Abstract
We characterized the NADPH-dependent metabolism of 17β-estradiol (E2) by liver microsomes from 21 male and 12 female human subjects. A large number of radioactive estrogen metabolite peaks were detected following incubations of [3H]E2 with male or female human liver microsomes in the presence of NADPH. The structures of 18 hydroxylated or keto estrogen metabolites formed by these microsomes were identified by gas chromatography/mass spectrometry analysis. 2-Hydroxylation (the formation of 2-OH-E2 and 2-OH-E1) was the dominant metabolic pathway with all human liver microsomes tested. The average ratio of 4-OH-E2 to 2-OH-E2 formation was ∼1:6. A new monohydroxylated E2 metabolite (chemical structure unidentified) was found to be one of the major metabolites formed by human liver microsomes of both genders. 6β-OH-E2 and 16β-OH-E2 were also formed in significant quantities, but products of estrogen 16α-hydroxylation (16α-OH-E2 + 16α-OH-E1) were quantitatively minor metabolites. In addition, many other estrogen metabolites such as 6-keto-E2, 6α-OH-E2, 7α-OH-E2, 12β-OH-E2, 15α-OH-E2, 15β-OH-E2, 16β-OH-E1, and 16-keto-E2 were also formed in relatively small quantities. The overall profiles for the E2 metabolites formed by male and female human liver microsomes were similar, and their average rates were not significantly different. The activity of testosterone 6β-hydroxylation (a selective probe for CYP3A4/5 activity) strongly correlated with the rate of formation of 2-OH-E2, 4-OH-E2, and several other hydroxyestrogen metabolites by both male and female liver microsomes. The dominant role of hepatic CYP3A4 and CYP3A5 in the formation of these hydroxyestrogen metabolites was further confirmed by incubations of selectively expressed human CYP3A4 or CYP3A5 with [3H]E2 and NADPH.
Endogenous estrogens [such as 17β-estradiol (E2) and estrone (E1)] can be hydroxylated at multiple positions (as illustrated in Fig. 1) by drug-metabolizing enzymes present in liver as well as in extrahepatic organs (reviewed by Martucci and Fishman, 1993; Zhu and Conney, 1998a). Cytochrome P450 (CYP) family enzymes are the major enzymes that catalyze the NADPH-dependent oxidative metabolism of estrogens to various hydroxylated or keto metabolites (Martucci and Fishman, 1993;Zhu and Conney, 1998a). In most animals as well as in humans, the hepatic tissues contain the highest levels of total CYP-dependent drug-metabolizing enzymes and possibly also the largest numbers of different CYP isoforms. By using a versatile HPLC separation method coupled with radioactivity detection, investigators recently showed that incubations of radiolabeled E2 with rat or mouse liver microsomes (a crude preparation containing many different CYP isoforms) resulted in the formation of at least 15 estrogen metabolites (Suchar et al., 1995, 1996; Zhu et al., 1998).
Structures of E2 and estrone E1. The R is a β-OH group for E2 and is a keto group for E1. The possible positions for the NADPH-dependent α- or β-hydroxylation or keto formation catalyzed by CYP enzymes are indicated by arrowheads. The filled arrowheads indicate the C-positions where either α- or β-hydroxylation or keto formation may take place, whereas the unfilled arrowheads indicate the C-positions where only a hydroxylation (single configuration) may occur. For the C-18 position, either hydroxylation or aldehyde formation may occur.
Several earlier studies have also examined the NADPH-dependent metabolism of E2 and E1 by microsomal preparations from human liver (Ball et al., 1990; Kerlan et al., 1992; Shou et al., 1997; Huang et al., 1998; Yamazaki et al., 1998). In these studies, however, only a few hydroxylated estrogen metabolites (i.e., products of estrogen 2-, 4-, and 16α-hydroxylation) were determined. As discussed in a recent review article (Zhu and Conney, 1998a), a large number of hydroxylated or keto metabolites of E2 and E1are known to be present in the biological samples (e.g., tissues, blood, or urine) from animals or humans. For years, researchers have postulated that 4-OH-E2 and 16α-OH-E1 may play an important role in estrogen-induced hormonal carcinogenesis in animal models and also humans, and there is growing evidence in support of such a role for these bioactive estrogen metabolites, in particular the 4-hydroxylated estrogen metabolites (Fishman et al., 1984; Bradlow et al., 1986, 1995;Cavalieri et al., 2000; Liehr, 2000; Newbold and Liehr, 2000). Recently, it has been suggested that some of the other estrogen metabolites (such as 2-methoxyestradiol and 15α-OH-E2) may also have unique biological actions that are different from the parent hormones E2 and E1 (Zhu and Conney, 1998a,b).
As part of a continuing effort to identify the multiple endogenous estrogen metabolites formed by human tissues, we characterized in this study the NADPH-dependent metabolism of [3H]E2 to various hydroxylated or keto metabolites by male and female human liver microsomes. Since it is known that human CYP3A family enzymes account for up to 60% of the total hepatic CYP enzymes present (Shimada and Guengerich, 1989; Guengerich and Kim, 1990), we also studied the NADPH-dependent metabolism of [3H]E2 by selectively expressed human CYP3A4 and CYP3A5, and their profiles for E2 metabolism were compared with the metabolic profiles obtained with human liver microsomes. Detailed knowledge of the CYP-dependent formation of multiple estrogen metabolites, in particular the bioactive estrogen metabolites, in human liver as well as in extrahepatic target tissues may greatly enhance our understanding of the potential diverse biological actions of endogenous estrogens in the body.
Materials and Methods
Chemicals.
E2, NADPH, and ascorbic acid were purchased from the Sigma Chemical Co. (St. Louis, MO). 7β-OH-E2 was a generous gift from Dr. I. Yoshizawa of Hokkaido College of Pharmacy in Japan. The sources of 6β-OH-E1, 7α-OH-E2, 12β-OH-E2, 12-keto-E2, 14-OH-E1, 14-OH-E2, 15α-OH-E2, and 15β-OH-E2 were described in an earlier paper (Suchar et al., 1995). Several hydroxy-E1metabolites, including 6α-OH-E1, 7α-OH-E1, 7β-OH-E1, 12β-OH-E1, 15α-OH-E1, 15β-OH-E1, and 16β-OH-E1, were metabolically formed from their respective hydroxy-E2 metabolites by human liver microsomes in the presence of NAD+ as a cofactor. The products formed were extracted with ethyl acetate and then isolated by HPLC. The reference compounds for all the other estrogen metabolites used in this study were obtained from Steraloids, Inc. (Newport, RI). BSTFA containing 1% TMCS was obtained from Pierce Chemical Co. (Rockford, IL). All the organic solvents used in this study were of HPLC grade and obtained from Fisher Scientific (Atlanta, GA).
