchased from LGC Promochem (Wesel, Germany). 4-(Trifluoromethyl) umbelliferone-glucuronide (MUG) and uridine-5′-diphosphate-β,d-glucuronic acid (UDPGA) were obtained from Sigma-Aldrich (Deisenhofen, Germany). Irisolone [7-(4-hydroxyphenyl)-9-methoxy-[1,3]dioxolo[4,5-g]chromen-8-one] was isolated from the rhizome of Iris germanica (Weleda AG, Schwäbisch Gmünd, Germany). 7-Hydroxy-4-(trifluoromethyl) coumarin and the human recombinant UGTs 1A1, 1A8, 1A9, and 2B15 were purchased from BD Biosciences (Heidelberg, Germany). Human recombinant UGTs 1A7 and 1A10 were obtained from PanVera Corp./Invitrogen (Karlsruhe, Germany). Pooled human male and female liver microsomes (HLM) were obtained from Advacell (Barcelona, Spain), and pooled liver microsomes from male Sprague-Dawley rats were prepared according to standard procedures as described by Lake (1987). Porcine liver microsomes (PLM) prepared from female pigs were provided by Prof. M. Metzler (Karlsruhe Institute of Technology, Karlsruhe, Germany). All chemicals were of the highest grade available.
Activity of Microsomes and Recombinant UGTs.
The catalytic activity of the human recombinant UGTs used was measured according to the UGT Batch Data Sheet provided by the manufacturer. 7-Hydroxy-4-(trifluoromethyl) coumarin was used as a reference substrate, which is metabolized to MUG by the UGT. MUG was quantified by HPLC-UV using an external calibration curve. To confirm the catalytic activity of the microsomes, an analogous procedure was chosen (results are summarized in Supplemental Table 1).
Glucuronidation Assay.
The glucuronidation assay was carried out as described elsewhere (Matern et al., 1994; Fisher et al., 2000) with slight modifications. HLM, rat liver microsomes (RLM), and PLM or human recombinant Supersomes were used. For a standard incubation, 0.075 mg of supersomal and 0.028 to 0.058 mg of microsomal proteins were mixed with 5 μg of alamethicin in 90 μl of 0.1 M potassium phosphate buffer (pH 7.4) and kept on ice for 10 min. MgCl2 (final concentration, 10 mM), the β-glucuronidase inhibitor d-saccharic acid 1,4-lactone (5 mM), and the substrate IRI dissolved in dimethyl sulfoxide (final dimethyl sulfoxide concentration, 1.0%) were added, and the mixture was preincubated at 37°C for 5 min. To initiate the reaction, UDPGA (final concentration, 4 mM) was added, resulting in a final volume of 200 μl. The incubation was carried out for an additional 90 min. The assays were terminated by adding 200 μl of 0.7 M glycine/HCl buffer (pH 1.2). Alamethicin and the glucuronidase inhibitor were also added to the UGT experiments to ensure comparable conditions in all incubations. All experiments were carried out in duplicate.
The incubation mixture was extracted with 2 × 800 μl of ethyl acetate, the extract was evaporated to dryness, and the residue was dissolved in methanol/water (80:20, v/v) and analyzed by HPLC-DAD to measure IRI and its glucuronides. For determining kinetic data, the substrate concentration for IRI ranged from 1 to 75 μM for the supersomal incubations and from 2.5 to 150 μM for the liver microsomal incubations.
HPLC-DAD-Mass Spectrometry Analysis.
