A Novel Testosterone Catabolic Pathway in Bacteria

Abstract

Forty years ago, Coulter and Talalay (A. W. Coulter and P. Talalay, J. Biol. Chem. 243:3238–3247, 1968) established the oxygenase-dependent pathway for the degradation of testosterone by aerobes. The oxic testosterone catabolic pathway involves several oxygen-dependent reactions and is not available for anaerobes. Since then, a variety of anaerobic bacteria have been described for the ability to degrade testosterone in the absence of oxygen. Here, a novel, oxygenase-independent testosterone catabolic pathway in such organisms is described. Steroidobacter denitrificansDSMZ18526 was shown to be capable of degrading testosterone in the absence of oxygen and was selected as the model organism in this study. In a previous investigation, we identified the initial intermediates involved in an anoxic testosterone catabolic pathway, most of which are identical to those of the oxic pathway demonstrated in Comamonas testosteroni. In this study, five additional intermediates of the anoxic pathway were identified. We demonstrated that subsequent steps of the anoxic pathway greatly differ from those of the established oxic pathway, which suggests that a novel pathway for testosterone catabolism is present. In the proposed anoxic pathway, a reduction reaction occurs at C-4 and C-5 of androsta-1,4-diene-3,17-dione, the last common intermediate of both the oxic and anoxic pathways. After that, a novel hydration reaction occurs and a hydroxyl group is thus introduced to the C-1α position of C₁₉steroid substrates. To our knowledge, an enzymatic hydration reaction occurring at the A ring of steroid compounds has not been reported before.

INTRODUCTION

Isoprenoids are built up from one or more five-carbon units and constitute a large group of natural compounds. Among them cholesterol, a triterpenoid, is present in the membranes of eukaryotes, and its major role is to modulate the toughness and permeability of membranes. In addition, cholesterol also serves as a precursor to all steroid hormones, vitamin D, and bile acids (3). One of its derivatives, testosterone, belongs to the C₁₉androgen group and is primarily secreted by the testes of males and ovaries of females. Compared to cholesterol, the aliphatic side chain present on C-17 is absent from testosterone (for its structure, see Fig. 1), which makes it more water soluble than its biosynthetic precursor. Mammals are able to synthesize, but cannot degrade, testosterone and other steroid hormones. After a slight modification to enhance the solubility, these steroid compounds are excreted into the environment through the urinary tract (3).

Fig. 1.

Initial steps involved in the catabolism of testosterone by bacteria. (A) Established oxic pathway (7, 17). (B) Proposed initial steps in the anoxic pathway, as studied in S. denitrificansDSMZ18526. The marked WPS3 and WPS4 are the presumed intermediates that were not found in the present study. The ring identification system (A to D) and carbon numbering system of steroids is as shown for testosterone.

A variety of androgens and estrogens have been detected in effluents of American, Brazilian, Canadian, and German wastewater treatment plants and in surface water in American and Dutch rivers at concentrations in the ng liter⁻¹range (2, 18, 20, 32). Most steroid hormones present in the environment are produced by humans and livestock (25). One of the major concerns about these natural steroid hormones is their ability to alter the sexual behavior and endocrine systems of wildlife (22, 31). Because of the negative environmental impacts of steroid hormones, the removal of these compounds from the environment has attracted increasing interest (1). Recent studies indicated that anoxic riverbed sediments and soil have the potential to be a reservoir for steroid compounds (13, 33). Thus, in order to improve the removal of steroids from the environment, it is necessary to understand the biochemical processes involved in anoxic mineralization of steroid hormones.

In addition, because of their diverse physiological functions in the human body, steroid compounds may be ranked among the most widely marketed chemicals by the pharmaceutical industry (11). This interest is due to the biotechnological applications of steroid-transforming enzymes with high regio- and stereospecificity in the industrial synthesis of steroids (19). The complex structure of steroid compounds and the high regio- and stereoselectivity of the enzymatic reactions render the utility of biocatalysts from microbial sources particularly fascinating.

The complete oxic mineralization of testosterone by various species such as Comamonas testosteroniwas studied in some detail (7, 16, 17) (Fig. 1A). The oxic catabolism of testosterone is initiated by dehydrogenation of the 17β-hydroxyl group to produce androst-4-en-3,17-dione, which then undergoes another dehydrogenation to form androsta-1,4-diene-3,17-dione. The subsequent cleavage of the core ring system is catalyzed by several oxygenases that utilize oxygen as a cosubstrate (12, 16, 17, 26) (see also Fig. 1A).

