Abstract
Recent studies in our laboratory in the human hepatic and intestinal cell lines Hep G2 and Caco-2 have demonstrated induction of UGT1A1 by the flavonoid chrysin (5,7-dihydroxyflavone) using catalytic activity assays and Western and Northern blotting. In the present study, we examined which features of the flavonoid structures were associated with induction of UGT1A1 and whether common drug-metabolizing enzyme inducers also produce this induction. We also determined whether flavonoid treatment affected sulfate conjugation and CYP1A1 activity. We used intact Hep G2 cells for these studies, with chrysin as the model substrate. Both glucuronidation and sulfation were measured. Hep G2 cells were pretreated for 3 days with 25 μM concentrations of 22 flavonoids (n = 4–12). Only four flavonoids demonstrated induction of glucuronidation similar to that of chrysin (i.e., 3–5-fold in the intact cells). These were acacetin, apigenin, luteolin, and diosmetin, all of which, like chrysin, are 5,7-dihydroxyflavones with varying substituents in the B-ring. 5-Hydroxy-7-methoxyflavone and 5-methyl-7-hydroxyflavone produced a modest 1.5 to 2-fold induction, whereas all other flavonoids examined were without effect. None of the flavonoids caused more than a modest change in sulfation activity (60–140% of control). In contrast, all tested 5,7-dihydroxyflavones and -flavonols induced CYP1A1 activity (ethoxyresorufin deethylation). Of seven common drug-metabolizing enzyme inducers only 3-methylcholanthrene and oltipraz showed modest induction of chrysin glucuronidation but not 2,3,7,8-tetrachlorodibenzo-p-dioxin or phenobarbital. Together, these results strongly suggest that the flavonoid induction of UGT1A1 is through a novel nonaryl hydrocarbon receptor-mediated mechanism.
Glucuronic acid conjugation is a main route of elimination for many drugs and other xenobiotics, such as carcinogens and phytochemicals, in addition to endogenous substrates such as bilirubin, steroids, and bile acids (de Wildt et al., 1999; Radominska-Pandya et al., 1999; Tukey and Strassburg, 2000). This reaction is carried out by a number of isoforms in the two UDP-glucuronosyltransferase (UGT1) subfamilies UGT1A and UGT2B (Mackenzie et al., 1997; Radominska-Pandya et al., 1999; Tukey and Strassburg, 2000). Although induction of UGTs, interestingly, has a very low profile in review articles, several UGT isoforms have been shown to be inducible in human cell cultures. UGT1A6 and UGT1A9 have been shown to be regulated by aryl hydrocarbon receptor (AhR) agonists, such as β-naphthoflavone and 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) (Abid et al., 1995; Bock et al., 1999; Münzel et al., 1999), and UGT1A6, UGT1A9, and UGT2B7 by antioxidant type inducers, such ast-butylhydroquinone (Münzel et al., 1999) in Caco-2 cells. UGT1A1 has also been shown to be inducible mainly by 3-methylcholanthrene (3-MC) and to a small extent by phenobarbital and oltipraz in fresh human hepatocytes (Ritter et al., 1999). Recent studies in our laboratory of the human hepatic cell line Hep G2 and the human intestinal cell line Caco-2 have demonstrated a high level of induction of UGT1A1 in these cells by the flavonoid chrysin (5,7-dihydroxyflavone), using catalytic activity assays and Western and Northern blotting (Galijatovic et al., 2000, 2001; Walle et al., 2000). This induction response seemed quite specific because UGT1A6, UGT1A9, and UGT2B7 were not affected by chrysin treatment. UGT1A1 is the main isoform responsible for the glucuronidation of the endogenous toxin bilirubin (Ritter et al., 1999). It is also involved in the glucuronidation of a variety of other exogenous and endogenous compounds (Senafi et al., 1994; King et al., 1996), as well as carcinogens such as 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine, where it may play an important bioinactivating role (Galijatovic et al., 2001; Malfatti and Felton, 2001; Yueh et al., 2001).