HPLC separation of IRI and its metabolites was carried out on a Prontosil (120 × 3.0 mm i.d.; particle size, 3 μm; pore size, 120 Å) reversed-phase column (Bischoff, Leonberg, Germany). The solvent system consisted of 0.05% formic acid in water (solvent A) and acetonitrile (solvent B). The following gradient was applied with a flow rate of 0.6 ml/min: 0 to 50 min linear from 15 to 35% B, 50 to 60 min linear from 35 to 40%, and 60 to 63 min linear from 40 to 90% B, followed by a reconditioning step for 5 min. Twenty-five–microliter aliquots were used as standard injection volume. The analysis was performed on an HP 1100 series HPLC (Agilent Technologies, Waldbronn, Germany) equipped with an autoinjector, quaternary HPLC pump, column heater, DAD detector, and HP Chem Station for data collection and handling. The eluent was recorded with a DAD at 270 nm. Observed peaks were scanned between 190 and 400 nm. The HPLC was interfaced to an HP series 1100 mass selective detector equipped with an atmospheric pressure ionization electrospray chamber. The conditions for metabolite analysis in the negative electrospray ionization mode were as follows: capillary voltage, 3.5 kV; fragmentor voltage, 80 V; nebulizing pressure, 50 psi; drying gas temperature, 350°C; and drying gas flow, 12.5 l/min. Data were collected using the scan mode, and spectra were scanned over a mass range of m/z 50 to 650.
Quantification of IRI was based on external calibration of the DAD signal at 270 nm. The detection limit, defined as 3-fold baseline noise, was 3.4 pmol for IRI. Because of the lack of IRI monoglucuronides as reference substances for quantification, it was assumed that the monoglucuronides have molar extinction coefficients similar to that of their aglycone. This assumption was based on the observation that the sum of the peak areas of the IRI glucuronides and the remaining unconjugated IRI were comparable with the peak area for the IRI aglycone determined in the control incubation without the cofactor UDPGA. The areas obtained in both cases were almost identical. Furthermore, the recovery rate of IRI extracted from the control incubations with microsomal protein was almost identical to that of a matrix-free standard solution.
Kinetic Data Analysis.
Each data point represents the mean of duplicate measurements. The apparent kinetic parameters Km and Vmax were calculated from the untransformed data by least-squares regression using GraFit 7.0.0 (Erithacus Software, Horley, Surrey, UK). Data were fitted to the equations of the following kinetic models:
The Michaelis-Menten equation where v is the rate of reaction, Vmax is the maximal velocity, Km is the Michaelis-Menten constant (substrate concentration at 0.5 Vmax), and S is the substrate concentration.
The Hill equation, which describes sigmoidal kinetics where S50 is the substrate concentration, resulting in 50% of Vmax in Hill kinetic profiles, and n is the Hill coefficient.
The substrate inhibition model equation where Ksi is the constant describing the substrate inhibition interaction.
The two-site model equation where Ks is the substrate dissociation constant and α and β are binding factors that reflect changes in Ks and product formation (Kp), respectively (Houston and Kenworthy, 2000).
For reactions exhibiting Michaelis-Menten and substrate-inhibited kinetics, intrinsic clearance (CLint) was calculated as Vmax/Km.
The goodness of the fit was determined by comparison of the statistical parameter χ2 between the models as well as the S.E.s of the various parameter estimates. The resulting data for the best fits are given in Table 1 as mean ± S.E. of fit.
Results
The in vitro glucuronidation of IRI was investigated using liver microsomes from three different mammalian species (human, rat, and pig) as well as six recombinant human UGTs. A representative HPLC chromatogram gave rise to two product peaks in addition to the IRI peak. The negative electrospray ionization mass spectrum of the IRI peak exhibited a quasimolecular ions [M − H]− at m/z 297, whereas the electrospray ionization mass spectra of both product peaks had [M − H]− ions at m/z of 473 corresponding to IRI-O-monoglucuronides (Fig. 2). In addition, the IRI aglycon fragment ion at m/z 297 was found in the MS spectra of both product peaks resulting from the loss of the glucuronic acid moiety (176 Da). IRI bears a hydroxyl function in position C-4′ as well as in position C-5, thus it can be concluded that both IRI glucuronides were formed (Fig. 1). No additional peaks, except those that were also present in the control incubation, could be detected. There was no indication for the formation of an IRI diglucuronide.
