Ethinyl estradiol (EE) is a synthetic oral contraceptive that is taken by millions of women and is prone to interactions with coadministered drugs (Shenfield, 1993). These interactions fall into one of two categories, involving agents that decrease circulating EE levels or those that increase EE levels. An example of the former class is rifampicin, which can induce EE metabolism, potentially leading to contraceptive failure (Bolt et al., 1977). Alternatively, impairment of EE metabolism can elevate plasma levels and increase the potential for hypertension and vascular disease (Ahluwalia et al., 1977; Stadel, 1981). Examples of compounds that have been documented to increase EE plasma levels include ascorbic acid, acetaminophen, and fluconazole (Back et al., 1981; Rogers et al., 1987a; Sinofsky and Pasquale, 1998).

EE is well absorbed, but its oral bioavailability is roughly 42% (Back et al., 1979). Data in the literature suggest that the lower systemic bioavailability of EE is due to first pass metabolism in both the intestinal tract and the liver (Back et al., 1982). Furthermore, these investigators have estimated that sulfation accounts for as much as 60% of EE first pass metabolism and that the intestine is twice as effective as the liver in sulfating EE.

Although SULT-catalyzed EE 3-O-sulfation has been evaluated in vitro using human hepatocytes, human intestinal tissue (Ussing chambers), subcellular fractions (e.g., intestinal and liver cytosol and 9000g supernatant fraction), and purified and recombinant SULTs, to our knowledge no attempt has been made to identify the SULT(s) involved in the metabolism of EE or to characterize the in vitro kinetics of 3-O-sulfation at low concentrations of EE (≤500 nM) (Rogers et al., 1987b; Pacifici and Back, 1988; Forbes-Bamforth and Coughtrie, 1994; Falany et al., 1995; Li et al., 1999). One study by Falany et al. (1995) examined the sulfation of EE in a recombinant SULT1E1 system and found that the substrate-velocity profile saturated at concentrations lower than 1 μM. However, the kinetic parameters for this reaction were not reported. Pacifici and Back (1988) determined that the K_m for EE-SULT in subcellular fractions was 7.2 μM. However, these investigators used a substrate concentration range of 3 to 50 μM, substantially higher than low nanomolar concentrations.

Examination of the EE substrate-velocity curve at nanomolar concentrations is critical since plasma concentrations of EE can range from 100 to 150 pg/ml, corresponding to ∼ 0.4 nM (Sinofsky and Pasquale, 1998). Although it is difficult to estimate intracellular concentrations from plasma values, it is likely that the relevant concentration in vivo will be in the low to subnanomolar range. During the course of our initial studies, we obtained data that implicated SULT1E1 and SULT1A3 as low K_m (≤20 nM) isoforms and concluded that these enzymes may play an important role in the first pass sulfation of EE. SULT1E1 is widely known as an “estrogen sulfotransferase;” however, the low K_m of SULT1A3 for EE-SULT was unexpected.

Materials and Methods

Reagents. [³H]EE (49.1 Ci/mmol), [³H]dopamine (60 Ci/mmol), and [³H]DHEA (60 Ci/mmol) were obtained from PerkinElmer Life and Analytical Sciences (Boston, MA). Insect cell (Sf9) cytosol containing cDNA-expressed human SULT1A1^*2, 1A2, 1A3, 1E1, and 2A1, as well as control cytosol (devoid of human SULTs), was purchased from PanVera Corp. (Madison, WI). Pooled (10 organ donors) tissue cytosols from liver and intestine were obtained from Tissue Transformation Technologies (Edison, NJ). SULT activity in the different cytosol preparations was measured using 4-nitrophenol (SULT1A1), 17β-estradiol (SULT1E1), dopamine (SULT1A3), and DHEA (SULT2A1; see below). Quercetin, estrone, 4-nitophenylsulfate, EE, and PAPS were purchased from Aldrich Chemical Co. (Milwaukee, WI). All other reagents were purchased from vendors at the best obtainable grade. EE 3-O-sulfate conjugate standard was obtained from Steraloids (Newport, RI).

Cloning and Expression of SULT1A1^*1. SULT1A1^*1 was amplified from human cDNA (BD Biosciences Clontech, Palo Alto, CA) by PCR using forward and reverse primers (5′→3′) 5′-GATGAATTCCACCATGGAGCTGATCCAGGACAC-3′ and 5′-ATCGAATTCACAGCTCAGAGCGGAACG-3, respectively. The gel-purified PCR product was ligated into pCR2.1 (Invitrogen, Carlsbad, CA). The complete coding sequence was recovered after digestion with EcoRI, and was ligated into pBlueBac4.5 (Invitrogen) and further subcloned into the pFASTBAC1 vector (Invitrogen). The plasmid was analyzed by sequence analysis using an Applied Biosystems 310 automated sequencer (Applied Biosystems, Foster City, CA). The sequence was identical to that deposited in the National Center for Biotechnology Information database under accession number NM_001055.

