Characterization of an NADPH-dependent 17β-hydroxysteroid dehydrogenase from a urinary tract bacterial isolate

Abstract

Androgens are a class of steroid hormones that play essential roles in somatic development, reproductive physiology, and anabolic processes in both males and females. Beyond their physiological importance, androgens are implicated in the pathogenesis of several diseases and contribute to the progression of hormone-sensitive malignancies such as prostate, breast, lung, and ovarian cancers. The metabolic conversion of androgens is primarily catalyzed by two major enzyme classes: cytochrome P450 monooxygenases and NAD(P)H-dependent hydroxysteroid dehydrogenases (HSDHs). While endogenous androgen biosynthesis occurs predominantly in the testes, ovaries, adrenal glands, and placenta, emerging evidence suggests that host-associated microbial communities, including those residing in the urinary tract, may also contribute to local androgen metabolism. In our previous work, we identified the desG gene from the urinary tract isolate Propionimicrobium lymphophilum strain API-1, which encodes a novel NADPH-dependent 17β-HSDH. This enzyme catalyzes the bidirectional conversion between androstenedione (AD) and testosterone (T), suggesting a microbial route for androgen production within the urinary tract. In the present study, we expand on these findings by conducting detailed kinetic and substrate-specificity analyses of this enzyme, alongside bioconversion assays using a broad panel of steroid substrates. These results shed light on the steroid-transforming potential of urinary bacteria.

Importance:

This work represents an important advance in the understanding of androgen metabolism by urinary tract bacteria through the characterization of an NADPH-dependent 17β-HSDH encoded by the desG gene. By elucidating kinetic properties, substrate specificity, and bioconversion capabilities of recombinant DesG, this research provides valuable insights into steroid hormone regulation.

1. Introduction

Steroid hormones serve fundamental roles in the regulation of metabolism, immune function and inflammation, salt and water balance, sexual development and differentiation [1–3]. Steroid hormones are produced from cholesterol through de novo steroidogenesis pathways in the adrenal cortex, gonads, and the placenta [2–4]. The conversion of cholesterol to diverse steroid hormones utilizes two functional classes of enzyme: cytochrome P450 monooxygenases and NAD(P)H-dependent hydroxysteroid dehydrogenases (HSDHs) [3–5]. The anatomical separation of cells expressing distinct steroidogenic enzymes in the adrenal cortex yields distinct classes of corticosteroids. The outer zone, the zona glomerulosa, produces mineralocorticoids such as aldosterone. The middle zone, the zona fasciculata, synthesizes cortisol. The inner zone, the zona reticularis, produces androgen precursors such as dehydroepiandrosterone (DHEA) and androstenedione (AD), and also converts AD to 11β-hydroxyandrostenedione (11β-OHAD) [3–5]. Adrenal steroids are taken up into peripheral target tissues from circulation. Host enzymes convert these precursor steroids into active nuclear receptor ligands (e.g., androgen receptor, estrogen receptor), convert them back to inactive precursors after receptor agonism, and release them back into circulation in a process described by the concept of “intracrinology” [6,7].

A new paradigm is emerging from the rapidly advancing field of microbiome science in which the metabolism of host steroids by host-associated microbial communities contributes to the composition of the host steroid metabolome [8–10]. Many of the peripheral reactions in host tissues, such as oxido-reductive metabolism of hydroxyl groups at C3 and C20, and the potential for 5β-H (A/B cis) ↔ 3-oxo-Δ⁴ ↔ 5α-H (A/B trans) epimerization, are also catalyzed by host-associated bacterial enzymes [8,11]. Yet, bacteria have the unique capacity to generate 11β-OHAD by removing the side-chain from cortisol in a reaction known as steroid-17,20-desmolase [12]. Indeed, this microbial function was first observed in the middle of the 20th century by British clinician scientists who measured urinary 17-ketosteroids after rectal infusion of cortisol in the treatment of ulcerative colitis [13,14]. The first isolates of fecal bacteria capable of steroid-17,20-desmolase and accessory NADH-dependent steroid 20μ-HSDH [15] or 20β-HSDH activities [16] were obtained and initially characterized in the 1980’s. In 2013, a gene cluster was identified in Clostridium scindens ATCC 35704 that encodes NADH-dependent steroid 20μ-HSDH (desC) and predicted to encode steroid-17,20-desmolase (desAB) [12]. The identification of desAB genes in C. scindens ATCC 35704 allowed subsequent identification of desAB genes in Clostridium cadavaris and Butyricicoccus desmolans whose genome sequences had recently been sequenced, and had previously been shown to have both steroid-17,20-desmolase and steroid 20β-HSDH activities [16]. We identified the gene encoding steroid 20β-HSDH (desE) from B. desmolans [17] and determined the structure of DesE from Bifidobacterium adolescentis [18]. During these studies, phylogenetic analysis of DesE led to the identification of the urinary tract bacterium, Propionimicrobium lymphophilum ACS-093-V-SCH encoding desE as part of the desAB gene cluster [17,19]. This provided the unexpected insight that bacteria inhabiting the urinary tract may metabolize host steroids; urine being a major source of steroid excretion from the body [4].