The radiolabeled E2 used in this study, [2,4,6,7,16,17-3H]E2(numerically labeled, specific radioactivity = 123.0 Ci/mmol), was purchased from the PerkinElmer Life Science Products (Boston, MA). There is no published information available on whether each of the designated positions was evenly labeled. A comparison of several tritium-labeled E2 products such as [6,7-3H]E2, [2,4,6,7,-3H]E2, and [2,4,6,7,16,17-3H]E2prepared by the same company showed that their highest specific activities (Ci/mmol) increased almost proportionally with increasing positions labeled with tritium, suggesting that each position likely was quite evenly labeled. In addition, earlier we conducted a comparative analysis by using 50 μM [2,4,6,7,16,17-3H]E2 or 50 μM [4-14C]E2 as substrate and the rat liver microsomes as the enzyme source. The profile of the multiple E2 metabolites formed and their average rates were found to be very similar with either [2,4,6,7,16,17-3H]E2 or [4-14C]E2 as substrate. This data suggested that using the [2,4,6,7,16,17-3H]E2 as substrate would not markedly skew the rates of formation for certain metabolites as a result of tritium loss during CYP-mediated hydroxylation of E2 at its radiolabeled positions.
Human Liver Microsomes and Selectively Expressed Human CYP3A4 and CYP3A5.
Liver microsomes of 33 human subjects (21 males and 12 females) were obtained from Human Biologics International (Scottsdale, AZ). The average age of the donors was 50 ± 14 years (mean ± S.D.). According to the supplier, the liver tissues were the autopsy samples from the donors, and the average cold ischemia time was 18.9 ± 12.4 h (mean ± S.D.). The main causes of death of these donors included head trauma, intracerebral bleeding, and/or intracranial hemorrhage. The protein content of each microsomal preparation was adjusted to 20 mg/ml by the supplier. The catalytic activities for several CYP isoforms in human liver microsomes (summarized in Table 1) were also determined by the supplier by analyzing the metabolism of selective probe substrates. Note that although the drug probes utilized are useful for estimating the levels of certain CYP isoforms in different liver microsomal preparations, the probes may not be entirely specific for a single CYP isoform.
Information on human liver microsomal preparations used in the study
The selectively expressed human CYP3A4 and CYP3A5 were purchased from GENTEST Co. (Woburn, MA). According to the supplier, these two human CYP isoforms were expressed in insect cells selectively transfected with a baculovirus expression system containing the cDNA for human CYP3A4 or CYP3A5. The total CYP content and the microsomal protein concentration were 2000 pmol/ml and 2.5 mg/ml, respectively, for CYP3A4, and were 2000 pmol/ml and 3.6 mg/ml, respectively, for CYP3A5. The catalytic activities for testosterone 6β-hydroxylation by the expressed CYP3A4 and CYP3A5 were 7.0 and 3.6 pmol of product formed/pmol of P450/min, respectively.
Assay of the NADPH-Dependent Metabolism of [3H]E2 by Human Liver Microsomes or Human CYP Isoforms.
The reaction mixture consisted of 1 mg/ml of human liver microsomal protein or 140 pmol/ml of human CYP3A4 or CYP3A5, a desired concentration of E2 (containing 2 μCi of [3H]E2), 2 mM NADPH, and 5 mM ascorbic acid in a final volume of 0.5 ml of Tris-HCl (0.1 M)/HEPES (0.05 M) buffer, pH 7.4. The presence of 5 mM ascorbic acid in the incubation mixture has previously been shown to protect catechol estrogen metabolites from oxidative degradation without significantly altering the enzyme activity (Hersey et al., 1981). The enzymatic reaction was initiated by addition of microsomal protein, and the incubations were carried out for 20 min at 37°C with mild shaking. The reaction was arrested by placing test tubes on ice followed by addition of 10 μl of 100 μM nonradiolabeled E2. The mixture was then immediately extracted with 8 ml of ethyl acetate, and the supernatants were transferred to another set of test tubes and dried under a stream of nitrogen. The resulting residues were redissolved in 100 μl of methanol, and an aliquot (50 μl) was injected into HPLC for analysis of estrogen metabolite composition.
Note that all the glass test tubes used in our study were silanized with 5% (v/v) dimethyldichlorosilane in toluene for 10 min followed by rinses in pure toluene twice and pure methanol three times. The test tubes were allowed to dry at room temperature and then thoroughly rinsed with distilled water. Our trial analyses of the NADPH-dependent [3H]E2 metabolism by human and rat liver microsomes using seven different types of unsilanized glass test tubes obtained from three different manufacturers showed that even under exactly the same incubation, extraction, and HPLC assay conditions, the results were very different for each of the hydroxylated [3H]E2metabolites detected. Based on measuring the radioactivity associated with [3H]2-OH-E2 and [3H]4-OH-E2 peaks, their overall recoveries with unsilanized test tubes were only 30 to 67% of the recoveries with the silanized test tubes. The increased recoveries of hydroxyestrogen metabolites with silanized glass tubes probably is because pretreatment of the glassware surface with dimethyldichlorosilane deactivates active chemical groups, thereby reducing physical adsorption of the hydroxylated estrogen metabolites to the test tubes.
HPLC Analysis of [3H]E2Metabolites.
Analysis of [3H]E2 metabolites was performed with an HPLC system coupled with in-line UV and radioactivity detection as described previously (Suchar et al., 1995). The HPLC system consisted of a Waters 2690 separation module, a Waters UV detector (model 484), an IN/US β-RAM radioactivity detector, and an Ultracarb 5 ODS column (150 × 4.60 mm, Phenomenex, Torrance, CA). The solvent system for separation of E2 and their metabolites consisted of acetonitrile (solvent A), 0.1% acetic acid in water (solvent B), and 0.1% acetic acid in methanol (solvent C). The solvent gradient (solvent A/solvent B/solvent C) used for eluting estrogen metabolites was as follows: 8 min of isocratic at 16:68:16, 7 min of a concave gradient (curve number 9) to 18:64:18, 13 min of a concave gradient (curve number 8) to 20:59:21, 10 min of a convex gradient (curve number 2) to 22:57:21, 13 min of a concave gradient (curve number 8) to 58:21:21, followed by a 0.1-min step to 92:5:3 and an 8.9-min isocratic period at 92:5:3. The gradient was then returned to the initial condition (16:68:16) and held for 10 min before analysis of the next sample.