Recombinant human SULT1A1^*1 baculovirus was generated using the Bac-to-Bac expression system (Invitrogen). Briefly, DH10BAC cells were transformed with pFASTBAC1-SULT1A1^*1 plasmid to generate SULT1A1^*1 bacmid. Generation of recombinant bacmid was confirmed by PCR amplification. Sf9 cells were transfected with SULT1A1^*1 recombinant bacmid and the resulting recombinant virus was amplified after two consecutive rounds of infection. Cytosol isolated from Sf9 cells infected with recombinant SULT1A1^*1 was evaluated for expression by immunoblot detection using human SULT1A1 polyclonal antibody (PanVera Corp.).

Incubations. Stock solutions of [³H]EE (ethanol), potassium phosphate buffer (50 mM, pH 7.4), MgCl₂ (5 mM), and enzyme were mixed to give a final incubation volume of 0.45 ml. The amount of enzyme added was adjusted so that no more than 10% substrate depletion occurred over the course of a typical incubation (2–12 μg of protein/ml for intestinal and hepatic cytosols; 1.5–8 μg of protein/ml for cytosol containing cDNA-expressed SULT). The mixture was preincubated for 5 min in a shaking water bath at 37°C. Reactions were initiated with the addition of PAPS (50 μl of 200 μM solution) and continued for 10 to 40 min under linear conditions. Inhibitors were dissolved in ethanol and the final volume of solvent in the incubation mixture did not exceed 0.5% (v/v). The final concentration range of EE used in incubations with SULTs 1A1^*1 (0.200–4 μM), 1A1^*2 (0.2–4 μM), 1A2 (1–10 μM), 1A3 (0.5–100 nM), 1E1 (0.5–100 nM), and 2A1 (50–1000 nM) was adjusted to yield optimal substrate-velocity data sets for each isoform. The concentration range of EE for intestinal and hepatic incubations was 1 to 2000 nM. No significant substrate inhibition was observed over the concentrations of EE tested.

All reactions were terminated by the addition of 100 μl of acetonitrile, followed by vortexing and centrifugation. The resultant supernatant was transferred into analysis vials and analyzed by HPLC without further workup. The identification of the 3-O-sulfate of EE was based both on mass spectrometry analysis and the comparison of HPLC retention time with retention time of an authentic standard.

In the expressed system hepatic and jejunum cytosols, control incubations (5–15 min) were performed with 4-nitrophenol (4 μM), dopamine (50 μM), 17β-estradiol (50 nM), and DHEA (5 μM). The incubations and analysis were performed as previously reported (Ma et al., 2003).

HPLC Analysis. HPLC analyses were conducted on a Hewlett Packard HP1050 or 1100 gradient system (Hewlett Packard, Palo Alto, CA). Separation and quantification of [³H]EE 3-O-sulfate was achieved on a reverse phase C18 column (4.6 × 250 mm or 4.6 × 150, 5 μm) using a mobile phase consisting of A, 25 mM ammonium formate (pH 3); and B, 0.1% formic acid in acetonitrile (constant flow rate of 1.0 ml/min). The following gradient was used: 0 min, 65% A; 8 min, 20% A; 10 min, 20% A; 11 min, 65% A; 16 min, 65% A. Alternatively, a second gradient was also used: 0 min, 65% A; 8 min, 20% A; 11 min; 65% A. In each case, the column was equilibrated at 65% A for at least 5 min before the next injection.

Radioactivity was quantitated postcolumn using a Radiomatic Flo-One Model A-200 detector (Radiomatic Instruments, Tampa, FL). Flowscint II scintillation cocktail (PerkinElmer Life and Analytical Sciences) was utilized postcolumn at a rate of 3.0 ml/min. The 3-O-sulfate of ethinyl estradiol was identified by monitoring radioactivity and by comparison of retention time with standard.

Data Analysis. Analysis of the data and curve-fitting (to obtain K_m and V_max values) was performed using nonlinear regression and the algorithms contained in the program Sigma Plot (SPSS Inc., Chicago, IL). Substrate-velocity data were analyzed assuming Michaelis-Menten kinetics. Equation 1 was used to estimate the contribution of high and low K_m components to total EE-SULT in cytosol preparations. IC₅₀ values were determined by fitting a curve to percentage remaining activity (relative to solvent control) using eq. 2. The term I_max in eq. 2 is equal to the percentage of maximal, saturable inhibition observed in IC₅₀ experiments (e.g., 95% inhibition, relative to control). The term s in eq. 2 is a modulator of curve shape.