We then confirmed that P. lymphophilum ACS-093-V-SCH expressed both NADH-dependent 20β-HSDH and steroid-17,20-desmolase activity [19]. Most recently, we isolated urinary tract P. lymphophilum strains from human prostatectomy tissue (Fig. 1A) and urine that converted 11-deoxycortisol to androstenedione as well as testosterone, indicating the presence of steroid 17β-HSDH activity in some strains of P. lymphophilum (Fig. 1B) [9]. We recently identified the desG gene encoding a novel NADPH-dependent 17β-HSDH in P. lymphophilum strain API-1 and confirmed bidirectional conversion between AD and T [9]. These studies have established that urinary tract bacteria are capable not only of metabolizing host steroids, but also of changing the class of steroid.

Fig. 1.

Propionimicrobium lymphophilum encodes a pathway for androgen formation from glucocorticoid precursors. A. Propionimicrobium lymphophilum strain API-1 isolated from human prostatectomy tissue. B. Proposed testosterone biosynthesis pathway in P. lymphophilum. DesAB catalyzes the conversion of 11-deoxycortisol to androstenedione. Recombinant DesG (rDesG), a 17β-hydroxysteroid dehydrogenase (17β-HSDH), catalyzes the NADPH-dependent reduction of androstenedione to testosterone. C. SEM image of P. lymphophilum.

In this report, we present enzyme kinetic and substrate-specificity analyses of recombinant DesG (rDesG). We demonstrate broad substrate-specificity as well as substrate recognition beyond that of human 17β-HSDHs.

2. Materials and methods

2.1. Bacterial strains and materials

Genomic DNA was isolated from P. lymphophilum API-1, amplification and cloning of the desG gene in the pET-51b(+) vector have been described in our previous study [9]. The pET51_desG was transformed into E. coli BL21 CodonPlus (DE3) RIPL chemically competent cells, as detailed in our earlier work [9]. For the current study, the E. coli BL21 cells were derived from in-house 50% glycerol stock cultures and cultured overnight in aerobically in LB medium supplemented with ampicillin (100 μg/mL) at 37°C.

3. Gene expression and protein purification by FPLC

E. coli BL21(DE3) pET51_desG cells were inoculated into LB broth supplemented with ampicillin (100 μg/mL) and incubated aerobically at 37°C with shaking for 6 h. The culture was then transferred to fresh LB broth (1 liter), also supplemented with ampicillin (100 μg/mL) and aerated at 220 rpm at 37°C for an additional 2 h. rDesG expression was induced by adding IPTG to a final concentration of 0.1 mM, and the temperature was reduced to 25°C for 16 h. Recombinant DesG bearing an N-terminal Strep-tag was purified using ÄKTAxpress FPLC system (Cytiva, USA). Cell pellets were harvested by centrifugation and resuspended in strep equilibration buffer (100 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, pH 8.0) containing 25 μL of β-mercaptoethanol, 250 units of benzonase (25–29 units per μL), and 15 mg of lysozyme (100 mg/ mL), and lysed by homogenization. The lysate was clarified by centrifugation at 12,000 rpm for 45 min at 4°C. The supernatant was filtered and applied to a 1 mL StrepTrap XT column (Cytiva, Marlborough, MA, USA) pre-equilibrated with 10 column volumes of equilibration buffer. Protein was loaded at a flow rate of 1.0 mL/min. After washing with 20 column volumes of equilibration buffer, rDesG was eluted with an elution buffer containing 20 mM biotin. Eluted fractions (2 mL) were collected and analyzed by 12% SDS-PAGE, with visualization of the bands using Coomassie Brilliant Blue G-250 staining. Protein concentrations were determined using the Thermo Scientific^™ NanoDrop^™ 2000/2000c Spectrophotometer based on the protein extinction coefficients and molecular weights (10220/28.9 w/strep tag). Fractions containing rDesG were pooled for further purification.

