Chenodeoxycholic Acid and Deoxycholic Acid Inhibit 11 (cid:1) Hydroxysteroid Dehydrogenase Type 2 and Cause Cortisol-induced Transcriptional Activation of the Mineralocorticoid Receptor*

Inappropriate activation of the mineralocorticoid re- ceptor (MR) results in renal sodium retention and potassium loss in patients with liver cirrhosis. Recent evidence suggested that this MR activation is, at least in part, a result of bile acid-dependent reduction in 11 (cid:1) -hydroxy-steroid dehydrogenase type 2 (11 (cid:1) HSD2) activity, an enzyme preventing cortisol-dependent activation of MR by converting cortisol to cortisone. Here, we investigated the molecular mechanisms underlying bile acid-mediated MR activation. Analysis of urinary bile acids from 12 patients with biliary obstruction revealed highly elevated concen- trations of chenodeoxycholic acid (CDCA), cholic acid (CA), and deoxycholic acid (DCA), with average concen- trations of 50–80 (cid:2) M . Although CDCA and DCA both mediated nuclear translocation of MR in the absence of 11 (cid:1) HSD2 and steroids in transiently expressing HEK-293 cells, the transcriptional activity of MR was not stimu- lated. In contrast, CDCA and DCA both inhibited 11 (cid:1) HSD2 with IC 50 values of 22 and 38 (cid:2) M , respectively and caused cortisol-dependent nuclear translocation and increased transcriptional activity of MR. LCA, the bile acid that most efficiently inhibited 11 (cid:1) HSD2, was present at very low concentrations Cells staining positively for the corresponding steroid hor- mone receptor were divided into three categories: N, predominantly nuclear; C/N, cytoplasmic/reticular and nuclear; C, predominantly cy- toplasmic/reticular. Results were obtained from at least three independent transfection experiments and represent the percentage of flu- orescent cells relative to total cells, whereby 200–300 fluorescent cells were determined. Our results demonstrate that the three bile acids CA, CDCA, and DCA are present in high concentrations in urine from patients with cholestasis Two of them, CDCA and DCA, inhibited 11 (cid:1) HSD2 at concentrations present in these patients. CA, the third bile acid found at high concentrations during cholestasis, was a weak inhibitor of 11 (cid:1) HSD2. LCA, the potent inhibitor of 11 (cid:1) HSD2, was found in very low in cholestatic and is not expected to 11 it is bile acid the cortical duct to the urine, to they are the cortical collecting duct is at the end of the nephron. Our results suggest that CDCA, and to a lesser extent DCA, are for the ratio of (THF (cid:1) 5 an in vivo 11

Inappropriate activation of the mineralocorticoid receptor (MR) results in renal sodium retention and potassium loss in patients with liver cirrhosis. Recent evidence suggested that this MR activation is, at least in part, a result of bile acid-dependent reduction in 11␤-hydroxysteroid dehydrogenase type 2 (11␤HSD2) activity, an enzyme preventing cortisol-dependent activation of MR by converting cortisol to cortisone. Here, we investigated the molecular mechanisms underlying bile acid-mediated MR activation. Analysis of urinary bile acids from 12 patients with biliary obstruction revealed highly elevated concentrations of chenodeoxycholic acid (CDCA), cholic acid (CA), and deoxycholic acid (DCA), with average concentrations of 50 -80 M. Although CDCA and DCA both mediated nuclear translocation of MR in the absence of 11␤HSD2 and steroids in transiently expressing HEK-293 cells, the transcriptional activity of MR was not stimulated. In contrast, CDCA and DCA both inhibited 11␤HSD2 with IC 50 values of 22 and 38 M, respectively and caused cortisol-dependent nuclear translocation and increased transcriptional activity of MR. LCA, the bile acid that most efficiently inhibited 11␤HSD2, was present at very low concentrations in cholestatic patients, whereas the weak inhibitor CA did not cause MR activation. In conclusion, these findings indicate that CDCA, and to a lesser extent DCA, by inhibiting 11␤HSD2, mediate cortisol-dependent nuclear translocation and transcriptional activation of MR and are responsible at least for a part of the sodium retention and potassium excretion observed in patients with biliary obstruction.