Results and Discussion

Under linear reaction conditions, the 3-O-sulfation of EE (1–2000 nM) in both pooled human intestinal (jejunum) and liver cytosol conformed to a two-K_m Michaelis-Menten system that yielded biphasic Eadie-Hofstee plots (Fig. 1). The low K_m values for both intestine (8 nM) and liver (24 nM) were similar, implicating a high-affinity SULT(s) in both tissues (Table 1). The high K_m values were similar to those reported by Pacifici and Back (1988).

EE is extensively metabolized and is known to undergo appreciable first pass metabolism in both the intestine and liver. In particular, sulfation plays a major role and can account for as much as 60% of the first pass metabolism (Back et al., 1982). The remaining 40% is a composition of hydroxylation, methylation, and 3-O-glucuronidation. The extent of hydroxylation averages 25% but can be significantly higher in some cases (Bolt et al., 1973). EE has been reported to be oxidized by a number cytochromes P450 including CYP3A, CYP2C, and CYP2E (Guengerich, 1988; Ball et al., 1990). In addition, it has been shown that EE is a substrate for UGT1A1 (Ebner et al., 1993). Although any of these enzymes may be involved in metabolic pathways that potentially form the basis for drug-drug interactions, sulfation is the major pathway for first pass metabolism and is uniquely poised to modulate the bioavailability of EE.

Plasma levels of EE are low (∼0.5 nM), and it is likely that intracellular concentrations will also be in the same range. Therefore, the low K_m components of the biphasic substrate velocity curves noted above were the focus of our studies. Identification of the SULT isozyme(s) contributing to the lower concentrations in the EE substrate-velocity curve would determine the isoform(s) primarily responsible for first pass EE sulfation at clinically relevant concentrations.

Chen et al. (2003) have found evidence for four SULT isoforms in the intestine, SULT1A1, SULT1A3, SULT1E1, and SULT2A1. In addition to these isoforms, the liver has been shown to potentially contain significant contributions from SULT1B1 (4-nitrophenol sulfation; Tabrett and Coughtrie, 2003) and has also been shown to contain low levels of SULT1A2 expression (Ozawa et al., 1998). However, investigators have reported that expressed SULT1B1 did not sulfate a number of estrogenic compounds (Adjei and Weinshilboum, 2002), and the high K_m observed for SULT1A2-mediated EE sulfation (Table 1), coupled with low expression for this isoform in the liver, suggests that SULT1A2 will also not contribute significantly to EE sulfation. Finally, it has been reported that individuals homozygous for the SULT1A1^*2 allele have lower levels of SULT1A1 activity and that the reduced biological half-life of the SULT1A1^*2 protein is potentially responsible for this phenomenon (Raftogianis, 2001). As a consequence, the SULT isoforms studied in the present set of inhibition experiments were confined to SULT1A1^*1, SULT1A3, SULT1E1, and SULT2A1.

SULT1E1 is present in the intestine and liver and is widely known as an “estrogen sulfotransferase” (Falany et al., 1995; Honma et al., 2002). Therefore, it was expected that the low K_m component of the substrate-velocity curve would correspond to this isoform. Table 1 reports the K_m values for the four SULTs noted above, and as predicted, SULT1E1 was characterized with a low K_m for 3-O-sulfation of EE. Similar single-K_m substrate-velocity data were obtained with cDNA-expressed SULT1A3 and the other recombinant SULTs (Table 1). One surprising result was the low SULT1A3 K_m for EE sulfation. These results suggest that both SULT1E1 and SULT1A3 are high affinity isoforms for EE sulfation. It is therefore possible that either SULT (or both) could contribute significantly to EE sulfation during first pass metabolism. Thus, inhibition experiments were performed in human tissue cytosol to establish the relative contribution that each isoform, SULT1E1 and SULT1A3, would have toward EE sulfation.

Quercetin has previously been demonstrated to be a selective inhibitor of SULT1A1 with IC₅₀ values in the range of 100 nM, when using p-nitrophenol as a marker substrate (Walle et al., 1995). These investigators also noted that quercetin concentrations in excess of 1000 μM were required to inhibit estrone and EE sulfation (presumed markers of SULT1E1 activity). In the present set of studies, Fig. 2 illustrates that 1 μM quercetin selectively inhibits EE sulfation (at 3 nM) in the presence of cDNA-expressed SULT1A1 and SULT1E1. In contrast, the activities of SULT1A3 and SULT2A1 were unaffected, as described by Walle et al. (1995).

