Research article

Hepatic reduction of the secondary bile acid 7-oxolithocholic acid is mediated by 11β-hydroxysteroid dehydrogenase 1

Alex Odermatt, Thierry Da Cunha, Carlos A. Penno, Charlie Chandsawangbhuwana, Christian Reichert, Armin Wolf, Min Dong, Michael E. Baker


The oxidized bile acid 7-oxoLCA (7-oxolithocholic acid), formed primarily by gut micro-organisms, is reduced in human liver to CDCA (chenodeoxycholic acid) and, to a lesser extent, UDCA (ursodeoxycholic acid). The enzyme(s) responsible remained unknown. Using human liver microsomes, we observed enhanced 7-oxoLCA reduction in the presence of detergent. The reaction was dependent on NADPH and stimulated by glucose 6-phosphate, suggesting localization of the enzyme in the ER (endoplasmic reticulum) and dependence on NADPH-generating H6PDH (hexose-6-phosphate dehydrogenase). Using recombinant human 11β-HSD1 (11β-hydroxysteroid dehydrogenase 1), we demonstrate efficient conversion of 7-oxoLCA into CDCA and, to a lesser extent, UDCA. Unlike the reversible metabolism of glucocorticoids, 11β-HSD1 mediated solely 7-oxo reduction of 7-oxoLCA and its taurine and glycine conjugates. Furthermore, we investigated the interference of bile acids with 11β-HSD1-dependent interconversion of glucocorticoids. 7-OxoLCA and its conjugates preferentially inhibited cortisone reduction, and CDCA and its conjugates inhibited cortisol oxidation. Three-dimensional modelling provided an explanation for the binding mode and selectivity of the bile acids studied. The results reveal that 11β-HSD1 is responsible for 7-oxoLCA reduction in humans, providing a further link between hepatic glucocorticoid activation and bile acid metabolism. These findings also suggest the need for animal and clinical studies to explore whether inhibition of 11β-HSD1 to reduce cortisol levels would also lead to an accumulation of 7-oxoLCA, thereby potentially affecting bile acid-mediated functions.

  • bile acid
  • glucocorticoid
  • 11β-hydroxysteroid dehydrogenase
  • liver
  • metabolism
  • 7-oxolithocholic acid


Bile acids play an essential role in the processing and uptake of dietary lipids and fat-soluble vitamins, and in the elimination of cholesterol and toxic lipophilic compounds from the body. Impaired regulation of the composition and concentration of bile acids and bile salts has been associated with hepatobiliary and digestive diseases [1]. Thus it is important to identify the proteins involved in the maintenance of bile acid homoeostasis.

Bile acids are synthesized from cholesterol by CYP (cytochrome P450)-mediated oxidative type I biotransformation reactions. In addition to CYP enzymes, oxidoreductases and peroxisomal oxidases are involved in bile acid synthesis [2]. The major bile acids present in human bile are CDCA (chenodeoxycholic acid) (35–50%), its 12-hydroxylated derivative CA (cholic acid) (30–45%) and DCA (deoxycholic acid) (10–20%), the bacterial 7-deoxy metabolite of CA (for structures, see Figure 1) [3,4]. Several type II biotransformation reactions of bile acids occur in the liver. Unconjugated bile acids, either newly synthesized or reaching the liver after bacterial deconjugation via the enterohepatic circulation [5], are subjected to reamidation with taurine and glycine, and, to a lesser extent, by sulfation and glucuronidation. In some species, the liver catalyses the CYP-mediated rehydroxylation of the secondary bile acids DCA and LCA (lithocholic acid), which are formed from CA and CDCA through 7-dehydroxylation by bacterial enzymes in the colon during the enterohepatic circulation [5]. Furthermore, hepatic enzymes convert iso- or 3-oxo bile acids into the preferred 3α-hydroxy derivatives and 7-oxo bile acids into the 7α- and 7β-hydroxy forms [6]. Whereas several hydroxylating CYPs and conjugating liver enzymes have been identified and characterized, the enzyme(s) involved in the hepatic oxidoreduction of 7-oxo bile acids remained unknown.

Figure 1 Structures of bile acids

In the human colon, several bacterial strains, including Escherichia coli, Bacteroides fragilis and Bacteroides intestinalis, express 7α-HSD (7α-hydroxysteroid dehydrogenase) enzymes that generate 7-oxoDCA (7-oxodeoxycholic acid) from CA, and 7-oxoLCA (7-oxolithocholic acid) from CDCA and UDCA (ursodeoxycholic acid) [7,8]. The gut microbiota also contains hydroxysteroid dehydrogenases that catalyse the epimerization of 7α- to 7β-hydroxy bile acids with the generation of a stable oxo-bile acid intermediate [911], thus contributing to the formation of the UDCA found in bile and faeces. The secondary bile acids DCA and LCA, as well as the 7-oxo bile acids 7-oxoDCA and 7-oxoLCA, are eliminated by the faeces. However, a substantial fraction of these secondary bile acids is not excreted, but reabsorbed in the distal intestine and transported back to the liver.