The HPLC retention times for all authentic estrogen metabolites were determined by in-line UV detection, whereas the [3H]E2 metabolite peaks formed with human liver microsomes were determined by in-line radioactivity detection. The calculation of the amount of each estrogen metabolite formed was based on the amount of radioactivity detected for each corresponding metabolite peak.
Structural Identification of E2 Metabolites Formed by Human Liver Microsomes or Selectively Expressed Human CYP3A4 and CYP3A5.
The identity of each of the major E2metabolites formed by human liver microsomes was confirmed through comparison of its HPLC retention time, its GC/MS retention time, and its mass fragmentation spectrum with each of the 41 authentic reference compounds (listed in Table 2). For the purpose of comparison, the mass spectrum for each trimethylsilylated reference compound was obtained with our GC/MS system under the same analytical conditions.
Structural identification of E2 metabolites
For GC/MS analysis, the collected HPLC fractions were first evaporated to dryness under a stream of nitrogen gas and then incubated at 60–65°C for 30 min in the presence of 50 μl of BSTFA containing 1% TMCS. A Hewlett Packard 5890 gas chromatograph and a 5970 mass spectrometer (GC/MS) were used with an RTX-5 MS capillary column (0.25 mm × 30 m; 0.25-μm film thickness, Restek Corporation, Bellefonte, PA) with helium as the carrier gas. The mass spectrometer was operated in the electron impact mode (70 eV), and the mass abundance was determined by scanning masses from 50 to 600m/z at 1.4 times/s. The injector and detector temperatures were 260 and 280°C, respectively. During analysis, the column temperature was increased from 180–260°C at a rate of 4°C/min and then maintained isothermal at 260°C for the remainder of the run. The retention time and the mass spectrum for each of the major E2 metabolite peaks were compared against a library of 41 authentic standards compiled in this study. We used the built-in library search function of our GC/MS system for spectrum match-up between the E2 metabolites enzymatically formed and the known standards.
Results
Structural Identification of E2 Metabolites Formed by Human Liver Microsomes
A representative HPLC profile for the multiple [3H]E2 metabolite peaks detected after incubation of [3H]E2 with human liver microsomes and NADPH is shown in Fig. 2A. The identity of each of the major E2 metabolites formed was confirmed through comparison of its HPLC retention time, its GC/MS retention time, and its mass fragmentation spectrum with each of the 41 authentic reference compounds (results are summarized in Table2). For the purpose of comparison, the mass spectrum for each of the known standards was determined with our GC/MS system under the same analytical conditions.
A representative HPLC profile for the multiple [3H]E2 metabolites formed by human liver microsomes (A) and the mass fragmentation spectra (B–E) for the TMS derivatives of the authentic 16β-OH-E2 as well as the metabolically formed 16β-OH-E2,y-OH-E2, and y-OH-E1. The methods for the NADPH-dependent metabolism of E2 by human liver microsomes, for the HPLC separation of the estrogen metabolites, and for the GC/MS analysis of the trimethylsilylated estrogen metabolites were described under Materials and Methods. Peak 3 was designated as y-OH-E2 because our GC/MS data (D) indicated that this metabolite is a monohydroxy-E2, but its structure did not match any of the known standards. Peak 5 was suggested to bey-OH-E1 (the 17-dehydrogenated product ofy-OH-E2) based on the evidence discussed in the text. Peaks U1, U2, and U3 are the unidentified radioactive metabolite peaks formed from [3H]E2. Our GC/MS analysis indicated that they were not the monohydroxylated E2 or E1 metabolites.
Most of the identified estrogen metabolites matched closely with the HPLC retention time for the authentic reference compounds (Table 2). Also, the GC/MS retention time of the TMS-derivative of each isolated metabolite matched well with that of the authentic reference compound. The final definitive verification of each isolated metabolite was based on the match-up between the mass fragmentation spectrum of its TMS-derivative with that of the authentic standard, which was done by using the built-in mass spectrum library search function of the instrument alongside with manual comparison of their mass spectra.
In this study, we confirmed with high degrees of confidence the structural identities of the following E2metabolite peaks: E1, 2-OH-E2, 2-OH-E1, 4-OH-E2, 4-OH-E1, 6α-OH-E2, 6β-OH-E2, 6β-OH-E1, 6-keto-E2, 7α-OH-E2, 12β-OH-E2, 15α-OH-E2, 15β-OH-E2, 16α-OH-E2, 16β-OH-E2, 16α-OH-E1, 16β-OH-E1, and 16-keto-E2 (summarized in Table 2). As shown in Fig. 2, B and C, for example, a representative mass spectrum for the TMS-derivative of 16β-OH-E2 (an uncommon hydroxy-E2 metabolite formed by human liver microsomes) matched almost perfectly with the mass spectrum for the authentic reference compound.
The chemical structure for one of the quantitatively major metabolites (peak 3 in Fig. 2A) was not fully identified. To determine its structure, we collected this estrogen metabolite from HPLC and examined the mass spectrum of its TMS-derivative (shown in Fig. 2D) by GC/MS. The mass spectrum showed that its TMS-derivative has a molecular ion (m/z) of 504, suggesting that it is a monohydroxy-E2 metabolite (designated asy-OH-E2 for convenience). However, careful comparison of its HPLC and GC/MS retention times as well as its mass fragmentation spectrum did not show a consistent match with any of the 41 authentic estrogen standards listed in Table 2. Since the monohydroxy-E2 metabolites that were not studied here include only 1-OH-E2, 8-OH-E2, 9-OH-E2, 12α-OH-E2, and 18-OH-E2, it is thus likely that y-OH-E2 is one of them.
The GC/MS analysis of the radioactive metabolite peak 5showed that its TMS-derivative has a molecular ion (m/z) of 430 (Fig. 2E), suggesting that it is a monohydroxy-E1 metabolite. Furthermore, this compound was formed as one of the major metabolites when [3H]E1 was used as the substrate (data not shown), providing additional support for this metabolite as a monohydroxy-E1. However, its HPLC and GC/MS retention times as well as its mass spectrum did not consistently match any of the 41 authentic estrogen standards listed in Table 2. We noted that the ratio of this hydroxy-E1 metabolite toy-OH-E2 formed at different E2 substrate concentrations was almost the same as the ratio of 2-OH-E1 to 2-OH-E2. This finding led us to suggest that the metabolite peak 5 very likely isy-OH-E1.
