The glucocorticoid-activating enzyme 11β-hydroxysteroid dehydrogenase type 1 catalyzes the activation of testosterone

Testosterone biosynthesis from its precursor androstenedione is thought to be exclusively catalysed by the 17  -hydroxysteroid dehydrogenases — HSD17B3 in testes, and AKR1C3 in the ovary, adrenal and peripheral tissues. Here we show for the first time that the glucocorticoid activating enzyme 11  - hydroxysteroid dehydrogenase type 1 (HSD11B1) can also catalyse the 17  -reduction of androstenedione to testosterone, using a combination of in vitro enzyme kinetic assays, mathematical modelling, and molecular docking analysis. Furthermore, we show that co-expression of HSD11B1 and AKR1C3 increases testosterone production several-fold compared to the rate observed with AKR1C3 only, and that HSD11B1 is likely to contribute significantly to testosterone production in peripheral tissues.


Introduction
The enzyme 11β-hydroxysteroid dehydrogenase type 1 (HSD11B1) is predominantly expressed in glucocorticoid target tissue, such as liver and adipose, where it catalyzes systemic and local glucocorticoid activation [1,2].This reaction involves the reduction of the 11-keto group in cortisone to produce an 11β-hydroxyl group in the active glucocorticoid, cortisol.The NADPH co-substrate required for this reductase activity is provided by the co-localized expression of hexose-6-phosphate dehydrogenase (H6PDH), which catalyzes the first two steps of the pentose-phosphate pathway and maintains a high local NADPH/NADP + ratio within the lumen of the endoplasmic reticulum (ER) [2][3][4][5][6].
In addition to its role in glucocorticoid metabolism, HSD11B1 has more recently been shown to play an essential role in the peripheral metabolism of 11-oxygenated androgens [7,8].Like the glucocorticoids cortisol and cortisone, 11-oxygenated androgens contain either an 11β-hydroxyl or 11-keto moiety, which makes them distinct from the classic androgens.As a result, the 11β-hydroxyl and 11-keto forms can be interconverted by HSD11B1 and 11β-hydroxysteroid dehydrogenase type 2 (HSD11B2) [3,9].Notably, while HSD11B1 catalyzes the activation of glucocorticoids, it catalyzes the inactivation of 11-oxygenated androgens as the 11β-hydroxyl forms are less androgenic than their 11-keto counterparts [9,10] (Figure 1A).Indeed, co-expression of HSD11B1 with the androgen activating enzyme aldo-keto reductase 1C3 (AKR1C3, also known as HSD17B5) in adipose tissue prevents the accumulation of the potent 11oxygenated androgen 11-ketotestosterone (11KT) and therefore protects against local androgen excess [7,8].

J o u r n a l P r e -p r o o f
To date, and to the best of our knowledge, HSD11B1 has never been shown to play a role in the metabolism of classic androgens.While investigating the interplay of HSD11B1 and AKR1C3 and their effect on 11-oxygenated androgen metabolism [8], we serendipitously observed an HSD11B1 mediated conversion of the classic androgen pathway precursor androstenedione (A4) to testosterone, suggesting that HSD11B1 may also act as a 17-hydroxysteroid dehydrogenase.Here we comprehensively investigated this putative activity and confirm that HSD11B1 catalyses the activation of A4 to testosterone.

Materials and methods
Experiments were performed in HEK293 cells and the human Simpson-Golabi-Behmel syndrome (SGBS) preadipocyte cell line as previously described [11].Steroid quantification was performed using an established UHPLC-MS/MS method [12].The limits of quantification for all steroid analytes are listed in Supplementary Table 1 and provide proof of the analytical validity for all kinetic experiments.UHPSFC-MS/MS was employed to separate testosterone from epitestosterone to confirm the stereochemistry of the 17β-reductase activity.Kinetic characterisation and the construction of a computational model was performed as previously described [8].The binding orientation of A4 to HSD11B1 was assessed by docking the substrate to the crystal structure of HSD11B1 (PDB ID: 2BEL) using Schrödinger Maestro software (Version 13.1).All detailed methods are provided in the Supplementary Material.