Early studies on the metabolism of radiolabelled 7-oxoLCA in rats with bile fistulas indicated the preferential formation of UDCA and lower amounts of CDCA and its metabolites [12]. Later, Fromm et al. [13] reported the preferential conversion of radiolabelled 7-oxoLCA into CDCA, with approximately 10% UDCA after a single hepatic passage following i.v. (intravenous) administration in humans. Similarly, 7-oxoLC-Gly (7-oxolithocholylglycine) and 7-oxoLC-Tau (7-oxolithocholyltaurine) were converted into the 7α-hydroxy epimer. After small intestinal infusion, 7-oxoLCA was metabolized primarily to CDCA as observed after i.v. injection. CDCA and UDCA were not metabolized by the liver, suggesting that UDCA is mainly produced by bacterial enzymes as a result of the epimerization of CDCA via the 7-oxoLCA intermediate [9,14,15]. Using human liver preparations Amuro et al. [16] provided evidence that 7-oxoLCA is primarily reduced to CDCA and lower amounts of UDCA by NADPH-dependent microsomal enzyme(s).

In order to identify this missing link, i.e. the source of hepatic reduction of bacterially derived 7-oxoLCA and formation of UDCA, we aimed in the present study to identify the hepatic 7-oxo bile acid reductase. Initially, human liver microsomes were used to characterize the 7-oxo bile acid reductase activity, which provided evidence for an ER (endoplasmic reticulum) luminal localization of the enzyme. Then, using recombinant enzyme, we demonstrated for the first time that 11β-HSD1 (11β-hydroxysteroid dehydrogenase 1) catalyses the irreversible reduction of 7-oxoLCA.

11β-HSD1 functions in intact cells primarily as a reductase and converts inactive 11-oxoglucocorticoids into 11β-hydroxyglucocorticoids using co-substrate NADPH, which is provided by H6PDH (hexose-6-phosphate dehydrogenase) in the ER [17]. Owing to the adverse metabolic effects of elevated 11β-HSD1-dependent glucocorticoid activation in tissues such as liver, skeletal muscle and adipose, 11β-HSD1 has emerged as a promising therapeutic target to treat metabolic diseases, and there is considerable effort to develop therapeutic inhibitors (reviewed in [1821]). However, regarding safety aspects of such inhibitors, it is necessary to identify other substrates of 11β-HSD1 [22], including bile acids as in the present study, and to understand the role of 11β-HSD1 in their metabolism.


Microsomal preparations and activity assays using human liver microsomes

Human liver microsomes (InVitro CYP H-class™ microsomes from a male donor, Celsis International) were thawed on ice and used immediately for activity assays. Microsomes, 0.2 mg per reaction, were incubated at 37 °C for 0–40 min in a total volume of 500 μl containing TS2 buffer (100 mM NaCl, 1 mM EGTA, 1 mM EDTA, 1 mM MgCl2, 250 mM sucrose and 20 mM Tris/HCl, pH 7.4), a final concentration of 500 μM NADPH, 1 μM substrate and vehicle or 5 μM 11β-HSD1 inhibitor as indicated. To assess the latency of 7-oxoLCA reduction and dependence on cofactor regeneration, microsomes were incubated in reaction mixture supplemented with the detergent Nonidet P40 (final concentration of 0.5%), G6P (glucose 6-phosphate) or G6S (glucose 6-sulfate) (final concentration of 1 mM), or the glucose-6-phosphate translocase inhibitor S3483 (obtained from Sanofi-Aventis; final concentration of 20 μM). Reactions were started by adding microsomes into freshly prepared reaction mixture and stopped by rapid freezing in solid CO2.

Rat liver microsomes and microsomes of HEK (human embryonic kidney)-293 cells transfected with human 11β-HSD1 or mock-transfected were prepared as described previously [23]. For immunoblotting, 50 μg of total microsomal proteins was separated by 12% PAGE, followed by transfer of proteins on to PVDF membranes. 11β-HSD1 was detected using primary anti-(human 11β-HSD1) antibody (Cayman Chemical).

Reduction of 7-oxoLCA by recombinant human 11β-HSD1

To assess 7-oxoLCA reductase activity in intact HEK-293 cells stably expressing recombinant human 11β-HSD1 or co-expressing 11β-HSD1 and H6PDH (AT6 and HHH7 clones respectively) [24], 20000 cells per well were distributed in 24-well plates and allowed to adhere for 16 h. Cells were then incubated with doubly charcoal-treated DMEM (Dulbecco's modified Eagle's medium), and the 7-oxidoreduction of 7-oxoLCA was measured at a final concentration of 1 μM after incubation for up to 24 h at 37 °C in a total volume of 1 ml. To determine the apparent Km of 11β-HSD1, frozen cell lysates were thawed, sonicated and immediately incubated for 10 min at 37 °C in a total volume of 500 μl containing 500 μM NADPH and 7-oxoLCA at concentrations between 62.5 nM and 4 μM. Reactions were terminated by freezing in solid CO2.

Impact of bile acids on the interconversion of glucocorticoids

The conversion of cortisone into cortisol using cell lysates was measured as described previously [25] (see the Supplementary Online Data at To assess the effect of 7-oxoLCA on the ratio of active to inactive glucocorticoids in intact cells at steady state, cells (30000 cells per well of a poly-L-lysine-coated 96-well plate, Becton-Dickinson) expressing 11β-HSD1 or co-expressing 11β-HSD1 and H6PDH were incubated for 24 h in a total volume of 40 μl of steroid-free DMEM in the presence of either 200 nM cortisone or cortisol and various concentrations of 7-oxoLCA.