For the radioactive peaks U1, U2, and U3 shown in Fig. 2A, the mass spectra of their TMS-derivatives did not match any of the 41 reference compounds, and no characteristic estrogen-related mass fragments (m/z of 504 or 414 for hydroxy-E2 metabolites; 430 or 340 for keto-E2 or hydroxy-E1metabolites) were found in their mass spectra. Also, small amounts of these metabolite peaks were formed nonenzymatically when 20 μM [3H]E2 was incubated with 2 mM NADPH in the absence of liver microsomes (data not shown). It is likely that these radioactive peaks may not be the hydroxylated or keto estrogen metabolites.
Optimization of Assay Conditions and pH Dependence
Extraction Efficiency and Reproducibility.
An average of 93.7 ± 1.4% (mean ± S.D.) of the total radioactivity was recovered when the incubation mixture was extracted once with 8 ml of ethyl acetate. Statistical analysis of triplicate determinations of several major [3H]E2metabolites (2-OH-E2, 2-OH-E1, 4-OH-E2,y-OH-E2, and E1) and the rate of overall [3H]E2 metabolism by three representative human male liver microsomes (HBI-101, HBI-102, and HBI-105) and three representative female liver microsomes (HBI-112, HBI-217, and HBI-226) showed highly consistent rates, with average intra-assay variations <5%.
Effect of Incubation Time and Microsomal Protein Concentration.
When 50 μM [3H]E2 was used as substrate, the rate of formation of 2-OH-E2, 2-OH-E1, 4-OH-E2,y-OH-E2, and E1, as well as the rate of overall E2 metabolism by a representative human liver microsomal preparation, was dependent on the incubation time (roughly linear up to 20 min; data not shown). The rate of formation of most hydroxy-E2 metabolites essentially remained constant after 20 min of incubation. However, the rate of formation of E1 and 2-OH-E1 from [3H]E2 increased almost linearly for at least 60 min.
When different concentrations (from 0.5–2.0 mg/ml) of microsomal protein were incubated with 50 μM [3H]E2 for 20 min, the rate of formation of most hydroxy-E2 metabolites decreased at >1.0 mg/ml of microsomal protein. However, the rate of formation of E1 and 2-OH-E1from [3H]E2 showed a linear increase up to the highest microsomal protein concentration (2.0 mg/ml) tested (data not shown).
Effect of Incubation pH.
We examined the effect of pH (from 4.0–10.0) on the NADPH-dependent metabolism of [3H]E2 by human liver microsomes. The formation of most E2 metabolites (such as 2-OH-E2, 4-OH-E2, and y-OH-E2) showed the highest velocity at pH 7 to 8, but the conversion of E2to E1 and the formation of some E1 metabolites such as 2-OH-E1 was optimal at a more basic pH (Fig.3).
Effect of the incubation pH on the formation of several [3H]E2 metabolites by HBI-115 human liver microsomes. The incubation mixtures consisted of human liver microsomes (1 mg/ml of microsomal protein), 20 μM E2(containing 2 μCi [3H]E2), 2 mM NADPH, and 5 mM ascorbic acid in a final volume of 0.5 ml of Tris-HCl/HEPES (0.1:0.05 M) buffer with incubation pH as indicated on thex-axis. The incubations were carried out for 20 min at 37°C, and the formation of estrogen metabolites was determined by HPLC analysis as described under Materials and Methods. Each point is the mean of duplicate determinations.
Effect of Different E2 Concentrations on Metabolite Formation
The effect of different [3H]E2 concentrations on the rate of metabolite formation by three representative human liver microsomes (HBI-112, HBI-115, and HBI-229) is shown in Figs.4 and 5. When the [3H]E2 concentration increased from 25 to 200 μM, the overall profile of radioactive E2 metabolites formed looked very similar, with 2-OH-E2 andy-OH-E2 as the quantitatively major hydroxyestrogen metabolites (representative HPLC traces for 25 and 200 μM [3H]E2 as substrate are shown in Fig. 4, middle and bottom panels). However, when a relatively low concentration (3.1 μM) of [3H]E2 was used as substrate, the overall estrogen metabolite profile looked very different, and 2-OH-E1 andy-OH-E1 were formed at higher rates than 2-OH-E2 andy-OH-E2, respectively (Fig. 4, top panel). These data suggest that the 17β-hydroxysteroid dehydrogenase contained in human liver microsomes is highly active during incubations with 3.1 μM E2. Double reciprocal analysis of the conversion of [3H]E2to E1 by three human liver microsomal preparations suggested an apparent KMvalue of 11.5 to 20.0 μM for this catalytic activity.
Representative HPLC profiles for the multiple estrogen metabolites formed by human liver microsomes at different [3H]E2 substrate concentrations. The incubation mixtures consisted of HBI-115 human liver microsomes (1 mg/ml of microsomal protein), [3H]E2, 2 mM NADPH, and 5 mM ascorbic acid in a final volume of 0.5 ml of Tris-HCl/HEPES (0.1:0.05 M) buffer, pH 7.4. The incubations were carried out for 20 min at 37°C, and the formation of estrogen metabolites was determined by HPLC analysis as described underMaterials and Methods. Although a different final concentration of E2 was in each test tube, the same amount of radioactive [3H]E2 (2 μCi) was present. The total radioactivities detected in the top, middle, and bottom HPLC profiles were 698,120, 727,751, and 707,805 cpm, respectively.
Effect of different substrate concentrations on the rate of NADPH-dependent metabolism of [3H]E2by human liver microsomes. The incubation mixtures consisted of human liver microsomes (0.5 mg of protein), a different concentration of E2 (containing 2 μCi of [3H]E2), 2 mM NADPH, and 5 mM ascorbic acid in a final volume of 0.5 ml of Tris-HCl/HEPES (0.1:0.05 M) buffer at pH 7.4. The final substrate concentration was 3.1, 6.3, 12.5, 25, 50, 100, or 200 μM. The incubations were carried out for 20 min at 37°C, and the formation of metabolites or overall E2 metabolism was determined by HPLC analysis as described under Materials and Methods. Each point is the mean of duplicate determinations.
