Dehydroepiandrosterone (3β-hydroxy-androst-5-ene-17-one; DHEA²) is a 19-carbon steroid derived from cholesterol by a series of cytochrome P450 monooxygenase- and hydroxysteroid dehydrogenase-dependent reactions (Conley and Bird, 1997). In its sulfated form, DHEA is the most abundant circulating steroid in humans and is a precursor to the sex steroids, estrogen and testosterone. Levels of DHEA-S in the circulation are high during fetal development (1–5 μM) but fall rapidly after birth and remain low for the first 5 years of life. DHEA and DHEA-S levels in blood then rise and peak during the second decade (∼10 μM), followed by an age-dependent decline for individuals aged 30 or above (Herbert, 1995). The developmental changes in circulating levels of DHEA and DHEA-S in the blood are not paralleled by other steroid hormones, suggesting that the mechanisms regulating DHEA formation in adrenal are unique (Rainey et al., 2002). In contrast, serum cholesterol levels tend to increase with age, whereas DHEA levels decline with age. The decline in circulating levels of DHEA and its sulfate derivative appears to be inversely correlated to the rise in cholesterol and the pathophysiological effects of aging (Barrett-Connor et al., 1999).

Treatment with exogenous DHEA has been shown to have beneficial effects in lowering body fat and modulating the effects of diabetes, atherosclerosis, and obesity in rodent models (Yoneyama et al., 1997). Additionally, DHEA has cancer chemopreventative actions when administered to rodents in low doses (Rao et al., 1992; Lubet et al., 1998). However, at higher doses, DHEA can cause peroxisome proliferation, resulting in hepatomegaly (Frenkel et al., 1990) and subsequent development of hepatocarcinomas (Rao et al., 1992). With the current utilization of DHEA as a proposed dietary supplement to protect against diabetes, atherosclerosis, obesity, and arthritis, the mechanism of biological action of this sterol and its metabolites has become important to study.

Since the rat adrenal does not express CYP17, the rat does not produce DHEA in the adrenal (Voutilainen et al., 1986; Prough et al., 1990). However, DHEA is formed in the human adrenal and is a precursor to sex steroids (Fig. 1). In humans, DHEA circulates as the 3β-sulfate conjugate DHEA-S until taken up by target tissues, where it is then converted to DHEA by sulfatases (Burstein and Dorfman, 1963). In steroidogenic tissues, DHEA is metabolized to androgens and estrogens by hydroxysteroid dehydrogenase reactions. However, other oxidative pathways of DHEA metabolism have not been extensively studied.

The beneficial effects resulting from exogenous administration of DHEA may involve the metabolism of DHEA to multiple biologically active species (Fitzpatrick et al., 2001; Marwah et al., 2002). Fitzpatrick et al. (2001) used LC/MS to identify 7α- and 16α-OH-DHEA as the major metabolites produced by the human, along with another monohydroxylated DHEA species whose position of hydroxylation was unknown. The purpose of this study was to quantify the liver microsomal metabolism of DHEA by various species and elucidate the P450s responsible for the metabolism of DHEA. A sensitive GC/MS method was developed to identify and quantify all the metabolites produced by the metabolism of DHEA. The results of this study provide a method for quantifying the microsomal metabolism of DHEA and demonstrate the regio- and stereoselectivity of specific P450s that account for the unique DHEA metabolite profiles formed by various species.

Materials and Methods

Chemicals. DHEA, 7α-hydroxy-DHEA, 7β-hydroxy-DHEA, 16α-hydroxy-DHEA, androstenedione, and etiocholanolone were purchased from Steraloids (Wilton, NH). Human liver samples were kindly provided by F. Peter Guengerich (Center for Molecular Toxicology, Vanderbilt University School of Medicine, Nashville, TN). The use of these human tissue samples was approved by the Institutional Review Boards of the University of Louisville and Vanderbilt University. The human P450 baculovirus system used to provide functional P450 preparations was designed to express both P450s and P450 oxidoreductase using a suspension culture of baculovirus-infected insect cells (Rushmore et al., 2000). Membranes fractions were freshly prepared at Merck Research Laboratories and the metabolic assays were performed at the University of Louisville.