Analysis of non-labelled steroids and bile acids by LC (liquid chromatography)–MS

Frozen samples from reactions using intact cells, cell lysates or microsomes were thawed, and a fixed amount of deuterated CDCA (0.5 nmol) or corticosterone (0.2 nmol) was added as an internal standard, followed by mixing and centrifugation at 3000 g for 5 min. Supernatants were loaded on to Oasis HBL SPE cartridges (pre-conditioned with 1 ml of methanol and 1 ml of water), followed by washing with 2 ml of water and elution with 2 ml of methanol. The solvent was evaporated and the residue was reconstituted in 100 μl of methanol.

7-OxoLCA and its metabolites were separated on an Atlantis T3 column (Waters) using an Agilent Technologies model 1200 liquid chromatograph (see the Supplementary Online Data). The liquid chromatograph was interfaced to an Agilent 6410 triple-quad mass spectrometer, operated in atmospheric pressure electrospray positive-ionization mode. Data acquisition was performed using MassHunter workstation software (version B.01.04).

Metabolites were identified by comparing their retention times and mass to charge ratios (m/z) with those of authentic standards. UDCA, 7-oxoLCA and CDCA were detected in the selected positive-ionization MS Scan2 mode (mass range, m/z 300–500). They were typically eluted at 5.2, 6.2 and 7.3 min, and were monitored at m/z 357.3, 373.2 and 357.3 (dehydrated bile acids) respectively. Quantitative determination of bile acids was performed by positive-ionization and MRM (Multiple Reaction Monitoring). Deuterated CDCA was used as internal standard (m/z 361.2).

Quantitative analysis of glucocorticoids was performed similarly by MRM. Cortisone (precursor and product ion at m/z 361 and 163) and cortisol (precursor and product ion at m/z 363 and 121) were eluted at 9.8 and 9.5 min respectively. Corticosterone (m/z 347.2) was used as internal standard and was eluted at 12.1 min.

Metabolites were quantified from calibration curves of the peak area ratio of the authentic standard and internal standard incubated in lysates of untransfected HEK-293 cells at a total protein concentration identical with that of the experimental setting and plotted against the concentration of authentic standards (normalization using internal standard).

Calculation of enzyme kinetic parameters

Enzyme kinetics was analysed by non-linear regression using four-parameter logistic curve fitting. For statistical comparisons, the ratio t-test in GraphPad Prism 5 software was used. Results (means±S.D.) were obtained from at least three independent experiments.

Molecular modelling of 11β-HSD1 with bile acids

Mouse 11β-HSD1 (PDB code 1Y5R) [26] was extracted from the PDB for use as a template to investigate the interactions of 7-oxo, 7α-hydroxy and 7β-hydroxy bile acids with 11β-HSD1. We used PDB code 1Y5R because it contains both corticosterone and NADP+, unlike other three-dimensional structures of 11β-HSD1 in the PDB. Human and mouse 11β-HSD1 have 79% sequence identity, which allows mouse 11β-HSD1 to be a good model for the interaction of bile acids with human 11β-HSD1.

To obtain 11β-HSD1 complexed with 7-oxoLCA, we superimposed the crystal structure of E. coli 7α-HSD [27] complexed with 7-oxoLC-Gly (PDB code 1FMC) with PDB code 1Y5R and extracted 7-oxoLC-Gly from PDB code 1FMC. Then we inserted 7-oxoLCA into 11β-HSD1. For conversion of the 7-oxo into either a 7α-hydroxy or a 7β-hydroxy, we used the Biopolymer option in Insight II. The glycine-conjugated bile acids also were constructed with Biopolymer. The final three-dimensional model of 11β-HSD1 with each bile acid was refined using Discover 3, which was run for 10000 iterations with a distant dependent dielectric constant of 2.


Reduction of 7-oxoLCA by human liver microsomes

Previous studies provided evidence for the existence of one or more hepatic enzymes catalysing the 7-oxo reduction of 7-oxoLCA to CDCA and/or UDCA [12,13,16,28,29]. These earlier studies suggested that the 7-oxo bile acid reductase is a microsomal enzyme preferentially using NADPH [16]. To identify this 7-oxo bile acid reductase, we first incubated human liver microsomes with 7-oxoLCA and studied the properties of the enzymatic reaction. After incubation for 40 min, approximately 70% of 7-oxoLCA was converted, mainly into CDCA and into approximately three times lower amounts of UDCA (Figure 2A). In addition, some minor products, including muricholic acids, were observed, but were not analysed further. The 7-oxoLCA reduction was approximately ten times more efficient in the presence of NADPH compared with NADH, and no 7-oxoLCA formation could be detected when microsomes were incubated with CDCA or UDCA and either NADP+ or NAD+ respectively, suggesting that the enzyme acts exclusively as a reductase.