The flavonoid chrysin, although present in honey (Siess et al., 1996) and marketed as an androgen-boosting supplement (Kao et al., 1998), is not one of the most abundant dietary flavonoids. Because of the presence of large amounts of structurally diverse flavonoids in fruits, vegetables, and plant-derived beverages, it was important to characterize the features in the flavonoid structure necessary to produce the UGT1A1 induction. The structures of the flavonoids examined in this study are shown in Table 1. The study was carried out in Hep G2 cells as the model system. For comparison, the effects of these flavonoids on the activities of sulfate conjugation and CYP1A1-mediated oxidation were also determined.
Materials and Methods
Chemicals.
Acacetin, apigenin, baicalein, biochanin A,tert-butylhydroquinone, (+)-catechin, chrysin, dexamethasone, diosmin, epigallocatechin gallate, galangin, genistein, hesperetin, 5-, 6-, and 7-hydroxyflavone, kaempferol, luteolin, 3-methylcholanthrene, β-naphthoflavone, naringenin, phenobarbital (sodium salt), quercetin, and d-saccharic acid 1,4-lactone were obtained from Sigma-Aldrich (St. Louis, MO). Diosmetin, 5-hydroxy-7-methoxyflavone, and 7-hydroxy-5-methylflavone were purchased from Indofine Chemical Co. (Somerville, NJ), and isorhamnetin was obtained from Extrasynthese (Geney, France). TCDD was obtained from the NCI Chemical Carcinogen Reference Standard Repository at the Midwest Research Institute (Kansas City, MO). Oltipraz was a gift from Aventis (Strasbourg, France).
Cell Culture and Treatment.
Hep G2 human hepatoma cells obtained from American Type Culture Collection (Rockville, MD) were maintained in Williams' Medium E with 10% fetal bovine serum, l-glutamine, and antibiotic/antimycotic solution in a humidified 37°C incubator with 5% carbon dioxide. When cells in six-well plates were about 90% confluent (4–6 days after seeding), they were treated with potential inducers in dimethyl sulfoxide (DMSO; 0.3% of final volume) or the same volume of DMSO for 3 days (Galijatovic et al., 2000), except TCDD (1 or 3 days). In all experiments, two control wells (DMSO) and two chrysin-treated wells (positive control) were included. All flavonoids were used at a concentration of 25 μM, the optimum concentration for UGT1A1 induction by chrysin in Hep G2 cells (Walle et al., 2000). The concentration dependence of induction was confirmed in experiments using 5, 10, 15, and 25 μM apigenin, chrysin, and luteolin. Maximum induction of both UGT and CYP1A activity was reached with 25 μM concentrations for each flavonoid. Other inducers were used at the concentrations used to induce other UGT isoforms or cytochrome P450 [i.e., 10 nM TCDD (Bock et al., 1999; Münzel et al., 1999), 1–2 μM 3-MC (Chung and Bresnick, 1994; Donato et al., 1995; Ritter et al., 1999; Runge et al., 2000), 2 mM phenobarbital (Doostdar et al., 1993; Donato et al., 1995; Ritter et al., 1999; Runge et al., 2000), 50 μM β-naphthoflavone (Abid et al., 1995; Runge et al., 2000), 1 μM dexamethasone (Doostdar et al., 1993; Donato et al., 1995), 50 μM oltipraz (Ritter et al., 1999), and 50 μMt-butylhydroquinone (Münzel et al., 1999)]. The medium was changed every 24 h, and the cells were used for in situ metabolism assays 24 h after the last medium change.
Catalytic Assays.
Our strategy for these assays in the intact Hep G2 cells was to first do the CYP1A1 oxidation of ethoxyresorufin by fluorometry followed by glucuronidation and sulfation with chrysin as the substrate.
CYP1A1 fluorometric assay.