Animals. Male Sprague-Dawley rats (225 g, HSD:SD) from Harlan (Indianapolis, IN) were maintained on a control diet (AIN-76A; ICN Biomedicals Inc., Costa Mesa, CA) for 5 days. Animals were anesthetized with CO₂ and the livers were perfused with 0.9% sodium chloride before dissection from the body. Livers were cut into small pieces and then homogenized in a Potter-Elvehjem homogenizer containing 4 volumes of 50 mM potassium phosphate buffer, pH 7.4, containing 0.25 M sucrose per gram of liver. Microsomal fractions were isolated by differential centrifugation as described by Remmer et al. (1966). Microsomal fractions were resuspended in 0.1 M Tris-HCl buffer (pH 7.4), containing 0.25 M sucrose, and sedimented a second time. The final preparation was resuspended in Tris-HCl buffer containing sucrose and 10% glycerol and stored at -70°C for up to 3 months without loss of activity. Protein concentrations were determined by measuring formation of bicinchoninic acid Cu¹⁺ complex at 562 nm.

NADPH/Cytochrome c Oxidoreductase Assay. The baculovirus expression system allows coexpression of both P450 and its flavoprotein oxidoreductase (Rushmore et al., 2000), and NADPH/cytochrome c oxidoreductase activity was measured to characterize the enzymatic efficiency in this baculovirus expression system. The reactions were carried out at 25°C in 0.05 M potassium phosphate buffer, pH 7.4, containing 100 μM NADPH, 40 μM cytochrome c, and aliquots of the P450 sample being characterized. The absorbance change at 550 nm was monitored at 25°C with a Cary 50 Bio UV-visible spectrophotometer, assuming a molar absorptivity of 21,100 M^-1 cm^-1 (Masters et al., 1967). The P450/P450 oxidoreductase ratios for CYP3A4, CYP3A5, CYP3A7, CYP2B6, and CYP2B1 preparations are shown in Table 1. The ratios for all P450s prepared were near 1, indicating that the content of P450 oxidoreductase in the preparations was likely not rate-limiting in the reaction.

DHEA Metabolism. Hepatic microsomal protein fractions or recombinant P450s were incubated in 2-ml reaction mixtures containing 0.1 M Tris-HCl buffer, pH 7.5, 1 mM EDTA, 10 mM MgSO₄, and an NADPH-regenerating system consisting of 1 mM β-NADPH, 0.8 mM isocitrate, and 0.1 U/ml of isocitrate dehydrogenase. The samples were oxygenated by blowing pure O₂ into each tube for 15 s. The microsomal fractions and regenerating system were preincubated for 4 min at 37°C before addition of 50 μM DHEA. After incubation for specified times at 37°C in a shaking water bath, the reactions were terminated at various times by adding equal volumes of chilled ethyl acetate. The rates of product formation were measured in the linear portion of the time course. The metabolites were extracted from the aqueous phase three times with ethyl acetate and dried under a stream of N₂ gas at room temperature.

Derivatization of Samples. DHEA and its metabolites were prepared for GC/MS analysis by adding 50 μl of MOX to the dried metabolites overnight at room temperature to derivatize any oxo functional groups. The sample was dried under a stream of N₂ gas at room temperature, 50 μl of BSTFA-TMS was added, and the solution was incubated at 70°C to derivatize hydroxyl groups. An internal standard, etiocholanolone, was added to each sample before extraction with ethyl acetate and analysis by GC/MS.