HSD11B1 catalyses the 17β-reduction of androstenedione
Non-steroidogenic HEK293 cells transfected to co-express HSD11B1 and H6PDH catalysed the conversion of 10 nM A4 to 2.99 nM testosterone during a 24-hour incubation period (30% conversion) (Figure 1B).In comparison, the known 17β-hydroxysteroid dehydrogenases, AKR1C3 and 17β-hydroxysteroid dehydrogenase type 3 (HSD17B3) yielded 25% and 100% conversion of A4 to testosterone over the same period (Figure 1B), which is in line with their known 17β-hydroxysteroid dehydrogenase activities [9,13].UHPSFC-MS/MS analysis of the product confirmed that the conversion of A4 by HSD11B1 generated testosterone but not the 17α-isomer epitestosterone (Figure 1D & E).Characterisation of HSD11B1 activity towards A4 yielded an apparent K m (K m, app ) and V max (V max, app ) values of 4.2 µM and 0.04 µM/h, respectively (Supplementary Figure 1).No conversion of A4 to testosterone was observed in HEK293 cells transfected with the empty plasmid vector (negative), thereby confirming that conversion was dependent J o u r n a l P r e -p r o o f on the expression of HSD11B1 (Figure 1B).Notably, joint incubation with A4 and the physiologically more abundant glucocorticoid substrate, cortisone, resulted in only a modest reduction in 17β-reductase activity.However, cortisone itself was metabolised by HSD11B1, thereby reducing its concentration over time and potentially reducing any competitive inhibitory effects (Supplementary Figure 2).While AKR1C3 and HSD17B3 also catalysed the conversion of 11-ketoandrostenedione (11KA4) to 11KT as previously reported [9], HSD11B1 converted 11KA4 to 11β-hydroxyandrostenedione (11OHA4) only (Figure 1C).
Taken together, these data show that HSD11B1 functions as an 11β-hydroxysteroid steroid dehydrogenase towards 11KA4 which contains an 11-keto group, but catalyses the 17β-reduction of A4, which lacks this moiety.

J o u r n a l P r e -p r o o f
This finding was confirmed by molecular docking simulations which showed that 11KA4 docked in a similar orientation to the glucocorticoid substrate cortisone (Figure 2A&B).Both substrates docked with the C11 keto group facing the hydroxyl groups of the catalytic residues Ser170 and Tyr183, while NADPH was positioned behind the C11 keto group thus leading to the formation of a 11β-hydroxyl group in each substrate after catalysis [7,14,15].Conversely, A4 docked to HSD11B1 in an orientation that was rotated 77° counter-clockwise relative to the binding orientation of cortisone (Figure 2C).Notably, the C17 keto group was positioned facing the hydroxyl groups of the catalytic residues, Ser170 and Tyr183, with the C4 carbon of the NADPH positioned behind the C17 keto group of A4.This conformation supports the observation that catalysis result in a 17β-conformation, while the higher docking score (-6.337) is in agreement with a 4-fold higher K m, app than for the 11-keto containing substrates (docking scores: cortisone, -7.978; 11KA4, -8.043).Considering the observed 17β-reductase activity towards A4, we also considered that HSD11B1 might show similar activity towards additional classic androgen precursors, such as 5α-androstanedione (5α-dione) and androsterone (AST).However, only low conversion of 5α-dione to its 17β-reduced product, 5α-dihydrotestosterone (DHT) was observed (Supplementary Figure 3A), while no conversion of AST to the expected product, 5α-androstane-3α,17β-diol (3α-adiol) could be detected by UHPLC-MS/MS (data not shown).Molecular docking again supported this finding as 5α-dione was found to dock in a similar orientation to A4, but with a 1.2-fold and 1.8-fold greater distance between the 17-keto group of the substrate and the two catalytic hydroxyl groups of Ser170 and Tyr183 in agreement with lower catalytic activity towards this substrate (Supplementary Figure 3C).Docking of AST yielded an orientation in which J o u r n a l P r e -p r o o f the distance between the 17-keto group of AST and catalytic hydroxyl group of Tyr183 was greater than 5 Å, while the distance between the C4 carbon of NADPH and the C17 carbon was greater than 4 Å thus offering a feasible explanation for the lack of observed conversion to 3α-adiol (Supplementary Figure 3D).

Both AKR1C3 and HSD11B1 contribute to testosterone biosynthesis in cells co-expressing both enzymes
Given that HSD11B1 is co-expressed with the androgen activating enzyme AKR1C3 in tissues such as adipose [16], we investigated if HSD11B1 would contribute towards testosterone biosynthesis in cells coexpressing HSD11B1 and AKR1C3.Increasing the relative expression ratio of HSD11B1 to AKR1C3 using a previously validated protocol [8] resulted in a corresponding significant increase in testosterone production (Figure 3A).Indeed, we observed a 3-fold increase in testosterone produced between the lowest (0.05:1) and highest (2:1) expression ratios tested (from 7.8 nM to 22.8 nM), demonstrating that both enzymes were contributing to testosterone biosynthesis.Integration of the contribution of the two enzymes, using a computational model based on the enzyme kinetic parameters determined for the individual enzymes [8,17] resulted in an accurate prediction of the increased testosterone production with increasing HSD11B1:AKR1C3 ratios (Figure 3A).
To further confirm the contribution of both HSD11B1 and AKR1C3 to testosterone biosynthesis in cells expressing both enzymes, we treated cells expressing HSD11B1 and AKR1C3 (1:1 ratio) with A4 in the absence or presence of the HSD11B1 and AKR1C3 inhibitors, carbenoxolone (CBX) and indomethacin (INDO), respectively (Figure 3B).Both CBX and INDO individually reduced the production of testosterone 2-fold.Only when both inhibitors were used together was testosterone production completely abolished thus demonstrating that both HSD11B1 and AKR1C3 contribute to testosterone biosynthesis when co-expressed.Control reactions with 11KA4 as substrate confirmed the specificity of the inhibitors (Supplementary Figure 4).Similarly, we found that the selective HSD11B1 inhibitor AZD4017 [18] significantly reduced the conversion of A4 to testosterone in differentiated SGBS adipocytes that endogenously express HSD11B1 and AKR1C3 [7,19] (Figure 3C), thereby confirming the contribution of HSD11B1 to the 17β-reduction of A4. 11KA4, on the other hand, was predominantly converted to 11OHA4 (92.1 nM) in the absence of AZD4017, while the inhibition of HSD11B1 resulted in 11KT (22.1 nM) as the primary product (Figure 3D) as previously reported [14].