Buffer exchange with anion exchange buffer (50 mM Tris-HCl, 50 mM NaCl, pH 8.0) was performed using a HiPrep 26/10 desalting column (26 mm x 100 mm; Cytiva, Marlborough, MA, USA). The pooled sample was subsequently loaded onto a 5 mL HiTrap Q HP column (Cytiva, Marlborough, MA, USA) pre-equilibrated with anion exchange equilibration buffer. The column was washed with 10 column volumes of the same buffer, and the bound rDesG protein was eluted using step elution with buffer containing 50 mM Tris-HCl, 500 mM NaCl, pH 8.0. 1 mL fractions were collected, and the purity of rDesG was assessed by SDS-PAGE. Fractions containing purified rDesG were pooled and concentrated for downstream analyses.

Determination of recombinant DesG subunit and native molecular weight:

To assess the oligomeric state of rDesG, gel filtration chromatography was performed using a Superdex 200 Increase 10/300 GL column (Cytiva, Marlborough, MA, USA) equilibrated with 50 mM Tris-HCl, 150 mM NaCl, pH 8.0. A total of 250 μL of rDesG at concentrations of 2,3, and 4 mg/mL was injected. Elution profiles were recorded, and retention volumes were analyzed. Gel filtration standards (Bio-Rad Gel Filtration Standard) were run under the same conditions. A calibration curve was generated by plotting log molecular weight against elution volume. The molecular weight of rDesG was estimated by interpolation from the standard curve to infer its oligomeric state.

Determination of pH optima:

To determine the pH optima of rDesG in both the oxidative and reductive directions, a range of buffers containing 100 mM NaCl and various buffering agents were employed: 50 mM Na acetate (pH 5.0–5.5), 50 mM sodium phosphate (pH 6–7.5), 50 mM Tris base (pH 8–9), and 50 mM glycine (pH 9.5–10.0). For testing the pH optimum in the reductive direction, a master mix consisted of a buffer (pH 5.0–10.0), 150 μM NADPH, and 150 μM androstenedione. For assays in the oxidative direction, a similar solution was prepared with buffer (pH 5.0–10.0), 150 μM NADP+ , and testosterone. The reaction mixtures were pipetted into a Nunc^™ MicroWell^™ 96-Well, Flat-Bottom Microplate along with 10 nM rDesG enzyme, resulting in a final volume of 200 μL per well. Absorbance was measured over time at 340 nm using the Agilent BioTek Epoch 2 Absorbance Microplate Spectrophotometer.

DesG kinetics and substrate-specificity assays:

Kinetic analysis was performed using the pH optima determined in the previous experiment using the extinction coefficient for NADPH (ε₃₄₀ nm 6220 M⁻¹cm⁻¹). In the oxidative direction, testosterone was used to assess the linearity of purified recombinant rDesG activity over time. Reactions were conducted aerobically, and absorbance was measured continuously at 340 nm for 10 min. All assays were performed under aerobic conditions at room temperature. The reaction was linear with respect to time and enzyme concentration under these conditions. Steroid substrates were dissolved in DMSO, and final reaction concentrations were 5% DMSO v/v. The reaction mixture consisted of a buffer containing 50 mM NaPi and 100 mM NaCl (pH 6), 150 μM NADP⁺, testosterone concentrations ranging from 5 μM to 300 μM, and 10 nM of rDesG. Background absorbance was subtracted using no enzyme control. Heat-inactivated enzyme controls did not display activity. In the reductive direction, androstenedione was used to evaluate rDesG activity under similar conditions. The reaction mixture contained a buffer consisting of 50 mM Tris base and 100 mM NaCl, 150 μM NADPH, androstenedione concentrations ranging from 5 μM to 300 μM, and 10 nM of rDesG.

Substrate specificity of rDesG was determined using spectrophotometric assays under the optimal pH conditions identified above. In the reductive direction (pH 6.0), reaction mixtures contained 50 mM sodium phosphate buffer with 100 mM NaCl, 150 μM NADPH, 150 μM steroid substrate, and 10 nM rDesG enzyme. The following substrates were tested: androstenedione (AD), 5α-androstenedione (5α-AD), 11β-hydroxyandrostenedione (11β-OHAD), 11-ketoandrostenedione (11KAD), 1,4-androstenedione (AT), dehydroepiandrosterone (DHEA), and estrone (E1).

In the oxidative direction (pH 9.0), reaction mixtures contained 50 mM Tris base with 100 mM NaCl, 150 μM NADP⁺, 150 μM steroid substrate, and 10 nM rDesG enzyme. The following substrates were tested: testosterone (T), dihydrotestosterone (DHT), 11β-hydrox-ytestosterone (11OHT), 11-ketotestosterone (11KT), androstenediol (5- AL), and 17β-estradiol (E2). Steroid substrates were dissolved in DMSO, and the final DMSO concentration in all reactions was 5% (v/v). Reactions were assembled in a Nunc^™ MicroWell^™ 96-well flat-bottom microplate to a final volume of 200 μL per well and initiated by the addition of enzyme. Absorbance at 340 nm was monitored for 10 min using an Agilent BioTek Epoch 2 microplate spectrophotometer. Under these conditions, product formation was linear with respect to time and enzyme concentration.