Gas Chromatography/Mass Spectrometric Analysis. GC/MS was utilized to resolve and quantify the DHEA metabolites, relative to etiocholanolone. Initial experiments assessed linearity of the reaction with time and protein concentration. Reactions were carried out with microsomes from rat, pig, and hamster, as well as five different human samples, to assess potential interindividual variability in product formation. Derivatized DHEA and metabolites were analyzed with an HP5890/HP5973 GC/MS system (Hewlett Packard, Palo Alto, CA). Separation was achieved by using a bonded-phase capillary column (DB-17MS, 15 m × 0.25 mm i.d. × 0.25 μm film thickness) from J&W Scientific (Agilent Technologies, Palo Alto, CA). The GC injection port and interface temperature was set to 280°C, with helium carrier gas maintained at 14 psig. Injections were made in the splitless mode with the inlet port purged for 1 min after injection. The GC oven temperature was held initially at 100°C for 0.5 min, increased at a rate of 30°C min^-1 to 325°C, increased at a rate of 2°C min^-1 to 325°C, and then held for 5 min. Eluate from GC was analyzed under 70-eV electron ionization with full mass scan. The mass scan range measured was m/z 50 to 550. The peak area of each metabolite standard relative to that of the added internal standard, etiocholanolone, was determined for selected ion retrieval chromatograms to establish a standard curve for quantitating DHEA metabolite formation. An internal standard curve was prepared for each compound of interest spanning the concentrations above and below those observed in the biological samples measured.

Statistical Analysis. Experiments were conducted in triplicate, and mean ± S.D. was determined. Statistical significance was determined using a two-tailed Student's t test with p ≤ 0.05 as the criterion for significance.

Results

Analysis of DHEA and Its Metabolites using GC/MS. Fitzpatrick et al. (2001) utilized LC/MS to separate and quantify DHEA and its resulting oxidative metabolites. DHEA was found to be converted by human microsomal fractions to 7α-OH-DHEA, 16α-OH-DHEA, and an unknown monohydroxylated compound. 7-oxo-DHEA was also observed when longer incubation times were utilized (Fitzpatrick et al., 2001; Robinzon et al., 2003). This method was hindered by poor ionization efficiencies of DHEA and its metabolites under conditions of chemical ionization at atmospheric pressure. For the current studies, the possibility of attaining better sensitivity and resolution of DHEA and metabolites using GC/MS was examined. Therefore, a GC/MS method, utilizing derivatization, was developed to separate and quantitate known DHEA metabolites.

DHEA and its metabolite standards contain keto and hydroxyl functional groups that can be derivatized to form stable and more ionizable molecules. To stabilize the compounds and improve their separation by GC, MOX was added to the commercial standards or samples to derivatize oxo functional groups (i.e., prevent keto-enol tautomerization), followed by the addition of BSTFA-TMS to derivatize hydroxyl groups (Fig. 2A). The standards were then separated by GC/MS after conditions for baseline separation of all metabolites were achieved (Fig. 2B). The identity of the compounds produced was determined by comigration with authentic standards and identical electron ionization mass spectra for each compound as shown for DHEA (Fig. 2C). The retention times and MS data are shown in Table 2. Etiocholanolone, which has been previously shown not to be a direct metabolite of DHEA under these conditions, was used as an internal standard. The peak areas of each standard relative to etiocholanolone were used to prepare a standard curve to quantify metabolite production.

Quantification of DHEA and Metabolites by GC/MS. To study the liver microsomal hydroxylation of DHEA in various species, microsomal protein fractions (0.5 mg/ml) from rat, hamster, or pig were incubated with 50 μM DHEA and an NADPH-regenerating system consisting of sodium isocitrate, isocitrate dehydrogenase, and MgSO₄ for up to 20 min. Extracts of the microsomal incubation mixtures were derivatized and then analyzed using GC/MS. To quantify and confirm metabolite identities, two or three characteristic ions for each steroid were selected on the basis of their mass fragmentation. The peak areas of the selected ions of each metabolite were obtained and compared with that of the internal standard, and the absolute values were calculated using calibration curves from the standards.

Figure 3A shows a representative chromatogram of the total ion current for rat liver microsomal metabolism of DHEA at 0 min. DHEA was metabolized by rat liver microsomes to 7α-OH-DHEA and 16α-OH-DHEA in 10 min as indicated by the presence of two metabolite peaks corresponding in retention times to the authentic compounds (Fig. 3B). Moreover, NADPH was required for microsomal metabolism of DHEA, since no metabolite peaks were formed in the absence of an NADPH-regenerating system (data not shown).