The apparent discord between the present data set and previously published results can be explained by the use of excessively high concentrations of substrate when measuring IC₅₀ values. An early report by Falany et al. (1995) noted that estrone sulfation by SULT1E1 saturated at approximately 20 nM. Initial experiments in our laboratory found that the K_m of estrone sulfation with expressed SULT1E1 was 6 nM (data not shown), similar to EE. Walle et al. (1995) reported using a concentration of 5 μM estrone, or roughly 1000 times the K_m for their IC₅₀ experiments with quercetin. Thus, the discrepancy between the current results and those published by Walle et al. (1995) can be rationalized by the use of a high concentration of substrate which “out-competes” quercetin for the active site. Moreover, it is probable that at high concentrations of estrone, multiple SULTs contribute to total sulfation.

Given that estrone has high affinity for SULT1E1, the inhibition of EE sulfation by estrone (40 nM) was examined, and it was found that SULT1A3 and SULT1E1 were inhibited, whereas SULT1A1 and SULT2A1 were unaffected (Fig. 2). Both quercetin and estrone were then used as inhibitors (1 μM and 40 nM, respectively) with human intestinal and liver cytosol (Table 2). Because either quercetin or estrone can potentially serve as substrate for sulfotransferases, the time interval for inhibition experiments was intentionally short (<10 min) to minimize depletion of “inhibitor.” Taken as a whole, quercetin/estrone inhibition data yield insight into the relative contribution of SULT1E1 and/or SULT1A3 to EE sulfation at low nanomolar concentrations.

Both quercetin and estrone inhibit roughly 75 to 80% of EE sulfation in all tissues examined (Table 2). Figure 2 demonstrates that 1 μM quercetin will selectively inhibit expressed SULT1A1 and SULT1E1. Thus, the inhibition of tissue EE sulfation by quercetin (1 μM) can be rationalized by inhibition of SULT1A1 and/or SULT1E1. This deduction assumes that inhibition results obtained with recombinant enzyme can be extrapolated directly to human SULT activity in human tissue cytosol. Additional inhibition experiments with DCNP (Table 3) show that both liver and intestinal cytosols yield IC₅₀ values consistent with SULT1E1, not SULT1A1. These data implicate SULT1E1 as the primary isoform involved in EE sulfation at nanomolar concentrations.

By analogy to the experiments outlined above, a similar set of studies was extended to estrone (40 nM), which selectively inhibits both cDNA-expressed SULT1E1 and SULT1A3 (Fig. 2). As noted above, when estrone inhibition was examined in both liver and intestinal cytosol, it was observed that EE sulfation was inhibited by approximately 75 to 80% (Table 2). This result was consistent with the involvement of SULT1E1, and not SULT1A3, because the contribution of SULT1A3 was ruled out by quercetin inhibition (vide supra). Moreover, the IC₅₀ of quercetin in intestinal cytosol was comparable to the value generated with expressed SULT1E1 and not SULT1A3 (Table 3). Finally, Fig. 3 shows the effect of estrone (40 nM) on an Eadie-Hofstee plot generated from the EE substrate-velocity data collected in intestinal cytosol. In the absence of estrone, the plot is clearly biphasic, with both high and low K_m components. In contrast, when estrone was included in the incubations, the high-affinity component of the curve was inhibited, resulting in a single-phase (high K_m) Eadie-Hofstee plot. Again, SULT1E1 is implicated as the high-affinity enzyme responsible for EE sulfation at nanomolar concentrations in tissue cytosol.

Based on the results presented here, it is concluded that SULT1E1 plays a predominant role in the 3-O-sulfation of EE in the intestine and liver. Overall, the factors that govern EE drug-drug interactions are likely to be complex and could include a classical mechanism of enzyme inhibition, enzyme induction, or depletion of required cofactor, such as PAPS. For example, it has been reported in hepatocytes that chrysin induces UGT1A1 and increases glucuronidation of EE 7-fold (Walle et al., 2000). In addition, the sulfation of 1 nM EE increased 1.5- to 3.3-fold in hepatocytes after treatment with rifampicin (Li et al., 1999). This last observation is supported by a recent study in which SULT1A1 and CYP3A4 gene expression was induced by rifampicin or phenobarbital (Maglich et al., 2002). The effect of rifampicin on SULT1E1 gene expression was not examined. Finally, it has been demonstrated that the administration of acetaminophen can dramatically attenuate levels of intracellular PAPS in rat liver hepatocytes (Sweeny and Reinke, 1988). This observation is consistent with the idea that concurrent administration of two sulfotransferase substrates can decrease intracellular PAPS and therefore limit the rate of sulfation of one or both compounds.

As noted, the complete metabolic scheme for EE includes potential contributions from UGT1A1, CYP2C9, CYP3A4, CYP2E1, and SULT1E1. Because the bioavailability of EE is governed by first pass metabolism, predominantly sulfation, it is therefore likely that intestinal and liver SULT1E1 may serve as a locus for drug-drug interactions. These interactions may be mediated by direct inhibition of SULT1E1, but could also include induction and cofactor (PAPS) depletion.