On the contrary, little is known about the anoxic degradation of steroids (4). It is obvious that anaerobic microorganisms would have to utilize a novel oxygen-independent catabolic pathway to degrade testosterone in the absence of molecular oxygen. Studies on anoxic testosterone metabolism should reveal many unprecedented reactions and novel enzymes with many potential applications. In the last decade, a few bacteria that can mineralize steroids under denitrifying conditions were isolated and characterized (9, 10, 15, 30). All of them are proteobacteria and have a relatively narrow substrate spectrum. One of them, Steroidobacter denitrificansDSMZ18526, can anaerobically utilize certain C₁₈estrogen or C₁₉androgen steroids as the sole carbon and energy source via an unknown catabolic pathway (10). Recently, we reported on the initial steps in the anoxic catabolism of testosterone by this denitrifying bacterium (5). We demonstrated that under anoxic conditions, S. denitrificansinitially oxidizes testosterone to 1-dehydrotestosterone, which is then transformed to androsta-1,4-diene-3,17-dione. Moreover, it seems that testosterone can also be transformed to androst-4-en-3,17-dione by Steroidobactercells. In general, the initial steps of anoxic testosterone degradation by S. denitrificansare very similar to those of the oxic pathway demonstrated in Comamonas testosteroni(7, 16, 17). In the present study, we report on subsequent intermediates of the anoxic testosterone catabolic pathway in S. denitrificans. From our current data, a novel pathway for testosterone catabolism is proposed (Fig. 1B).

MATERIALS AND METHODS

Materials and bacterial strain.

[4C-¹⁴C]testosterone and ¹⁸O-labeledwater (99 atom%) were respectively obtained from Perkin-Elmer and Sigma. The chemicals used were of analytical grade and were purchased from Mallinckrodt Baker, Merck, or Sigma-Aldrich. Steroidobacter denitrificansDSMZ18526 (10) was obtained from the Deutsche Sammlung für Mikroorganismen und Zellkulturen (Braunschweig, Germany).

The denitrifying growth of bacteria.

S. denitrificanswas grown anaerobically at 28°C under a nitrogen atmosphere. Large-scale fed-batch cultures were carried out in 5-liter glass bottles sealed with rubber stoppers. The medium and method used for the cultivation of S. denitrificansunder denitrifying conditions are described elsewhere (5). Cells were harvested by centrifugation in the exponential growth phase at an optical density at 600 nm (OD₆₀₀) of 0.8 to 1.0 (optical path, 1 cm), and the cell pellet was then stored at −80°C.

Bacterial culture grown anaerobically on [4C-¹⁴C]testosterone.

In a small-scale fed-batch culture (250 ml), S. denitrificanscells were incubated with 2.5 mM testosterone under denitrifying conditions, to which [4C-¹⁴C]testosterone (10⁸dpm) was added as a tracer. The fed-batch culture was carried out in a 300-ml glass bottle sealed with a rubber stopper. The headspace (ca. 50 ml) of the culture was connected to a 120-ml glass tube containing 100 ml of 3 M NaOH, which trapped ¹⁴CO₂produced by S. denitrificanscells during the mineralization of [4C-¹⁴C]testosterone. A Tygon tube (2 mm in inner diameter, 25 cm long) was used for the connection, one end of which was connected to a ceramic cylinder (15 by 8 mm) to minimize the size of the CO₂bubbles. The ceramic cylinder was located near the bottom of the 120-ml glass tube containing a 3 M NaOH solution. After different time intervals of incubation (0, 8, 16, 24, 32, 40, and 48 h), samples (1.5 ml) were withdrawn from the 3 M NaOH solution (100 ml). One hour before each sampling, nitrogen gas (ca. 300 ml) was used as the carrier gas to expel the residual ¹⁴CO₂from the 300-ml glass bottle with a flow rate of ca. 5 ml min⁻¹. The amount of the trapped ¹⁴CO₂was determined as described below. At the same time intervals, samples (5 ml) were withdrawn from the bacterial culture to measure the growth of bacterial cells (measured as total proteins), the residual amount of nitrate and testosterone in the medium, the amount of ¹⁴C remaining in the growth medium, and the amount of ¹⁴C assimilated in the biomass. Separation of S. denitrificanscells from the residual testosterone by centrifugation (10,000 × gfor 15 min) was not successful. Therefore, the culture samples (0.5 ml) were extracted with the same volume of ethyl acetate three times to isolate the residual [4C-¹⁴C]testosterone from the water fraction. After centrifugation (10,000 × gfor 10 min) the biomass, including cell debris and lipids, remained in the water phase and interface, whereas [4C-¹⁴C]testosterone remained in the ethyl acetate phase. The ethyl acetate fractions were combined, evaporated, and the residue was redissolved in 0.5 ml of ethanol. The amount of ¹⁴C remaining in the 0.5-ml water phase (mainly the assimilated biomass) and ¹⁴C extracted by ethyl acetate (mostly the remaining [4C-¹⁴C]testosterone) were determined as described below.

Measurement of protein content.

Culture samples (0.2 ml) were centrifuged at 10,000 × gfor 10 min. After centrifugation, the pellet was resuspended in 1 ml of reaction reagent (Pierce BCA protein assay kit; Thermo Scientific). The protein content in the culture samples and in cell extracts were determined using a BCA protein assay according to manufacturer's instructions with bovine serum albumin as the standard.

Measurement of testosterone and nitrate concentrations.

Testosterone was quantified by using high-performance liquid chromatography (HPLC) as described below. Culture samples of 0.5 ml were extracted three times with an equal volume of ethyl acetate. After evaporation of ethyl acetate under a vacuum, the residue was dissolved in 0.5 ml of 2-propanol for the HPLC analysis. Nitrate was determined by using the 2,6-dimethylphenol photometric method as described previously (6).