Rat, hamster, pig, and human liver microsomal fractions all metabolized DHEA. DHEA was rapidly metabolized in rat (7.2 nmol/min/mg) and hamster (18.9 nmol/min/mg). Rat liver microsomes produced two major monohydroxylated metabolites, 7α-OH-DHEA (4.6 nmol/min/mg) and 16α-OH-DHEA (2.6 nmol/min/mg). In the hamster, DHEA was converted to 7α-OH-DHEA (7.4 nmol/min/mg) and 16α-OH-DHEA (0.26 nmol/min/mg), as well as 11 unidentified metabolites that accounted for a rate of DHEA conversion of 11.2 nmol/min/mg. Pig microsomal metabolism of DHEA displayed lower rates of conversion than rat and hamster metabolism and produced three metabolites, 7α-OH-DHEA (0.70 nmol/min/mg), 16α-OH-DHEA (0.16 nmol/min/mg), and ADIONE (0.26 nmol/min/mg). Although ADIONE has been shown to be formed in the cytosolic fractions of other species with NAD⁺, the formation of ADIONE by pig liver microsomal fractions required NADPH, but not NAD⁺ or NADP⁺ (data not shown), indicating the presence of a 3β-hydroxysteroid dehydrogenase enzyme activity, not only in cytosolic fractions, but also in anabolic liver microsomal fractions of the pig (Fig. 4; Table 3). Future studies will evaluate the role of P450s in this reaction.

Upon incubation with 50 μM DHEA, one human liver microsomal fraction (HL110) hydroxylated DHEA at a rate of 7.8 nmol/min/mg. As in rat, hamster, and pig, 7α-OH-DHEA (0.66 nmol/min/mg) and 16α-OH-DHEA (3.6 nmol/min/mg) were produced (Fig. 5; Table 3). Unlike the other species, the human also converted DHEA to 7β-OH-DHEA at a significant rate (3.5 nmol/min/mg). The identity of the unique metabolite, 7β-OH-DHEA, was established based on its GC retention time and a mass spectrum identical (Fig. 6A) to that of the 7β-OH-DHEA standard (Fig. 6B), but distinct from other DHEA metabolite standards including 11β-OH-DHEA (data not shown). Not all human microsomal fractions tested oxidized DHEA as well as sample HL110. In fact, although four other human liver microsomal fractions displayed the same metabolite profile as HL110, the other human fractions metabolized less than 2 nmol/min/mg DHEA in 10 min (Fig. 5, inset), indicating interindividual variability of DHEA metabolism among humans.

Cytochrome P450 Metabolism of DHEA. To establish which cytochrome P450 was responsible for DHEA metabolite production, 50 μM DHEA was incubated with membrane fractions from baculovirus-infected insect cells that express both a specific P450 and its flavoprotein oxidoreductase, NADPH/cytochrome P450 oxidoreductase. CYP3A4 and CYP3A5 apparently are responsible for the production of 7α-OH-DHEA, 16α-OH-DHEA and 7β-OH-DHEA, with CYP3A4 exhibiting the highest rate of product formation. CYP3A7 is not expressed in adult liver but is expressed in fetal liver (Hakkola et al., 1994); it also formed 16α-OH-DHEA and 7β-OH-DHEA, but no detectable 7α-OH-DHEA (Table 4). CYP2D1 was the rat P450 that most extensively converts DHEA to 16α-OH-DHEA. CYP2B1 and CYP2C11 also contributed to 16α-OH-DHEA metabolite production, whereas CYP3A23 was the rat P450 apparently responsible for 7α-OH-DHEA formation.