Figure 2 Reduction of 7-oxoLCA by human liver microsomes

(A) Human liver microsomes were incubated for 40 min with 7-oxoLCA (1 μM), NADPH (500 μM) and either vehicle or hexose 6-phosphate (1 mM G6P or G6S) and 11β-HSD1 inhibitor [5 μM glycyrrhetinic acid (GA), Merck-544 (T0504) or BNW16] as indicated. Samples were analysed for the amount of unconverted substrate 7-oxoLCA (white bars) and the products CDCA (black bars) and UDCA (hatched bars). Results (n=3) are means±S.D. (B) Expression of 11β-HSD1 in liver microsomes and in microsomes of transfected HEK-293 cells. A total amount of 50 μg of microsomal proteins was separated by gel electrophoresis, and proteins were blotted on to PVDF membranes and probed with an anti-(human 11β-HSD1) antibody. Asterisks indicate non-specific bands, and arrows indicate 11β-HSD1. Molecular masses are indicated in kDa.

Ketoconazole (5 μM) had no effect either on the amount of CDCA and UDCA formed or on their ratio, suggesting that CYPs play a minor role in the metabolism of 7-oxoLCA to its 7-hydroxylated forms. Experiments using the detergent Nonidet P40 suggested latency of the 7-oxo bile acid reductase; however, prolonged incubation with the detergent also seemed to inhibit the enzyme activity (results not shown).

In the ER lumen, NADPH is regenerated by H6PDH, which, under physiological conditions, is primarily dependent on G6P [30]. We therefore tested whether the 7-oxo bile acid reductase is stimulated in the presence of the hexose phosphate. In the presence of G6P, the 7-oxoLCA supplied was almost completely metabolized. Comparable stimulation was observed in the presence of G6S (Figure 2A), which is a specific substrate of the luminal H6PDH, but not the cytoplasmic G6PDH [30]. Furthermore, the glucose-6-phosphate translocase inhibitor S3483 abolished the G6P- and G6S-induced stimulation of 7-oxoLCA reduction (results not shown).

To our knowledge, the only currently known luminal oxidoreductase using NADPH is 11β-HSD1, which is a reversible enzyme and catalyses the interconversion of glucocorticoids and some other substrates, including 7-oxycholesterol, 7-oxydehydroepiandrosterone, 11-oxyprogesterone and 11-oxyandrogen metabolites [22,3133]. An antibody raised against human 11β-HSD1 detected a single band at approximately 35 kDa and confirmed the high expression in human liver microsomes [3436] (Figure 2B). A band at approximately 33 kDa was detected in rat liver microsomal preparations. The size difference can be explained by the presence of three glycosylation sites in human 11β-HSD1 and two in the rat enzyme [37]. The occurrence of three non-specific bands in rat liver microsomes and in HEK-293 microsomal preparations indicates some cross-reactivity of the antibody. The recombinant enzyme was constructed with a C-terminal FLAG epitope, resulting in a slightly slower migration of the protein in gel electrophoresis. Probing the blot with anti-FLAG antibody resulted in a single band at 35 kDa (results not shown).

To test whether 11β-HSD1 might catalyse the reduction of 7-oxoLCA, we used human liver microsomes and studied the effect of three structurally unrelated 11β-HSD1 inhibitors, i.e. glycyrrhetinic acid, T0504 (also known as Merck-544) and BNW16 [25]. All three inhibitors abolished the conversion of 7-oxoLCA into CDCA and UDCA (Figure 2A). Next, we compared the reduction of 7-oxoLCA and cortisone. The human liver microsomes (0.2 mg in a reaction volume of 500 μl) converted approximately 50 and 80% of 7-oxoLCA (1 μM) after 10 and 20 min respectively, and 7-oxoLCA was almost completely metabolized after 40 min (Figure 3). In comparison, in analogous experiments, 37, 56 and 67% of cortisone was converted, indicating a higher capacity to metabolize 7-oxoLCA compared with cortisone.

Figure 3 Comparison of the reduction of 7-oxoLCA and cortisone by human liver microsomes

Human liver microsomes were incubated at a final concentration of 1 μM 7-oxoLCA (A) or 1 μM cortisone (B) for 0–40 min, followed by analysis by LC–MS of the amounts of residual substrate and products formed. One of three representative experiments is shown.

Reduction of 7-oxoLCA by recombinant 11β-HSD1

To verify that the reduction of 7-oxoLCA indeed is catalysed by 11β-HSD1, experiments in lysates of HEK-293 cells expressing the recombinant enzyme were performed. 11β-HSD1 efficiently catalysed the reduction of 7-oxoLCA with an apparent Km of 980±210 nM and a Vmax of 2.8±0.4 nmol·mg−1·h−1 as calculated by four-parametric non-linear regression (Figure 4A). Comparable values were obtained using the Hanes–Woolf equation (Figure 4B). No conversion of 7-oxoLCA was observed in untransfected HEK-293 control cells. The taurine- and glycine-conjugated 7-oxo bile acids, 7-oxoLC-Tau and 7-oxoLC-Gly, were similarly converted into the 7α-hydroxylated CDC-Tau (chenodeoxycholyltaurine) and CDC-Gly (chenodeoxycholylglycine) with minor amounts of UDC-Tau (ursodeoxycholyltaurine) and UDC-Gly (ursodeoxycholylglycine) respectively, demonstrating that 11β-HSD1 accepts both unconjugated and conjugated 7-oxoLCA as substrate.