After the 3-day incubation of the Hep G2 cells with flavonoid or vehicle (DMSO), the cells were washed once with medium and incubated with 0.6 μM ethoxyresorufin for 30 min in the presence of 1.5 mM salicylamide (Ciolino et al., 1998). The formation of resorufin was measured fluorometrically directly in the cell culture medium with excitation at 530 nm and emission at 590 nm. The results were adjusted for protein content as above.
Chrysin conjugation assay.
After the assay above, the medium containing ethoxyresorufin and salicylamide was replaced, and the cells were incubated for 6 h with 3 ml of medium containing 25 μM chrysin. This incubation time was chosen to give easily measurable chrysin glucuronide concentrations, with the main fraction of the parent compound still intact. Preliminary experiments showed that both glucuronidation and sulfation were linear with time from 0.5 to 6 h, both in control and chrysin-treated cells. Also, cellular β-glucuronidase did not seem to confound this assay, as d-saccharic acid 1,4-lactone (5 mM), an effective inhibitor of this enzyme (Thomasic, 1978), had no effect on either glucuronidation or sulfation in three separate experiments (data not shown). The medium was then collected, subjected to solid-phase extraction with Oasis cartridges (Waters, Milford, MA) as previously described (Galijatovic et al., 1999), and analyzed by HPLC. Quantitation of chrysin glucuronide and chrysin sulfate was based on peak areas compared with a standard curve obtained by injecting known amounts of chrysin. All data were adjusted for the amount of cellular protein in each well, as measured by the Lowry assay after digestion of the cells with 0.5 M NaOH (Lowry et al., 1951). The prior CYP1A1 assay conducted as above had no effect on the glucuronidation and sulfation of chrysin.
In experiments pretreating Hep G2 cells with galangin and chrysin as above, the flavonoid concentrations in the cells 24 h after the last treatment were analyzed by homogenization of the cells, methanol extraction of the homogenate, and HPLC. The residual concentrations for both flavonoids were found to be <0.5 μM. To determine whether these residual flavonoid concentrations in the Hep G2 cells could inhibit glucuronidation of chrysin (25 μM) in this assay, we used galangin (1–50 μM) as the inhibitor in an assay with recombinant UGT1A1 (25 μg of protein) (GENTEST, Woburn, MA) as the enzyme source (Walle et al., 2000). Our results demonstrated an IC50value of about 10 μM (n = 3).
Data Analysis.
The induction of conjugation activity by each compound was measured in two to six separate experiments with duplicate wells, except for chrysin (n = 40). The differences between treated and control cell activities in the same experiments were analyzed by unpaired Student's t tests with a significance level ofP < 0.05.
Results
In this study, we used the human hepatoma cell line Hep G2 to examine the regulation of glucuronidation and sulfation of the model compound chrysin by a total of 22 flavonoids (Table 1). Molecularly specific detection and quantitation of the two chrysin conjugates formed was done by reversed-phase HPLC following solid-phase extraction (Fig. 1A). Similarly to previous studies in Hep G2 (Walle et al., 2000) and Caco-2 cells (Galijatovic et al., 2000), we observed a 4-fold increase in chrysin glucuronidation after pretreating the cells with 25 μM chrysin for 3 days, with no effect on sulfate conjugation (Fig. 1B; Table2).
Of the total 21 additional flavonoids examined, only six gave rise to induction of glucuronidation. Four of these flavonoids (i.e., acacetin, apigenin, diosmetin, and luteolin) showed similar induction response to chrysin (i.e., 2.6- to 4.1-fold), whereas 5-hydroxy-7-methoxyflavone and 7-hydroxy-5-methylflavone showed a lower level of response (i.e., 1.7- to 2-fold) (Table 2). None of the flavanes, including epigallocatechin gallate (structure not shown), or isoflavones had any effect on the glucuronidation of chrysin. Only genistein showed a small but statistically significant increase in the sulfate conjugation of chrysin. In contrast, several flavonoids slightly decreased sulfate conjugation. One of these, 6-hydroxyflavone, also decreased the glucuronidation of chrysin.
As the assay for glucuronidation was done at 24 h after the last dose of flavonoid, it was important to determine that the residual intracellular flavonoid concentrations could not cause inhibition of chrysin glucuronidation. In experiments with galangin, which was not an inducer of glucuronidation, <0.5 μM remained in the cells at the time of the glucuronidation assay, considerably less than the IC50 of about 10 μM for the inhibition of chrysin glucuronidation by galangin. In addition, we did experiments with microsomes from galangin-treated versus control Hep G2 cells. Even in this preparation, where chrysin pretreatment showed a 14-fold induction of glucuronidation (Walle et al., 2000), galangin was without effect.
The effects of pretreatment with other well known drug-metabolizing enzyme inducers on the glucuronidation and sulfation of chrysin in intact Hep G2 cells are summarized in Table3. The concentrations used, noted together with literature references, have previously been shown to produce induction of various cytochromes P450 and other enzymes in cell culture. TCDD pretreatment was for 1 day versus 3 days for the other six compounds. Additional experiments with TCDD treatment for 3 days gave the same results. All observations of activities were compared with that of vehicle-treated (0.3% DMSO) cells. 3-MC and oltipraz both induced glucuronidation about 2-fold, whereas neithert-butylhydroquinone, dexamethasone, β-naphthoflavone, phenobarbital, nor TCDD had any significant effect. None of these compounds affected the sulfation of chrysin.
Because flavonoids have been shown to interact with CYP1A1 expression in various ways (Tsyrlov et al., 1994; Ciolino et al., 1999; Ashida et al., 2000), we also examined the effect of some of these compounds on the CYP1A1 activity by measuring the ethoxyresorufin deethylation (EROD) (Fig. 2B) in the same cells in which glucuronidation and sulfation of chrysin were determined (Fig.2A). Pretreatment with the two flavonoids apigenin and chrysin, which showed clear UGT1A1 induction, resulted in an 11- to 15-fold increase in EROD activity. A very similar CYP1A1 induction response was seen for galangin and isorhamnetin, two flavonoids that had no effect on the UGT1A1 activity. All tested flavones and flavonols with hydroxyl substituents in the 5- and 7-positions showed at least a 3-fold induction of EROD activity, whereas the corresponding flavanes and isoflavones did not. For comparison, both 3-methylcholanthrene and TCDD increased the EROD activity by 15- to 23-fold, as expected.
Discussion
The finding that the flavonoid chrysin could induce UGT1A1 substantially in both liver (Walle et al., 2000) and colonic (Galijatovic et al., 2000, 2001) cells may have practical implications in prevention of disease. In the liver, elevated expression of UGT1A1 could facilitate the glucuronidation of bilirubin, thereby normalizing the circulating levels of this endogenous toxin in unconjugated hyperbilirubinemia (Ritter et al., 1999). In the colon, elevated expression of UGT1A1 could increase the glucuronidation of the colon carcinogenN-hydroxy-2-amino-1-methyl-6-phenylimidazo-[4,5-b]pyridine, thereby protecting this tissue from carcinogenesis (Malfatti and Felton, 2001). For these reasons, it was important to extend our findings with chrysin to flavonoid molecules in general.
We selected to use glucuronidation of chrysin as a sensitive measure of induced UGT1A1 activity in Hep G2 cells because of our previous investigations (Walle et al., 2000). A low level of chrysin glucuronidation in uninduced Hep G2 cells is probably due to UGT1A6, an isoform that 1) is clearly expressed in these cells (Western analysis), 2) is using chrysin as a substrate, as shown with recombinant enzyme, and 3) is not induced by flavonoids (Western analysis) (Walle et al., 2000). After pretreatment with chrysin and other flavonoids, the increased UGT activity is a direct reflection of the magnitude of UGT1A1 expression, as confirmed by both Western and Northern analyses (Walle et al., 2000). Chrysin is also a good substrate for UGT1A9, which is not, however, expressed or induced in Hep G2 or Caco-2 cells (Galijatovic et al., 2001).
Our observations in this study from the 22 flavonoids examined provided a detailed picture of the structural elements necessary for effective induction of UGT1A1. Maximum (3- to 4-fold) induction was obtained with very few of the compounds studied, all of which can be classified as flavones (Fig. 3). It is clear from these data that the two hydroxyl groups in the 5- and 7-positions of the A-ring are essential. With only one hydroxyl group in either the 5- or 7-position, no induction was obtained (Table 2). Also, when the 7-hydroxyl group was methylated or the 5-hydroxyl group was replaced with a methyl group, the degree of induction was diminished. The lack of induction by diosmin (diosmetin-7-rutinoside) was not surprising. The bulky sugar substituent in a critical position is also very polar, greatly diminishing the cellular uptake of this flavonoid.
In contrast, substitutions in the B-ring, whether involving one hydroxyl group (apigenin), two hydroxyl groups (luteolin), or methoxy and hydroxyl groups (acacetin and diosmetin), had no effect on the induction. This is in sharp contrast to the structural requirements for the best known biological property of the flavonoids (i.e., the antioxidant activity) in which hydroxyl substitutions in the B-ring are essential for activity (Rice-Evans, 2001; Yang et al., 2001). On the other hand, several structural features in the C-ring seem to be critically important. Thus, the addition of a hydroxyl group in the 3-position effectively abolished the induction potential (e.g., in galangin compared with chrysin in kaempferol compared with apigenin and in quercetin compared with luteolin). Also, saturation of the 2,3-double bond, such as in naringenin compared with apigenin, abolished the response. The role of the keto group in the 4-position could not be addressed with the flavonoids examined. For the isoflavonoid genistein, with the p-hydroxylated B-ring attached in the 3-position rather than in the 2-position of the C-ring, as in apigenin, there was no induction. Similarly, the isoflavonoid biochanin A was not an inducer of UGT1A1, although it has been shown to induce UGT2B15, a testosterone-glucuronidating isoform, in prostate cancer cells (Sun et al., 1998). False-negative results due to inhibition of chrysin glucuronidation by residual intracellular flavonoids from the pretreatment could be excluded because these levels were much lower than the IC50 (<0.5 μM versus 10 μM).
The approximate 4-fold maximal induction of chrysin glucuronidation observed in both the present study and previous studies (Galijatovic et al., 2000, 2001; Walle et al., 2000), using an intact cell assay, was as much as 14-fold when examining activity in cell homogenates or microsomes (Galijatovic et al., 2000; Walle et al., 2000). The reason for this difference is at present not known but emphasizes that our findings in this study may be underestimating the induction response. Microsomes from cells treated with galangin, a flavonoid that did not induce glucuronidation in the in situ assay, were indistinguishable from control microsomes.
The analytical approach used in this study also permitted determination of the effect of the flavonoids on sulfate conjugation of the substrate chrysin. Previous studies have indicated that this reaction is most efficiently catalyzed by SULT1A1 (P-PST) (Galijatovic et al., 1999), the major sulfotransferase isoform expressed in Hep G2 cells (Walle et al., 1994). Genistein was the only flavonoid in this study to induce sulfation of chrysin, but only modestly so. Because genistein is a well known selective inhibitor of tyrosine kinase (Kim et al., 1998), it is tempting to suggest that the induction may be related to this property. A number of flavonoids exerted a modest inhibitory effect on sulfotransferase activity. Previous studies have demonstrated that flavonoids can indeed be potent inhibitors of SULT1A1 (Walle et al., 1995; Eaton et al., 1996).
The mechanism by which the 5,7-dihydroxyflavones induce UGT1A1 has not been addressed. Because 3-MC, an AhR agonist and potent CYP1A1 inducer, has been shown to induce UGT1A1 in human hepatocytes (Ritter et al., 1999), it was of interest to determine the effect of flavonoids on the CYP1A1 activity in the Hep G2 cells. Most flavonoids examined in this study, including all 5,7-dihydroxyflavones, substantially induced CYP1A1 activity (Fig. 2B). This is somewhat surprising because most flavonoids have been indicated to be potent antagonists but weak agonists of the AhR (Ciolino et al., 1999; Ashida et al., 2000). Of greatest importance regarding UGT1A1 induction was the finding that two of the flavonoids with high induction of CYP1A1, galangin and isorhamnetin, had no effect on the UGT1A1 activity, suggesting that the inducing effect of UGT1A1 is not related to the AhR. Also, previous studies in Caco-2 cells had clearly demonstrated that AhR agonists induced UGT1A6, which is not inducible by the flavonoids (Walle et al., 2000; Galijatovic et al., 2001). Consistent with a previous study (Ritter et al., 1999), we observed induction of UGT1A1 by 3-MC but not by TCDD (Fig. 2B), suggesting that the effect of 3-MC on UGT1A1 may not be due to stimulation of the AhR.
Except for the modest ability of 3-MC and oltipraz to induce UGT1A1 in the Hep G2 cells, general drug-metabolizing enzyme inducers were ineffective. In an earlier study in Hep G2 cells, phenobarbital-induced UGT activity, as measured by bilirubin glucuronidation (Doostdar et al., 1993), the prototypical UGT1A1 substrate. However, dexamethasone, rifampicin, and 1,2-benzanthracene had the same effect on bilirubin glucuronidation, indicating a nonselective effect. In cultured human primary hepatocytes, UGT1A1 mRNA was marginally phenobarbital-responsive (Ritter et al., 1999), whereas UGT1A4, a second isoform capable of metabolizing bilirubin (Ritter et al., 1991), was more phenobarbital-responsive. A recent study has localized a phenobarbital distal enhancer sequence in the UGT1A1 gene, which was activated by the nuclear orphan receptor hCAR (Sugatani et al., 2001). However, hCAR was not expressed in the Hep G2 cells and was transfected into these cells. Because our studies used unmodified Hep G2 cells, this does not seem likely to be the mechanism of UGT1A1 induction by the flavonoids and suggests that the induction of UGT1A1 by the flavonoids follows a novel, distinct pattern of signaling events. Thus, the mechanism of induction of UGT1A1 by flavonoids should be an important part of future studies. Moreover, it will be critically important to establish the UGT1A1-inducing effect of the flavonoids in vivo. Our findings in the present study will allow the selection of the one with the greatest bioavailability from a small group of five flavonoids. Also, the information gained in the present study should help identify the diets with a particularly high content of UGT1A1-inducing flavonoids.
Acknowledgments
We thank Yoko Otake and Lori Grismore for expert technical assistance.
Footnotes
-
This work was supported by National Institutes of Health Grant GM55561 and by U.S. Department of Agriculture Cooperative State Research, Education, and Extension Service (CSREES) Grant 00-35200-9071. This study was presented in part at the 10th North American meeting of the International Society for the Study of Xenobiotics in Indianapolis, IN, October 24–28, 2000.
- Abbreviations used are::
- UGT
- UDP-glucuronosyltransferase
- AhR
- aryl hydrocarbon receptor
- TCDD
- 2,3,7,8-tetrachlorodibenzo-p-dioxin
- 3-MC
- 3-methylcholanthrene
- DMSO
- dimethyl sulfoxide
- HPLC
- high-performance liquid chromatography
- EROD
- ethoxyresorufin deethylase
- Received July 6, 2001.
- Accepted January 31, 2002.
- The American Society for Pharmacology and Experimental Therapeutics