Relative activity was calculated from the initial rate (ΔA340/min) and normalized to androstenedione (reductive direction) or testosterone (oxidative direction), which were set to 100%. Standard deviation was calculated from initial velocity measurements prior to normalization and expressed as the percentage of relative activity. Data represent the mean ± SD from four independent experiments (n = 4).

DesG enzyme bioconversion assay:

Enzyme reactions were prepared using 50 μM of a variety of substrates, including androstenedione (AD), testosterone (T), 5α-androstenedione (5α-AD), dihydrotestosterone (DHT), 11β-hydroxyandrostenedione (11OHAD), 11β-hydroxytestosterone (11OHT), 11-ketoandrostenedione (11KAD), 11-ketotestosterone (11KT), 1,4-androstenedione (AT), epi-testosterone (Epi-T), and 5α-epitestosterone (5α-epi-T), along with 150 μM NADPH/NADP+ , 50 nM of rDesG enzyme, and phosphate-buffered saline (PBS) buffer in triplicate, for a final reaction volume of 1 mL. Reactions were allowed to proceed for 24 h before being terminated. Metabolites were extracted by adding two volumes of ethyl acetate and vortexing for 1 min. The organic phase was recovered, and the metabolites were dried under nitrogen gas. The residue was dissolved in 200 μL of LC-MS grade methanol and sent for analysis.

Liquid chromatography-mass spectrometry:

Liquid Chromatography-Mass Spectrometry (LC-MS) analysis was performed at the UIUC Mass Spectrometry Laboratory using a Waters Acquity UPLC system coupled with a Waters Synapt G2-Si ESI mass spectrometer. Chromatographic separation was achieved on a Waters Acquity UPLC BEH C18 column (2.1 mm × 50 mm, 1.7 μm particle size) at a flow rate of 0.5 mL/min. The mobile phases consisted of 95% water, 5% acetonitrile (ACN), and 0.1% formic acid (FA) for eluent A and 95% ACN, 5% water, and 0.1% FA for eluent B. The gradient program was as follows: 0.0–0.5 min, 100% A; 0.5–6.0 min, linear decrease to 30% A and 70% B; 6.0–7.0 min, linear decrease to 0% A and 100% B; 7.0–8.0 min, 100% B; 8.1–10.0 min, return to 100% A for re-equilibration. The injection volume was 0.2 μL. This method facilitated the separation and detection of analytes across a wide polarity range.

Proteomics:

Purified rDesG was submitted to the Proteomics Core Facility at the University of Illinois Urbana-Champaign. The protein sample was digested with Trypsin (Thermo Scientific) (1:50 w/w) overnight at 37 ° in 25 mM TEAB pH 8.0. Following digestion, samples were desalted with Stagetips and dried down in a Speedvac.

Peptide digests were analyzed using a Thermo UltiMate 3000 UHPLC system coupled to a high-resolution Thermo Q Exactive HF-X mass spectrometer. Peptides were separated by reversed-phase chromatography using a 25 cm Acclaim PepMap 100 C18 column maintained at 50°C with mobile phases of 0.1% formic acid (A) and 0.1% formic acid in 80% acetonitrile (B). A two-step linear gradient from 5% B to 35% B over the course of 45 min and 35% B to 50% B over 10 min was employed for peptide separation, followed by additional steps for column washing and equilibration. The MS was operated in a data- dependent manner in which precursor scans from 350 to 1500 m/z (120,000 resolution) were followed by higher-energy collisional dissociation (Stepped NCE of 30) of the 15 most abundant ions. MS2 scans were acquired at a resolution of 15,000 with a precursor isolation window of 1.2 m/z and a dynamic exclusion window of 60 s.

The raw chromatograms were processed with Mascot Distiller (vers 2.8.5) (Matrix Science) against Uniprot using Mascot Daemon (Vers 2.8) with the following settings: Report top hits (Auto), Decoy (Checked), Enzyme (Trypsin), Variable Mods (Acetyl Protein N-terminal, Oxidation M), Max missed cleavages (2), Monoisotopic, Peptide charge (2 +, 3 +), Peptide tol ± (10 ppm), MS/MS ions search (checked), MS/MS tol ±(0.4 Da), Target PSM FDR (no target), Instrument (ESI-Trap), Quantitation (None). Processing options were optimized for Orbitrap HFX. The resulting peak list files were batch submitted to Mascot (vers 2.8.3) by Mascot Daemon. The peak lists were searched against Uniprot E. coli K12 and the expressed protein sequence with and without the N-ter-minal tag.