Figure 4 Enzyme kinetic profile of 7-oxo reduction of 7-oxoLCA by recombinant human 11β-HSD1

Enzymatic activity of human 11β-HSD1 was determined in lysates of transfected HEK-293 cells as described in the Experimental section. (A) Reaction rate against substrate concentration plot of 10 min incubations with 7-oxoLCA. (B) Hanes–Woolf plot of the data shown in (A). Results (n=3) are means±S.D.

Next, we studied the impact of H6PDH on 11β-HSD1-dependent reduction of 7-oxoLCA in intact HEK-293 cells stably expressing either human 11β-HSD1 alone or co-expressing 11β-HSD1 and H6PDH. Co-expression with H6PDH stimulated the 7-oxo reductase activity of 11β-HSD1 (Figure 5). In cells co-expressing 11β-HSD1 and H6PDH, the reaction was almost completed after 24 h, resulting in the formation of approximately 90% CDCA and 10% UDCA. In contrast, only approximately 50% of 7-oxoLCA was converted in cells expressing solely 11β-HSD1, and it took more than 48 h until the reaction was completed (results not shown). No oxidation of CDCA and UDCA was detected, independent of the cell line used, confirming the observation from human liver microsomes and showing that 11β-HSD1 catalyses the irreversible conversion of 7-oxoLCA into CDCA and lower amounts of UDCA.

Figure 5 Impact of H6PDH on 11β-HSD1-dependent 7-oxo reduction of 7-oxoLCA in living cells

Intact HEK-293 cells stably expressing human 11β-HSD1 (A) or 11β-HSD1 and H6PDH (B) were incubated for 0–24 h with 1 μM 7-oxoLCA, followed by quantification of bile acids by LC–MS. Results (n=3) are means±S.D.

Interference of bile acids with the metabolism of glucocorticoids by 11β-HSD1

Several bile acids, including CDCA and LCA, have been found in previous studies to inhibit 11β-HSD1 and 11β-HSD2 respectively [3840]. We therefore compared the effect on 11β-HSD1 activities of 7-oxoLCA and its taurine- and glycine-conjugated forms with that of other relevant bile acids (Table 1). Whereas the 7α-hydroxylated CDCA and its conjugated derivatives CDC-Tau and CDC-Gly showed a more than 10-fold preference to inhibit the dehydrogenase over the reductase activity of 11β-HSD1, 7-oxoLCA displayed a slight preference to inhibit the reduction of cortisone, an effect that was more pronounced for the conjugated derivatives. The 7β-hydroxylated bile acids UDCA and UDC-Tau preferentially inhibited 11β-HSD1 dehydrogenase activity; however, they were approximately 10-fold less potent than the 7α-hydroxylated forms.

View this table:
Table 1 Inhibition by bile acids of the oxidation and reduction of glucocorticoids by 11β-HSD1

11β-HSD activities were measured in lysates of HEK-293 cells expressing the respective human recombinant enzyme as described in the Experimental section. Results are IC50 values in μM (means±S.D.) from four independent experiments.

Impact of 7-oxoLCA on the ratio of cortisol to cortisone at steady state

Bile acids can reach high concentrations in the liver and may affect not only initial rates of conversion, but also steady-state ratios of cortisol to cortisone controlled by 11β-HSD1. We therefore determined the effect of 7-oxoLCA on the steady-state ratio of cortisol to cortisone in HEK-293 cells stably expressing 11β-HSD1 or co-expressing 11β-HSD1 and H6PDH. As shown in Figure 6(A), approximately 40% cortisol was produced in 11β-HSD1-expressing cells, whereas over 90% of initially supplied cortisone was converted into cortisol upon co-expression with H6PDH, in line with earlier observations [33]. A mirror image was obtained when cells were incubated initially with cortisol (Figure 6B). Co-incubation of the cells with the respective glucocorticoid and increasing concentrations of 7-oxoLCA resulted in diminished cortisol production when cortisone was supplied and enhanced cortisone formation when cortisol was supplied initially, thus reflecting a shift from the active to the inactive glucocorticoid at steady state in the presence of high concentrations of 7-oxoLCA.

Figure 6 Effect of 7-oxoLCA on the steady-state ratio of cortisol to cortisone

(A) Intact HEK-293 cells expressing 11β-HSD1 (white bars) or co-expressing 11β-HSD1 and H6PDH (black bars) were incubated for 24 h with 200 nM radiolabelled cortisone and increasing concentrations of 7-oxoLCA, followed by determination of cortisol formation. (B) Similarly, cells were incubated with 200 nM radiolabelled cortisol, followed by measuring cortisone formation. Results representative of three independently performed experiments are shown.

Analysis of binding of bile acids to 11β-HSD1 by three-dimensional modelling

We used the crystal structure of 11β-HSD1 with corticosterone [26] (see Supplementary Figure S1 at as a benchmark for comparison of the three-dimensional models of 11β-HSD1 with 7-oxoLCA, CDCA and UDCA (Figure 7). In the crystal structure of 11β-HSD1 in complex with corticosterone, the phenolic group of Tyr183 on 11β-HSD1 and C4 on NADP+ are 2.8 and 3.9 Å (1 Å=0.1 nm) respectively from the 11β-hydroxy group on corticosterone. The hydroxy group on Tyr183 is hydrogen-bonded with the 2′-hydroxy group on the nicotinamide ribose. The ∊-amino group on Lys187 is 3.2 Å from the 2′- and 3′-hydroxy groups on the nicotinamide ribose. This stabilizes the orientation of the nicotinamide ribose [26]. N∊ and Nη on the guanidinium group of Arg66 have an electrostatic interaction with the 2′-phosphate of NADP+. This electrostatic interaction is characteristic of NADP(H)-dependent SDRs (short-chain dehydrogenase/reductases), including 17β-HSD1 [4143].