A screening revealed that IRI is conjugated by all tested UGTs to at least one glucuronide, with the exception of UGT2B15, which did not lead to product formation under the incubation conditions. At an IRI concentration of 25 μM, UGT1A1, UGT1A7, and UGT1A8 catalyzed the formation of both monoglucuronides; however, at a lower substrate concentration of 2.5 μM, only one conjugate was detected. The UGTs 1A9 and 1A10 catalyzed the formation of one single monoglucuronide each independently of the IRI concentration. UGT1A9 gave rise to the product with the shorter retention time, whereas UGT1A10 exclusively formed the product eluting several minutes later in the reversed-phase chromatography. This UGT selectivity was used for the structure elucidation of the two IRI monoglucuronides.
Structural Considerations.
A structural elucidation of the two IRI monoglucuronides was not feasible by NMR because of the low yield of glucuronide metabolites. To obtain the necessary structural information, glucuronidation of irisolone, the 5-methoxy-irilone derivative, with the UGTs 1A9 and 1A10 was carried out. In the case of irisolone, glucuronidation is possible only at the 4′-hydroxyl position (Fig. 1). As expected, only one of the two UGTs, namely 1A10, metabolized irisolone, and this IF was not converted by UGT1A9. In conclusion, only UGT1A10 catalyzed glucuronidation of irisolone in position C-4′. We assume that UGT1A10 also catalyzed the C-4′-glucuronidation of IRI because of the high structural similarity. Thus, UGT1A9 converting IRI, but not irisolone, should be responsible for the glucuronidation in position C-5, which is not accessible in the irisolone structure. In the following sections, the monoglucuronides will be labeled according to this assumption with IRI-O-5-monoglucuronide (IRI-G5) for the glucuronide conjugated in position C-5 and IRI-O-4′-monoglucuronide (IRI-G4′) for the C-4′ conjugate, respectively.
Kinetic Profiles of the IRI Glucuronidation Derived from Microsomal Experiments.
Incubation of IRI with each of the mammalian liver microsomes in the presence of UDPGA led to the formation of IRI-G5 and IRI-G4′. A significant difference in the pattern of the glucuronide formation could be observed, depending on the mammalian species as well as the substrate concentration. Remarkably, in all initial microsomal incubations in which only two substrate concentrations were used, the intensity of the IRI-G4′ formation increased more strongly with higher substrate concentrations compared to the IRI-G5 formation. Therefore, IRI glucuronidation activities of the three mammalian liver microsomes were determined over a concentration range from 2.5 to 150 μM IRI (135 μM for HLM).
In Fig. 3, the results are presented as plots of the glucuronidation activity versus the IRI concentration for the formation of the two glucuronides. As expected from the initial experiments, none of the plots showed typical Michaelis-Menten kinetics and thus did allow a satisfactorily display of the kinetic data using Lineweaver-Burk plots. In addition, results are presented as Eadie-Hofstee plots to examine the atypical kinetic behavior by direct visual inspection (Miners et al., 2010). The formation of IRI-G4′ and IRI-G5 by HLM as well as of IRI-G5 by RLM exhibited substrate-inhibited kinetics (Fig. 4, A–C). In contrast, the glucuronidation of IRI in position C-4′ catalyzed by RLM showed sigmoidal or autoactivation kinetics (Fig. 4D). Formation of IRI-G5 by PLM showed a Michaelis-Menten kinetic profile with slight substrate inhibition characteristics (Fig. 4E). Unfortunately, the formation of the dominating product IRI-G4′ by PLM led to an Eadie-Hofstee plot that cannot be evaluated (Fig. 4F). In summary, once the three mammalian species were compared, significant differences were observed.
In addition to the glucuronidation kinetics, the activities for the formation of the two glucuronides also varied among the different species. Although HLM and RLM predominantly formed IRI-G5, PLM catalyzed the formation of IRI-G4′ to a much higher extent. This distinction is reflected in the kinetic parameters obtained by fitting of the untransformed data to the equations of the four kinetic models. The calculated kinetic parameters for the model that provided the best fit according to least-squares regression analysis are summarized in Table 1.