The identified peptides from the Mascot search were examined for the presence and absence of both the unique N-terminal tag as well as the N-terminal peptide from the processed tag-free protein. The relative numbers of the identified N-termini were used to judge the relative proportions of the tagged and processed proteins.

4. Results

Overexpression, purification, and oligomeric composition of rDesG.

Our recent study established that P. lymphophilum strain API-1 converts cortisol to 11-hydroxy-testosterone, and that the desG gene encodes a novel 17β-HSDH enzyme [9]. To further investigate the enzymatic activity, desG was overexpressed in E. coli and purified using fast protein liquid chromatography (FPLC). Initial purification was performed by affinity chromatography on a Strep-Tactin column using step elution with 25 mM biotin. SDS-PAGE was performed following affinity purification to evaluate protein purity and confirm the expected molecular weight. The eluted protein was pooled and passed through a HiPrep 26/10 desalting column to exchange into the anion exchange buffer, eluting as a sharp peak followed by ion exchange chromatography using a 5 mL HiTrap Q HP column to increase the protein concentration (Supplementary Figure 1). In a final step, we further purified the recombinant DesG (3 mg/mL) by gel filtration chromatography. rDesG eluted across ten fractions (Fig. 4A; B). Each fraction was evaluated for enzymatic activity, confirming NADPH and NADP⁺-dependent oxidoreduction by rDesG using androstenedione or testosterone, respectively, as substrates (Fig. 4C). The recombinant strep-tagged DesG has a theoretical subunit molecular mass of 28.9 kDa according to Expasy Compute pI/Mw tool, and the subunit molecular mass of the strep-tagged recombinant DesG was determined to be 30.2 ± 0.4 kDa (n = 3 independent SDS-PAGE gels) (Fig. 2A). A single peak was observed during gel filtration (Supplementary Figure 3), and a single band was observed by SDS-PAGE (Fig. 2A). Following the calculation of the subunit molecular mass, proteomic analysis was performed to assess the relative abundance and sequence coverage of rDesG. Mass spectrometry analysis confirmed a subunit molecular mass of 28.91 kDa (Fig. 3A). The purified rDesG had an emPAI score of 3375, while the next most abundant protein emPAI score was 12.85, indicating that rDesG was approximately 260-fold more abundant than the next most abundant protein (Supplementary Table 1). LC-MS analysis also revealed 91% sequence coverage of rDesG, including confident identification of both the N- and C-termini (Supplementary Figure 2).

Fig. 4.

rDesG elutes as a single enzymatically active protein complex. A. Gel filtration chromatogram of purified rDesG obtained by FPLC, showing the elution profile monitored at 280 nm. B. SDS-PAGE analysis of representative gel filtration fractions, confirming the presence and purity of DesG. C. Enzyme activity of selected gel filtration fractions, tested with either androstenedione and NADPH or testosterone and NADP⁺ as substrate/cofactor pairs. D. Gel filtration standard curve generated using protein standards of known molecular weights, used to estimate the apparent molecular weight and oligomeric of rDesG.

Fig. 2.

Recombinant DesG is purified as a ~28.9 kDa protein and catalyzes androgen interconversion. A. SDS-PAGE analysis of purified rDesG overexpressed in E. coli BL21 and affinity-purified using a StrepTrap XT column. A single band corresponding to ~28.9 kDa was observed, consistent with the calculated theoretical molecular mass of rDesG. A total of 2 μg of protein was loaded. B. LC-MS analysis of enzyme bioconversion assays demonstrating the conversion of androstenedione and testosterone in the presence of the corresponding cofactors (NADPH or NADP⁺).

Fig. 3.

DesG exhibits substrate saturation kinetics and distinct pH optima for reductive and oxidative reactions. A. Enzyme activity of rDesG measured at varying concentrations of androstenedione and fitted to the Michaelis–Menten equation to determine apparent Km and Vmax. B. Enzyme activity of rDesG measured at varying concentrations of testosterone and fitted to the Michaelis–Menten equation to determine apparent Km and Vmax. Error bars represent the standard deviation from four independent experiments. C. pH optimization of rDesG activity in reductive (NADPH-dependent, red circles) and oxidative (NAD⁺-dependent, blue squares) directions. The pH optimum was 6 for the reductive reaction and 9 for the oxidative reaction.