Discussion

Since DHEA is considered a natural product/dietary supplement with many proposed benefits to humans (Kroboth et al., 1999) and is available as an over the counter supplement, the mechanism of action of this sterol and its metabolites becomes important to study. DHEA is metabolized to androgens and estrogens in steroidogenic tissues; however, the metabolism of DHEA in other tissues has not been extensively studied. Fitzpatrick et al. (2001) utilized LC/MS to identify the DHEA metabolites formed by rodent and human liver microsomal fractions in the presence of NADPH and O₂. Hydroxylated metabolites, principally, 7α-OH-DHEA, 7-oxo-DHEA, and 16α-OH-DHEA, were identified in both species. However, the major DHEA metabolite produced in humans was monohydroxylated at an unknown position. We also demonstrated that formation of these products was inhibited by miconazole, indicating the role of cytochromes P450 in the metabolism of DHEA. With human liver microsomal fractions, the DHEA hydroxylation was shown to be due to CYP3A, since its metabolism to several products was strikingly inhibited by troleandomycin (ca. 80% inhibition), whereas this inhibitor was less effective in inhibiting DHEA hydroxylation in rat liver microsomal fractions (ca. 16% inhibition). Our results with these inhibitors demonstrate that human liver microsomal hydroxylation of DHEA is predominantly due to the role of CYP3A, whereas in rat, other P450s account for significant conversion to 16α-OH-DHEA (CYP2B1, 2C11, 2D1, 2E1). In addition, α-naphthoflavone (inhibitor of CYP1), quinidine (inhibitor of CYP2D), and chlorzoxazone (inhibitor of CYP2E1) also inhibited DHEA hydroxylation by rat liver microsomes, NADPH, and O₂ by 24%, 12%, and 13%, respectively (Fitzpatrick et al., 2001), demonstrating that several rat P450s are involved in DHEA hydroxylation. We have also shown that DHEA and its cytosolic metabolites induce CYP3A23 expression (native gene in rat hepatocytes and reporter gene constructs in HepG2 cells), showing that DHEA can induce its own metabolism to the 7α-OH-DHEA by induction of CYP3A through action of the pregnane X receptor in rats (Ripp et al., 2002). This increase in 7α-hydroxylase over 16α-hydroxylase activity is also due to the negative regulation of CYP2C11, a 16α-hydroxylase, that is negatively regulated by DHEA (Ripp et al., 2003), demonstrating a complex metabolic scheme when contrasting metabolism across species.

The purpose of the current study was to further identify the unknown metabolite formed by the human liver microsomal metabolism of DHEA and to identify the specific P450s responsible for production of various DHEA metabolites. Although LC/MS allowed for the identification of most of the DHEA metabolites, quantification of DHEA metabolism was difficult to attain due to low ionization efficiency of metabolites under conditions of chemical ionization at atmospheric pressure (Fitzpatrick et al., 2001). The current study, a GC/MS method, was developed to provide a more sensitive method for identification and quantification of the liver microsomal metabolism of DHEA, especially the unknown human metabolite.

The current study examined the oxidative metabolism of DHEA by rodent, hamster, pig, and human microsomal fractions. Each species extensively converted DHEA into monohydroxylated metabolites. ADIONE was also produced in pig liver microsomal fractions in the presence of NADPH and oxygen. ADIONE is an anabolic steroid that is converted to testosterone, resulting in an increase in growth and development of muscle tissue. Since it has been reported to promote lean muscle growth, ADIONE is used frequently by athletes interested in increasing muscle mass (Ziegenfuss et al., 2002). Cytosolic 3β-hydroxysteroid dehydrogenases convert DHEA to ADIONE in the presence of NAD⁺. Pigs are primarily bred for lean muscle production, suggesting a possible role for enhanced levels of an NADPH-dependent microsomal 3β-hydroxysteroid dehydrogenase activity in pig liver. Hamster liver microsomal fractions also converted DHEA into 7α-OH-DHEA and 16α-OH-DHEA, as well as 11 unidentified hydroxylated DHEA species that are possibly secondary metabolites. These results suggest that several cytochrome P450 enzymes may play a role in the DHEA metabolism in the hamster and demonstrate the significant species differences in the metabolism of DHEA.

Metabolism of DHEA by human microsomal fractions yielded both 7α-OH-DHEA and 16α-OH-DHEA; however, the human was the only species to produce 7β-hydroxy-DHEA. This monohydroxylated species, namely 7β-OH-DHEA, is the unknown compound previously reported by Fitzpatrick et al. (2001) and was recently shown to be formed by CYP3A4 (Stevens et al., 2003). Not all human microsomal fractions tested exhibited extensive oxidative metabolism of DHEA (Table 3). Although one human microsomal fraction (HL110), previously noted by Guengerich et al. (1991) to contain high levels of CYP3A, metabolized DHEA at a high rate (7.8 nmol/min/mg), fractions from four other humans hydroxylated DHEA at much lower rates (≤2 nmol/min/mg of DHEA) but exhibited similar metabolite profiles (Table 3). The various rates in DHEA metabolism among humans could be attributed to differences in P450 expression or various P450 polymorphisms.

Although 7α-, 7β-, and 16α-OH-DHEA were produced in human liver microsomal fractions, 7-oxo-DHEA was also formed, albeit at later time points (Fitzpatrick et al., 2001). Additionally, we have found that upon treatment with 50 μM 7-oxo-DHEA, human liver fractions can convert 7-oxo-DHEA into 7α- and 7β-OH-DHEA, indicating a complex metabolic pathway for DHEA in the liver that includes 11β-hydroxysteroid dehydrogenase activity (Robinzon et al., 2003).

The human CYP3A family plays a dominant role in the metabolic elimination of more drugs than any other biotransformation enzyme (Lamba et al., 2002). In adult human liver, the predominant CYP3 is CYP3A4, but low amounts of CYP3A5 have been observed; CYP3A7 is expressed nearly exclusively in the human neonate (Stevens et al., 2003). Fitzpatrick et al. (2001) reported that selective P4503A inhibitors were able to inhibit DHEA metabolite production in the human. The current study utilized insect cells infected with baculovirus expression vectors to examine the P450s responsible for the liver microsomal metabolism of DHEA. Recombinant CYP3A4 was responsible for the conversion of DHEA into 7α-OH-DHEA, 16α-OH-DHEA, and 7β-OH-DHEA. CYP3A5 also converted DHEA into the same metabolites as CYP3A4; this was different from the activity of expressed CYP3A5 reported by Stevens et al. (2003). However, Rushmore and coworkers have shown that expressed CYP3A4 and 3A5 have similar catalytic activities toward testosterone, as we have shown with DHEA as a substrate. The hepatic fetal enzyme, CYP3A7, was found to convert DHEA to 16α-OH-DHEA and to lower amounts of 7β-OH-DHEA. The rat CYP2D1 converted DHEA to 16α-OH-DHEA, as did CYP2C11 and CYP2B1. However, CYP3A23, a major constitutive P450 in rat liver, was the P450 responsible for 7α-OH-DHEA production in the rat. This pattern of hydroxylation of rat CYP3A23 is strikingly different from that of human CYP3A4, 3A5, or 3A7.

The current study utilized recombinant P450 expressed in insect cells to examine DHEA metabolism. The assay of purified P450s requires that they be reconstituted with NADPH/cytochrome P450 reductase in a complex mixture that includes detergent, phospholipids, and reduced glutathione (Gillam et al., 1995). Some in vitro reconstitution experiments have shown that for a number of P450s, the inclusion of cytochrome b₅ can significantly increase substrate turnover by the monooxyenase system by improving the coupling between the cytochromes P450 and NADPH/cytochrome P450 reductase (Gorsky and Coon, 1986; Holmans et al., 1994; Bell and Guengerich, 1997). Cytochrome b₅ is a heme protein whose mechanism of action in reconstituted systems is not clear. It has been suggested that cytochrome b₅ plays a role in donating electrons from NADPH/cytochrome P450 oxidoreductase to P450 (Morgan and Coon, 1984; Yamazaki et al., 1996; Bell and Guengerich., 1997). Although these authors acknowledge that there is evidence that the inclusion of cytochrome b₅ in bacterial membranes may enhance CYP3A4 activity, inclusion of this cytochrome b₅ did not enhance DHEA metabolism by recombinant CYP3A4 under the conditions of our assay (K. K. Michael Miller and R. A. Prough, unpublished data).

In conclusion, this study demonstrates that different species exhibit unique DHEA metabolite profiles due to the stereospecificity of hydroxylation by the various P450s that metabolize DHEA. The unknown major metabolite produced by the human, previously reported by Fitzpatrick et al. (2001), was shown to be 7β-OH-DHEA. DHEA and some of its metabolites are known to interact with certain nuclear receptors and to activate P450 transcription. This could explain the mechanism of some beneficial effects that have been reported with the administration of DHEA. Assessment of the biological activity of DHEA and metabolites in activating other novel nuclear receptors is under way.