Discussion
The role of HSD11B1 in modulating stress response, blood pressure and energy metabolism by regulating local glucocorticoid bioavailability in glucocorticoid target tissue is well established [2].HSD11B1 is however a promiscuous enzyme with broad substrate specificity.In addition to its role in glucocorticoid activation, which includes exogenous synthetic glucocorticoids such as prednisolone and betamethasone [20], HSD11B1 has also been shown to detoxify xenobiotics such as metyrapone and 4-nitrobenzaldehyde [21], to convert the secondary bile acid 7-oxo-lithocholic acid to chenodeoxycholic acid [22], and to catalyse the 7-reduction of 7-ketocholesterol [2,23,24].HSD11B1 also plays an important role in the inactivation of 11-oxygenated androgens [3].Notably, while investigating the interplay between HSD11B1 and AKR1C3 and their effect on 11-oxygenated androgen metabolism, we observed that HSD11B1 appeared to contribute towards the biosynthesis of testosterone from its precursor A4, thus suggesting an additional 17β-hydroxysteroid dehydrogenase functionality.Here we have provided evidence for an extension of the repertoire of reactions catalysed by HSD11B1 by showing that it catalyses the 17β-reduction of the classic C19 androgen precursor, A4.This enzymatic activity has previously only been reported for the reductive 17β-hydroxysteroid dehydrogenase enzymes, which include AKR1C3 (also known as HSD17B5) and HSD17B3 [13,25].Interestingly, while HSD17B3 is predominantly expressed in the testes, AKR1C3 is expressed in peripheral tissue such as adipose, which co-expresses HSD11B1 [26,27].
We have previously shown that this co-expression of HSD11B1 and AKR1C3 modulates the biosynthesis of the active 11-oxygenated androgen 11KT, thus preventing accumulation of this potent androgen and protecting against androgen excess [8].Notably, here we show that co-expression of HSD11B1 and AKR1C3 leads to increased local testosterone biosynthesis and that both enzymes need to be inhibited to abolish the activation of A4 to testosterone.Interestingly, neither enzyme is particularly efficient in catalysing the 17-reduction of A4.Indeed, we have previously shown that 11-oxygenated androgens containing an 11-keto moiety are in fact catalysed significantly more efficiently by AKR1C3 than the corresponding classic androgen precursors [7,17].The HSD17B3-mediated 17-reduction of A4 is also substantially more efficient than the AKR1C3-catalysed reaction [9,13].Similarly, here we show that the 17-reduction of A4 catalysed by HSD11B1 is substantially less efficient than its 11-reductase activity which has previously been characterised in detail [3,8].The docking analyses showed that A4 needs to bind in an orientation which is rotated 77˚ relative to cortisone and 11KA4 in order for the 17-keto group to be positioned opposite the catalytic residues, Ser170 and Tyr183, and this binding is less energetically favourable than that of cortisone or 11KA4, which likely contributes to the lower observed enzymatic J o u r n a l P r e -p r o o f efficiency.Notably, it is well known that the 11-reductase activity of HSD11B1 is reliant on the co-expression of H6PDH for the production of co-substrate NADPH [2].Our docking studies demonstrate that the 17-reductase activity towards A4 almost certainly makes use of the same catalytic mechanism.
All our catalytic analyses were performed in HEK293 cells which we co-transfected to express both HSD11B1 and H6PDH.
Our finding may be particularly pertinent to women where activation of androgens relies greatly on peripheral metabolism [28][29][30].In particular, following the cessation of ovarian steroidogenesis at menopause, adrenal-derived androgen precursors become the primary source of sex steroids [31,32] and require peripheral activation [33], where HSD11B1 may feed into the active androgen pool.In contrast, men are substantially less reliant on peripheral androgen activation due to high levels of HSD17B3-mediated testosterone biosynthesis by the testes [13].
Clinical studies with HSD11B1 inhibitors cannot be used to investigate the contribution of HSD11B1 to peripheral testosterone biosynthesis as the reduction of cortisol activation results in reduced negative feedback on the Hypothalamic-pituitary-adrenal axis, increased adrenocorticotropin hormone and increased adrenal A4 biosynthesis resulting in increased peripheral AKR1C3-mediated testosterone production.As such, testosterone levels tend to be unaffected or increased following HSD11B1 inhibition in women [34,35].A more appropriate way of determining the contribution of HSD11B1 to circulating testosterone concentrations may be by the inhibition of AKR1C3.A recent study by Gashaw et al. investigated the effect of AKR1C3 inhibition in pre-and post-menopausal women.AKR1C3 inhibition was found to lead to increased AST levels in both pre-and post-menopausal women, indicative of a buildup of androgen precursors.However, no significant decrease in testosterone levels were observed in either groups [36].Assuming relatively robust inhibition of AKR1C3, these findings therefore support possible testosterone production via an AKR1C3-independent mechanism and based on our findings we propose this to be mediated by HSD11B1.Given the relatively similar contributions of HSD11B1 and AKR1C3 towards testosterone biosynthesis when co-expressed, the contribution of HSD11B1 is likely dependent on its relative expression level to AKR1C3.It is well established that HSD11B1 expression increases significantly with age therefore further suggesting a potential role for HSB11B1 in peripheral androgen activation in postmenopausal women [37][38][39].Notably, the HSD11B1-mediated activation of A4 may also be more pronounced in conditions associated with increased adrenal A4 output, such as polycystic ovary syndrome (PCOS) or congenital adrenal hyperplasia (CAH).The role in CAH may be of particular interest J o u r n a l P r e -p r o o f