For the three human liver microsomes tested, the rate of overall E2 metabolism increased continuously with increasing E2 concentrations (Fig. 5). The rate of conversion of [3H]E2to most oxidative metabolites by HBI-112 liver microsomes didn't show saturation at the highest substrate concentration (200 μM) tested. In comparison, the formation of 2-OH-E2, 4-OH-E2, and 16β-OH-E2 by HBI-115 and HBI-229 liver microsomes plateaued abruptly at >100 μM E2, while the formation of some other metabolites showed a linear increase up to the highest E2concentration tested.
Note that the ratios of y-OH-E1 toy-OH-E2 formed with these three liver microsomes at different E2 substrate concentrations are almost the same as the ratios of 2-OH-E1 to 2-OH-E2 (data not shown). This observation suggests that the same 17β-hydroxysteroid dehydrogenase present in human liver microsomes may catalyze both the conversion of 2-OH-E2 to 2-OH-E1 and the conversion ofy-OH-E2 toy-OH-E1.
NADPH-Dependent Metabolism of [3H]E2 by Male and Female Human Liver Microsomes
The results for the NADPH-dependent metabolism of 20 μM [3H]E2 by human male and female liver microsomes are summarized in Table3. The average rates of formation of several hydroxy-E2 metabolites by female liver microsomes were somewhat faster than the rates by male liver microsomes. However, these differences were not statistically significant. In both male and female liver microsomes, 2-OH-E2 was the major hydroxyestrogen metabolite detected, followed by y-OH-E2 and 2-OH-E1. The average rate of estrogen 2-hydroxylation (formation of 2-OH-E2 plus 2-OH-E1) by all 33 liver microsomes was 57.9 ± 32.5 pmol/mg of protein/min, and the average rate of estrogen 4-hydroxylation (the formation of 4-OH-E2 plus 4-OH-E1) was 10.9 ± 4.8 pmol/mg of protein/min. The average ratio of 4-OH-E2 to 2-OH-E2 was ∼1:6. The average rate ofy-OH-E2 formation was 12.4 ± 8.3 pmol/mg of protein/min. Several other hydroxylated metabolites such as 6β-OH-E2 and 16β-OH-E2were also formed in significant quantities. The combined rate of formation of 6β-OH-E1, 6-keto-E2, 16α-OH-E1, 16β-OH-E1, and 16-keto-E2(five coeluted metabolites) was only ∼7% of the rate of 2-OH-E2 formation (Table 3). Additional GC/MS analysis of the collected radioactive HPLC peak (from 23.7–25.5 min) corresponding to these five estrogen metabolites showed a ratio of approximately 2:1:4:4:1 for 6β-OH-E1, 6-keto-E2, 16α-OH-E1, 16β-OH-E1, and 16-keto-E2. The formation of 6α-OH-E2, 7α-OH-E2, 12β-OH-E2, 15α-OH-E2, and 15β-OH-E2 was also detected with our HPLC system, but their rate of formation was not precisely quantified because these metabolites were formed at very small quantities (<2.0 pmol/mg of protein/min). The formation of large amounts of E1 from E2 was also observed.
The rate of metabolic conversion of [3H]E2 to major estrogen metabolites catalyzed by male and female liver microsomes
Several nonpolar estrogen metabolites (collectively labeled asX) were consistently detected. Quantitatively, these nonpolar metabolites constitute a very significant fraction of the total amount of [3H]E2substrate metabolized (Table 3). The structures of these nonpolar estrogen metabolites were not identified.
In summary, the major hydroxyestrogen metabolites formed were 2-OH-E2,y-OH-E2, and 2-OH-E1 when [3H]E2 was incubated with either male or female human liver microsomes and NADPH. Several other E2 metabolites, including 4-OH-E2, 4-OH-E1, 6β-OH-E2, 16α-OH-E2(E3), and 16β-OH-E2, were formed in substantial quantities. Small amounts of 6α-OH-E2, 6β-OH-E1, 6-keto-E2, 7α-OH-E2, 12β-OH-E2, 15α-OH-E2, 15β-OH-E2, 16α-OH-E1, 16β-OH-E1, and 16-keto-E2were also formed. The overall profiles of E2metabolites formed by male or female human liver microsomes were not significantly different.
Correlation between the Activity of CYP Isoforms and the Rate of [3H]E2 Metabolite Formation
To probe which CYP isoform(s) may be responsible for E2 hydroxylation at specific positions, we determined in all 33 human liver microsomes the correlation coefficients between the rate of formation of each E2 metabolite and the rate of metabolism of the selective probe substrate by the targeted CYP isoform. These data are summarized in Table 4 and Fig.6.
Correlation coefficients (r) for comparing the formation of various E2 metabolites with the activities of CYP1A2, 2B6, 3A4, or 3A4/5 by measuring the metabolism of probe substrates in 33 human liver microsomes (21 male and 12 female)
Analysis of the correlation coefficiency between the formation of three major hydroxy-E2 metabolites (2-OH-E2, 4-OH-E2, and y-OH-E2) and the activities for hepatic CYP3A4/5, CYP3A4, and CYP2B6. The rate of formation of 2-OH-E2, 4-OH-E2, andy-OH-E2 was determined by incubation of human liver microsomes with 20 μM [3H]E2and 2 mM NADPH as described under Materials and Methods. The Pearson correlation coefficients (r) for liver microsomes from all 21 male subjects (M), from all 12 female subjects (F), and from all male and female subjects combined (T) were determined by using the SAS software (SAS Institute Inc., Cary, NC).
The total CYP content in human liver microsomes showed a high degree of correlation with the rate of formation of most hydroxyestrogen metabolites, suggesting that CYP-dependent enzymes constitute the major catalytic activity in human liver microsomes for the NADPH-dependent metabolism of E2. Among the CYP isoforms examined, the catalytic activity of CYP3A4 (according to dextromethorphan N-demethylation) or CYP3A4/5 (according to testosterone 6β-hydroxylation) showed high degrees of correlation with the rate of formation of several E2metabolites (Table 4; Fig. 6). In addition, the catalytic activity of CYP2B6 (according to S-mephenytoinN-demethylation) also showed a good correlation with the rate of formation of y-OH-E2, E3, and 2-OH-E2. Overall, liver microsomes from female human subjects showed a somewhat higher degree of correlation between the rate of formation of most E2 metabolites and the catalytic activity of the corresponding CYP isoform as compared with male liver microsomes.
