Appropriate Function of 11 (cid:1) -Hydroxysteroid Dehydrogenase Type 1 in the Endoplasmic Reticulum Lumen Is Dependent on Its N-terminal Region Sharing Similar Topological Determinants with 50-kDa Esterase*

By interconverting glucocorticoids, 11 (cid:1) -hydroxy-steroid dehydrogenase type 1 (11 (cid:1) -HSD1) exerts an important pre-receptor function and is currently considered a promising therapeutic target. In addition, 11 (cid:1) HSD1 plays a potential role in 7-ketocholesterol metabolism. Here we investigated the role of the N-terminal region on enzymatic activity and addressed the relevance of 11 (cid:1) -HSD1 orientation into the endoplasmic reticulum (ER) lumen. Previous studies revealed that the luminal orientation of 11 (cid:1) -HSD1 and 50-kDa ester-ase/arylacetamide deacetylase ( E 3) is determined by their highly similar N-terminal transmembrane do-mains. Substitution of Lys 5 by Ser in 11 (cid:1) -HSD1, but not of the analogous Lys 4 by Ile in E 3, led to an inverted topology in the ER membrane, indicating the existence of a second topological determinant. Here we identified Glu 25 /Glu 26 in 11 (cid:1) -HSD1 and Asp 25 in E 3 as the second determinant for luminal orientation. Our results suggest that the exact Oxoreduction of 7KC was determined as described (8). Briefly, freshly prepared lysates were suspended in buffer TG1 and 400 (cid:2) M NADP (cid:2) , 400 n M 7KC, and 30 nCi of 3 H-labeled 7KC added, followed by incubation at 37 °C for 15 or 30 min. Intact cells were incubated in steroid-free medium in the absence of cofactor for 30 or 60 min. The reactions were stopped by adding an excess of unlabeled oxycholesterols in methanol, followed by separation by TLC and scintillation counting.

In humans, 11␤-HSD1 1 catalyzes the reduction of biologically inactive cortisone to active cortisol, thereby playing an essential role in the local activation of the glucocorticoid receptor. Recent animal experiments provided insight into the pathophysiological role of 11␤-HSD1. Mice deficient of 11␤-HSD1 were resistant to hyperglycemia induced by obesity or stress (1), whereas transgenic mice overexpressing 11␤-HSD1 developed visceral obesity with insulin resistance and dyslipidemia. In addition, overexpression of 11␤-HSD1 in adipose tissue caused salt-sensitive hypertension mediated by an activated renin-angiotensin system (2,3). Experiments in obese and diabetic mice treated with a specific 11␤-HSD1 inhibitor showed reduced blood glucose levels and increased insulin sensitivity (4,5). Therefore, 11␤-HSD1 is currently considered a promising drug target for the treatment of cognitive dysfunction in elderly men and patients with obesity and type 2 diabetes mellitus (6,7). However, whether 11␤-HSD1 is indeed a suitable target for therapeutic treatment of excessive glucocorticoid actions remains to be tested.
Despite its importance, relatively little is known about the molecular mechanisms by which 11␤-HSD1 exerts its physiological function. 11␤-HSD1 belongs to the family of short-chain dehydrogenases-reductases, characterized by a core domain with conserved regions, including the Rossmann fold for binding of the cofactor NADP(H) and the Tyr-(Xaa)3-Lys motif in the catalytic site, and less conserved N-and C-terminal sequences (17,18). Several studies using 11␤-HSD1 constructs with N-terminal deletions demonstrated that this part of the enzyme is not only essential to anchor the enzyme to the ER membrane but also for stability and catalytic activity (19 -22). The residues involved, however, were not identified.
We have shown previously (19,23) that the single N-terminal transmembrane helix is responsible for the luminal orientation of the catalytic moiety of 11␤-HSD1. However, in contrast to 11␤-HSD2, whose cytoplasmic orientation is required to tether the mineralocorticoid receptor to the ER membrane in the absence of steroid and that prevents its occupation by cortisol through conversion of cortisol to cortisone (24), the physiological role of the luminal orientation of 11␤-HSD1 remained unclear. In our previous study (19), we demonstrated that substitution of Lys 5 by Ser in the short cytoplasmic N terminus of 11␤-HSD1 inverted its topology in the ER membrane. Surprisingly, no difference was found for the oxoreduction of 11-dehydrocorticosterone between the mutant enzyme with cytoplasmic orientation and wild-type 11␤-HSD1 upon expression in HEK-293 cells.