IRI-G5 formation followed substrate-inhibited kinetics for the microsomes from all three species. The Vmax values were 1.37 nmol/min · mg for RLM, 0.43 nmol/min · mg for HLM, and 0.28 nmol/min · mg for PLM. In contrast, the Km value was higher for the PLM (Km = 16.8 μM) than for the HLM (Km = 9.8 μM) but still lower than the one for RLM (Km = 34.0 μM).
The substrate inhibition decreased in the row RLM (Ksi = 65.3 μM), which were strongly inhibited, followed by HLM (Ksi = 96.9 μM), which were less inhibited, and PLM, which showed almost no inhibition (Ksi = 183.4 μM). However, for the formation of IRI-G5 catalyzed by RLM, the fitting to the two-site model equation resulted in a goodness of fit that was as good as the fitting according to the substrate inhibition equation. The following parameters were calculated: Ks, 91.5 ± 11.4 μM; Vmax, 3.46 ± 0.33 nmol/min · mg; α, 0.13 ± 0.06; and β, 0.06 ± 0.02.
Although HLM also catalyzed the IRI-G4′ formation in a substrate inhibition manner (Vmax, 0.34 nmol/min · mg; Km, 64.7 μM; and Ksi, 50.9 μM), the type of the kinetic profile could not be accurately estimated for RLM and PLM, based either on the fitting of the untransformed data or the Eadie-Hofstee plots (Fig. 4, D and F). For IRI-G4′ formation by RLM, the data were fitted to the Hill as well as to the substrate inhibition equation, resulting in a similar goodness of fit (χ2 of 0.00037 and 0.00016, respectively) in both cases. The additional fit, according to the equation of the two-site model, led to a slightly higher χ2 value. For IRI-G4′ formation catalyzed by PLM, all fits showed rather poor χ2 values with the best one being 0.0023 for the substrate-inhibited kinetics resulting in a very high Vmax value of 4.87 nmol/min · mg, a Km of 88.6 μM, and a Ksi of 49.1 μM. The fit to substrate-inhibited kinetics of IRI-G4′ formation by RLM resulted in the following parameters: Vmax, 1.33 nmol/min · mg; Km, 124.5 μM; and Ksi, 44.7 μM. An assumed autoactivated formation of IRI-G4′ by RLM showed a Vmax of 0.294 nmol/min · mg, an S50 value of 14.8 μM, and a degree of sigmoidity with n = 2.1. Thus, remarkable differences between the species were observed, especially regarding the Vmax values derived for the IRI-G4′ formation (Table 1).
Kinetic Profiles of the IRI Glucuronidation Derived from UGTs.
Six human recombinant UGTs (1A1, 1A7, 1A8, 1A9, 1A10, and 2B15) were selected and tested for their ability to glucuronidate IRI. This screening showed that IRI was a suitable substrate for all the tested enzymes, except for UGT2B15, which did not metabolize IRI. All tested members of the UGT1A subfamily metabolized IRI to at least one glucuronide. UGT1A10 exclusively catalyzed the formation of IRI-G4′, whereas UGT1A9 and 1A7 solely formed IRI-G5 independent of the IRI concentration. In case of UGT1A1 and 1A8, the IRI concentration clearly influenced the distribution pattern of the two possible glucuronide isomers: at a rather low concentration of 2.5 μM, only the IRI-5G formation was catalyzed, whereas at a 10-fold higher IRI concentration, IRI-G4′ was additionally formed.
For elucidating the activity of the UGT isoforms in a substrate concentration range from 1 to 75 μM, the kinetic profiles for the glucuronidation of IRI were assessed. For almost all UGTs, a Michaelis-Menten–type kinetic was observed for concentrations <20 μM. The enzymatic activity started to decrease at IRI concentrations >20 μM, indicating substrate-inhibited kinetic profiles (Fig. 5). Thus, the evaluation of the kinetic parameters was carried out by fitting, according to the different kinetic model equations described under Materials and Methods.