NADPH-Dependent Metabolism of [3H]E2 by Human CYP3A4 and CYP3A5
Incubations of 20 or 50 μM [3H]E2 with selectively expressed human CYP3A4 or CYP3A5 in the presence of NADPH resulted in the formation of multiple hydroxylated estrogen metabolites. Figure7 shows the representative HPLC profiles for the radioactive estrogen metabolites formed by human CYP3A4 and CYP3A5 at a 20 μM [3H]E2 substrate concentration. The overall profiles formed with these two CYP isoforms were very similar. 2-OH-E2 was the quantitatively major hydroxy-E2 metabolite formed by either CYP3A4 or CYP3A5, followed by y-OH-E2and 4-OH-E2. Several other hydroxy-E2 metabolites (6β-OH-E2, 16α-OH-E2, and 16β-OH-E2) were also formed in substantial quantities by CYP3A4 and CYP3A5. In addition, large amounts of nonpolar metabolites (collectively labeled as X in Fig. 7) were formed during incubations of E2 with CYP3A4- or CYP3A5-expressed microsomes. The overall profiles for the hydroxylated and nonpolar E2 metabolites formed by CYP3A4 or CYP3A5 looked quite similar to those formed by male or female human liver microsomes (compare Fig. 7 with Fig. 2). In contrast, the CYP3A4 or CYP3A5-expressed microsomes contained much lower activity for the formation of E1 and several hydroxy-E1 metabolites when compared with human liver microsomes.
Representative HPLC profiles for the NADPH-dependent [3H]E2 metabolites formed by selectively expressed human CYP3A4 and CYP3A5. The incubation mixtures consisted of 140 pmol/ml CYP3A4 or CYP3A5, 20 μM E2 (containing 2 μCi of [3H]E2), 2 mM NADPH, and 5 mM ascorbic acid in a final volume of 0.5 ml of Tris-HCl/HEPES (0.1:0.05 M) buffer, pH 7.4. The incubations were carried out for 20 min at 37°C, and the formation of estrogen metabolites was determined by HPLC analysis as described under Materials and Methods. For the descriptions of the peaks labeledy-OH-E2, U1, U2, and U3, refer to the legend to Fig. 2.
Quantitatively, the rates for 2-OH-E2 formation (mean ± S.D. of triplicate determinations) by CYP3A4 and CYP3A5 were 276 ± 34 and 84 ± 19 pmol/nmol of P450/min, respectively, and the rates for 4-OH-E2 formation were 53 ± 9 and 43 ± 4 pmol/nmol of P450/min, respectively. Therefore, the ratio of 4-OH-E2 formation to 2-OH-E2 formation was ∼1:5 with CYP3A4, and it was ∼1:2 with CYP3A5.
The structural identities for the major hydroxy-E2 metabolites (2-OH-E2, 4-OH-E2, 6β-OH-E2, 16α-OH-E2, 16β-OH-E2, andy-OH-E2) formed by CYP3A4 and CYP3A5 were also confirmed by GC/MS analysis (data not presented).
In summary, multiple hydroxy-E2 metabolites were formed by selectively expressed human CYP3A4 and CYP3A5. The overall profiles for hydroxylated and nonpolar E2metabolites formed by human CYP3A4 and CYP3A5 were similar to the metabolites formed by either male or female human liver microsomes. It is of great interest to note that the ratio of 4-OH-E2 to 2-OH-E2formation by CYP3A5 (a polymorphic hepatic CYP isoform) was much higher than the ratio for CYP3A4.
Discussion
Liver Microsomal Hydroxylation of E2 at Various Positions.
In the present study, we characterized the NADPH-dependent metabolism of [3H]E2 by 33 human liver microsomal preparations. By using a versatile HPLC separation method coupled with radioactivity detection, we identified over 18 hydroxylated or keto metabolites that were formed with [3H]E2 as substrate. The structural identities of all major [3H]E2 metabolites formed by human liver microsomes were confirmed by our triple match-up approach according to their HPLC retention times, the GC/MS retention times of their TMS-derivatives, and the mass fragmentation spectra of their TMS-derivatives (data summarized in Table 2).
Earlier studies reported that 2-hydroxylation of E2 to a catechol is a major metabolic pathway in rodent and human livers, whereas 4-hydroxylation to a different catechol represents a quantitatively minor pathway (usually <15% of 2-hydroxylation) in this organ (Dannan et al., 1986; Kerlan et al., 1992; Zhu et al., 1993; Suchar et al., 1995). The data from our present study also showed that E2 2-hydroxylation is by far the major hydroxylation pathway in human liver microsomes, and the rate of E2 4-hydroxylation was 1:4 to 1:6 of that for E2 2-hydroxylation. The potential importance of 4-hydroxylated estrogen metabolites (4-OH-E2and 4-OH-E1) in hormonal carcinogenesis has recently received considerable attention (reviewed recently by Liehr, 2000). These catechol estrogens not only serve as intermediates for the generation of reactive chemical species (Liehr and Roy, 1990; Cavalieri et al., 2000; Liehr, 2000), but 4-OH-E2 may also have its own signal transduction pathway that is refractory to E2 (Das et al., 1997).
The NADPH-dependent 6α- and 6β-hydroxylations of E2 are known metabolic pathways in both animals and humans (Mueller and Rumney, 1957; Breuer et al., 1966). A partially purified CYP preparation from rat brain showed significant catalytic activity for E2 6α- and 6β-hydroxylation (Sugita et al., 1987), and larger amounts of 6α-OH-E2 were detected than 6β-OH-E2 based on thin-layer chromatographic analysis. Similarly, an earlier study showed that 6α- and 6β-OH-E1 were the major metabolites of E1 formed by porcine uterine endometrial tissues (Maschler et al., 1983), and 6α-OH-E2 was found to be a major metabolite of E2 formed by human placental microsomes (B. T. Zhu, M. X. Cai, and A. H. Conney, unpublished observations). The results of our present study showed that 6α- and 6β-OH-E2 were very minor metabolites formed by either male or female human liver microsomes. The formation of 6α-OH-E2 was quantitatively less than the formation of its β isomer.