The renal sodium retention and potassium loss observed in patients suffering from liver cirrhosis is caused by the activation of the mineralocorticoid receptor (MR). 1 Due to excessive intra-abdominal fluid sequestration, the kidney is hypoperfused, leading to a compensatory secondary hyperaldosteronism (1,2). However, evidence provided by several studies indicate that the extent of sodium retention and potassium loss cannot be completely explained by the elevated aldosterone concentrations in cirrhotic patients and in many situations, renal sodium retention precedes ascites formation (1,(3)(4)(5)(6). A series of recent studies provided strong evidence that inhibition of 11␤-hydroxysteroid dehydrogenase type 2 (11␤HSD2) by bile acids causes 11␤-hydroxyglucocorticoid-induced activation of MR, leading to sodium retention and potassium wasting in the cholestatic state (7)(8)(9)(10)(11)(12)(13).
Although serum 11␤-hydroxyglucocorticoid concentrations are in a 100-fold excess over aldosterone in vivo and both glucoand mineralocorticoids bind with similar affinities to the MR, the transcriptional activity of the MR is normally regulated by aldosterone (14 -18). By converting biologically active 11␤-hydroxyglucocorticoids (cortisol in humans, corticosterone in rodents) into inactive 11-ketosteroids (cortisone in humans, 11dehydrocorticosterone in rodents), 11␤HSD2 prevents access of 11␤-hydroxyglucocorticoids to the MR and renders specificity of the MR for aldosterone (for review see Refs. 19 and 20). Patients with genetic mutations in the gene encoding 11␤HSD2 suffer from severe hypertension because of cortisol-induced MR activation (for review see Refs. 19 -21). A similar form of hypertension is observed in patients with excessive consumption of licorice, containing the 11␤HSD2 inhibitor glycyrrhetinic acid (22)(23)(24)(25).
That cortisol-induced MR activation due to reduced 11␤HSD2 activity upon inhibition by bile acids can account for at least part of the sodium retention and potassium loss in cirrhotic rats and in patients with cholestasis was suggested by the following findings. (i) Bile acid concentrations can increase by a factor of 100 under cholestatic conditions (26,27). (ii) Bile acid-dependent inhibition of the activity of 11␤HSD enzymes was demonstrated using total renal microsomes (7,8), transfected COS-1 cells (11), or isolated rat cortical collecting tubules (12). (iii) In bile duct-ligated rats, an animal model of liver cirrhosis, the urinary ratio of (tetrahydrocorticosterone ϩ 5␣-tetrahydrocorticosterone)/11-dehydrotetrahydrocorticosterone, an in vivo measure of 11␤HSD2 activity, was significantly increased, indicating inhibition of 11␤HSD2 (12). (iv) Rats treated with chenodeoxycholic acid (CDCA) developed increased blood pressure (10), and adrenalectomized rats treated with CDCA showed enhanced renal sodium retention and urinary potassium excretion (9). (v) A study in 12 patients with cholestasis, with an average total bile acid concentration of 65 M, showed a significantly increased urinary ratio of (tetrahydrocortisol ϩ 5␣-tetrahydrocortisol)/tetrahydrocortisone indicating impaired activity of 11␤HSD2; after removal of biliary obstruction both bile acid concentrations and glucocorticoid metabolites normalized in all 12 patients, suggesting that the observed reduction of 11␤HSD2 activity during cholestasis was caused by bile acids (13).