Figure 7 Three-dimensional models of mouse 11β-HSD1 complexed with CDCA, 7-oxoLCA and UDCA

In the complexes of 11β-HSD1 with CDCA and 7-oxoLCA, Tyr183 and C-4 on NADP(H) have favourable contacts with the 7α-hydroxy group on CDCA and 7-carbonyl on 7-oxoLCA. The phenolic hydroxy group on Tyr183 on 11β-HSD1 is 4.1 Å from the 7β-hydroxy group on UDCA. The ∊-amino group on Lys187 has stabilizing interactions with the ribose hydroxy groups on NADPH for 11β-HSD1 complexed with 7-oxoLCA. However, this is not the case with NADP+ and 11β-HSD1 complexed with CDCA and UDCA. Arg66 has an electrostatic interaction with the 2′-phosphate on NADP(H) in all three complexes.

Although all three bile acid substrates have favourable interactions with 11β-HSD1, there are differences in some interactions that can explain differences in the observed binding and catalytic activity. In the three-dimensional model of 11β-HSD1 in complex with 7-oxoLCA, the phenolic group on Tyr183 on 11β-HSD1 and C-4 on the nicotinamide ring of NADPH are 2.9 and 3.5 Å respectively from the 7-oxo group on 7-oxoLCA. The ∊-amino group on Lys187 has two favourable contacts with the 2′- and 3′-ribose hydroxy groups on NADPH, and the hydroxy group on Tyr183 is 3.5 Å from the 2′-hydroxy group of the nicotinamide ribose. Furthermore, Arg66 and other residues in the N-terminal end of 11β-HSD1 have favourable contacts with the adenosine on NADPH. Together, this three-dimensional model indicates that 7-oxoLCA fits into 11β-HSD1 in an orientation that favours reduction of the 7-oxo group to an alcohol.

In the three-dimensional model of 11β-HSD1 in complex with CDCA, Tyr183 on 11β-HSD1 and C-4 on NADP+ are 2.9 and 3.4 Å respectively from the 7α-hydroxy group on CDCA. However, Lys187 has an asymmetric orientation to the ribose hydroxy groups. The ∊-amino group on Lys187 is 4.6 Å from 3′-hydroxy group, which is too far to form a hydrogen bond. Also, the hydroxy group on Tyr183 is 4.4 Å from the 2′-hydroxy group of the nicotinamide ribose. Thus the nicotinamide ribose lacks two stabilizing interactions that are present in 11β-HSD1 in complex with 7-oxoLCA. Arg66 and other residues in the N-terminal end of 11β-HSD1 have favourable contacts with the adenosine on NADP+.

In the three-dimensional model of 11β-HSD1 in complex with UDCA, Tyr183 on 11β-HSD1 and C-4 on NADP+ are 4.1 and 3.3 Å respectively from the 7β-hydroxy group on UDCA. Lys187 has an asymmetric orientation to the ribose hydroxy groups. The ∊-amino group on Lys187 is 4.5 Å from 3′-hydroxy group, which is too far to form a hydrogen bond. Arg66 and other residues in the Nterminal end of 11β-HSD1 have favourable contacts with the adenosine on NADP+ (see also the Supplementary Online Data).

We also constructed three-dimensional models of 11β-HSD1 in complex with the glycine conjugates of 7-oxoLCA, CDCA and UDCA, as shown in Supplementary Figure S2 at All three glycine–bile acid conjugates have stabilizing contacts with various backbone nitrogens or oxygens in 11β-HSD1. Examination of these three-dimensional models reveals an unexpected coulombic interaction between the glycine carbonyl group and Nη2 on Arg66, which also has a key electrostatic interaction with the 2′-ribose phosphate on NADP+. Thus Arg66 has two important stabilizing interactions in the complexes of 11β-HSD1 with glycine conjugates of bile acids.

Thus the three-dimensional models of 11β-HSD1 with bile acids (Figure 7) reveal that 7-oxoLCA has the most favourable interaction with the catalytic site in 11β-HSD1 and that glycine–bile acid conjugates also can fit into 11β-HSD1 in which they have a coulombic interaction with Arg66, a key residue in the stabilization of NADPH binding to 11β-HSD1.


By catalysing the biotransformation of gut bacteria-derived secondary bile acids, the liver plays a key role in damage and repair; damage being changes in the steroid nucleus by bacterial enzymes, and repair being rectification of these changes by the hepatocyte. Several bacterial strains express 7α-HSDs to yield 7-oxoLCA from CDCA and 7-oxoDCA from CA [7]. 7-OxoLCA can be metabolized further by reversible bacterial 7β-HSDs, or taken up actively via sodium-dependent transporter [SLC10A2 (solute carrier 10A2)] from the lumen of the ileal segment or passively in the colon. Although 7-oxoLCA is readily detectable in faeces and portal blood, it cannot be detected at substantial levels in bile and plasma [44,45], suggesting efficient hepatic metabolism.