All UGTs showed similar substrate-inhibited kinetic profiles except for UGT1A9, which catalyzed the IRI-G5 formation with an almost Michaelis-Menten–like kinetics over the whole substrate concentration range. The apparent substrate inhibition constant Ksi was >500 μM and therefore almost negligible. The fitting of the untransformed data to the equation for the substrate-inhibited kinetics for the formation of IRI-G5 by UGTs 1A1, 1A7, and 1A8 led to widely varying kinetic parameters. The highest Vmax was observed for UGT1A8, which was almost 30-fold higher than the value obtained for UGT1A7. Although UGT1A8 led to the formation of both glucuronides, IRI-G5 was dominantly formed, and Vmax was 2-fold higher compared with the IRI-G4′ formation.
UGT1A10, which exclusively formed IRI-G4′, did not allow a good fit for any of the kinetic models considered. However, the best fit expressed by the lowest χ2 was obtained for the substrate-inhibited kinetics. Table 1 summarizes the kinetic parameters obtained by fitting of the various glucuronidation data.
Because of the weak product formation, the conjugation in position C-4′ mediated by UGT1A1 could not be fitted satisfactorily to any of the equations. Hence, these data were excluded from the kinetic analysis.
Instrinsic Clearance.
The values for the CLint represent the sum of the values for both IRI glucuronides. In almost all cases, the goodness of fit for substrate inhibition was at least gradually better than for the Hill kinetics. Therefore, only CLint values derived from the substrate inhibition fitting model are given. With the recombinant enzymes, IRI clearance by UGT1A10 exhibited the highest value (99.9 ± 46.5 ml/min · mg protein). CLint of the other enzymes decreased in the following order: UGT1A9, 53.3 ± 16.6 ml/mg · min; UGT1A1, 27.2 ± 8.9 ml/mg · min; UGT1A8, 20.2 ± 3.9 ml/mg · min; and UGT1A7, 3.9 ± 1.2 ml/mg · min. Thus, the catalytic efficacy of UGT1A10 was 20-fold higher than the one measured for 1A7. The clearances found for the three mammalian microsomes were all in a comparable range with a CLint 71.8 ± 24.4 ml/mg · min for PLM, 51.0 ± 17.4 ml/mg · min for RLM, and 48.5 ± 10.4 ml/mg · min for HLM.
Discussion
The formation of the two possible monoglucuronides, the C-4′ glucuronide and the C-5 glucuronide, was documented by HPLC-MS and tentatively assigned to the two new product peaks observed. This assignment was made based on the assumption that UGTs 1A9 and 1A10 show the same regioselectivity for IRI than for its monomethylated derivative irisolone.
All liver microsomes used in the study as well as the UGTs 1A1 and 1A8 catalyzed the formation of both glucuronides, whereas the UGTs 1A7, 1A9, and 1A10 only gave rise to one site-selective product under the assay conditions used in this study. UGT2B15 did not convert IRI to any glucuronide. The UGTs 1A7 and 1A9 led to the formation of the C-5 conjugate, whereas UGT1A10 exclusively formed the C-4′ glucuronide. Because only a set of the six most important UGTs was studied, it cannot be ruled out that further UGTs are able to glucuronidate IRI. In particular, the UGTs 1A3 and 2B7 should be investigated in the future because they are known to metabolize estradiol (Lepine et al., 2004).
The microsomal glucuronidation of IRI in position C-5 was the dominating reaction for HLM and RLM, whereas the glucuronidation in position C-4′ prevailed for PLM. Although three of the tested recombinant human UGTs (1A1, 1A8, and 1A10) gave rise to the metabolite conjugated in position C-4′, very low amounts of IRI-G4′ were detected after incubation with HLM. This result may be explained by the fact that UGT1A10, as the main IRI-G4′-producing enzyme, represents an extrahepatic isoform not being expressed in human liver tissue (Tukey and Strassburg, 2000). The predominant occurrence of IRI-G5 correlates well with the fact that UGTs 1A9 and 1A1 are commonly expressed in the human liver (Tukey and Strassburg, 2000), either exclusively (UGT1A9) or at least predominantly (UGT1A1) forming the C-5 glucuronide. One might conclude that in vivo, the C-4′ glucuronidation occurs in considerable amounts in the intestine, where UGT1A10 is predominantly expressed.