Figure 1 .
Figure 1.HSD11B1 catalyzes the conversion of androstenedione to testosterone.(A) Schematic showing the reactions catalyzed by AKR1C3 and HSD11B1.Conversion of (B) 10 nM androstenedione and (C) 10 nM 11KA4 by HEK293 cells expressing AKR1C3, HSD17B3 or HSD11B1 after 24 hours.Steroid concentrations are shown as mean ±SEM of three independent experiments.A paired t test was used to compare the product formation of each enzyme to that of the negative control (*p<0.05,**p<0.005,and ***p<0.001).(D) Separation of testosterone and epitestosterone by UHPSFC-MS/MS.(E) UHPSFC-MS/MS chromatogram showing the production of testosterone and not epitestosterone by HSD11B1.

Figure 2 .
Figure 2. Molecular docking shows that androstenedione binds to HSD11B1 in an orientation that allows for 17-reduction.Cortisone (A), 11KA4 (B) and androstenedione (C) were docked to the crystal structure of HSD11B1 (PDB ID: 2BEL) using Schrödinger Maestro software (Version 13.1).Wolfram Mathematica was used for visualisation of the PDB file created by the Schrödinger Maestro software.

Figure 3 .
Figure 3.Both AKR1C3 and HSD11B1 contribute to the conversion of androstenedione to testosterone in cells expressing both enzymes.(A) Increased HSD11B1:AKR1C3 ratios lead to the increased conversion of androstenedione to testosterone.The conversion of 100 nM androstenedione to testosterone by increasing ratios of HSD11B1:AKR1C3 expressed in HEK293 cells after 24 hours.The experimental data from the ratio experiments are shown as points with error bars, while the computational models' predictions are shown with the solid connecting lines.(B) Both AKR1C3 and HSD11B1 need to be inhibited to abolish testosterone production in HEK293 cells expressing both enzymes.Conversion of 100 nM androstenedione to testosterone after 24 hours is shown in the absence and presence of CBX and/or INDO.A Tukey's multiple comparison paired t-test was performed for statistical analysis.Letters a, b and c represent significant differences between treatments (p<0.05), while values which do not differ significantly are assigned the same letter.Control reactions are shown in Supplementary Figure 4. (C) Inhibition of HSD11B1 endogenously expressed in SGBS cells by the selective inhibitor AZD4017 leads to a significant reduction in testosterone production from androstenedione (100 nM).(D) Inhibition of HSD11B1 by AZD4017 in SGBS cells is confirmed by the inhibition of 11OHA4 and 11β-hydroxytestosterone (11OHT) production by HSD11B1, while AKR1C3 activity is unaffected resulting in 11KT production.A paired t test was performed for statistical analysis (*p<0.05, and ***p<0.001).All data is shown as mean ±SEM from three independent experiments.J o u r n a l P r e -p r o o f