In the present paper, we report the identification of 11β-HSD1 as a hepatic 7-oxoreductase, providing an explanation for the low circulating 7-oxoLCA concentrations. Human and rodent liver expresses high levels of 11β-HSD1 [34,35] (Figure 2B). Previous studies demonstrated that 11β-HSD1 purified from human or rodent liver catalyses the NADPH-dependent conversion of cortisone into cortisol and 11-dehydrocorticosterone to corticosterone respectively [36]. In addition, it was shown that 11β-HSD1 purified from rabbit and hamster liver accepts not only glucocorticoids as substrates, but also 7-oxycholesterol metabolites [46,47]. We now demonstrate that both human liver microsomes and recombinant human 11β-HSD1 expressed in HEK-293 cells catalysed the NADPH-dependent 7-oxo reduction of 7-oxoLCA to form preferentially the 7α-hydroxy bile acid CDCA and to a lesser extent the 7β-hydroxy isomer UDCA (10–20%). The ratio of CDCA to UDCA observed in our experiments with human liver microsomes as well as recombinant human enzyme is in line with earlier observations with human liver preparations and measurements in blood following i.v. administration [13,16]. Importantly, the 7-oxo reduction of 7-oxoLCA in liver microsomes was completely abolished by the 11β-HSD1 inhibitors glycyrrhetinic acid, T0504 and BNW16 (Figure 2A). Although we cannot exclude the existence of another enzyme that catalyses the 7-oxo reduction of 7-oxoLCA in the liver, it is highly unlikely that such an enzyme would be completely inhibited by all of the three structurally unrelated compounds. Thus the results provide strong evidence that 11β-HSD1 is the major enzyme catalysing the 7-oxo reduction of 7-oxoLCA in humans.

Analysis of the kinetic properties revealed that 11β-HSD1 efficiently catalyses 7-oxo reduction of 7-oxoLCA with approximately 2-fold lower affinity, but 2-fold higher Vmax compared with reduction of cortisone. The conversion of both 7-oxoLCA and cortisone was latent, dependent on H6PDH and stimulated to a similar extent by addition of G6P to the reaction mixture. 11β-HSD1 accepted the taurine- and glycine-conjugated forms as substrates, with catalytic efficiencies comparable with those for the free bile acids. This is consistent with predictions of the three-dimensional models of the three bile acids conjugated to glycine, which uncovered an unexpected interaction between the glycine carbonyl group and Nη2 on Arg66 on 11β-HSD1 (see Supplementary Figure S2). Arg66 has an important role in neutralizing the negative charge on the 2′-phosphate on NADP(H) in 11β-HSD1 and other SDRs that use NADP(H) as a cofactor [26,4143]. Thus Arg66 has two key coulombic interactions in 11β-HSD1 complexed with the three glycine-conjugated bile acids.

Unlike other steroid and sterol substrates, 11β-HSD1 irreversibly catalyses the 7-oxo reduction of 7-oxoLCA, and the stereoselectivity for the bile acid metabolites formed is just opposite of that observed for the metabolites of 7-oxocholesterol [31,32], 7-oxodehydroepiandrosterone and 7-oxopregnenolone [33]. Neither CDCA nor UDCA, even upon prolonged incubation and at high concentrations, were converted into 7-oxoLCA, and there was also no isomerization of CDCA to UDCA or vice versa, as has been observed for 7α- and 7β-hydroxyepiandrosterone [48] and 7α- and 7β-hydroxydehydroepiandrosterone [33] respectively. Our results are in line with an earlier report on the metabolism of radiolabelled CDCA in rats with bile fistulas [12], where conversion of CDCA into trihydroxylated metabolites and minor amounts of UDCA, but no formation of 7-oxoLCA was observed. Furthermore, in humans, after a single hepatic passage following i.v. administration, neither CDCA nor UDCA was modified on the steroid ring.

The three-dimensional models indicate that only 7-oxoLCA has optimal binding of substrate and cofactor to Tyr183 and Lys187 (Figure 7), which is necessary for reduction of 7-oxoLCA to CDCA. In contrast, in the three-dimensional models of 11β-HSD1 with CDCA and UDCA, the ∊-amino group on the key catalytic residue Lys187 is too far from the 3′-ribose hydroxy group on NADP+ to form a stabilizing hydrogen bond required for catalytic activity. The distance between the 7β-hydroxy group on UDCA and the phenolic group on Tyr183 is 4.1 Å, which indicates weaker binding than found for the similar interaction between Tyr183 and either CDCA or 7-oxoLCA. This may explain the preference for CDCA as a product in the reduction of 7-oxoLCA by 11β-HSD1. Although only 7-oxoLCA was metabolized by 11β-HSD1, the three-dimensional models predicted binding of all three bile acids and their conjugates, and supported the more potent inhibitory effect of free and conjugated forms of CDCA compared with UDCA on cortisol oxidation (Table 1).