Regarding the oligomeric form of rDesG, the rDesG eluted at a V_e = 13.19, corresponding to a native M_r = 111.59. With a subunit mass of 28.9 kDa, gel filtration chromatography indicates that the DesG is a homotetramer (3.86 subunits) (Supplementary Figure 3).

pH optimization of DesG:

To optimize the enzymatic activity of rDesG, we monitored the conversion of pyridine nucleotides at 340 nm across a range of buffer solutions with pH values from 5.0 to 10.0 (Fig. 3A). When rDesG activity was measured using AD as the substrate and NADPH as the cofactor across this pH range, the optimum was found to be pH 6.0. In contrast, the optimum pH in the oxidative direction, with T as the substrate and NADP+ as the cofactor, was determined to be between 9.0 and 10.0.

Kinetic analysis and substrate-specificity of rDesG:

Kinetic analysis was conducted with the optimum pH values identified in the previous experiment with saturating levels of the corresponding cofactors. In the reductive direction, NADPH was held constant at 150 μM, and 11 final concentrations of androstenedione were varied from 5 μM to 300 μM. rDesG exhibited a [S] vs. velocity curve that is consistent with Michaelis-Menten kinetics. The enzyme exhibited a V_max of 9.2 ± 0.7 μmoles•min⁻¹•mg⁻¹, a K_M of 105.8 ± 1.7 μM, and a turnover number (k_cat) of 251.2 ± 1.8 s⁻¹ (Fig. 3B; Table 1).

Table 1.

Steady-state kinetic constants for rDesG.

Substrate	Vmax (µmoles⋅min−1⋅mg−1)	KM (µM)	kcat (s−1)	kcat / KM (s−1⋅µM−1)
Androstenedione	9.20 ± 1.09	105.81 ± 2.88	251.19 ± 1.77	2.37 ± 0.04
Testosterone	15.88 ± 2.11	34.65 ± 1.90	433.50 ± 5.77	12.55 ± 0.71
NADPH	3.28 ± 0.20	13.33 ± 0.88	89.55 ± 0.55	6.34 ± 0.45
NADP+	4.53 ± 0.31	25.49 ± 0.97	123.65 ± 0.84	4.86 ± 0.19

In the oxidative direction, NADP⁺ was held constant at 150 μM, and 11 final concentrations of testosterone varied from 5 μM to 300 μM. The enzyme exhibited a V_max of 15.8 ± 2.1 μmoles• min⁻¹ mg⁻¹, a K_M of 34.6 ± 1.9 μM, and a turnover number (k_cat) of 433.5 ± 5.7 s⁻¹ (Fig. 3C; Table 1). The specificity constant of the enzyme (k_cat/K_M) was determined to be 2.4 ± 0.04 s⁻¹•μM⁻¹ in the reductive direction, and an order of magnitude larger in the oxidative direction at 12.6 ± 0.7 s⁻¹•μM⁻¹(Table 1). This preference for the oxidative direction is in line with our prior observation that a mixture of 11β-OHAD and 11β-OHT was formed following metabolism of cortisol by P. lymphophilum strains [9].

We also assessed the relative activity of derivatives of androstenedione based on the LC/MS results. In the reductive direction, androstenedione displayed the highest activity. The 5α-reduced derivative (5α-AD) resulted in a loss of a quarter (75.6 ± 2.64%) of the activity compared to androstenedione, and hydroxylation at C-11 (11β-OHAD) displayed about half (50.59 ± 3.64%) the activity. Activity against 11-KAD was slightly lower than that of 11β-hydroxyandrostenedione (46.1 ± 5.14%). Both DHEA and the side-chain cleavage product of the pharmaceutical glucocorticoid prednisone, 1,4-androstenedione, displayed 62 ± 10.9% relative activity. Estrone (3-hydroxyestra-1,3,5 (10)-trien-17-one) displayed the lowest relative activity at 6.97 ± 1.43% (Table 2).

Table 2.

Substrate specificity for rDesG.

In the oxidative direction, testosterone had the highest relative activity. Interestingly, 11β-OHT retained 94.93 ± 0.55% activity. Dihydrotestosterone exhibited 48.7 ± 2.6% relative activity, which was comparable to activity with 11-keto-testosterone (49.3 ± 7.1%) (Table 2). Androstenediol was found to be oxidized at 25.5 ± 7.24% of testosterone. Finally, 17β-estradiol showed the lowest relative activity at 9.8 ± 7.9% (Table 2). To confirm product formation in the substrate specificity assays, we performed LC/ESI-MS analysis (see Supplementary Figure 4).