However, several important questions regarding the mechanisms by which the presence of increased bile acid concentrations lead to inappropriate activation of MR remain to be answered. First, is the observed sodium retention and potassium excretion caused by direct activation of MR by bile acids, or indirectly by cortisol-induced MR activation due to the inhibition of 11␤HSD2? Second, does cortisol in the presence of bile acids lead to transcriptional activation of MR, and third, which of the bile acids present in patients with cholestasis are responsible for the observed effects? Therefore, we investigated the molecular mechanisms underlying the bile acid-dependent inhibition of 11␤HSD2 and the subsequent activation of MR.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media and LipofectAMINE Plus reagent were purchased from Invitrogen. NAD ϩ , cortisol, and mouse monoclonal anti-FLAG M2 antibody were from Sigma Chemical. [1,2,6, H]cortisol was from Amersham Biosciences. The solvent was evaporated and cortisol was resolved in methanol. Bile acids and aldosterone were purchased from Steraloids Inc., Newport, Rhode Island. High affinity rat monoclonal anti-hemagglutinin (HA) antibody was from Roche Diagnostics, Rotkreuz, Switzerland. Highly cross-absorbed fluorescent goat anti-mouse IgG ALEXA-488 and goat anti-rat IgG ALEXA-594 were from Molecular Probes Inc., Eugene, Oregon. The Dual-Light kit was from Tropix, Bedford, Massachusetts. The SV40 luciferase plasmid was from Promega, Wallisellen, Switzerland. Plasmids for MMTV-lacZ, HA-tagged MR and FLAG-tagged 11␤HSD2 were described previously (28 -30). The green fluorescent protein-MR expression construct (MR-GFP) was obtained from A. Naray-Fejes-Toth (31). In all our experiments no significant differences between wild-type MR, HA-tagged MR, or MR-GFP were detected, indicating that epitope tags did not affect MR function in these experiments. The N-terminally FLAG epitope-tagged human AR construct was provided by J. J. Palvimo (32). Expression constructs encoding rat 3␣-hydroxysteroid dehydrogenase (33), human steroid 5␣-reductase type I (34), mouse CYP7A1 cholesterol 7␣-hydroxylase (35), and mouse CYP7B1 oxysterol 7␣-hydroxylase (36) were a gift from D. W. Russell.
Patients-The 12 patients (8 women and 4 men) with obstructive jaundice showed a mean serum bilirubin of 140 M (105-305 M, 4ϫ the upper range) and were described previously (13). Choledocholithiasis was the underlying disease state in nine patients, while malformation of the bile duct, carcinoma of pancreas, and carcinoma of choledochus was the etiology in one patient each. Urinary total bile acids were quantitated by gas chromatography-mass spectrometry as described recently (13).
Immunofluorescence Detection of Nuclear Translocation of MR-HEK-293 cells (300,000 per well) were grown on glass cover slips in 6-well plates containing 2 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transient transfections were performed 24 h later by using the calcium phosphate precipitation method with 1 g of MR cDNA and 0.5 g of 11␤HSD2 cDNA per well. 6 h post-transfection, cells were washed twice with medium that was charcoal-stripped twice followed by incubation in this medium for 14 h. Cells were then incubated in the presence or absence of the appropriate concentrations of bile acids for 5 min as indicated, followed by the addition of 10 nM cortisol. Cells were incubated for 45 min at 37°C, washed once with buffer NAPS (150 mM sodium phosphate, pH 7.4, 120 mM sucrose), followed by fixation with 4% paraformaldehyde for 10 min.
Immunostaining was performed as described (29) using mouse anti-FLAG antibody M2 and ALEXA-488 antibody to detect FLAG-tagged 11␤HSD2 and rat anti-HA antibody and ALEXA-594 to detect HAtagged MR, respectively. In coexpression experiments with AR or GFP-MR, human 11␤HSD2 was detected with a rabbit polyclonal antibody (provided by Z. N. Kyossev,Ref. 37) and secondary ALEXA-594 antirabbit antibody.
Immunofluorescence was detected using a Carl Zeiss confocal microscope LSM410 (Carl Zeiss, Goettingen, Germany). The intracellular localization of steroid receptors in 11␤HSD2-positive cells was determined by distinguishing three different categories: C, for predominant staining of cytoplasm or endoplasmic reticulum membrane; C/N, when cytoplasmic/reticular and nuclear staining were of comparable intensity; N, for predominantly nuclear staining. Results were obtained from at least three independent transfection experiments, whereby between 200 and 300 stained cells of each sample were determined by an observer who was blinded to the cell treatment procedures.
Determination of Transcriptional Activity of MR Using a Chemiluminescent Reporter Gene Assay-CHO cells (100,000 per well) were grown in 12-well plates containing 1 ml of an equal mixture of Dulbecco's modified Eagle's medium and F12 medium, supplemented with 10% fetal calf serum. Transient transfection was performed 24 h later with LipofectAMINE Plus reagent in Optimem using 0.05 g of SV40-luciferase plasmid and 0.3 g of MMTV-lacZ plasmid as reporters and 0.2 g of MR and 0.1 g of 11␤HSD2 or empty pcDNA3 vector, respectively. 4 h post-transfection, cells were washed twice with medium that was charcoal-stripped twice followed by incubation in this medium for 20 h. Cells were then incubated with 100 M of the corresponding bile acid for 5 min before adding 10 nM aldosterone or 100 nM cortisol and incubated for another 24 h. Cells were then harvested and luciferase and galactosidase activities determined using the Dual-Light kit according to the manufacturer's instructions. The ratio of galactosidase to luciferase was calculated in all experiments.
Test for Cell Viability and Toxicity-The cell viability was analyzed by incubating transfected HEK-293 cells and CHO cells for 1 h with the corresponding concentration of bile acid and staining with trypan blue. The toxicity of bile acids was analyzed using the tetrazolium salt MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) according to the cell proliferation kit I from Roche Diagnostics. No significant differences between control and bile acid-treated cells were obtained in both tests.
Determination of 11␤HSD2 Activity and Inhibition by Bile Acids-11␤HSD2 enzyme activity was measured in cell lysates as described (29). Briefly, transfected HEK-293 cells, incubated in charcoal-treated Dulbecco's modified Eagle's medium for 24 h, were washed once with Hanks' solution and resuspended in a buffer containing 100 mM NaCl, 1 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, 250 mM sucrose, 20 mM Tris-HCl, pH 7.4. Cells were lysed by freezing in liquid nitrogen. Dehydrogenase activity was measured in a final volume of 20 l containing the appropriate concentration of bile acid, 30 nCi of [ 3 H]cortisol, and unlabeled cortisol to a final concentrations of 50 nM. The reaction was started by mixing cell lysate with the reaction mixture. Alternatively, endoplasmic reticulum microsomes were prepared from transfected HEK-293 cells according to a procedure described previously (7) and incubated with reaction mixture containing various concentrations of cortisol and CDCA. Incubation proceeded for 20 min, and the conversion of cortisol to cortisone was determined by thin layer chromatography (TLC). Because of the inaccuracy of the TLC method at low conversion rates and the end-product inhibition of 11␤HSD2 at conversion rates higher than 60 -70%, only conversion rates between 10 and 60% were considered for calculation. The inhibitory constant IC 50 was evaluated using the curve-fitting program from MDL Information Inc., San Leandro, California. Results are expressed as means Ϯ S.E. and consist of at least four independent measurements.
To determine whether 11␤HSD2 metabolizes CDCA, supernatants from cultures of HEK-293 cells transiently expressing 11␤HSD2 were analyzed by gas chromatography-mass spectrometry. As a recovery standard 4 g of 23-nor-deoxycholic acid was added to 1 ml of cell culture supernatant, and the sample was extracted on a Sep-Pak C 18 column. To this extract 4 g of 5␤-cholestan-3␤-ol was added as a standard for derivatization and gas chromatography, and the sample was derivatized to form the methylester trimethylsilyl ether. The excess of silylating agent was removed by gel filtration on a Lipidx-5000 column. Samples were analyzed by gas chromatography-mass spectrometry using a Hewlett Packard gas chromatograph 6890 equipped with a mass selective detector 5973 by selected ion monitoring (SIM, programmed for 5 different bile acids) and in the scan mode to detect any potentially new metabolite formed from CDCA.

RESULTS
To identify the major bile acids present in urine of patients with biliary obstruction, we determined by gas chromatography-mass spectrometry the concentrations of various bile acids in the urine of 12 cholestatic patients (13) (Table I). Whereas urine from control individuals or from the 12 individuals after removal of biliary obstruction contained only negligible amounts of bile acids, the samples collected during cholestasis contained significantly increased concentrations of bile acids. CA, CDCA, and DCA were the three major constituents, with similar average concentrations between 50 and 80 M. Compared with these three major bile acids, LCA and UDCA, although significantly increased compared with control individuals, were present at only low concentrations.
Using transient coexpression of 11␤HSD2 and MR in HEK-293 cells and fluorescence microscopic detection, we have recently shown that CDCA causes nuclear translocation of MR in the presence of low concentrations of cortisol (13). To understand the molecular mechanisms underlying the bile acid-dependent increase in MR activity, we addressed now the question of whether bile acids can activate the MR directly, in the absence of cortisol, or whether the activation of MR is caused by cortisol upon inhibition of 11␤HSD2. As reported previously, the MR showed a heterogeneous distribution in the absence of both 11␤HSD2 and steroids, with approximately one-third of cells showing nuclear localization, one-third mixed cytoplasmic/reticular and nuclear distribution, and one-third showing cytoplasmic/reticular localization (30). Interestingly, the addition of CDCA or LCA led to the translocation of MR from the cytoplasm/endoplasmic reticulum membrane into the nucleus, with more than 90% of receptor molecules located in the nucleus at 100 M of the corresponding bile acid (Table II). DCA was less efficient in mediating nuclear translocation of MR with effective concentrations of 200 M and higher. In contrast, CA and UDCA, as well as the conjugated bile acids GCDCA and TCDCA did not affect the intracellular localization of MR at concentrations up to 200 M. The observed effect of CDCA and LCA on MR was receptor-specific because neither CDCA, LCA, nor any other bile acid used in this study altered the intracellular localization of either AR or GR. Importantly, CDCA and LCA were only able to mediate nuclear translocation of MR in the absence of 11␤HSD2 but not when the receptor was coexpressed with 11␤HSD2. This protective effect of 11␤HSD2 on MR was specific because coexpression of MR with several other steroid-metabolizing enzymes with reticular distribution and cytoplasmic orientation did not prevent CDCA-mediated nuclear MR translocation (Table II). We also tested whether 11␤HSD2 can metabolize CDCA. Lysates of cells expressing 11␤HSD2 were incubated in the presence of NAD ϩ and CDCA, followed by analysis using gas chromatography/mass spectrometry. No decrease in the amount of CDCA could be detected, and there were no detectable peaks that may indicate a novel metabolite (not shown), suggesting that the protective effect of 11␤HSD2 on bile acid-mediated nuclear translocation of MR is not due to enzymatic inactivation of CDCA. The nuclear trans-  absence of cortisol HEK-293 cells were transfected with MR-GFP, epitope-tagged GR, AR, or MR or cotransfected with HA-tagged MR and FLAG-tagged 11␤HSD2 or unrelated steroid metabolizing enzymes. Transfected cells were grown for 14 h in steroid-free medium prior to the addition of bile acids and incubated for another 45 min. Immunostaining using antibodies against the HA-and FLAG-epitopes was performed as described under "Experimental Procedures." Cells staining positively for the corresponding steroid hormone receptor were divided into three categories: N, predominantly nuclear; C/N, cytoplasmic/ reticular and nuclear; C, predominantly cytoplasmic/reticular. Results were obtained from at least three independent transfection experiments and represent the percentage of fluorescent cells relative to total cells, whereby 200 -300 fluorescent cells were determined. We further investigated the impact of various bile acids present in patients with cholestasis on the nuclear translocation of MR after coexpression with 11␤HSD2 and in the presence of 10 nM cortisol ( Fig. 1 and Table III). Whereas the two major bile acids CDCA and DCA similarly mediated cortisoldependent nuclear translocation of MR, the third major bile acid CA did not alter the intracellular distribution of the receptor. The most potent effect on cortisol-dependent nuclear MR translocation was exerted by the less abundant bile acid LCA, whereas UDCA and the conjugated bile acids GCDCA and TCDCA showed no effect.
We next analyzed the transcriptional activity of MR in the absence of both 11␤HSD2 and steroids but upon incubation of cells with bile acids. None of the bile acids tested was able to stimulate the transcriptional activity of MR under conditions where nuclear translocation of MR was observed ( Fig. 2A). This was in contrast to cortisol or aldosterone, which both significantly enhanced transcriptional activity of MR. Although the presence of 100 M CDCA induced almost complete nuclear translocation of the MR and an antagonistic effect of this bile acid on corticosteroid-mediated MR activation was expected, the transcriptional activity of MR in the presence of 100 nM cortisol or 100 nM aldosterone was not affected by CDCA (not shown).
Furthermore, bile acids at concentrations up to 100 M did not stimulate the transcriptional activity of GR or of MR upon coexpression with 11␤HSD2 but in the absence of steroids (not shown). Also, the presence of 100 nM cortisol in the absence of bile acids did not enhance the transcriptional activity of MR due to conversion of cortisol to cortisone by 11␤HSD2 (Fig. 2B). However, there was a significant increase in the transcriptional activity of MR upon coincubation of cells coexpressing MR and 11␤HSD2 with 100 nM cortisol and 100 M of either CDCA, DCA, or LCA, suggesting cortisol-dependent receptor activation due to inhibition of 11␤HSD2 (Fig. 2B). In contrast, CA, UDCA, as well as the conjugated bile acid GCDCA did not alter the transcriptional activity of MR at this concentration.
To assess the direct inhibition of 11␤HSD2 by bile acids, we measured the conversion of cortisol to cortisone in an assay using cell lysates and determined the apparent IC 50 values of bile acids (Fig. 3). The inhibitory potential of LCAϾ CDCAϾDCA with IC 50 values of 7, 22, and 38 M, respectively is in line with the observed potential of these bile acids to mediate cortisol-dependent nuclear translocation of MR. Moreover, only a weak inhibitory effect, with IC 50 values of over 100 M, was detected for bile acids that did not alter the intracellular localization of MR in the nuclear translocation assay, e.g. for CA, UDCA, and the conjugated bile acids TCDCA and GCDCA. The fact that bile acid concentrations required to mediate cortisol-dependent nuclear translocation of MR were somewhat higher than concentrations for inhibition of 11␤HSD2 may reflect the higher concentrations required in intact cells to reach the corresponding intracellular concentrations. The mode of inhibition of 11␤HSD2 by CDCA was analyzed by incubating endoplasmic reticulum microsomes with various concentrations of cortisol and CDCA, respectively. Kinetic analysis using both the Lineweaver-Burk method (1/v plotted against 1/s) and the Dixon method (1/v plotted against inhibitor concentration) provided evidence for a competitive mode of inhibition (Fig. 4). DISCUSSION Chronically elevated bile acid concentrations result in sodium retention; however, the mechanisms involved remain un-  bile acid-dependent inhibition of 11␤HSD2 HEK-293 cells were cotransfected with HA-tagged MR and FLAGtagged 11␤HSD2. At 6 h after transfection, culture medium was replaced by steroid-free medium and cells were incubated for another 14 h. Intracellular distribution of MR was analyzed after a 5 min preincubation of cells with various concentrations of bile acids as indicated, followed by the addition of 10 nM cortisol and another 45 min of incubation. Immunostaining using antibodies against the HA and FLAG epitopes was performed as described under "Experimental Procedures." Cells staining positively for the corresponding steroid hormone receptor were divided into three categories: N, predominantly nuclear; C/N, cytoplasmic/reticular and nuclear; C, predominantly cytoplasmic/reticular. Results were obtained from at least three independent transfection experiments and represent the percentage of fluorescent cells relative to total cells, whereby 200 -300 fluorescent cells were determined.
Incubation conditions a N C/N C clear. Although recent reports provided evidence that bile acidinduced inhibition of 11␤HSD2 may lead to cortisol-dependent activation of MR (7)(8)(9)(10)(11)(12)(13), these studies did not address the question of whether bile acids directly activate the MR or whether the MR is activated by cortisol as a result of the inhibition of 11␤HSD2. Here, we demonstrate that the MR is not directly activated by bile acids. Although in the absence of 11␤HSD2 the MR translocated from the cytoplasm into the nucleus after addition of LCA, CDCA, or DCA, the transcriptional activity of the receptor was not increased under these conditions. The bile acid-dependent nuclear translocation of MR in absence of 11␤HSD2 was receptor-specific because the intracellular localization of GR or AR was not affected by bile acids. In contrast to the study by Miura et al. (38) reporting Cortisol-induced transcriptional activation of MR due to bile acid-dependent inhibition of 11␤HSD2. CHO cells were cotransfected with expression plasmids for MR (0.2 g) and 11␤HSD2 (0.1 g) or empty pcDNA3 vector (0.1 g), in combination with the reporter plasmid for MMTV-lacZ (0.3 g) and the internal control SV40-luciferase (0.05 g). 4 h after transfection, culture medium was replaced by steroid-free medium, and cells were incubated for another 20 h. Cells were then incubated with 100 M of the corresponding bile acid for 5 min before adding 100 nM cortisol and incubation for another 24 h. Cells were assayed for galactosidase and luciferase activities. Galactosidase reporter activity was normalized to the internal luciferase control, and the data were plotted as fold induction relative to a control that contained instead of bile acid an equal volume of solvent. nuclear translocation of GR after incubation with UDCA, we did not observe nuclear translocation of GR after a 45-min incubation with 100 M UDCA. This discrepancy may be explained by prolonged incubation for 6 h and higher UDCA concentrations (200 M) in their experiment. Nevertheless, UDCA did not stimulate the transcriptional activity of GR either in their or our experiments, supporting our finding that bile acids do not directly activate the transcription rate of corticosteroid receptors. Furthermore, 11␤HSD2, which determines the intracellular localization of MR in the absence of hormone, prevented nuclear translocation of MR mediated by bile acids. This effect was only observed for 11␤HSD2. Several other steroid metabolizing enzymes, including a mutant 11␤HSD1 with cytoplasmic orientation, did not prevent bile acid-induced nuclear translocation of MR. These findings suggest that in the absence of hormones, bile acids do not alter the intracellular distribution of MR in native tissues such as renal cortical collecting ducts or colon, expressing both MR and 11␤HSD2 (37). The mechanism by which 11␤HSD2 prevents nuclear translocation of MR is unknown. It is possible that in the absence of 11␤HSD2 the bile acids CDCA, DCA, and LCA bind to the MR and induce a conformational change that results in the exposure of the nuclear localization signal, thus causing nuclear translocation. We hypothesize an interaction between 11␤HSD2 and MR that either prevents binding of these bile acids to the MR or blocks the conformational change required for exposure of the nuclear localization signal. The hypothesis of an interaction between 11␤HSD2 and MR is also supported by our recent study, showing that glucocorticoids abolish the activation of MR by aldosterone by a mechanism that is strictly dependent on 11␤HSD2 (30). Because aldosterone, but not CDCA or LCA, induces nuclear translocation of MR in the presence of 11␤HSD2 and because these bile acids do not prevent aldosterone-induced MR activation, the binding sites for aldosterone and for CDCA and LCA within the MR are most likely different. This may be in analogy to the existing evidence for the distinct binding of dexamethasone and UDCA in the ligand-binding domain of the GR (38).
Our results demonstrate that the three bile acids CA, CDCA, and DCA are present in high concentrations in urine from patients with cholestasis (Table I). Two of them, CDCA and DCA, inhibited 11␤HSD2 at concentrations present in these patients. CA, the third bile acid found at high concentrations during cholestasis, was a weak inhibitor of 11␤HSD2. LCA, the most potent inhibitor of 11␤HSD2, was found in very low concentrations in cholestatic patients and is not expected to contribute significantly to the inhibition of 11␤HSD2 during biliary obstruction. Although it is unknown whether bile acid concentrations in the cortical collecting duct are similar to the concentrations measured in urine, it is reasonable to assume that they are similar because the cortical collecting duct is at the end of the nephron. Our results suggest that CDCA, and to a lesser extent DCA, are responsible for the increased ratio of (THF ϩ 5␣-THF)/THE, an in vivo indicator of reduced 11␤HSD2 activity, in patients with biliary obstruction (13) and of (THB ϩ 5␣-THB)/THA in rats with biliary cirrhosis (12). The fact that inhibition of 11␤HSD2 by CDCA or DCA caused cortisol-induced nuclear translocation and transcriptional activation in the cell assay further suggests that these bile acids mediate cortisol-induced MR activation in vivo, resulting finally in an increased activity of Na ϩ /K ϩ -ATPase and ENaC in the renal cortical collecting duct and causing sodium retention and potassium excretion. These findings are supported by a previous study by Latif et al. (9) who infused CDCA or CA in rats and observed increased sodium retention and potassium excretion after infusion of CDCA but no alterations after infu-sion of CA. In a recent study, Wu et al. (39) reported significantly elevated blood pressure in rats that were treated orally with CA for one month. However, the primary bile acid CA is converted to DCA by bacteria-mediated 7␣-dehydroxylation in the intestine, reabsorbed in the ileum and transported back to the liver via the portal circulation (40,41). Thus, oral administration of CA may result in increased concentrations of DCA in the circulation, leading to inhibition of 11␤HSD2.
Previous studies on cirrhotic rats, induced by bile duct ligation, showed a reduction in 11␤HSD2 mRNA by 25%, but the protein content was not changed. 11␤HSD2 activity measured in kidney microsomes from these animals was reduced by 27% (11). Similarly, 11␤HSD2 activity measured on isolated intact cortical collecting ducts from cirrhotic rats revealed a 10 -20% reduction compared with control (12). A reduced 11␤HSD2 activity of 10 -20% due to reduced expression is unlikely to explain entirely the 2.5-fold increase in the ratio of (THBϩ5␣-THB)/THA observed in cirrhotic rats. Because the increased ratio of (THBϩ5␣-THB)/THA in cirrhotic rats and of (THFϩ5␣-THF)/THE in cholestatic patients is in line with their increased bile acid concentrations and these bile acid concentrations are in the range where they inhibit 11␤HSD2, the reduced 11␤HSD2 activity in liver cirrhosis and in cholestasis are mainly explained by a direct inhibition of 11␤HSD2 by bile acids rather than by the reduced expression.
Interestingly, the bile acids that mediated nuclear translocation of MR in the absence of 11␤HSD2 and steroids, also inhibited 11␤HSD2. The similarity of the structure of bile acids and corticosteroids suggests that bile acids may occupy the substrate binding site of 11␤HSD2 and may act as competitive inhibitors. Indeed, our kinetic analysis supports this hypothesis and provides evidence for a competitive mode of inhibition of 11␤HSD2 by CDCA (Fig. 4). A more detailed analysis of the kinetics of 11␤HSD2 activity and of its inhibition will require data obtained from purified and functionally active enzyme. A comparison of the structures of bile acids (Fig. 5) and their inhibitory effects on 11␤HSD2 indicates that the presence of two hydroxyl groups in position 7 and 12 decreases the inhibitory potential, whereas LCA, without a hydroxyl group in these positions is the strongest inhibitor. In bile acids with a hydroxyl group in position 7 only, its orientation is critical. Whereas CDCA has the hydroxyl group in ␣-orientation and inhibits 11␤HSD2 with an IC 50 of 22 M, the hydroxyl group in position 7 is in ␤-orientation in UDCA, which has a high IC 50 of 271 M.
The inhibitory effect of CDCA on 11␤HSD2 and subsequent inappropriate MR activation disfavors the use of CDCA for clinical applications such as the treatment of gallstones that was performed earlier (42). In contrast, UDCA, which does not mediate transcriptional activation of corticosteroid receptors and which is reported to suppress NF-B-dependent transcription (38), showed significant improvements in the treatment of patients with primary biliary cirrhosis (43).
Previous reports on the inhibitory effect of TCDCA on 11␤HSD2 are controversial. Whereas a study with transfected COS-1 cells showed maximal inhibition of 11␤HSD2 by both 50 M CDCA or TCDCA (11), measurements on isolated rat cortical collecting ducts yielded an IC 50 for CDCA of 20 M and for TCDCA of 80 M (12). We obtained values in line with the second study with an IC 50 for CDCA of 22 M and for TCDCA of 140 M. Conjugation in position 24 by taurine or glycine abolishes the inhibitory potential of the corresponding bile acid.
In conclusion, our findings demonstrate that chenodeoxycholic acid and deoxycholic acid are highly increased in patients with biliary obstruction. Inhibition of 11␤HSD2 by these bile acids mediates cortisol-dependent nuclear translocation and transcriptional activation of MR. This may contribute to the sodium retention and potassium excretion observed in patients with liver cirrhosis or cholestasis.