Alterations in the availability of bile acids, which reach high concentrations in the hepatocyte in cholestatic liver disease, may affect the hepatic activation of glucocorticoids. In intact HEK-293 cells expressing 11β-HSD1, but not H6PDH, CDCA and its conjugates preferentially inhibited 11β-HSD1 dehydrogenase activity and stimulated cortisone reduction. However, upon co-expression with H6PDH, which reflects the situation in hepatocytes, CDCA displayed weak inhibitory activity on 11β-HSD1. In contrast, 7-oxoLCA preferentially inhibited 11β-HSD1 reductase activity, and the presence of high concentrations of 7-oxoLCA stimulated cortisol oxidation and shifted the ratio of active to inactive glucocorticoids under steady-state conditions, probably by altering the ratio of NADPH to NADP+ in the ER lumen. It was shown previously that a ratio of NADPH to NADP+ greater than 10 is required for 11β-HSD1 to efficiently reduce cortisone [49]. The presence of high concentrations of 7-oxo bile acids, 7-oxo cholesterols or 7-oxo steroids may thus result in decreased ER luminal NADPH levels and lower concentrations of active glucocorticoids, thereby modulating redox signalling pathways and glucocorticoid-dependent adaptive responses.

Further research is needed to elucidate the physiological role for the rapid hepatic removal of 7-oxoLCA. Distinct effects of 7-oxoLCA, CDCA and UDCA on bile acid sensing receptors may affect the regulation of genes involved in lipid metabolism and inflammation. CDCA has been found to be a potent activator of the FXR (farnesoid X receptor)/RXR (retinoid X receptor) heterodimeric receptor [50,51]. In contrast, UDCA showed no or very little effect and 7-oxoLCA was a modest activator of human FXR, but did not activate mouse FXR. Thus, by converting the weak activator 7-oxoLCA into the more potent CDCA, 11β-HSD1 might play a role in modulating FXR activity. However, the amount of CDCA from de novo synthesis in the liver probably exceeds that from conversion of bacterially derived 7-oxoLCA, and the relative contribution remains to be determined. Also, there are currently no data available on potential effects of 7-oxoLCA on other nuclear receptors, including LXR (liver X receptor), VDR (vitamin D receptor) and PXR (pregnane X receptor).

Nevertheless, the results may be relevant regarding the current development of 11β-HSD1 inhibitors for treatment of metabolic diseases [1921]. Inhibition of 11β-HSD1 is expected to abolish hepatic metabolism of 7-oxoLCA, thereby leading to elevated hepatic and circulating 7-oxoLCA levels, similar to the observed accumulation of 7-oxocholesterol following 11β-HSD1 inhibition in rats [32]. 11β-HSD1 inhibition is not expected to affect bacterially derived production of UDCA and its metabolism in the liver; however, as shown in Figure 5, some UDCA is formed in the 11β-HSD1-dependent reduction of 7-oxoLCA, and inhibition of the enzyme might lower the local availability of UDCA. Clearly, further studies in 11β-HSD1-knockout mice and pre-clinical and clinical studies using selective inhibitors are needed to elucidate the impact of 11β-HSD1 on bile acid composition and function.


Alex Odermatt had the responsibility for the overall planning and conduct of the work, performed inhibitor experiments, analysed data and wrote the paper. Thierry Da Cunha and Carlos Penno developed the LC–MS protocol, performed enzyme activity experiments and analysed data. Christian Reichert performed activity experiments. Min Dong and Armin Wolf assisted in the design of experiments with human liver microsomes and analysis of protein expression. Charlie Chandsawangbhuwana performed three-dimensional modelling and analysed data, and Michael Baker performed three-dimensional modelling, analysed data and wrote the paper. All authors read and approved the final paper.


This work was supported by the Swiss National Science Foundation [grant number 31003A-124912 to A.O.]. A.O. has a Chair in Molecular and Systems Toxicology by the Novartis Research Foundation.


We thank Dr Alan F. Hofmann (University of California, San Diego, San Diego, CA, U.S.A.) for providing some of the bile acids for this study, and also for critical comments on the paper. M. D. and A. W. are employees of Novartis AG, Basel, Switzerland. The employer did not influence the design and interpretation of the data of this study.

Abbreviations: CA, cholic acid; CDCA, chenodeoxycholic acid; CDC-Gly, chenodeoxycholylglycine; CDC-Tau, chenodeoxycholyltaurine; CYP, cytochrome P450; DCA, deoxycholic acid; DMEM, Dulbecco's modified Eagle's medium; ER, endoplasmic reticulum; FXR, farnesoid X receptor; G6P, glucose 6-phosphate; G6S, glucose 6-sulfate; HEK, human embryonic kidney; H6PDH, hexose-6-phosphate dehydrogenase; 7α-HSD, 7α-hydroxysteroid dehydrogenase; 11β-HSD, 11β-hydroxysteroid dehydrogenase; i.v., intravenous; LC, liquid chromatography; LCA, lithocholic acid; MRM, Multiple Reaction Monitoring; 7-oxoDCA, 7-oxodeoxycholic acid; 7-oxoLCA, 7-oxolithocholic acid; 7-oxoLC-Gly, 7-oxolithocholylglycine; 7-oxoLC-Tau, 7-oxolithocholyltaurine; SDR, short-chain dehydrogenase/reductase; UDCA, ursodeoxycholic acid; UDC-Gly, ursodeoxycholylglycine; UDC-Tau, ursodeoxycholyltaurine


View Abstract