4.1. Sequence similarity network analysis and multiple sequence alignment of DesG

Using the DesG amino acid sequence as a pBLAST query, we retrieved 921 sequences with 922 nodes and 422,615 edges using an alignment score threshold of 25 (-log₁₀25), resulting in one large network (Fig. 5A). As stringency thresholds were increased, isofunctional clusters and singletons began to separate from the main network (Fig. 5B). A stringency threshold of 45 resulted in DesG moving to the periphery of the main network, now making edges with only 7 other SDR proteins (Fig. 5C). Further increases in stringency resulted in DesG as a singleton. These results indicate that DesG does not share close sequence identity and lacks isofunctional clustering with its top 1000 BLAST hits.

Fig. 5.

DesG clusters with homologous proteins in sequence similarity network analysis. A. DesG amino acid sequence pBLAST query resulted in 921 sequences with 922 nodes and 422,615 edges using an alignment score threshold of 25 (-log₁₀25) B. Protein network at a stringency threshold of 45 (-log₁₀45) C. Species and protein ID of nodes shared with DesG at a threshold of 45. D. Gene organization of nodes shared with DesG at threshold of 45.

A multiple sequence alignment between DesG and characterized SDR family 17β-HSDHs from bacteria (Comamonas testosteroni), fungi (Cochliolobus lunatus), snail (Nucella lapillus), nematode (Caenorhabditis elegans), trematode (Schistosoma japonicum, Clonorchis sinensis), fishes (Salmo salar, Ictalurus punctatus), amphibian (Xenopus tropicalis), old world Monkey (Macaca fascicularis, Macaca mulatta), and ape (Homo Sapien) was performed (Fig. 6). While the DesG shared at most 30% amino acid identity (with C. testosteroni) and as low as 18.8% with S. scrofa, all the proteins shared a conserved cofactor binding site (TGxxxGxG) and the catalytic tetrad (N157, S185, Y198, K202 in H. sapien).

Fig. 6.

DesG shares conserved SDR catalytic and cofactor-binding motifs with known 17β-HSDHs. Multiple sequence alignment between DesG and characterized SDR family 17β-HSDHs from Comamonas testosteroni (C.testosteroniP19871), Cochliolobus lunatus (C.lunatus3IS3A), Nucella lapillus (N.lapillusAFV95595.1), Caenorhabditis elegans (C.elegansQ09517.2), Schistosoma japonicum (S.japonicumCAX70720), Clonorchis sinensis (C.sinensisGAA55022.1), Salmo salar (S.salarNP_001135118.1), Ictalurus punctatus (I.punctatusNP_001187), Xenopus tropicalis (X.tropicalisNP_00101), Macaca fascicularis (M.fascicularisAAY845), Macaca mulatta (M.mulat-taEHH22972.1), and Homo Sapien (H.sapiensNP_000188.1). *Denotes conserved pyridine nucleotide binding residues, ^Denotes conserved active site amino acids.

5. Discussion

Mammals synthesize cholesterol de novo from acetate, and all steroid hormones are synthesized from cholesterol by two classes of enzyme: HSDHs and cytochrome P450 monooxygenases [1]. Hydroxysteroid dehydrogenases are pyridine nucleotide-dependent stereo- and regio-selective, or specific enzymes in the short-chain dehydrogenase/reductase and aldo-keto reductase superfamilies that regulate the biological activity of steroids at the pre-receptor level [4]. 17β-HSDHs regulate the potency of estrogens and androgens; 17-keto forms are inactive, while 17β-forms are active agonists of [20]. To date, there are 12 isoforms of 17β-HSDHs identified in humans that vary in protein superfamily (AKR vs. SDR), intracellular location, tissue expression, substrate and cofactor-specificity, and directionality (i.e. oxidative vs. reductive) [20–22]. 17β-HSDHs are largely grouped into in vivo oxidative enzymes (17β-HSDH types 2, 4, 6, 8, 9, 10, 11, 14) which catalyze the NAD+ dependent inactivation of AR and ER ligands, and in vivo reductive enzymes (17β-HSDH types 1, 3, 5, 7) which are NADPH dependent and whose reactions lead to AR and ER receptor ligands [23]. Several 17β-HSDHs also display other steroid dehydrogenase (e.g. 3α-HSDH, 20α/β-HSDH activities), and other enzymes such as 11β-HSD type 1 also have reported 17β-HSDH activity [24].