The regiospecificity of the IRI glucuronidation by UGTs is supported by the findings of Joseph et al. (2007) who examined the glucuronidation of the red clover IF prunetin (4′,5-dihydroxy-7-methoxy-isoflavone). As observed in our study for IRI, the UGTs 1A10, 1A8, and 1A1 (in the order of decreasing activity) were also mainly responsible for the 4′-glucuronidation of prunetin. The formation of the prunetin-5-O-glucuronide was catalyzed by UGTs 1A9, 1A8, 1A7, and 1A1, very similar to the results presented for IRI in this study. We assume that this specificity of the UGTs is characteristic for the glucuronidation of IF, bearing only two available hydroxyl functions in the positions C-4′ and C-5. However, this behavior is no longer observed as soon as an additional hydroxyl group is present in position C-7 of the IF skeleton as was demonstrated for genistein (4′,5,7-trihydroxy-isoflavone), where a dominant formation of the 4′- and 7-, but not of the 5-O-glucuronide was observed (Doerge et al., 2000).
Kinetic analysis of the glucuronidation of IRI, including the CLint estimates, allows to draw some conclusions regarding the in vivo situation. In our glucuronidation experiments, a tendency toward a substrate-inhibited kinetic profile was observed in almost all cases. However, some exceptions were observed: for the formation of IRI-G4′ by RLM but also by PLM, an autoactivated kinetic profile cannot be completely ruled out, based on the goodness of the fit of the untransformed data and of the Eadie-Hofstee plots. The kinetic parameters calculated for the glucuronidation by microsomes can vary tremendously (depending on the model chosen), which is not observed in case of single recombinant UGTs. This result can be explained by the fact that microsomes contain multiple forms of UGT enzymes.
Moreover, UGT1A9 catalyzed the formation of IRI-G5 in an almost Michaelis-Menten–type kinetic with a tendency toward substrate inhibition. UGTs 1A1 and 1A8, both forming two glucuronides, exerted a different kind of atypical kinetics: at low substrate concentrations, both isoforms did not catalyze an IRI-G4′ formation, whereas at >2.5 μM, in case of UGT1A8, and 7.5 μM, in case of UGT1A1, the C-4′ glucuronide could be detected. For UGT1A8, mathematical fitting gave evidence for substrate-inhibited kinetics with a rather strong Ksi and a weak Km for both glucuronides. A similar tendency was observed for UGT1A1. This atypical kinetic cannot explain the delayed formation of IRI-G4′. Thus, for those enzymes giving rise to both products, an interference of the formation of IRI-G4′ caused by the conversion of IRI to IRI-G5 must be considered. One possible explanation may be a higher binding affinity for a particular binding pose of the IRI molecule into the active site of the specific UGT, which results in the formation of IRI-G5. IRI-G4′ is formed only at substrate concentrations close to saturation (or Vmax) of the stronger binding site.
Furthermore, there is compelling evidence indicating that UGTs are oligomeric enzymes (Finel and Kurkela, 2008). Thus, atypical kinetic profiles could also result from interactions between the single monomer units in the (hetero)-oligomer. This activity might also lead to a modulation of the IRI-G4′ formation caused by IRI-G5 formation in a second monomer of the complex. When the IRI concentration exceeds approximately 20 μM, substrate-inhibited kinetics could be observed in most cases. The IF concentration in blood plasma, after ingestion of usual doses of dietary supplements, does not reach this level (King and Bursill, 1998; Richelle et al., 2002). However, after intake of very high doses of IF, plasma levels higher than 15 μM were reported (Takimoto et al., 2003). The IF plasma concentrations, according to the manufacturers' recommended intake of red clover-based dietary supplements, are as high as 1 μM (Howes et al., 2002; Maul and Kulling, 2010), and, therefore, the formation of the C-5 glucuronide should be dominating in the liver. Formation of the IRI C-4′ glucuronide by the UGTs investigated in this study can hardly be expected in the liver. However, UGT1A10, which mainly catalyzed the IRI-G4′ formation, is predominantly located in the small intestine. Therefore, glucuronidation in the intestinal epithelium directly after absorption might be possible.