Measurement of the amount of ¹⁴C.

The amount of ¹⁴C in water-soluble samples (0.5 ml) was determined by liquid scintillation counting (liquid scintillation analyzer, Tri-Carb 2900 TR [Perkin-Elmer]) using 1.5 ml of an Ultima Gold high flash-point LSC scintillation cocktail (Perkin-Elmer). In contrast, the amount of ¹⁴C in ethanol-soluble samples (0.5 ml) was determined by liquid scintillation counting using the Ecoscint O scintillation cocktail (National Diagnostics). The counting efficiency was determined via the channel ratio method.

In vivotransformation of 1-testosterone.

S. denitrificanscells grown anaerobically with testosterone in a 2-liter glass bottle were harvested by centrifugation (10,000 × gfor 10 min, 4°C) when the culture was incubated until an OD₆₀₀of 0.9 (optical path, 1 cm) was obtained. The pellet with a wet weight of 2.1 g was resuspended in 23 ml of ice-cold 50 mM potassium phosphate buffer (pH 7) containing 15 mM sodium nitrate and 1 mM 1-testosterone. The assay (25 ml) was performed under anoxic conditions at 28°C for 1 h under a nitrogen gas phase. After that, the cell suspension was kept on ice, and steroid compounds in the 25-ml cell suspension were immediately extracted three times with equal volumes of ethyl acetate. The extracts were combined, and the solvent was evaporated to dryness. The residue was redissolved in 2 ml of methanol. The extremely nonpolar contaminants were removed by passing the methanol-soluble samples (2 ml) through a solid-phase extraction cartridge (Bakerbond [C₁₈] SPE, 40-μm particles, 1 ml [J. T. Baker]). The presence of 1-testosterone-derived steroid intermediates in the samples was detected by ultraperformance liquid chromatography-atmospheric pressure chemical ionization-mass spectrometry (UPLC-APCI-MS) as described below.

Preparation of cell extracts.

All steps used for preparation of cell extracts were performed at 4°C under anoxic conditions. Frozen cells were suspended in twice the volume of 150 mM Tris-HCl buffer (pH 9) containing 0.1 mg of DNase I ml⁻¹. Cells were broken by passing the cell suspension through a French pressure cell (Thermo Fisher Scientific) twice at 137 MPa. The cell lysate was fractionated by two steps of centrifugation: the first step involved centrifugation for 30 min at 20,000 × gto get rid of the cell debris, unbroken cells, and residual testosterone. The supernatant (crude cell extract) was then centrifuged at 100,000 × gfor 1.5 h to separate soluble proteins from membrane-bound proteins.

In vitrobiotransformation assays.

In vitroassays were routinely performed under anoxic conditions at 30°C for 16 h under a nitrogen gas phase. The assay mixtures (1.5 ml) for exclusively producing the steroid products, WPS1 and WPS2, contained 100 mM Tris-HCl buffer (pH 9), soluble proteins (7.5 mg) extracted from S. denitrificanscells, 4.5 mM NADH, and 100 μl of a 67.5 mM androsta-1,4-diene-3,17-dione solution (in 2-propanol). The final concentration of androsta-1,4-diene-3,17-dione in the reaction mixture was 4.5 mM. The final 2-propanol content was 6.67%. To produce the steroid products WPS5α, WPS5β, and WPS6, the reaction mixture (1.5 ml) contained 100 mM Tris-HCl buffer (pH 9), soluble proteins (7.5 mg) extracted from S. denitrificanscells, 75 μl of a 90 mM androsta-1,4-diene-3,17-dione solution (in 2-propanol), and 25 μl of a 90 mM testosterone solution (in 2-propanol). The final concentrations of androsta-1,4-diene-3,17-dione and testosterone in the reaction mixture were 4.5 and 1.5 mM, respectively. The final 2-propanol content was also 6.67%. In the biotransformation assays to produce WPS5α, WPS5β, and WPS6, no artificial electron acceptor was added. To produce the steroid products on a large scale, the reaction mixture was enlarged to 100 ml.

¹⁸O-incorporation experiments.

To elucidate the origin of the oxygen atom at the C-1 position of three hydroxylated steroid products (WPS5α, WPS5β, and WPS6), two in vitroassays were performed. The two reaction mixtures (3 ml for each assay) were prepared anaerobically and incubated at 30°C for 16 h with shaking. Three hydroxylated steroid products were purified from the assays and then analyzed by UPLC-APCI-MS as described below.

(i) Control assay.The 3-ml reaction mixture contained 100 mM Tris-HCl buffer (pH 9), soluble proteins (15 mg) of S. denitrificans, 4.5 mM androsta-1,4-diene-3,17-dione, and 1.5 mM testosterone. The stock solution (in 2-propanol) of testosterone and androsta-1,4-diene-3,17-dione was prepared as described in the biotransformation assays for producing the intermediates, WPS5α, WPS5β, and WPS6. The 2-propanol content of the control assay was 6.67%.

(ii) ¹⁸O-labeled water-treated assay.¹⁸O-labeled water (1.5 ml) was added to 1.5 ml of 200 mM Tris-HCl buffer (pH 9) containing soluble proteins of S. denitrificans(15 mg). The final ¹⁸O-labeled water content was ca. 50%. The reaction was begun by adding 1.5 mM testosterone and 4.5 mM androsta-1,4-diene-3,17-dione to the anoxic assay. The 2-propanol content of the ¹⁸O-labeled water-treated assay was also 6.67%.

TLC.

Steroid products were first extracted three times with an equal volume of ethyl acetate, and the ethyl acetate-soluble fraction was concentrated under a vacuum. The steroid standards and extracted products were separated on silica gel aluminum thin-layer chromatography (TLC) plates (Silica gel 60 F₂₅₄; thickness, 0.2 mm; 20 by 20 cm [Merck]). The following developing solvent system was used: dichloromethane-ethyl acetate-ethanol (14:4:1 [vol/vol]). The steroid compounds were visualized under UV light at 254 nm or visualized by spraying the TLC plates with 30% (vol/vol) H₂SO₄.

HPLC.

A reversed-phase Hitachi HPLC system was used for the separation, isolation, and identification of steroid standards and steroid products transformed from androsta-1,4-diene-3,17-dione.

(i) System 1.For the separation and isolation of WPS1 and WPS2, an analytical RP-C₁₈column [Luna C18(2), 5 μm, 150 by 4.6 mm (Phenomenex)] was used with flow rate of 0.4 ml min⁻¹at room temperature. The mobile phase was 70% (vol/vol) methanol. Steroid standards and products were detected with a UV detector (L-2400; Hitachi) at 225 nm.

(ii) System 2.The column used for the separation and isolation of WPS5α, WPS5β, and WPS6 was the same as that described for system 1. Separation was performed isocratically with 50% (vol/vol) methanol as an eluent and a flow rate of 0.4 ml min⁻¹. Steroid products were detected using two detectors: a UV detector (Hitachi) monitored at 205 nm and a refractive-index detector (Bischoff) in series.

UV-VIS spectroscopy.

Standards and HPLC-purified steroid products were dissolved in acetonitrile in the range of 5 to 10 μg ml⁻¹. UV absorption spectra of these compounds were obtained using a U-1900 UV/VIS spectrometer (Hitachi).

UPLC-APCI-MS.

The ethyl acetate-extractable samples or HPLC-purified steroid intermediates were analyzed by UPLC-MS with UPLC coupled to an APCI mass spectrometer. Mass spectral data were obtained using a Micromass ZQ quadrupole mass spectrometer (Waters) equipped with a standard APCI source operating in the positive-ion mode. Separation was achieved on a reversed-phase C₁₈column (Acquity UPLC BEH C₁₈, 1.7 μm, 100 by 1.0 mm [Waters]) with a flow rate of 0.1 ml min⁻¹at 35°C (column oven temperature). The mobile phase comprised a mixture of two solvents: solvent A (1% [vol/vol] acetonitrile containing 0.1% formic acid to enable good ionization in the APCI) and solvent B (methanol containing 0.1% formic acid). Separation was achieved with a linear gradient of solvent B from 90 to 95% in 17 min. In the APCI-MS analysis, the temperature of the ion source was maintained at 100°C. Nitrogen desolvation gas was set at a flow rate of 500 liters h⁻¹, and the probe was heated to 400°C. Nitrogen was used as an APCI carrier gas. The corona current was maintained at 20 μA, and the electron multiplier voltage was set to 650 eV. The parent scan was in the range of 250 to 350 (m/z).

ESI-MS.

Electrospray ionization (ESI)-MS data were recorded on a Bruker APEX II mass spectrometer running in the positive-ion mode.

Nuclear magnetic resonance (NMR) spectroscopy.

¹H- and ¹³C-NMR spectra were recorded at 27°C with a Bruker Avance-400 FT-NMR spectrometer. Chemical shifts (δ) were recorded and are shown as parts per million (ppm) values with deuterated chloroform (99.5%;¹H, δ = 7.26 ppm; ¹³C, δ = 77.0 ppm) as the solvent and internal reference.

RESULTS

Anoxic mineralization of testosterone by S. denitrificans.

In our previous study, we showed that S. denitrificansshould be able to anaerobically degrade testosterone using a stoichiometric method. However, the possibility of partial oxidation of testosterone by our model organism could not be excluded (5). We presented here clear and direct evidence to show that testosterone was mineralized to CO₂during the denitrifying growth of S. denitrificans(Fig. 2).

Fig. 2.

(A) Growth of S. denitrificansDSMZ18526 with 2.5 mM total testosterone under denitrifying conditions. Symbols: •, bacterial growth (measured as the total protein concentration in the culture); ▪, residual total testosterone; ▴, total nitrate consumption. (B) Assimilation and mineralization of [4C-¹⁴C]testosterone (original amount was 400,000 dpm ml⁻¹) with time in the same bacterial culture (250 ml). Symbols: □, residual ¹⁴C-labeled carbon in the medium; ○, assimilated ¹⁴C-labeled carbon in the biomass; ▿, trapped ¹⁴CO₂. The data are the average of three experimental measurements. After 48 h of incubation, testosterone (2.5 mM) was consumed to a residual concentration of 0.2 mM (8%), which was accompanied by the consumption of 25.5 mM nitrate. Hence, based on the nitrate consumption (see the equation in Results), 1.3 mM testosterone (52%) should have been completely degraded to CO₂, and 1.0 mM testosterone (40%) should have been assimilated in cells. The measured ¹⁴C distribution is in good agreement with this inference: 10% of the ¹⁴C remained in the medium (mainly as [4C-¹⁴C]testosterone), whereas 44% of the ¹⁴C was assimilated in the biomass.

As shown in Fig. 2A, bacterial growth (measured as the increase in the protein concentration) was accompanied by a decrease in the concentration of total testosterone and the consumption of nitrate. The results were very consistent with the theoretical stoichiometry for anoxic testosterone mineralization under denitrifying conditions, which follows the equation (see the explanation in the legend to Fig. 2): C₁₉H₂₈O₂+ 20 NO_3⁻+ 20 H⁺→ 19 CO₂+ 10 N₂+ 24 H₂O.

Furthermore, bacterial growth was accompanied by a decrease in the residual [4C-¹⁴C]testosterone in the medium, the accumulation of radioactive ¹⁴C in the bacterial cells, and an increase in the amount of trapped ¹⁴CO₂(Fig. 2B). However, it seemed that ¹⁴CO₂was not efficiently captured using 3 M NaOH as the trap. Thus, after 48 h of incubation, only 20% of the ¹⁴C (in the form of CO₂) had been trapped (Fig. 2B). After 48 h of anaerobic growth of S. denitrificans, the total ¹⁴C recovered was 74% (10% of the ¹⁴C remained mainly as testosterone in the medium, 44% of the ¹⁴C was assimilated in cells, and 20% the ¹⁴C was trapped by the 3 M NaOH solution; Fig. 2B). Compared to the stoichiometric results (Fig. 2A), it seems that most of the lost ¹⁴C had escaped in the form of CO₂. Our results clearly indicated that S. denitrificansis able to completely degrade testosterone to CO₂under anaerobic conditions.

In vitrotransformation of steroid substrates by cell extracts of S. denitrificans.

In our previous study (5), androsta-1,4-diene-3,17-dione was the last identified intermediate of the anoxic testosterone catabolic pathway. To identify subsequent intermediates, in vitrotransformation of androsta-1,4-diene-3,17-dione under anoxic conditions was carried out. In the first experiment, the crude cell extract of S. denitrificanswas fractionated by ultracentrifugation into soluble and membrane-bound proteins. These different protein fractions were incubated with androsta-1,4-diene-3,17-dione and testosterone without the addition of any artificial electron acceptor. When the total proteins of Steroidobactercells were incubated with androsta-1,4-diene-3,17-dione and testosterone, five intermediates, named WPS1, WPS2, WPS5α, WPS5β, and WPS6, were observed (Fig. 3, lane D). The same intermediates were also produced when soluble proteins extracted from Steroidobactercells were used (Fig. 3, lane F). It is noteworthy that the soluble proteins or membrane proteins alone (without the addition of steroid substrates) did not cause the accumulation of any apparent steroid product after the overnight incubation (Fig. 3, lanes E and G). The soluble proteins of S. denitrificanswere then incubated with the steroid substrates (androsta-1,4-diene-3,17-dione and testosterone) and different artificial electron acceptors (nitrate, ferric ion, NAD⁺, NADP⁺, 2,6-dichlorophenolineophenol [DCPIP], or methylene blue). However, no additional intermediates were observed (data not shown). Intermediate WPS2 highly overlapped with the wax contaminants on the TLC plates (Fig. 3, lanes D and F). Fortunately, because of their apparent UV absorption at ∼225 nm, WPS1 and WPS2 could be separated by applying the HPLC system with a UV detector (see the supplemental material). The intermediates, WPS1 and WPS2, were exclusively produced by incubating soluble proteins of Steroidobactercells with androsta-1,4-diene-3,17-dione and NADH. The transformation of androsta-1,4-diene-3,17-dione to WPS1 and WPS2 occurred, since soluble proteins of S. denitrificanswere present (see the supplemental material). The addition of NADH to the reaction mixture apparently improved the accumulation of WPS1 and WPS2. In an in vitrobiotransformation assay, different concentrations of NADH were added, and the production of the intermediates, WPS1 and WPS2, from androsta-1,4-diene-3,17-dione occurred in an NADH-dependent manner (data not shown).

Fig. 3.