The N-terminal transmembrane span of 11␤-HSD1 is highly similar to that of the 50-kDa esterase/arylacetamide deacetylase (E3) (23), two proteins that are otherwise unrelated (Fig.  1A). E3, which is highly expressed in liver and adrenal glands and to a lesser extent in small intestine, stomach, kidney, and pancreas (25), acts as an N-deacetylase catalyzing hydrolytic reactions and plays a potential role in the prevention of arylamine-induced carcinogenesis (26). E3 might be involved, like the putative triacylglycerol hydrolase E1, in the assembly of hepatic very low density lipoprotein in the ER lumen (25). A significantly reduced expression of E3 was observed in insulindeficient diabetes, which is characterized by a severe decrease in the secretion of hepatic very low density lipoprotein triacylglycerol. Unlike Lys 5 in 11␤-HSD1, substitution of the analogous Lys 4 in E3 had no effect on topology, suggesting the existence of a second determinant for the orientation of these enzymes in the ER membrane (19,23).
Here we tested the hypothesis whether the negatively charged residues on the luminal side of the transmembrane span might act as a second determinant for the orientation of these two enzymes in the ER membrane, and we studied the effect of charged amino acid residues on 11␤-HSD1 activity. In addition, we investigated the role of the luminal orientation of 11␤-HSD1 on the oxidation of cortisol and the oxoreduction of cortisone and 7KC.
Construction of Plasmids-The plasmid for expression of FLAG epitope-tagged human 11␤-HSD1 was constructed as described (19). Mutant 11␤-HSD1 constructs were generated by site-directed mutagenesis according to the QuickChange mutagenesis kit (Stratagene, La Jolla, CA). The rabbit E3-enhanced green fluorescent protein (EGFP) chimera, containing the N-terminal membrane anchor sequence of E3 with a C-terminal EGFP, was constructed as described (23). For immunofluorescence experiments, a construct containing the N-terminal 34 amino acids of rabbit E3 followed by a FLAG epitope tag was generated by three-piece ligation of an EcoRI-BamHI fragment containing the N-terminal 34 amino acids of E3, a BamHI-XbaI fragment encoding the FLAG epitope (MDYKDDDD), and an EcoRI-XbaI pcDN3 expression vector fragment. Mutant E3-FLAG and-EGFP constructs were made by site-directed mutagenesis. All constructs were verified by sequencing.
Cell Culture and Transient Transfection-HEK-293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 4.5 g/liter glucose, 50 units/ml penicillin, 50 mg/ml streptomycin, and 2 mM glutamine. For microscopy experiments, cells were seeded at ϳ10% confluence onto glass coverslips placed in 6-well plates. After growth for 24 h at 37°C under 5% CO 2 , cells were transfected by Ca 2ϩ -phosphate precipitation with 2 g/well of the corresponding ex-pression plasmid and 1 g/well EGFP control plasmid. After incubation for 8 h, the medium was replaced to remove the Ca 2ϩ -phosphate precipitate. For isolation of microsomes and activity analysis, cells were seeded to 50% confluence in 10-cm dishes and transfected, and medium was replaced or cells were washed three times with steroid-free Dulbecco's modified Eagle's medium (doubly charcoal-treated).
Selective Permeabilization and Immunofluorescence Analysis-Immunofluorescence analysis was performed 48 h post-transfection as described previously (24). Briefly, paraformaldehyde (4%) fixed cells coexpressing the corresponding FLAG-tagged construct and EGFP control were washed four times with the buffer NAPS (150 mM sodium phosphate, pH 7.4, 120 mM sucrose). For complete permeabilization of membranes, cells were blocked in NAPS buffer containing 1% milk powder and either 0.5% Triton X-100, for 11␤-HSD1 constructs, or 1% saponin, for E3 constructs. The fluorescence signal obtained for E3 constructs was much stronger when saponin was used instead of Triton X-100, an observation made previously with N-terminally FLAG-tagged 11␤-HSD1. The reason is unknown, but it seems that Triton X-100 interferes with antibody-epitope interactions in close proximity to the membrane. For semipermeabilization of the plasma membrane, 25 M digitonin was used (19). FLAG-tagged constructs were detected using anti-FLAG antibody M2 as primary antibody and ALEXA-594 goat anti-mouse secondary antibody. EGFP served as a transfection efficiency control. Following incubation with antibodies in NAPS containing 0.1% milk, samples were washed four times with NAPS and treated with Slow Fade Antifade kit (Molecular Probes). Samples were analyzed on a Carl Zeiss confocal microscope LSM410 (Carl Zeiss, Goettingen, Germany).
Isolation of Microsomes from HEK-293 Cells-HEK-293 cells were transfected with 8 g of the corresponding expression plasmid per 10-cm dish. Cells from four dishes were collected 48 h post-transfection, washed with phosphate-buffered saline, and centrifuged at 150 ϫ g for 3 min, and the pellet was resuspended in 1.2 ml of buffer containing 10 mM Tris-HCl, pH 7.5, 0.5 M MgCl 2 , and protease inhibitor (Complete, Roche Diagnostics). Cells were sonicated; 1.2 ml of isotonic buffer (1 M sucrose, 10 mM Tris-HCl, pH 7.5, 2.5 M NaCl, 1 mM dithiothreitol) was added, and lysates were centrifuged at 1,000 ϫ g for 10 min at 4°C, followed by centrifugation at 11,000 ϫ g for 10 min at 4°C and 100,000 ϫ g for 1 h at 4°C. The pellet was resuspended in 300 l of buffer containing 10 mM Tris-HCl, pH 7.5, 0.5 M sucrose, 1.25 M NaCl, and 0.5 mM dithiothreitol. The protein concentration was adjusted to 1 mg/ml, and microsomal preparations were shock-frozen in liquid nitrogen and stored at Ϫ70°C until analysis.
Protease Protection Assay and Immunoblotting-Microsomes (20 g of proteins) were incubated in a total volume of 25 l with 0.25 g/l proteinase K (Roche Diagnostics) for 30 min on ice in the presence or absence of 0.5% Triton X-100. Proteinase K was inactivated by adding 2.5 l of 10 mM phenylmethylsulfonyl fluoride in isopropyl alcohol for 5 min, followed by solubilization of proteins with SDS sample buffer and boiling for 5 min. Proteins were subjected to SDS-PAGE and Western blot analysis using anti-FLAG antibody M2 or anti-EGFP antibody as primary antibodies and secondary horseradish peroxidase-conjugated antibody (Roche Diagnostics). Antibody binding was visualized using the enhanced chemiluminescence Western detection system (Pierce).
Assay for 11␤-HSD-11␤-HSD1-dependent oxidation and oxoreduction were measured as described previously (19,27). Briefly, the rate of conversion of cortisol to cortisone or the reverse reaction was determined in a final volume of 20 l in a TG1 buffer (20 mM Tris-HCl, pH 7.4, 1 mM EGTA, 1 mM EDTA, 1 mM MgCl 2 , 100 mM NaCl, 20% glycerol) supplemented with 400 M NADP ϩ or NADPH, 30 nCi of 3 H-labeled substrate, and unlabeled substrate at different concentrations ranging from 12.5 nM to 2 M. Cell lysates were prepared by washing transfected cells with phosphate-buffered saline, centrifugation for 3 min at 150 ϫ g, removal of the supernatant, and quick-freezing the cell pellet in a dry ice ethanol bath. Cell pellets were resuspended in buffer TG1, sonicated, and used immediately for activity assays. The reactions were started by mixing 10 l of cell extract corresponding to 2-5 g of total proteins with 10 l of reaction mixture and incubated for 10 -20 min at 37°C. When measuring activities in intact cells, steroid-free medium was used, and cofactor was omitted. To assess the effect of cyclosporin A, cells were preincubated for 15 min with 50 M cyclosporin A prior to addition of cortisone. The reactions were stopped by adding an excess of unlabeled steroids in methanol, followed by separation of steroids by TLC. Enzyme kinetics were analyzed by nonlinear regression using Data Analysis Toolbox (MDL Information Systems Inc.) assuming firstorder rate kinetics. Hill coefficients for measurements in lysates were ranging between 0.94 and 1.17. Data represent mean Ϯ S.D. of at least four independent transfections.
Oxoreduction of 7KC was determined as described (8). Briefly, freshly prepared lysates were suspended in buffer TG1 and 400 M NADP ϩ , 400 nM 7KC, and 30 nCi of 3 H-labeled 7KC added, followed by incubation at 37°C for 15 or 30 min. Intact cells were incubated in steroid-free medium in the absence of cofactor for 30 or 60 min. The reactions were stopped by adding an excess of unlabeled oxycholesterols in methanol, followed by separation by TLC and scintillation counting.

Sequence Comparison of the N Termini of 11␤-HSD1 and
E3-An alignment of the sequences of human 11␤-HSD1 and rabbit E3 revealed that both proteins share a similar N-terminal region (Fig. 1A) with a short and positively charged Nterminal cytoplasmic part, a single transmembrane span followed by negatively charged residues immediately downstream of the membrane helix, and a large C-terminal luminal domain. In previous studies, we have shown that substitution of Lys 5 by Ser led to inverted orientation of 11␤-HSD1 in the ER membrane (19), whereas substitution of the analogous Lys 4 by Ile in E3 had no effect on topology (23), indicating the existence of a second determinant for the topology of these two proteins.
The Cytoplasmic Lys 4 and the Luminal Asp 25 Determine the Topology of E3-To test the hypothesis that the negatively charged residues immediately downstream of the membrane span represent this second topology determining motif, we generated a fusion protein encoding the N-terminal 34 amino acids of E3 followed by a FLAG epitope for facilitated immunodetection, and we subjected this construct to site-specific mutagenesis (Fig. 1B). Cells coexpressing the corresponding FLAG-tagged construct and EGFP control protein were either completely permeabilized with 1% saponin or the plasma membrane was selectively permeabilized with 25 M digitonin, allowing restricted access of the anti-FLAG antibody to the cytosolic compartment. Cells were then incubated with anti-FLAG antibody and red fluorescent secondary antibody and analyzed by fluorescence microscopy. Cells expressing EGFP were analyzed for the presence of red fluorescence signal from the corresponding E3 construct, whereby 200 -300 cells were counted in a typical experiment. By using saponin, over 95% of cells expressing wild-type E3 stained positive but less than 5% when digitonin was used, in line with the luminal orientation of E3 (Fig. 2). Mutant K4I showed wild-type orientation, confirming previous findings (23). To investigate the role of negatively charged residues at the luminal side of the transmembrane helix, we replaced Asp 25 , Glu 28 , and Glu 29 to Lys in either wild-type E3 or mutant K4I (Fig. 1B). Substitution of the negatively charged residues by Lys led to inverted insertion into the ER membrane in mutant K4I/D25K/E28K/E29K but not in mutant D25K/E28K/E29K (Fig. 2). Additional mutagenesis revealed that substitution of Asp 25 /Glu 28 /Glu 29 to Asn 25 /Gln 28 / Gln 29 and a single substitution of Asp 25 by Lys in mutant K4I both were sufficient to invert the orientation in the ER membrane. All of the mutant proteins analyzed showed exclusive localization to the ER membrane, indicating that they are not important for ER retention. These results demonstrate that both Lys 4 and Asp 25 are essential for determining the topology of E3. The effects on topology were confirmed by subjecting microsomal vesicles expressing wild-type or mutant constructs to proteinase K digestion in the presence or absence of detergent (data not shown).
The Luminal Residues Glu 25 and Glu 26 Are Important Determinants for the Topology of 11␤-HSD1-In analogy to Asp 25 of E3, 11␤-HSD1 contains two glutamate residues (Glu 25 and Glu 26 ) immediately downstream of the membrane span. To investigate whether the luminal di-glutamate motif plays a role in determining the topology of 11␤-HSD1, we generated a series of constructs with a C-terminal FLAG epitope (Fig. 1C) and analyzed them in fully permeabilized (0.5% Triton X-100) or semipermeabilized (25 M digitonin) cells by immunofluorescence detection. Substitution of both Glu 25 and Glu 26 to Lys led to an inverted orientation of 11␤-HSD1 in the ER membrane (Fig. 3). Substitution of Glu 25 by Lys and Glu 26 by Gln (mutant E25K/E26Q) did not change the orientation of the mutant enzyme, indicating that the change from positive to negative charge at both positions is necessary to invert the topology of 11␤-HSD1.
We have shown previously that mutant K5S but not K6S showed inverted topology. To assess a potential role of Lys 6 in determining topology, we mutated Glu 25 and Glu 26 in mutant K6S. Independent of the residue at position 6, the orientation of the enzyme was only inverted when both Glu 25 and Glu 26 were changed to Lys (Fig. 1C and Fig. 3). These results indicate that Lys 6 is not involved in determining the topology of 11␤-HSD1. In addition, all mutations analyzed in the present study did not alter the restricted expression of 11␤-HSD1 in the ER membrane (not shown).
To confirm the effect of these mutations on 11␤-HSD1 topology, microsomal vesicles expressing FLAG-tagged wild-type or mutant 11␤-HSD1 constructs were subjected to proteinase K digestion in the presence or absence of detergents, followed by separation on SDS-PAGE and detection with anti-FLAG antibody on Western blots. As seen in Fig. 4, in intact vesicles mutants E25K/E26K and K6S/E25K/E26K were completely digested by proteinase K, whereas wild-type 11␤-HSD1 and all other mutants were protected from proteinase K-dependent digestion. Upon addition of Triton X-100, all constructs were digested. These results are consistent with the findings from fluorescence microscopic analyses of semipermeabilized cells, demonstrating an essential role of Glu 25 and Glu 26 in determining the orientation of 11␤-HSD1 in the ER membrane.
The Effect of Charged Residues in the N-terminal Region of 11␤-HSD1 on Enzymatic Activity in Lysates-To assess the effect of charged residues in the N-terminal region of 11␤-HSD1 on enzymatic activity, the reduction of cortisone to cortisol was measured in lysates of cells expressing wild-type or mutant 11␤-HSD1. None of the mutations had a significant effect on apparent K m values (Table I); however, replacement of both Glu 25 and Glu 26 by Gln or Lys led to a significant decrease in activity (p Ͻ 0.01), with a tendency for a loss of V max when substituting the negatively charged Glu residues first by polar Gln and then by positively charged Lys. The fact that mutant K6S/E25K/E26Q with luminal orientation and mutant K6S/ E25K/E26K with cytoplasmic orientation both had similarly reduced activities suggests that the di-glutamate motif at position 25/26 stabilizes enzymatic activity independent of the orientation of 11␤-HSD1 in the ER membrane. In addition, mutant K5S/K6S showed kinetic values comparable with that of wild-type 11␤-HSD1, indicating that the orientation in the ER membrane itself does not affect enzymatic activity. Similar effects of these mutations were observed for the oxidation of corticosterone (data not shown).
We also investigated the effect of the replacement of Lys 35 and Lys 36 by Ser immediately upstream of the conserved cofactor binding site on intracellular localization and enzymatic activity. Although expression level and intracellular distribution were not altered, this mutation led to a complete loss of function, demonstrating that Lys 35  elucidate the potential role of the luminal orientation of 11␤-HSD1 on its activity, we compared the interconversion of cortisone and cortisol in lysates and intact HEK-293 cells expressing either luminally oriented wild-type 11␤-HSD1 or cytoplasmically oriented mutant K5S/K6S. Measurements in lysates yielded similar kinetic parameters for the oxidation of cortisol and the oxoreduction of cortisone with both enzymes (Tables II and III). In intact cells, the oxoreduction of cortisone was not different between wild-type 11␤-HSD1 and mutant K5S/K6S; however, mutant K5S/K6S showed 50% lower V max (p Ͻ 0.01) than wild-type for the oxidation of cortisol without a change in K m . Approximately 50% lower V max but similar K m was also observed for the oxidation of corticosterone by mutant K5S/K6S (not shown).
Next, we investigated whether the inhibition of P-glycoprotein, which has been shown to catalyze the efflux of various steroids from the cytoplasm (28,29), by cyclosporin A (30) may differentially affect the kinetic parameters of wild-type 11␤-HSD1 and mutant K5S/K6S. As shown in Tables II and III, cyclosporin A led to a 60% increase in catalytic efficiency of the cytoplasmic mutant K5S/K6S (p Ͻ 0.05 compared with untreated cells). In contrast, the catalytic efficiency of wild-type 11␤-HSD1 increased only by 18% (not significant). The effect on K5S/K6S tended to be more pronounced than that on wildtype enzyme, although the difference did not reach statistical significance. These results suggest that the inhibition of the activity of endogenous P-glycoprotein in HEK-293 cells leads to an increase in cytoplasmic cortisone concentration, favoring the activity on the cytoplasmic side.
The Luminal Orientation of 11␤-HSD1 Is Required for the Oxoreduction of 7KC in Intact Cells-Recently, we have shown (8) that 11␤-HSD1 plays an essential role in the metabolism of 7KC in intact cells. Here we compared the ability of luminally oriented wild-type enzyme and cytoplasmic mutant K5S/K6S to catalyze the oxoreduction of 7KC in lysates and intact HEK-293 cells. Both enzymes readily converted 7KC to 7␤-hydroxycholesterol in lysates, but only wild-type 11␤-HSD1 catalyzed the oxoreduction of 7KC in intact cells (Fig. 5). Mutant K5S/ K6S showed background activity comparable with that of untransfected cells. These results demonstrate that the luminal orientation of 11␤-HSD1 is essential for the oxoreduction of 7KC. DISCUSSION To gain insight into the molecular mechanisms by which 11␤-HSD1 exerts its physiological functions, we investigated the impact of the N-terminal membrane anchoring region on 11␤-HSD1 function. The two type II anchored ER membrane proteins 11␤-HSD1 and E3 share a highly similar N-terminal region consisting of a conserved Lys in the short cytoplasmic sequence, a membrane spanning helix with a cluster of Tyr residues, and a segment of negatively charged amino acid residues C-terminal to the membrane span (Fig. 1A). Otherwise, the two proteins are unrelated. Previous analyses revealed that substitution of the conserved Lys 5 by Ser inverted the orientation of 11␤-HSD1, but the substitution of the analogous Lys 4 by Ile had no effect on the orientation of E3, indicating the existence of a second topological determinant (19,23). In the present study, we identified Glu 25 /Glu 26 in 11␤-HSD1 and Asp 25 in E3 as the second topological determinant.
According to the positive-inside rule, net charges on the polypeptide sequences flanking the transmembrane span determine the orientation, with the more positive of the two remaining on the cytoplasmic side of the membrane (31,32). The positive-inside rule correctly predicts both the topology of wild-type E3, with N-terminal charge ϩ1 and C-terminal charge Ϫ3 (ϩ1/Ϫ3), when considering six residues downstream of the membrane span, and 11␤-HSD1 (ϩ2/Ϫ2). However, the present analysis of several mutant proteins revealed that specific amino acid residues rather than net charge distribution determine the topology of these two enzymes.
We have shown previously that E3 mutant K4I (0/Ϫ3) still adopts wild-type topology (N cyt /C lum ) (23). Therefore, we now analyzed the role of luminal, negatively charged amino acid residues. Substitution by Lys of the three negatively charged residues that are closest to the membrane span (mutant D25K/ E27K/E28K (ϩ1/ϩ3)) did not alter its orientation in the ER membrane despite the excess of positive charges introduced at the luminal side (Fig. 2). Thus, the topology of this mutant clearly does not follow the positive-inside rule. However, substitution of Asp 25 by Lys (mutant K4I/D25K (0/Ϫ1)) or of Asp 25 by Asn and Glu 25 /Glu 26 by Gln (mutant K4I/D25N/E27Q/E28Q (0/0)) led to an inverted orientation in the ER membrane. This indicates that Lys 4 is the major determinant of E3 topology, whereby the introduction of several positively charged residues on the C-terminal end of the membrane span cannot overcome the dominant N-terminal signal. In mutants lacking the Nterminal signal, Asp 25 , the negatively charged residue closest to the membrane span, dictates the topology.
That the topology of 11␤-HSD1 is determined by the appropriate charge at a specific position rather than by the net charge distribution is demonstrated by the fact that mutant K5S (ϩ1/Ϫ2) adopts N lum /C cyt topology, whereas K6S shows wild-type N cyt /C lum orientation (19). The importance of Lys 5 is also reflected by its conservation in all known species, whereas Lys 6 is replaced by Thr in squirrel monkey and by Asn in mouse. Substitution of both Glu 25 and Glu 26 by Lys (Fig. 1C) led to an inverted N lum /C cyt orientation (Figs. 3 and 4), indicating that in 11␤-HSD1 the introduction of two positive charges at the C-terminal side of the membrane span is dominant over the N-terminal Lys 5 signal. This stands in contrast to E3, where the effect of Lys 4 is not affected by the substitution of three negative by positive charges at the C-terminal side of the membrane span. In mutant K6S, like in wild-type 11␤-HSD1, both Glu 25 and Glu 26 had to be replaced by Lys to invert its topology, further demonstrating that Lys 6 is not important for topology.
In addition to determining the luminal orientation in the ER membrane, the N-terminal regions of 11␤-HSD1 and E3 are sufficient to mediate retention in the ER membrane. We show that removal by mutagenesis of the cytoplasmic Lys signal, the luminal negatively charged residues immediately downstream of the membrane helix or of Lys 35 and Lys 36 in 11␤-HSD1, does not affect retention in the ER membrane. Similarly, all of the constructed mutants of the fusion between the N-terminal 34 residues of E3 and green fluorescent protein showed restricted localization to the ER membrane, suggesting that intrinsic properties of the transmembrane sequence may be responsible. Furthermore, the truncated splice variant 11␤-HSD1B, lacking the transmembrane span, was found to be attached to the ER membrane, suggesting that the hydrophobic domain located at amino acid residues 136 -158 also sufficiently mediates retention to the ER membrane (21).
Previous studies provided evidence that the N-terminal region of 11␤-HSD1 has an important stabilizing effect on enzymatic activity (19 -22); however, the residues involved were not identified. A fusion protein containing the transmembrane anchor of 11␤-HSD2 followed by residues 40 -292 of 11␤-HSD1 as well as the truncated 11␤-HSD1 protein alone (residues 40 - ND e ND e a Apparent K m (nM) and apparent V max (nmol ϫ h Ϫ1 ϫ mg Ϫ1 of total protein) values were calculated by nonlinear regression using Data Analysis Toolbox (MDL Information Systems Inc.) assuming first-order rate kinetics (Hill coefficients ranging between 0.94 and 1.17).
b For calculation of V max , the amount of 11␤-HSD1 protein per mg of total proteins was determined, and values were normalized to the expression of FLAG-tagged wild-type 11␤-HSD1 by semiquantitative densitometric analysis of Western blots c p Ͻ 0.05 compared with wild type. d p Ͻ 0.01 compared with wild type. e Activity was not detectable despite normal expression levels.  (21) reported that the splice variant 11␤-HSD1B, lacking the first 30 amino acids but containing Lys 35 and Lys 36 , could be partially reactivated upon solubilization from the microsomal fraction. Substitution of Val 149 by Arg in this truncated splice variant rendered the protein more soluble; however, this protein was like 11␤-HSD1B and not stable, indicating that the more N-terminal residues are required for the stability, e.g. for full activity of the enzyme. In line with this assumption, Walker et al. (20) purified a truncated enzyme lacking the first 23 amino acids, which retained its activity. The present finding that mutagenesis of residues Glu 25 and Glu 26 led to a significantly reduced V max , independent of the orientation of 11␤-HSD1 in the ER membrane and not affecting expression level, suggests that these residues stabilize a conformationally active enzyme.
In a previous study, we demonstrated that the N-terminal region determines the topology of 11␤-HSD1 to the ER membrane (19), but no difference in oxoreduction of 11-dehydrocorticosterone was found between luminally oriented wild-type 11␤-HSD1 and a cytoplasmic mutant enzyme; thus the physiological consequences of the luminal orientation remained unclear. Here we addressed the question of the impact of luminal orientation on enzymatic function in more detail and compared the function of the luminally oriented wild-type 11␤-HSD1 with the cytoplasmic mutant K5S/K6S in intact HEK-293 cells not expressing endogenous 11␤-HSD activity. In line with the previous study, no difference was found for the oxoreduction of cortisone between wild-type and mutant enzyme. Similar results were also obtained when Chinese hamster ovary cells were used, a cell line not expressing endogenous 11␤-HSD activity (not shown). This is rather surprising because the cytoplasm is considered to have a more reducing environment than the ER lumen, and cofactor concentrations may be different between the two compartments. This finding also implies that the substrate has similar access to both compartments. However, regeneration systems exist for cofactors in the cytoplasm and in the ER lumen, and 11␤-HSD1 accepts both cofactors with a slight preference for NADP(H), and neither glycosylation nor disulfide bridges are required for the catalytic activity of the enzyme (19,33,34).
It was shown previously (28,29) that P-glycoprotein acts as an energy-dependent efflux pump that exports a wide range of substrates including corticosteroids. P-glycoprotein is endogenously expressed in HEK-293 cells, although at a relatively low level, and can be efficiently inhibited by cyclosporin A (30). 2 The more pronounced increase in catalytic efficiency of the oxoreduction of cortisone observed for the cytoplasmic mutant K5S/K6S compared with wild-type 11␤-HSD1 indicates that the inhibition of P-glycoprotein by cyclosporin A predominantly enhanced cortisone concentrations in the cytoplasm. The tendency of an increase of catalytic efficiency of wild-type 11␤-HSD1 upon cyclosporin treatment indicates that the substrate is exchanged between the cytoplasm and the ER lumen.
In contrast to the oxoreduction of cortisone, a 50% decrease in the oxidation of cortisol was observed for mutant K5S/K6S in the cytoplasm compared with wild-type 11␤-HSD1, thus demonstrating that the luminal orientation is essential for appro-2 David Clarke, University of Toronto, personal communication.   priate function of 11␤-HSD1. The oxidative activity of 11␤-HSD1 has been suggested to protect mature Leydig cells from the inhibitory effect of glucocorticoids on testosterone production (35). In addition, dehydrogenase activity of 11␤-HSD1 is predominant in adipose stromal cells, where it may represent an autocrine protective mechanism that facilitates preadipocyte proliferation, but upon differentiation to adipocytes 11␤-HSD1 switches to oxoreductase activity (36). Recent studies (37) provided evidence that this switch from oxidation to oxoreduction is dependent on the activity of hexose-6-phosphate dehydrogenase, which may act as a regulator of the redox potential in the ER. Thus, the luminal orientation may not only be essential for efficient dehydrogenase activity as demonstrated in the present work but also for adequate functional interaction with hexose-6-phosphate dehydrogenase. Moreover, we demonstrate that the luminal orientation of 11␤-HSD1 is required for oxoreduction of 7KC, because no conversion of 7KC to 7␤-hydroxycholesterol was detected for mutant K5S/K6S with cytoplasmic orientation. The molecular mechanism underlying the required luminal orientation of 11␤-HSD1 for oxoreduction of 7KC is unclear but may be dependent on other enzymes localized in the ER lumen. 7-KC has been suggested to play a role in atherosclerosis, for example, by inhibiting free cholesterol efflux from macrophages (38) and disturbing cholesterol trafficking between the ER membrane and the plasma membrane (39). In macrophages, the subcellular distribution of 7KC mirrors that of cholesterol, which is localized in the plasma membrane and intracellular membranes but also cycles constitutively between the plasma membrane and subcellular organelles (reviewed in Refs. 10 and 40). The cholesterol content of mammalian cells is controlled by the sterol regulatory element-binding protein pathway. This pathway is regulated by sterol regulatory element-binding protein cleavage-activating protein (SCAP) (41). Upon sterol accumulation, SCAP no longer moves from the ER to the Golgi, thus restricting the cleavage of sterol regulatory element-binding proteins and preventing the transcription of the cholesterolsynthesizing enzymes (42,43). Recently, Brown et al. (44) showed that among other sterols, 7␤-hydroxycholesterol but not 7KC reduces the activity of SCAP in the ER membrane presumably by causing cholesterol to translocate from the plasma membrane to the ER, where it interacts with the sterol sensing domain of SCAP and produces conformational change of the protein. In line with this hypothesis, conversion of 7KC to 7␤-hydroxycholesterol in the ER lumen by 11␤-HSD1 may have a beneficial effect by protecting cells from lipid accumulation and may play a role in macrophage pathophysiology in atherosclerosis (45). The molecular mechanism of the oxoreduction of 7KC by 11␤-HSD1 require further studies; nevertheless, this novel role of 11␤-HSD1 should be kept in mind for the design of specific inhibitor molecules for therapeutic purposes.
In conclusion, we identified the residues determining the orientation of 11␤-HSD1 and E3 toward the ER lumen. Analysis of the charged residues in the N-terminal region of 11␤-HSD1 revealed that the di-lysine motif at position 35/36 and the di-glutamate motif at position 25/26 are essential for enzymatic activity. Most important, we demonstrate that the luminal orientation of 11␤-HSD1 is essential for efficient oxidation of cortisol and for the oxoreduction of 7KC. In future studies, the characterization of the expression and activities of E3 and 11␤-HSD1 constructs with cytosolic orientations of their catalytic moiety should contribute further to the understanding of the physiological function of these proteins.