17β-HSDH activity in microorganisms has been known for 80 years, but few have been characterized. In soil, bacteria encode 17β-HSDH as part of larger steroid degradation gene clusters allowing mineralization of sterols to CO₂ under aerobic conditions [33]. Under anaerobic conditions and rapid transit times characteristic of the mammalian GI tract, steroid degradation pathways are less complex and typically relegated to oxidation-reduction of sterol hydroxyl groups [25], dehydroxylation of sterols [10,26,27], side-chain cleavage [9,19,28], and A/B ring oxidation and epimerization [29–31]. These reactions are through to be important for maintaining the cellular redox potential but may also allow interkingdom communication through androgen and estrogen receptor signaling in the intestinal epithelium [32], and cells of the gut-associated lymphoid tissue, which comprise an estimated 70–80% of the innate and adaptive immune cells in the body [33].

The far reach of our host-associated microbiomes in both health and disease, development in utero and from adolescence to senescence has led many to regard our inhabitants as the “second human genome” [34, 35]. Determining the role of host associated microbiomes in steroid metabolism, particularly androgen formation, is of fundamental importance in furthering our understanding of the human endocrine system.

The conversion of glucocorticoids to androgens or their precursors by bacterial steroid-17,20-desmolase may be considered an “endocrine” function of the gut (and potentially the urinary) microbiome as bacterial metabolites are absorbed into portal circulation and enter target tissues. Likewise, bacteria forming biofilms along the GI epithelium and the prostatic urethra may function in a “paracrine” manner by taking luminal sterols and converting them to androgen-receptor ligands, affecting adjacent cells. We speculate that the K_M values reported here for DesG in the micromolar level may suggest absorption and accumulation of steroids in urinary tract biofilms. The K_M value is often thought of as an indicator of the cytosolic concentration and thus a reflection of the environmental concentration of the substrate. If we assume this to be the case, it may seem that a K_M in the micromolar range is high compared with the hundreds of nanomolar range for steroids in the urine [4]. However, we previously found evidence of biofilm formation in that the urinary isolates we have obtained clump together in liquid broth, and under scanning electron microscopy (SEM), we observed dense networks of what appear to be extracellular polysaccharides linking cells in vast webs [9,36]. Urinary commensals attach to the urinary epithelium to persist in this environment [37]. Biofilms on microplastics have been shown to enhance absorption of hydrophobic organic compounds compared to abiotic surfaces indicating that the biofilm matrix can retain lipophilic substances [38]. While further work is required, we speculate that with the right partitioning coefficient in the biofilm, and deconjugation of sulfate and glucuronide, accumulation to micromolar levels of DesG substrates in biofilm microenvironments is reasonable. Arylsulfatases that target steroids are common in host-associated microbiota [39], as are β-glucuronidases [40]. Whether androgen formation through the concerted effort of DesAB and DesG occurs in certain regions of the urinary tract, or is distributed throughout is currently unclear, but studies have shown anatomically distinct microbiomes in the urinary tract [41]. Additional work will be required to determine the role of bacterial androgen formation on host physiology and pathophysiology.

Microbial DesG has substrate recognition distinct from human 17β- HSDH isoforms in the ability of DesG to convert 11OHAD to 11OH-testosterone. In addition, the human isoforms of 17β-HSDH prefer 11keto-AD over AD, while the opposite is the case with DesG [42–44]. In P. lymphophilum strains harboring both desAB and desG, the endogenous inputs would be cortisol and cortisone, which are converted to 11OHAD and 11keto-AD, respectively, by DesAB [19,28]. Both end products become substrates for the conversion by DesG to 11OH-testosterone and 11keto-T, respectively. Despite the higher catalytic efficiency (k_cat/K_M) in the oxidative direction at alkaline pH from 9.0 to 10.0, under physiological conditions in bacterial cytosolic pH from 6.0 to 8.0 [45], the reductive direction is expected to be favored.

Of 22 strains of P. lymphophilum, most were isolated in our recent study from human male urine samples by selecting colonies capable of steroid biotransformation ability [9]. Of these, roughly half (12 of 20 strains) of desABE positive strains also harbor desG. This indicates that strain variation with respect to steroid biotransforming capacity is important and measuring des genes rather than phylogenetic markers (i. e. 16S rDNA sequencing) will be important in distinguishing androgen-forming strains from non-androgen-forming strains of P. lymphophilum, and other steroid biotransforming taxa.

Abbreviations:

HSDH

hydroxysteroid dehydrogenases

AD

androstenedione

T

testosterone

DHEA

dehydroepiandrosterone

11β-OHAD

11β-hydroxyan-drostenedione

5α-AD

5α-androstenedione

11OHT

11β-hydroxytestosterone

11KAD

11-ketoandrostenedione

11KT

11-ketotestosterone

Epi-T

epi-testosterone

5α-epi-T

and 5α-epitestosterone

AT

1,4-androstenedione

DHT

dihydrotestosterone

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jsbmb.2026.106996.

Data availability

Data will be made available on request.