It is noteworthy that the PLM seems to exhibit a different UGT activity pattern in comparison with the HLM and the RLM, as demonstrated by a dominant C-4′ glucuronide formation. In contrast, the glucuronidation pattern obtained by incubations with RLM at least comes close to the results obtained for HLM. The finding that the glucuronidation pattern and kinetics are highly species dependent has already been described by Joseph et al. (2007), pointing out the differences among human, mouse, and rat liver microsomes.
In general, our data showed high values for the intrinsic clearance of IRI. Thus, it can be assumed that the compound will be rapidly conjugated after absorption. This is in agreement with previous results that show intense conjugative metabolism for the IF genistein and daidzein (Sfakianos et al., 1997; Doerge et al., 2000; Hendrich, 2002). Pritchett et al. (2008) investigated the conjugative metabolism of daidzein and genistein by a panel of HLM and found activities as high as 0.5 nmol/min · mg for UGT1A9 and 1A10 at 100 μM substrate concentration. Although kinetic parameters were not determined, the data for the highest conversion rates are very similar to our results for IRI. In contrast, Doerge et al. (2000) found an enzymatic activity for the conversion of the same IF, which was approximately one order of magnitude lower. Regardless of the substrate, it is generally difficult to compare enzyme kinetics when no standard reference substrate is included to determine the basic activity of the enzymes used. For example, Soars et al. (2003) and Fisher et al. (2000) both present data for the glucuronidation of 17β-estradiol, applying the same kinetic models but with rather deviating values for Vmax and Km/S50. This model emphasizes the need to determine the UGT activity, based on a commonly accepted reference substrate such as umbelliferone.
In conclusion, for the glucuronidation of IRI by almost all human UGTs, it can be stated that similar to prunetin but in contrast to genistein, the C-5 position appears to be the preferred site of conjugation. With increasing substrate concentrations, the C-4′ conjugation gains more importance. In all cases, the formation of both glucuronides can be best described according to substrate-inhibited kinetics. Because a strong glucuronidation activity of IRI by UGT1A10 was observed, the glucuronide conjugation should occur already in small intestine. Here, the formation of IRI-G4′ as the conjugation product of UGT1A10 seems to be feasible. The in vitro values obtained for the CLint of IRI by microsomes and UGTs in general are relatively high compared with other xenobiotics. Thus, a rapid conversion could be expected to occur in vivo.
Authorship Contributions
Participated in research design: Maul and Kulling.
Conducted experiments: Maul and Siegl.
Wrote or contributed to the writing of the manuscript: Maul and Kulling.
Acknowledgments
We thank the unknown reviewers for their helpful comments on the manuscript.
Footnotes
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.110.033076.
↵ The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material.
-
ABBREVIATIONS:
- IF
- isoflavone
- DAD
- diode array detector
- HLM
- human liver microsomes
- HPLC
- high-performance liquid chromatography
- IRI
- 5,4′-dihydroxy-6,7-methylenedioxy-IF
- IRI-G4′
- IRI-O-4′-monoglucuronide
- IRI-G5
- IRI-O-5-monoglucuronide
- HPLC-DAD
- high-performance liquid chromatography-diode array detector
- MUG
- 4-(trifluoromethyl) umbelliferone-glucuronide
- PLM
- porcine liver microsomes
- RLM
- rat liver microsomes
- UGT
- UDP glucuronosyltransferase
- UDPGA
- uridine-5′-diphosphate-β,d-glucuronic acid.
- Received March 2, 2010.
- Accepted December 21, 2010.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics