The Intracellular Localization of the Mineralocorticoid Receptor Is Regulated by 11 b -Hydroxysteroid Dehydrogenase Type 2*

11 b -Hydroxysteroid dehydrogenase (11 b -HSD) type 2 has been considered to protect the mineralocorticoid receptor (MR) by converting 11 b -hydroxyglucocorticoids into their inactive 11-keto forms, thereby providing specificity to the MR for aldosterone. To investigate the func-tional protection of the MR by 11 b -HSD2, we coexpressed epitope-tagged MR and 11 b -HSD2 in HEK-293 cells lacking 11 b -HSD2 activity and analyzed their subcellular localization by fluorescence microscopy. When expressed alone in the absence of hormones, the MR was both cytoplasmic and nuclear. However, when coexpressed with 11 b -HSD2, the MR displayed a reticular distribution pattern, suggesting association with 11 b -HSD2 at the endoplasmic reticulum membrane. The endoplasmic reticulum membrane localization of the MR was observed upon coexpression only with 11 b -HSD2, but not with 11 b -HSD1 or other steroid-metabolizing enzymes. Aldosterone induced rapid nuclear translocation of the MR, whereas moderate cortisol concentrations (10–200 n M ) did not ac- tivate the receptor, due to 11 b -HSD2-dependent oxidation to cortisone. Compromised 11 b -HSD2 activity (due to ge-netic mutations, the presence of inhibitors, or saturating cortisol concentrations) led to cortisol-induced nuclear accumulation of the MR. Surprisingly, coexpression secondary Immunofluorescence was detected using a Model LSM410 confocal microscope Quan- titation of intracellular localization of the MR or AR was performed by counting the number of cells within an area selected under the trans- mitted light and determination of positively staining cells that were classified into three different categories: predominantly cytoplasmic staining, comparable intensity of nuclear and cytoplasmic staining, and exclusively nuclear staining. Results were obtained from at least three independent transfection experiments, in which between 400 and 500 stained cells were determined for each sample.

The mineralocorticoid receptor (MR) 1 and the glucocorticoid receptor (GR) both belong to the steroid/thyroid receptor superfamily. These ligand-dependent transcription factors are generally located in the nucleus even in the absence of hormone, with the exception of the GR, the androgen receptor (AR), and the MR. The localization of the MR in the absence of hormone is controversial (1). Several investigators reported mainly cytoplasmic localization of the MR in different tissues or transfected cells using immunohistochemistry and immunofluores-cence analysis (2)(3)(4)(5). In contrast, others observed both nuclear and cytoplasmic distribution of the native MR in cells from normal or adrenalectomized animals (6 -9) or of the recombinant MR in transfected cell lines (10). The unliganded MR is part of a soluble hetero-oligomeric complex that includes hsp90, hsp70, and other associated proteins (11,12). The presence of hsp90 in the receptor complex seems to prevent activation of the unliganded MR. Binding of aldosterone initiates a conformational activation within the ligand-binding domain of the receptor that leads to the dissociation of several associated proteins from the receptor, followed by dimerization and nuclear translocation of the activated receptor. Variations in the expression level of associated proteins necessary for nuclear translocation of the MR or differences in the expression levels of proteins regulating intracellular free corticosteroid concentrations might be responsible for the observed controversial findings obtained from different tissues or cell systems. Furthermore, cross-reactivity of the applied anti-MR antibodies with other members of the steroid/thyroid receptor family cannot be excluded in some of the previous investigations.
The mineralocorticoid aldosterone and the glucocorticoids cortisol and corticosterone have similar affinities for binding to the MR (13)(14)(15)(16). Although the circulating concentrations of cortisol (in humans) and corticosterone (in rodents) are 100 -1000 times higher than those of aldosterone (17), cortisol and corticosterone do not lead to MR activation under normal conditions. This is due to the 11␤-hydroxysteroid dehydrogenase (11␤-HSD) activity present in most MR-expressing cells. Two 11␤-HSD enzymes, both localizing to the ER membrane, have been described so far (for review, see Refs. 18 and 19). 11␤-HSD1 is ubiquitously expressed, with the highest activity in the liver (20,21). In vivo, it acts predominantly as a reductase, but it efficiently oxidates 11␤-hydroxyglucocorticoids when measured in intact cells or in cell lysates. The catalytic domain of 11␤-HSD1 is directed toward the ER lumen (22,23). 11␤-HSD1 knockout mice were resistant to hyperglycemia provoked by obesity or stress, but these animals did not display severe defects in glucocorticoid metabolism (24). In contrast, 11␤-HSD2 is mainly expressed in the kidney and in other mineralocorticoid-sensitive tissues and catalyzes exclusively the oxidation of 11␤-hydroxyglucocorticoids (25). Its catalytic moiety protrudes into the cytoplasm (23,26). In patients exhibiting loss-of-function mutations in the gene encoding 11␤-HSD2, e.g. individuals suffering from the syndrome of apparent mineralocorticoid excess (AME) (18,19), or in mice lacking 11␤-HSD2 (27), deficiency of 11␤-HSD2 allows 11␤-hydroxyglucocorticoids to bind to the MR, leading to sodium retention, hypokalemia, and severe hypertension. A similar type of hypertension is induced by ingestion of licorice, which contains glycyrrhetinic acid, a potent inhibitor of 11␤-HSD2 (28 -31). Elevated concentrations of endogenous 11␤-HSD2 inhibitors, such as certain bile acids (32-34) and progesterone metabolites (35,36), and the uptake of exogenous inhibitors, such as glycyrrhetinic acid, carbenoxolone (28 -31), grapefruit flavonoids (37), and furosemide (38,39), have been causally linked with glucocorticoidinduced activation of the MR.
By binding to the MR, aldosterone and, under certain circumstances, glucocorticoids play a major role in the regulation of sodium and potassium homeostasis. Whereas aldosterone normally elicits sodium retention and kaliuresis, 11␤-hydroxyglucocorticoids cause kaliuresis without sodium retention. There is evidence from several in vivo studies that the aldosterone-induced salt retention is blunted by the co-administration of glucocorticoids, suggesting an MR antagonist-like action of glucocorticoids (40 -44). More recently, evidence for distinct physiological effects of aldosterone in aldosterone target tissues and non-epithelial tissues was presented (43). However, the mechanisms underlying these observations remain unclear.
In this study, we evaluated the impact of 11␤-HSD enzyme expression on the intracellular distribution of the MR and analyzed the corticosteroid-induced nuclear translocation of the receptor under various conditions. The results suggest that the MR is specifically associated with 11␤-HSD2 at the ER membrane in the absence of corticosteroids. Moreover, 11␤-HSD2 tightly regulated the access of aldosterone to the MR by inactivating 11␤-hydroxyglucocorticoids, whereby the formed 11-keto products blunted the aldosterone-induced nuclear translocation of the MR.

EXPERIMENTAL PROCEDURES
Expression Plasmids-Wild-type and FLAG epitope-tagged human 11␤-HSD1 and 11␤-HSD2 expression plasmids were constructed as described (23). Their activities were indistinguishable from those of their wild-type enzymes. Constructs expressing the mutant 11␤-HSD2 proteins ⌬L114,E115 (45) and R337C (46) were generated by polymerase chain reaction-based mutagenesis. The mutant 11␤-HSD1 protein K5S,K6S was described previously (23). The plasmid RShMR, expressing the full-length human MR, was a generous gift from Dr. R. M. Evans (Salk Institute, La Jolla, CA) (15). The tagged human MR was obtained by insertion of the nonapeptide hemagglutinin (HA) epitope with the sequence YPYDVPDYA at the carboxyl-terminal end just upstream of the stop codon. The human sarcolipin construct containing a FLAG tag at the C terminus was described previously (47). The N-terminally FLAG epitope-tagged human AR construct was generously provided by Dr. J. J. Palvimo (Institute of Biomedicine, Helsinki, Finland) (48). Expression constructs encoding rat 3␣-hydroxysteroid dehydrogenase (49), human steroid 5␣-reductase type I (50), mouse CYP7A1 cholesterol 7␣-hydroxylase (51), and mouse CYP7B1 oxysterol 7␣-hydroxylase (52) were a generous gift from Dr. D. W. Russell (Texas Southwestern Medical Center, Dallas). All constructs were verified by sequencing.
Cell Culture and Transfection-HEK-293 cells were grown on glass coverslips in six-well plates containing 2 ml of Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Transient transfections were performed using the calcium phosphate precipitation method with 0.5 g of MR cDNA and 0.5 g of 11␤-HSD2 cDNA per well. When expressing the MR alone, 0.5 g of empty pcDNA3 vector DNA was added as a carrier. Transient transfection resulted in 30 -35% of 11␤-HSD2-and MR-positive cells. Six hours post-transfection, cells were washed twice with Hanks' solution, followed by incubation in medium that was charcoal-stripped twice. Analysis of the medium by gas chromatography/mass spectrometry revealed the absence of corticosteroids. 2 Determination of the Transcriptional Activity of MR Constructs-HEK-293 cells were transiently cotransfected with 50 ng of plasmid MMTV-lacZ (53), carrying the bacterial ␤-galactosidase gene (lacZ) under the control of the mouse mammary tumor virus promoter, and either 450 ng of plasmid RShMR, encoding the human MR (15), or the HA-tagged MR cDNA construct. Dulbecco's modified Eagle's medium was replaced by steroid-free medium at 24 h post-transfection. Another 24 h later, cortisol dissolved in methanol or the corresponding amount of methanol was added, and cells were incubated for another 16 h prior to analysis by the Dual-Light assay (Tropix, Inc., Foster City, CA).   distribution of the MR HEK-293 cells were transfected with the HA-tagged MR or cotransfected with the HA-tagged MR and FLAG-tagged wild-type or mutant 11␤-HSD2, FLAG-tagged wild-type or mutant 11␤-HSD1, or unrelated steroid-metabolizing enzymes. Transfected cells were grown for 14 h in steroid-free medium prior to the addition of corticosteroid hormones or inhibitors of 11␤-HSD2 and incubation for another 45 min. Immunostaining using antibodies against the HA and FLAG epitopes was performed as described under "Experimental Procedures." Alternatively, the FLAG-tagged AR was expressed alone or with wild-type 11␤-HSD2 both in the absence and presence of testosterone. Cells staining positively for the MR were divided into three categories: N, predominantly nuclear; N/C, nuclear and cytoplasmic; C, predominantly cytoplasmic. Results represent the percentage of fluorescent cells relative to total cells, whereby 400 -500 fluorescent cells were determined. Determination of 11␤-HSD Enzyme Activity-11␤-HSD enzyme activity was measured in intact cells as described (23). Briefly, transfected HEK-293 cells incubated in charcoal-treated Dulbecco's modified Eagle's medium for 48 h were washed once with Hanks' solution and resuspended in prewarmed (37°C) steroid-free medium. Dehydrogenase activity was measured in a final volume of 20 l containing 30 nCi of [ 3 H]corticosterone and unlabeled corticosterone at different concentrations ranging from 25 nM to 2 M. The reaction was started by mixing the cell suspension with the reaction mixture. Incubation was for 10 -30 min, and the conversion of corticosterone to 11-dehydrocorticosterone was determined by thin-layer chromatography. Enzyme kinetics were analyzed by the Eadie-Hofstee linear transformation of the Michaelis-Menten equation. The protein expression of the different 11␤-HSD constructs was compared by semiquantitative Western blot analysis.
Visualization of Subcellular Localization of MR, AR, and 11␤-HSD2 Enzymes by Immunofluorescence-Transfected cells grown in charcoaltreated medium for 14 h were incubated in the presence or absence of the appropriate concentrations of 11␤-HSD2 inhibitors for 5 min as indicated, followed by the addition of corticosteroids. Cells were incubated for 45 min at 37°C and washed once with 150 mM sodium phosphate (pH 7.4) and 120 mM sucrose, followed by fixation with 4% paraformaldehyde for 10 min. Immunostaining was performed as described (23). FLAG epitope-tagged proteins and HA-tagged proteins were detected using mouse monoclonal anti-FLAG antibody M2 (Sigma, Buchs, Switzerland) and a rat monoclonal anti-HA antibody (Roche Molecular Biochemicals, Rotkreuz, Switzerland) as first antibodies, respectively, and anti-mouse antibody Alexa-488 and anti-rat antibody Alexa-594 (Molecular Probes, Inc., Leiden, The Netherlands) as secondary antibodies, respectively. In coexpression experiments with the AR, human 11␤-HSD2 was detected with a rabbit polyclonal antibody generously provided by Dr. Z. N. Kyossev (Division of Nephrology, Univer-sity of Arkansas, Little Rock, AK) (54) and anti-rabbit secondary antibody Alexa-594. Immunofluorescence was detected using a Model LSM410 confocal microscope (Carl Zeiss, Göttingen, Germany). Quantitation of intracellular localization of the MR or AR was performed by counting the number of cells within an area selected under the transmitted light and determination of positively staining cells that were classified into three different categories: predominantly cytoplasmic staining, comparable intensity of nuclear and cytoplasmic staining, and exclusively nuclear staining. Results were obtained from at least three independent transfection experiments, in which between 400 and 500 stained cells were determined for each sample.

RESULTS
Heterogeneous Distribution of the MR in the Absence of 11␤-HSD2 and Hormone-To investigate the intracellular distribution of the MR in the absence of hormone, we transiently expressed the C-terminally HA epitope-tagged MR in HEK-293 cells and analyzed receptor localization by immunostaining and confocal microscopy. We confirmed the lack of endogenous 11␤-HSD activity in HEK-293 cells (23). The transcriptional activities of wild-type and HA-tagged MRs, determined by a chemiluminescent reporter gene assay, were indistinguishable. No MR transcriptional activity was detected in untransfected cells (data not shown). The anti-HA antibody, specifically recognizing the nonapeptide epitope YPYDVPDYA, was used for fluorescence microscopy detection. Incubation of untransfected HEK-293 cells with the anti-HA antibody did not produce any signal; therefore, the problem of cross-reactivity with other FIG. 2. Colocalization of the MR and 11␤-HSD2 to the ER membrane in the absence of hormone. A-C, HEK-293 cells were cotransfected with the HAtagged MR and FLAG-tagged 11␤-HSD2 and grown on coverslips in steroid-free medium for 20 h at 37°C. Cells were duallabeled with rat anti-HA antibody and mouse anti-FLAG antibody, followed by visualization using anti-rat antibody Alexa-594 and anti-mouse antibody Alexa-488. Samples were analyzed by confocal microscopy. A, green fluorescence representing 11␤-HSD2 expression; B, red fluorescence showing MR expression; C, an overlay of A and B demonstrating colocalization of the MR and 11␤-HSD2 at the ER membrane in the absence of hormone (magnification ϫ 400). D-F, sections of A-C are shown at a higher magnification (ϫ1000), respectively. Localization of the MR at the ER membrane was confirmed by coexpressing untagged wild-type 11␤-HSD2 with the HA-tagged MR and FLAG-tagged sarcolipin, a known ER marker. G, an enlarged section of a cell indicating the reticular expression pattern of FLAG-tagged sarcolipin expression; H, the same section analyzed for MR distribution; I, an overlay of G and H demonstrating the 11␤-HSD2-dependent colocalization of the MR and sarcolipin at the ER membrane in the absence of hormone (magnification ϫ 800). In the presence of 10 nM aldosterone, 11␤-HSD2 (J) did not prevent nuclear translocation of the MR (K), as demonstrated in the overlay (L) (magnification ϫ 400).
receptors inherent with antibodies raised directly against the MR could be avoided. Cells transiently expressing the MR and grown in steroid-free medium exhibited both cytoplasmic and nuclear presence of the receptor (Table I). A heterogeneous distribution pattern with significant cell-to-cell differences was observed (Fig. 1). Low concentrations of aldosterone, cortisol, or corticosterone mediated complete nuclear translocation of the receptor (Table I).
11␤-HSD2-dependent Association of the MR with the ER Membrane in the Absence of Hormone-In most mineralocorticoid-sensitive cells, the MR is coexpressed with 11␤-HSD2. Therefore, we analyzed the impact of 11␤-HSD2 expression on the intracellular distribution of the MR. Dual staining of cells incubated in the absence of hormone and expressing both the HA-tagged MR and FLAG-tagged 11␤-HSD2 and subsequent analysis by confocal microscopy revealed a localization pattern that was dramatically different from that observed when the MR was expressed in the absence of 11␤-HSD2. The MR displayed a typically reticular distribution pattern, with no MR present in the nucleus, suggesting association of the receptor with the ER membrane in the absence of aldosterone (Table I and Fig. 2, A-C). An overlay image of the expression of the MR and 11␤-HSD2 at high resolution indicated the colocalization of both proteins at the ER membrane (Fig. 2, D-F). The ER membrane association of the MR was analyzed further by coexpressing the HA-tagged MR with untagged wild-type 11␤-HSD2 and FLAG-tagged sarcolipin, a small proteolipid known to be located exclusively in the ER membrane (47). Both the MR and sarcolipin showed almost identical reticular expression patterns (Fig. 2, G-I), indicating ER membrane association of the MR. Sarcolipin alone did not affect the intracellular distribution of the MR since, in the absence of 11␤-HSD2, a heterogeneous MR distribution comparable to that shown in Fig. 1 was observed. The MR also displayed exclusively reticular localization when coexpressed in the absence of corticosteroids with 11␤-HSD2 mutant ⌬L114,E115 or R337C, both of which were derived from patients suffering from AME and which retained only very low dehydrogenase activity (Table I) (45,46). Independent of the presence of 11␤-HSD2, the MR translocated into the nucleus upon addition of 10 nM aldosterone (Table I and Fig. 2, J-L). In contrast, the presence of low-tomoderate concentrations of the active 11␤-hydroxyglucocorticoids cortisol and corticosterone did not alter the intracellular distribution of the MR when coexpressed with 11␤-HSD2 due to their conversion to the inactive 11-keto products cortisone and dehydrocorticosterone, respectively (Table I).
Neither 11␤-HSD1 nor Unrelated Steroid-metabolizing Enzymes Affect the Intracellular Distribution of the MR-11␤-HSD1 and 11␤-HSD2 both efficiently catalyze the oxidation of 11␤-hydroxyglucocorticoids in intact cells (23,45). Dehydrogenase activity measurements performed on intact HEK-293 cells expressing 11␤-HSD2 yielded a K m of 56 nM and a V max of 0.77 nM/h/mg of total protein for the substrate corticosterone. 11␤-HSD1, which has been reported to act mainly as a reductase in vivo (55,56), efficiently oxidated the substrate corticosterone in intact cells and yielded a K m of 163 nM and a V max of 0.42 nM/h/mg of total protein. Although 11␤-HSD1 oxidated cortisol and corticosterone in intact cells, it did not protect the MR from cortisol or corticosterone binding, and low concentrations of glucocorticoids were sufficient to mediate nuclear translocation of virtually all of the MR (Fig. 3). Furthermore, coexpression of 11␤-HSD1 with the MR and incubation of the cells in the absence of hormone did not affect MR localization, resulting in a heterogeneous MR distribution similar to that presented in Fig. 1. To test whether a cytosolic oriented 11␤-HSD1 activity would mediate ER membrane association of the MR, we coexpressed the MR and 11␤-HSD1 mutant K5S,K6S, which is fully active, but displays an inverted topology with the catalytic moiety protruding into the cytoplasm (23). Unlike upon coexpression with 11␤-HSD2, the MR was not tethered to the ER membrane upon coexpression with 11␤-HSD1 mutant K5S,K6S or with other hydrophobic steroid-metabolizing enzymes such as 3␣-hydroxysteroid dehydrogenase (49) and steroid 5␣-reductase type I (50). For unknown reasons, coexpression with the cytochrome P450 enzymes CYP7A1 cholesterol 7␣-hydroxylase (51) and CYP7B1 oxysterol 7␣-hydroxylase (52) mediated enhanced nuclear MR localization (Table I).
The Intracellular Distribution of the AR Is Not Altered upon Coexpression with 11␤-HSD2-To test whether the observed 11␤-HSD2-dependent association of the MR with the ER membrane is receptor-specific, we expressed the FLAG-tagged AR either alone or together with 11␤-HSD2 (Table I and Fig. 4) and analyzed the intracellular distribution of the AR in both the absence and presence of the androgen testosterone. In the absence of testosterone, the AR displayed a heterogeneous distribution similar to the pattern observed for the MR. Unlike in case of the MR, coexpression of 11␤-HSD2 did not tether the AR to the ER membrane. Testosterone mediated complete nuclear translocation of the AR independent of the presence of 11␤-HSD2.
Glucocorticoid-induced Nuclear Translocation of the MR Due to Compromised 11␤-HSD2 Activity-11␤-HSD2 prevents activation of the MR by 11␤-hydroxyglucocorticoids by efficiently converting them into 11-ketoglucocorticoids. A limited function of 11␤-HSD2 is expected to result in increased intracellular 11␤-hydroxyglucocorticoid concentrations and may cause subsequent activation of the MR. Therefore, we investigated the glucocorticoid-induced nuclear translocation of the MR under various conditions of reduced 11␤-HSD2 activity. The most severe form of glucocorticoid-induced MR activation is observed in patients suffering from AME, where mutations in 11␤-HSD2 result in almost completely inactive enzymes (18,19). Both 11␤-HSD2 mutants ⌬L114,E115 and R337C were unable to FIG. 3. 11␤-HSD1 cannot prevent the cortisol-dependent nuclear translocation of the MR. HEK-293 cells coexpressing the HA-tagged MR and FLAG-tagged 11␤-HSD1 were grown in steroid-free medium prior to a 45-min incubation in medium supplemented with 10 nM cortisol. Cells were stained as described in the legend to Fig. 2. A, reticular expression pattern of 11␤-HSD1; B, nuclear localization of MR; C, an overlay of A and B demonstrating the cortisol-induced MR activation in the presence of 11␤-HSD1 (magnification ϫ 400). protect the MR from glucocorticoid binding even at low concentrations of cortisol (Fig. 5, A-C) or corticosterone (data not shown). At 10 nM cortisol, ϳ80% of MR molecules were present in the nucleus, whereas Ͼ95% were nuclear at 50 nM (Table I).
We next analyzed the effects of glycyrrhetinic acid and furosemide, two known inhibitors of 11␤-HSD2 (25,38,39,57), on the glucocorticoid-induced activation of the MR. Both glycyrrhetinic acid and furosemide did not alter the 11␤-HSD2-dependent tethering of the MR to the ER membrane in the absence of 11␤-hydroxyglucocorticoids (Table I). In contrast, in the presence of both cortisol and glycyrrhetinic acid, the MR was no longer associated with the ER membrane (Fig. 5, D-F), with an increasing amount of the receptor located in the nucleus in the presence of increasing inhibitor concentrations. Both the 11␤-HSD2 inhibitors glycyrrhetinic acid and furosemide mediated a dose-dependent glucocorticoid-induced activation of nuclear translocation of the MR (Fig. 6). Approximately 50% of the MR was nuclear at 2.5 M glycyrrhetinic acid, and almost all receptor molecules were located in the nucleus at 50 M. Much higher concentrations of furosemide were required to obtain similar glucocorticoid-induced activation of the MR, whereby half-maximal activation was observed at 250 M. Furosemide is a much less potent inhibitor of 11␤-HSD2 than glycyrrhetinic acid, and 10 nM cortisol did not lead to complete nuclear translocation of the MR, even in the presence of very high furosemide concentrations such as 1 mM (Fig. 6).
11␤-HSD2 has a relatively low K m for its substrates. Therefore, we investigated whether elevated glucocorticoid concentrations might cause saturation of the enzyme, thereby leading to glucocorticoid-induced nuclear translocation of the MR. Both the substrates cortisol and corticosterone displayed a saturation effect, with half-maximal nuclear translocation of the MR at 2 mM cortisol and 500 M corticosterone (Fig. 7). The concentrations required to obtain half-maximal activation of the MR were 5-10-fold higher than the concentrations required to obtain half-maximal activity of 11␤-HSD2 in intact HEK-293 cells.  -488 (A, B, E, and F). Wild-type 11␤-HSD2 was detected with a rabbit polyclonal anti-11␤-HSD2 primary antibody and anti-rabbit secondary antibody Alexa-594 (C and G). Overlay pictures of AR expression and 11␤-HSD2 expression are shown in D and H, respectively (magnification ϫ 100).

FIG. 5. Cortisol-induced nuclear translocation of the MR due to compromised 11␤-HSD2 activity.
A-C, HEK-293 cells were cotransfected with the HA-tagged MR and 11␤-HSD2 mutant ⌬L114,E115, which was derived from a patient with AME. Cells were grown on coverslips in steroid-free medium, followed by a 45-min incubation in medium containing 10 nM cortisol and immunodetection as indicated in the legend to

11␤-HSD2 Regulates the Intracellular Localization of the MR
Glucocorticoids Block the Aldosterone-induced Nuclear Translocation of the MR by an 11␤-HSD2-dependent Mechanism-In the absence of 11␤-HSD2, both the mineralocorticoid aldosterone and the 11␤-hydroxyglucocorticoids cortisol and corticosterone activated the MR and led to complete nuclear receptor translocation (Table I). The 11-ketoglucocorticoids cortisone and 11-dehydrocorticosterone did not activate the MR. When the MR and 11␤-HSD2 were coexpressed, cortisol and corticosterone did not cause receptor activation due to their efficient oxidation to cortisone and corticosterone, respectively. In contrast, aldosterone induced nuclear MR translocation in both the presence and absence of 11␤-HSD2. Surprisingly, when cells coexpressing the MR and 11␤-HSD2 were preincubated for 5 min with the 11-ketoglucocorticoid cortisone or 11-dehydrocorticosterone or with the 11␤-hydroxyglucocorticoid cortisol or corticosterone, then the aldosterone-dependent MR activation was almost completely blocked. This glucocorticoid-mediated blunting of aldosterone-dependent MR translocation was not observed when AME mutant 11␤-HSD2 proteins and the MR were coexpressed, indicating that a proper function of 11␤-HSD2 is required to tether the MR to the ER membrane. Furthermore, association of the MR with the ER membrane was clearly dependent on 11␤-HSD2 since neither cortisone nor 11-dehydrocorticosterone could block the aldosterone-dependent MR activation in the absence of 11␤-HSD2 (Table II). DISCUSSION Here, we transiently expressed the epitope-tagged MR in HEK-293 cells in the absence of hormone and 11␤-HSD2 and observed a heterogeneous MR distribution pattern (Fig. 1). This is in line with the findings of Fejes-Toth et al. (10), who reported a heterogeneous localization pattern of a green fluorescent protein-MR chimeric construct in the absence of steroids in transfected CV1 or Chinese hamster ovary cells, both lacking endogenous 11␤-HSD2 activity. 3 The reason for the heterogeneous MR distribution and the significant cell-to-cell differences is not clear at present. Fejes-Toth et al. (10) speculated that the MR distribution might be cell cycle-dependent.
In contrast to the mixed nuclear/cytoplasmic MR distribution observed in the absence of 11␤-HSD2 expression (Fig. 1), we obtained a clearly reticular localization pattern upon coexpression of the receptor with 11␤-HSD2 (Fig. 2). This is in contrast to the current literature reporting either nuclear or cytoplasmic localization of the MR. However, previous fluorescence microscopy studies were performed at relatively low resolution, 3 A. Odermatt, unpublished data. 11␤-HSD2 Regulates the Intracellular Localization of the MR or the recombinant MR was expressed in cell lines in the absence of 11␤-HSD2. In this study, dual staining of the epitope-tagged MR and 11␤-HSD2 or the ER marker sarcolipin (47), along with confocal microscopy and high resolution analysis, allowed us to distinguish between localization of the MR in the nucleus, the cytoplasm, or at the ER membrane. The following evidence suggested that 11␤-HSD2 specifically mediates association of the MR with the ER membrane in the absence of steroids and that this effect is dependent on a structural component of 11␤-HSD2 rather than on its dehydrogenase function. 1) Mutant 11␤-HSD2 enzymes ⌬L114,E115 and R337C, which are almost completely inactive and which are causal for severe hypertension in humans (45,46), are sufficient to mediate ER membrane association of the MR. 2) Although 11␤-HSD1 efficiently oxidated 11␤-hydroxyglucocorticoids in intact cells, both wild-type 11␤-HSD1, whose catalytic domain is oriented toward the ER lumen, and mutant K5S,K6S, which is fully active, but whose catalytic moiety is directed toward the cytoplasm, did not alter the heterogeneous intracellular distribution of the MR. 3) The coexpression of hydrophobic steroid-metabolizing enzymes that are located in the ER membrane and whose catalytic moieties are like those of 11␤-HSD2 protruding into the cytoplasm did not tether the MR to the ER membrane. 4) Association of the MR with the ER membrane seems to be receptor-specific since the intracellular distribution of the AR was not affected by coexpression with 11␤-HSD2. 5) The two known inhibitors of 11␤-HSD2, glycyrrhetinic acid and furosemide, did not affect MR localization in the absence of steroids. Regarding the "guardian role" of 11␤-HSD2 by inactivating 11␤-hydroxyglucocorticoids and rendering aldosterone specificity of the MR, a direct interaction of 11␤-HSD2 with the MR or localization of 11␤-HSD2 in close proximity to the MR would allow the most efficient protection of the MR from 11␤-hydroxyglucocorticoid binding. A possible interaction could occur either directly or via a protein that is associated with the unliganded MR complex.
At moderate concentrations of cortisol, wild-type 11␤-HSD2 fully protected the MR and retained the receptor at the ER membrane. The complete nuclear translocation of the MR upon addition of only 10 nM cortisol demonstrated the lack of functional protection of the MR by 11␤-HSD2 mutants ⌬L114,E115 and R337C (Fig. 5), thus reflecting the severe phenotype of AME patients bearing these mutations (45,46). Glucocorticoidinduced nuclear translocation of the MR was also observed in the presence of the two known 11␤-HSD2 inhibitors, glycyrrhetinic acid and furosemide (Figs. 5 and 6). By directly inhibiting 11␤-HSD2, both glycyrrhetinic acid, a constituent of licorice, and furosemide, which is widely used as a diuretic drug in clinical treatment, mediated glucocorticoid-induced nuclear translocation of the MR, with half-maximal nuclear translocation obtained at 2.5 and 500 M, respectively (Fig. 6). Previous experiments using cell lysates revealed 11␤-HSD2 K i values for glycyrrhetinic acid of between 5 and 10 nM (25,57) and for furosemide of ϳ30 M (39). The substantially higher inhibitor concentrations required for glucocorticoid-induced nuclear translocation of the MR may be explained by the lower actual intracellular concentration of these inhibitors due to the rather inefficient uptake by the cell. Both inhibitors did not alter the 11␤-HSD2-dependent association of the MR with the ER membrane or activate the MR in the absence of 11␤-HSD2 (data not shown). The characterization of substances with mineralocorticoid effects in the HEK-293 expression system offers the advantage of distinguishing between direct effects on the MR and indirect effects via inhibition of 11␤-HSD2. This assay also provides an initial estimation of the magnitude of MR activation caused by a given effector.
Our results demonstrate that saturation of 11␤-HSD2 at high concentrations of cortisol and corticosterone leads to nuclear translocation of the MR (Fig. 7). The half-maximal nuclear translocation of the MR was reached at glucocorticoid concentrations that were 5-10-fold higher than the concentrations required to attain half-maximal activity of 11␤-HSD2 in intact cells. The pulsatile nature of corticosterone release suggests that 11␤-HSD2 may be transiently saturated at the peak of glucocorticoid release in vivo (58). In addition, saturation of 11␤-HSD2 may play a critical role in situations of elevated glucocorticoid levels during an inflammatory response, upon administration of high doses of glucocorticoids for therapeutic reasons, or under conditions of reduced 11␤-HSD2 expression. Inhibition or saturation of 11␤-HSD2 is expected to result in 11␤-hydroxyglucocorticoid-induced activation of both the MR and GR. Using the highly selective GR agonist RU28362 in the presence of carbenoxolone, Funder et al. (59) demonstrated that occupancy of the GR in epithelial aldosterone target tissues is similarly followed by an electrolyte response equivalent to that of aldosterone.
Receptor translocation assays in the HEK-293 cell system revealed that 11-ketoglucocorticoids abrogate the aldosterone-TABLE II Glucocorticoids block the aldosterone-induced activation of the MR by an 11␤-HSD2-dependent mechanism HEK-293 cells were transfected with the HA-tagged MR or cotransfected with the HA-tagged MR and either FLAG-tagged wild-type 11␤-HSD2 or loss-of-function 11␤-HSD2 mutant ⌬L114,E115 or R337C. The glucocorticoid-mediated blocking of aldosterone-induced nuclear translocation of the MR was determined by the addition of the corresponding glucocorticoid for 5 min, followed by the addition of aldosterone and further incubation for 45 min. Receptor localization was analyzed as described under "Experimental Procedures" and as indicated in Table I induced nuclear translocation of the MR by an 11␤-HSD2-dependent mechanism (Table II). 11-Ketoglucocorticoids did not block aldosterone-dependent MR activation in the absence of 11␤-HSD2, suggesting an interaction between 11␤-HSD2 and the receptor complex consisting of the MR and associated proteins. Our finding that glucocorticoids can block aldosteroneinduced MR activation is supported, at least in part, by evidence obtained from animal studies. Young et al. (42) and Young and Funder (60) reported that the peripheral infusion of high doses of aldosterone in rats caused hypertension, cardiac hypertrophy, and cardiac fibrosis. These effects were substantially blocked by the co-administration of corticosterone at a 30-fold excess. In analogy with our results from the HEK-293 expression system, 11-dehydrocorticosterone, formed by 11␤-HSD2 upon injection of corticosterone, may have abrogated the aldosterone-induced effects. In addition, our observations that 11-keto-and 11␤-hydroxyglucocorticoids block aldosterone-induced MR translocation might allow the interpretation of results from animal studies conducted by Morris and co-workers (40,41,44). In these studies, co-administration of 11-keto-or 11␤-hydroxyglucocorticoids with aldosterone blunted aldosterone-dependent sodium retention in adrenalectomized rats (40,44) and attenuated sodium transport in isolated toad bladders (41). Recently, Morris et al. (44) mentioned in a review a preliminary set of unpublished experiments indicating that 11-dehydrocorticosterone can block, by an unknown mechanism, aldosterone-dependent MR activation in experiments using COS-7 cells transfected with the MR and a luciferase reporter system. This observation is in line with our study. Based on the colocalization of the MR with 11␤-HSD2 at the ER membrane in the absence of hormone and the observed 11-ketoglucocorticoid-dependent block of aldosterone-induced MR translocation, we hypothesize that 11␤-HSD2 functionally interacts with the receptor complex consisting of the unliganded MR and associated proteins. Cortisone and 11-dehydrocorticosterone may block a conformational activation of the MR and the subsequent dissociation of associated proteins that are required for dimerization and nuclear translocation of the receptor. The cortisone-mediated blocking of the activation of the MR by aldosterone was lost in both ⌬L114,E115 and R337C, suggesting a defect in conformational changes in mutant 11␤-HSD2. As a consequence of our hypothesis, the MR would be activated by aldosterone only in the nadir of pulsatile glucocorticoid secretion, when virtually no 11-ketosteroids are present. At periods of higher glucocorticoid concentrations, the 11-keto products formed by 11␤-HSD2 are expected to block the activation of the MR, and a reduced 11␤-HSD2 activity would allow the activation of the MR by 11␤-hydroxyglucocorticoids. The inhibition of aldosterone-induced MR translocation by the formation of 11-ketosteroids from the 11␤-hydroxyglucocorticoids might explain, at least in part, the puzzling observation that physiological 11␤-hydroxyglucocorticoids given in high doses act as MR antagonists in the kidney and heart (43). Thus, the elucidation of the molecular mechanisms for the interaction between 11-ketoglucocorticoids, 11␤-HSD2, and the MR in the future will be of practical relevance.