It has been suggested for many years that the 16α-hydroxylation of E2 and E1 plays an important role in mammary carcinogenesis (Fishman et al., 1984; Bradlow et al., 1986, 1995). The 16α-hydroxylated estrogens (i.e., 16α-OH-E1 and 16α-OH-E2) are not only hormonally active, but 16α-OH-E1 is also chemically reactive and may bind covalently to the estrogen receptor possibly resulting in sustained hormonal stimulation (Swaneck and Fishman, 1988; Hsu et al., 1991). Because of the unique biological properties of 16α-hydroxylated estrogens, several earlier studies have examined the NADPH-dependent 16α-hydroxylation of E2 and E1 in human subjects (Osborne et al., 1993) or by human hepatic and extrahepatic tissues or cells in vitro (Aoyama et al., 1990; Shou et al., 1997; Huang et al., 1998). Human liver homogenates were reported to metabolically convert E2 to 16α-OH-E2 at a relatively fast rate. A recent study by Yamazaki et al. (1998)estimated that the average rate of 16α-OH-E2formation from E2 by human liver microsomes was ∼100 pmol/nmol of P450/min, which was much faster than the rate of 4-OH-E2 formation. However, the results of our present study showed that only very minute amounts of 16α-OH-E2 and 16α-OH-E1were formed by human liver microsomes when [3H]E2 was used as substrate (Fig. 2). In our study, the average rate of 16α-OH-E2 formation by 33 human liver microsomes was 3.0 ± 1.1 pmol/mg of protein/min (Table 3); the rate of 16α-OH-E1 formation was even lower (<2.0 pmol/mg of protein/min) and could not be precisely quantified. The combined average rate of 16α-OH-E2 and 16α-OH-E1 formation from [3H]E2 substrate was much less than that of 4-OH-E2 and 4-OH-E1 formation. In additional studies, we found that 16α-hydroxylation of E2 by human placental microsomes was also a very minor metabolic pathway (B. T. Zhu, M. X. Cai, and A. H. Conney, unpublished observations). It should be noted that the identification of the estrogen metabolites in several earlier studies was only based on comparison of their TLC or HPLC retention times against a very limited number of available estrogen standards. It is possible that misidentification of other hydroxyestrogen metabolites as 16α-OH-E2 might have contributed to the high rates reported in some earlier studies.
In addition to the formation of several well known hydroxy-E2 metabolites (2-, 4-, 6α-/6β-, and 16α-OH-E2), the formation of several uncommon E2 metabolites (6-keto-E2, 7α-OH-E2, 12β-OH-E2, 15α-/15β-OH-E2, 16β-OH-E2, 16-keto-E2, and y-OH-E2) were identified following incubations of E2 with NADPH and human male and female liver microsomes. We also found that E2y-hydroxylation is a quantitatively major hydroxylation pathway, and it is only secondary to 2-hydroxylation. Candidates for the identity of this metabolite include 1-, 8-, 9-, 12α-, and 18-OH-E2.
We noted that several hydroxy-E1 metabolites (e.g., 2-OH-E1, 6β-OH-E1, and 16β-OH-E1) were also formed from [3H]E2 as substrate. Overall, the rates of formation of these metabolites were proportional to the formation of their respective hydroxy-E2metabolites. This observation suggests that the hydroxy-E1 metabolites were probably formed by 17β-oxidation of their respective hydroxy-E2metabolites catalyzed by 17β-hydroxysteroid dehydrogenase present in human liver microsomes. Notably, when a low 3.1 μM concentration of [3H]E2 was incubated with human liver microsomes, 2-OH-E1 was formed at a much higher rate than 2-OH-E2 (Fig. 4). The amount of [3H]E1 formed from 3.1 μM E2 as substrate accounted for 45% of the total radioactivity detected, but this phenomenon was not observed at higher [3H]E2concentrations, suggesting that 17β-hydroxysteroid dehydrogenase present in human liver microsomes has a relatively high affinity but low capacity for the conversion of E2 to E1. In addition, our results also showed that large interindividual variations existed for this estrogen-converting enzyme activity.
Finally, it should be noted that large amounts of several unidentified nonpolar metabolites (collectively labeled as X in Fig. 2A) were consistently detected following incubation of [3H]E2 with each of the 33 human liver microsomal preparations. These nonpolar estrogen metabolites were also formed by selectively expressed human CYP3A4 and CYP3A5 (Fig. 7). However, several other selectively expressed human CYP isoforms did not show catalytic activity for the formation of nonpolar estrogen metabolites (data not shown), suggesting that their formation is enzymatically catalyzed.
Interindividual Variation and Effect of Gender on Liver Microsomal Metabolism of E2.
It is well documented that large interindividual variations exist in the activities of certain human hepatic xenobiotic monooxygenases and CYP enzymes (Conney, 1982;Forrester et al., 1992; Shimada et al., 1994; Transon et al., 1996). Such variations are attributable to the effects of both genetic and environmental factors (Conney, 1982; Guengerich and Shimada, 1991;Wrighton and Stevens, 1992). In agreement with several earlier observations, the total CYP content and catalytic activities for the metabolism of several marker substrates that selectively probe certain CYP isoforms in 33 human liver microsomes showed very large interindividual variations (Table 1). Similarly, the rates of formation of various E2 metabolites also showed wide variations (Table 3), and their rates correlated closely with the total CYP content and particularly the CYP3A activity (as measured by testosterone 6β-hydroxylation) present in human liver microsomes (Table 4; discussed later).
It is known that some CYP isoforms present in the liver or extrahepatic tissues of experimental animals are gender-specific (MacGeoch et al., 1984; Waxman, 1984; Waxman et al., 1985; Bandiera et al., 1986). An earlier study suggested the “male-selective/specific” 15α- and 16α-hydroxylations of [3H]E2 in rats through measuring biliary estrogen metabolites (Maggs et al., 1992). A more recent study showed that liver microsomes from male rats were much more active than female liver microsomes for E2 2-, 15α-, and 16α-hydroxylations (Suchar et al., 1995). However, analysis of 60 human liver samples (30 Caucasians and 30 Japanese) showed a lack of sex-dependent differences in the content of several major CYP isoforms and the activity of several xenobiotic-metabolizing enzymes (Shimada et al., 1994). Despite the large quantitative variations that existed in our samples, we found that all identified hydroxy/keto estrogen metabolites were consistently formed with almost all human liver microsomes tested, and the overall profiles of E2 metabolites formed by both male and female human liver microsomes were similar. Last, we also noted that the average rate of formation of major E2 metabolites by female human liver microsomes appeared to be slightly higher than the rate with male human liver microsomes, but this difference was not statistically significant.