Thin-layer chromatograms showing the production of intermediates from androsta-1,4-diene-3,17-dione and testosterone by the cell extract, soluble proteins, and membrane proteins of S. denitrificansin 100 mM Tris-HCl (pH 9). Lanes: A, four steroid standards; B, negative control 1 (steroid substrates only, without the addition of the proteins of S. denitrificans); C, negative control 2 (cell extract only, without the addition of steroid substrates); D, cell extract with steroid substrates; E, negative control 3 (soluble proteins only, without the addition of steroid substrates); F, soluble proteins with steroid substrates; G, negative control 4 (membrane proteins only, without the addition of steroid substrates); H, membrane proteins with steroid substrates. The total protein concentrations of the cell extract, soluble protein, and membrane protein fractions in different assays were all diluted to 5 mg ml⁻¹. The assays (1.5 ml) containing different protein fractions, 4.5 mM androsta-1,4-diene-3,17-dione, and 1.5 mM testosterone were incubated at 30°C for 16 h under anoxic conditions. Products were extracted with ethyl acetate, separated by TLC, and visualized by spraying the TLC plates with 30% (vol/vol) H₂SO₄. WPS2*, WPS2 highly overlapped with the wax contaminants in the TLC system. Abbreviations: AD, androst-4-en-3,17-dione; ADD, androsta-1,4-diene-3,17-dione; DT, 1-dehydrotestosterone; T, testosterone.

Identification and structural elucidation of products derived from androsta-1,4-diene-3,17-dione.

The R_fvalues, retention times, UV absorption maxima, and molecular weights of HPLC-purified steroid products are summarized in Table 1. The ¹³C- and ¹H-NMR spectral data for these steroid products are shown in Table 2and Fig. 4, respectively. The structures of WPS1 and WPS2 were identified by comparing their spectral data to assignments reported in the literature (34).

Table 1.

TLC, HPLC, UV absorption, and mass assignments of HPLC-purified products and the authentic compound androsta-1,4-diene-3,17-dione^a

Steroid product	TLC behavior (Rfvalue)	HPLC behavior (mean retention time [min] ± SD)	UV absorption maximum (nm)	Mrmeasured by APCI- or ESI-MS
ADDb	0.81	9.8 ± 0.1	242	284
WPS1	0.69	21.1 ± 0.4	225	288
WPS2	0.92	16.5 ± 0.2	227	286
WPS5α	0.23	15.5 ± 0.2	<210	308
WPS5β	0.08	13.1 ± 0.4	<210	308
WPS6	0.15	13.5 ± 0.2	<210	306

^aTLC separation was performed using the developing solvent: dichloromethane-ethyl acetate-ethanol (14:4:1 [vol/vol]). Steroid standard and products were analyzed by HPLC in triplicate under two different separation conditions as described in Materials and Methods.

^bADD, androsta-1,4-diene-3,17-dione.

Table 2.

¹³C-NMR chemical shifts for androsta-1,4-diene-3,17-dione compared to those of HPLC-purified steroid products in CDCl₃

Chemical shift (δC) for ADDa	Chemical shift (δC)
	WPS1	WPS5α	WPS5β	WPS6
186.5	200.4	82.2	82.3	220.5
168.5	158.6	73.1	73.5	73.3
155.7	127.9	63.3	67.0	67.0
155.5	82.1	51.2	51.4	51.9
128.2	51.0	47.4	47.5	48.2
124.6	50.6	43.3	43.4	47.6
52.7	44.8	39.9	40.0	40.2
50.9	43.6	36.9	38.8	38.8
48.1	41.4	36.6	38.5	38.3
43.8	39.5	35.9	37.8	37.8
36.0	37.0	34.0	37.0	36.2
35.6	36.2	32.7	35.9	35.4
32.9	31.3	31.7	31.7	31.8
32.7	31.0	30.9	30.9	31.0
31.6	27.9	28.6	28.8	28.7
22.5	23.7	23.9	23.8	22.2
22.3	21.3	20.1	20.7	20.2
19.1	13.4	13.0	13.4	14.2
14.2	11.7	11.5	11.5	13.3

^aADD, androsta-1,4-diene-3,17-dione.

Fig. 4.

¹H-NMR spectra of androsta-1,4-diene-3,17-dione and HPLC-purified steroid products (400 MHz, CDCl₃).

In the ¹H-NMR spectrum of WPS6, two oxygenated methine protons were present at δ_H4.04 (1H, m, H-3) and 3.84 (1H, br.s, H-1) (Fig. 4). In the ¹³C-NMR spectrum of WPS6, one carbonyl signal revealed at δ_C220.5 (C-17) and two heteroatom-substituted carbons were observed at δ_C73.3 (C-1) and 67.0 (C-3) (Table 2). Final structural elucidation of WPS6 was carried out using two-dimensional (2D)-NMR (COSY, NOESY, HSQC, and HMBC). The location of a carbonyl group at C-17 was confirmed by the ³Jcorrelation signal between H-18 (δ_H0.83) and a carbonyl carbon (δ_C220.5). Hydroxylated C-1 was proven by the ²Jand ³Jcorrelation signals between H-1 (δ_H3.84)/C-10 (δ_C51.9) and H-19 (δ_H0.87)/C-1 (δ_C73.3), respectively. In the NOESY spectrum of WPS6, the presence of an NOE signal between H-1 and 19-CH₃indicated that H-1 should be in a β-conformation (data not shown). All of the data described above suggested that WPS6 was 1α,3β-dihydroxy-5α-androstan-17-one.

In the case of WPS5β, an additional oxygenated methine proton and a corresponding oxygenated carbon were present at δ_H3.64 (1H, t, J= 8.4 Hz, H-17; Fig. 4) and δ_C82.3 (Table 2), respectively. The final structural elucidation of WPS5β was performed using 2D-NMR. HMBC correlations between H-18 (δ_H0.76)/C-17 (δ_C82.3) and H-1 (δ_H3.84)/C-19 (δ_C13.4) suggested that WPS5β was androstan-1,3,17-triol. The NOE signal between H-1 and 19-CH₃was present, indicating the β-conformation of H-1 of this compound. These data suggested that WPS5β was androstan-1α,3β,17β-triol.

The ¹H and ¹³C-NMR spectra of WPS5β and WPS5α were very similar (Fig. 4, Table 2). An H-3 signal of WPS5α was present at δ_H4.14 (1H, br.s), and the signal of C-3 of WPS5α was upfield shifted to 63.3 ppm (Table 2) compared to that of WPS5β. From the results of the splitting pattern and chemical shifts in their NMR spectra, H-3 of products WPS5β and WPS5α was determined to be in the α- and β-conformations, respectively. Our data suggested that WPS5α was androstan-1α,3α,17β-triol.

Evidence for the hydration reaction at the A ring of testosterone.

To look for the origin of the hydroxyl groups at C-1 of WPS5α, WPS5β, and WPS6, two in vitrotransformation assays were performed: (i) an ¹⁸O-labeled water-treated assay contained ca. 50% ¹⁸O-labeled water (mol/mol) in the anoxic reaction mixture (3 ml), and (ii) a control assay (3 ml) was incubated under anoxic conditions without the addition of ¹⁸O-labeled water. After overnight incubation, the steroid products WPS5α, WPS5β, and WPS6 were purified from these assays, and their molecular weights were determined by APCI-MS (Fig. 5).

Fig. 5.

APCI-MS spectra (positive-ion mode) of WPS5α, WPS5β, and WPS6. (A) WPS5α purified from the anoxic control assay. (B) WPS5α purified from the ¹⁸O-labeled H₂O-treated assay. (C) WPS5β purified from the anoxic control assay. (D) WPS5β purified from the ¹⁸O-labeled H₂O-treated assay. (E) WPS6 purified from the anoxic control assay. (F) WPS6 purified from the ¹⁸O-labeled H₂O-treated assay.

The WPS5α samples purified from the control assay had a molecular weight of 308, and a protonated molecular ion ([M+H]⁺, m/zat 309) and a significant dehydrated fragment ion ([M-H₂O+H]⁺, m/zat 291) derived from WPS5α were observed (Fig. 5A). In contrast, around one-half of the product WPS5α purified from the ¹⁸O-labeled water-treated assay showed an increase in the molecular weight from 308 to 310 (Fig. 5B). An increase in the molecular weight from 306 to 308 was also observed in product WPS6 purified from the ¹⁸O-labeled water-treated assay (Fig. 5F). The proportion of products WPS5α and WPS6 with increased molecular weight (ca. 50%) exactly matched the proportion of ¹⁸O-labeled water (ca. 50%) present in the in vitrobiotransformation assay (¹⁸O-labeled water-treated assay). For WPS5β purified from the control assay, [M+H]⁺at m/z309 was omitted, and a significant [M-H₂O+H]⁺at m/z291 was present (Fig. 5C). Around one-third of the product WPS5β purified from the ¹⁸O-labeled water-treated assay showed an increase in the dehydrated fragment ion ([M-H₂O+H]⁺from m/zat 291 to 293 (Fig. 5D). According to the ESI-MS analysis, the molecular weights of WPS5α and WPS5β were both 308 (Table 1). It is worth mentioning that for product WPS5β, the proportion (ca. 33%) of the dehydrated fragment ion [M-H₂O+H]⁺(Fig. 5D) with an increased m/zratio exactly matched that of WPS5α (Fig. 5B). These results clearly demonstrate that the oxygen atom of the incorporated hydroxyl groups at C-1 of products WPS5α, WPS5β, and WPS6 originated from water and not from molecular oxygen.

DISCUSSION

Proposed initial reactions of the anoxic testosterone catabolic pathway.

According to data from our previous work (5) and the present study, an outline for the initial reactions of a novel testosterone catabolic pathway is proposed using S. denitrificansas the model organism (Fig. 1B). In our previous report (5), we showed that testosterone is oxidized to androsta-1,4-diene-3,17-dione via two dehydrogenation reactions at C-1/C-2 and the hydroxyl group at C-17 of testosterone. In the present study, we used androsta-1,4-diene-3,17-dione as the substrate and NADH as the electron donor in the in vitrotransformation assays, which resulted in the production and accumulation of WPS1 and WPS2. Compared to androsta-1,4-diene-3,17-dione, the double bond at C-4/C-5 of WPS1 and WPS2 was saturated by a reduction reaction. The fact that S. denitrificansis able to grow anaerobically on these steroid compounds (testosterone, 1-dehydrotestosterone, androst-4-en-3,17-dione, androsta-1,4-diene-3,17-dione, WPS1, or WPS2) as the sole carbon source corroborates the conclusion that these steroids should be true intermediates of an anoxic testosterone catabolic pathway. WPS1 and WPS2 were produced from androsta-1,4-diene-3,17-dione in an NADH-dependent manner, suggesting that the biocatalyst responsible for the reduction reaction at C4/C5 of steroids may be an NADH-dependent enzyme. Our NMR data showed that the hydrogen atom at C-5 of WPS1 and WPS2 is of the α-form. Thus, the transformation of androsta-1,4-diene-3,17-dione to WPS2 may be catalyzed by an enzyme from the steroid-5α reductase subfamily. Major research efforts were devoted to the study of steroid-5α-reductase (SRD5α) and steroid-5β-reductase (SRD5β) in mammalian model organisms because of their implications in numerous human diseases such as prostate carcinoma and hepatic dysfunction, and it seems that the two enzymes have different ancestors (21). In contrast, very little is known about nonmammalian species.

Missed steroid intermediates between WPS1/2 and WPS5/6.

In the proposed anoxic testosterone catabolic pathway (Fig. 1B), two presumed intermediates, WPS3 and WPS4, were still not found in the present study. According to current data, we supposed that WPS3 (for its structure, see Fig. 1B) is the direct precursor of WPS5β, which might be produced from WPS3 by an unidentified 3β/17β-hydroxysteroid dehydrogenase (3β/17β-HSD) of S. denitrificans. The same enzyme may also catalyze the transformation of WPS4 to WPS6 (for its structure, see Fig. 1B). Such an enzyme was purified and characterized from Comamonas testosteroni, which has the ability to degrade testosterone under oxic conditions (28, 29). On the other hand, WPS3 may be transformed to WPS5α by another unidentified enzyme of S. denitrificans, 3α-hydroxysteroid dehydrogenase. This enzyme was also purified and characterized from the same Gram-negative bacterium, C. testosteroni(24). Nevertheless, the biocatalysts of S. denitrificansresponsible for these functions remain to be identified.

Unprecedented hydration reaction of the steroid A ring.

In the present work, we identified three hydroxylated steroid products (WPS5α, WPS5β, and WPS6) produced from androsta-1,4-diene-3,17-dione by soluble proteins of S. denitrificansin the absence of molecular oxygen (for their structures, see Fig. 1B). All of these products have a hydroxyl group at the C-1α position. It is known that almost all of the hydroxyl groups of steroid compounds result from hydroxylation reactions catalyzed by cytochrome P450 monooxygenases (3). In addition, we previously reported that oxygen-independent hydroxylation catalyzed by a molybdenum-containing hydroxylase results in the addition of a hydroxyl group at C-25 of C27 steroid compounds, with water as the oxygen donor (6). We present here the first report concerning the unprecedented hydration reaction of the A ring of steroid compounds. The following lines of evidence suggest that the hydroxyl group at C-1 of WPS5α/β and WPS6 was introduced via a hydration reaction: (i) WPS1 (with a double bond at C-1/C-2) was transformed to WPS5α/β and WPS6 by a soluble protein fraction of S. denitrificans. (ii) The addition of electron acceptors (e.g., ferric ion) or electron donors (NADH or NADPH) did not improve the production of WPS5α/β or WPS6 in the in vitrobiotransformation, indicating that the introduction of a hydroxyl group at C-1 of WPS5α/β and WPS6 was catalyzed by neither a cytochrome P450 monooxygenase (which requires NADPH) nor an Mo-containing hydroxylase (which requires ferric ion, 8). (iii) The oxygen atom of the hydroxyl group at C-1 of steroid products WPS5α/β and WPS6 originated from water. Considering that a variety of anaerobes utilize enoyl coenzyme A hydratase and fumarase, respectively, involved in the β-oxidation pathway and citric acid cycle to oxidize different organic compounds, it is not surprising to see that anaerobes apply hydration reactions to oxidize and activate steroid substrates in the absence of oxygen. It is known that thiocyanate is a noncompetitive inhibitor of fumarase (8, 14, 23, 27). However, the addition of 500 mM thiocyanate to the in vitrobiotransformation assay did not inhibit the production of three hydroxylated steroid intermediates (data not shown). The hydration reaction implies a novel type of hydratase acting on C-1 and C-2 of steroid compounds. It is noteworthy that the hydration reaction at C-1/C-2 of steroids also occurred in the presence of molecular oxygen (data not shown), suggesting that the corresponding enzyme is not oxygen-labile.

In order to validate the roles of these hydroxylated steroids in the anoxic testosterone catabolism, we also conducted an in vivoassay with whole cells of S. denitrificansto transform 1-testosterone (WPS1). The ethyl acetate-extractable sample was then analyzed by UPLC-APCI-MS, and WPS2, WPS5β, and WPS6 were present in the extract (data not shown). The result corroborates the conclusion that these steroids should be true intermediates of the anoxic pathway.