Role of Human CYP3A4 and CYP3A5 in Hepatic E2Metabolism.
Isoforms of the CYP enzymes are the major catalysts for the NADPH-dependent oxidative metabolism of estrogens to various hydroxylated and keto metabolites in the body. The CYP3A family enzymes are the most abundant CYP isoforms present in human liver (Thummel and Wilkinson, 1998; reviewed by Guengerich, 1999). An earlier study suggested that CYP3A4 and CYP3A5 in human liver microsomes were responsible for up to 80% of estrogen 2- and 4-hydroxylase activities (Kerlan et al., 1992). Our data showed that testosterone 6β-hydroxylation activity (a selective marker reaction for CYP3A4/5) was strongly correlated with the formation of most hydroxylated E2 metabolites (r > 0.8 andP < 0.001, Table 4). Interestingly, the activity of testosterone 6β-hydroxylation showed a slightly higher degree of correlation with E2 hydroxylations than did dextromethorphan N-demethylation, a selective marker reaction for CYP3A4 (not for CYP3A5). These data suggest that CYP3A5 may also play an important role in hepatic estrogen metabolism. The lack of a perfect correlation between the metabolism of marker substrates for CYP3A4/5 and E2 hydroxylation suggests that other non-CYP3A isoforms also participate in liver microsomal hydroxylation of E2.
To further confirm the role for CYP3A4 and CYP3A5 in the hepatic hydroxylation of E2, we also analyzed in this study the NADPH-dependent metabolism of [3H]E2 by selectively expressed human CYP3A4 and CYP3A5. We found that the overall profiles of [3H]E2 metabolites formed by human CYP3A4 and CYP3A5 were similar to each other, and they were also similar to the profile of estrogen metabolites formed by human liver microsomes (refer to Figs. 2A and 7). E2 2-hydroxylation was the quantitatively dominant hydroxylation pathway for either CYP3A4 or CYP3A5, but a significant amount of 4-OH-E2 was also formed. The ratio of 4-OH-E2 formation to 2-OH-E2 formation by CYP3A4 was ∼1:5, but their ratio for CYP3A5 was ∼1:2, suggesting a crucial role for CYP3A5 in hepatic E2 4-hydroxylation in humans. Since human CYP3A5 is a polymorphic CYP isoform (Wrighton et al., 1989, 1990), it will be of interest to determine whether its polymorphism correlates with the amount of 4-hydroxylated estrogens formed in humans and also with the risk of estrogen-associated human cancers. In addition to catechol estrogen formation by the CYP3A family, significant amounts of 6β-OH-E2, 16β-OH-E2, and y-OH-E2 were also formed by CYP3A4 and CYP3A5, suggesting their role in the hepatic formation of 6β-, 16β-, and y-hydroxylated E2.
We noted that the formation of hydroxy-E1metabolites (e.g., 2-OH-E1 and 4-OH-E1) by microsomes containing selectively expressed human CYP3A4 or CYP3A5 was markedly less than their formation by human liver microsomes. This difference is most likely due to the low 17β-hydroxysteroid dehydrogenase activity contained in theCYP3A4- and CYP3A5-expressed microsomes. The proportionally reduced conversion of E2 to E1 observed with these microsomal preparations supports this explanation (Fig. 7).
In addition to the CYP3A family isoforms, CYP1A2 is another major CYP isoform that has high catalytic activity for NADPH-dependent metabolism of E2 to catechol estrogens (Shou et al., 1997;Cai et al., 1998; Yamazaki et al., 1998). However, our data showed that the hepatic CYP1A2 activity (using caffeine N3-demethylation as a selective probe) was only weakly correlated with the formation of several E2 metabolites (Table 4). This may be due to the relatively low levels of CYP1A2 present in most human liver microsomes as compared with other more abundant E2-metabolizing CYP isoforms.
Conclusions
The present study demonstrated the metabolism of E2 to a large number of hydroxylated or keto E2 metabolites by male and female human liver microsomes. The overall profiles for the E2metabolites formed by male or female human liver microsomes did not appear to be significantly different from each other. Quantitatively, the major hydroxyestrogen metabolites formed were 2-OH-E2,y-OH-E2, and 2-OH-E1. Several other E2metabolites, including 4-OH-E2, 4-OH-E1, 6β-OH-E2, 16α-OH-E2 (E3), and 16β-OH-E2, were also formed in substantial quantities. 6α-OH-E2, 6β-OH-E1, 6-keto-E2, 7α-OH-E2, 12β-OH-E2, 15α-OH-E2, 15β-OH-E2, 16α-OH-E1, 16β-OH-E1, and 16-keto-E2 were only formed in very minute quantities. Testosterone 6β-hydroxylation activity (a selective probe for CYP3A4/5) showed high degrees of correlation with the rate of formation of several hydroxy-E2 metabolites. The dominant role of CYP3A4 and CYP3A5 in hepatic 2-, 4-, andy-hydroxylation of [3H]E2 was further confirmed by studying [3H]E2 metabolism with selectively expressed human CYP3A4 and CYP3A5. It will be of considerable interest to determine the potential physiological or pathophysiological functions associated with any of the endogenous E2 metabolites identified in our present study. Research in this area may lead to an enhanced understanding of the diverse biological actions of the endogenous estrogens.
Footnotes
-
This study was supported by Grant CA 74787 from the National Institutes of Health. This study was presented in preliminary form at the 91st Annual Meeting of the American Association for Cancer Research, San Francisco, CA, April 2000 [Lee AJ, Conney AH and Zhu BT (2000) NADPH-Dependent metabolism of 17β-estradiol and estrone by microsomes from twenty-four human liver samples. Proc Am Assoc Cancer Res41:743].
-
1William M. and Myrle W. Garbe Professor of Cancer and Leukemia Research.
- Abbreviations:
- E2
- 17β-estradiol
- E1
- estrone
- E3
- estriol
- OH
- hydroxy
- CYP
- cytochrome P450
- NADPH
- β-nicotinamide adenine dinucleotide phosphate (reduced form)
- HPLC
- high-performance liquid chromatography
- GC/MS
- gas chromatography/mass spectrometry
- TMS
- trimethylsilyl
- BSTFA
- N,O-bis(trimethylsilyl)trifluoroacetamide
- TMCS
- trimethylchlorosilane
- Received January 9, 2001.
- Accepted April 17, 2001.
- The American Society for Pharmacology and Experimental Therapeutics
