Hexose-6-phosphate Dehydrogenase Modulates 11β-Hydroxysteroid Dehydrogenase Type 1-Dependent Metabolism of 7-keto- and 7β-hydroxy-neurosteroids

Background The role of 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) in the regulation of energy metabolism and immune system by locally reactivating glucocorticoids has been extensively studied. Experiments determining initial rates of enzyme activity revealed that 11β-HSD1 can catalyze both the reductase and the dehydrogenase reaction in cell lysates, whereas it predominantly catalyzes the reduction of cortisone to cortisol in intact cells that also express hexose-6-phosphate dehydrogenase (H6PDH), which provides cofactor NADPH. Besides its role in glucocorticoid metabolism, there is evidence that 11β-HSD1 is involved in the metabolism of 7-keto- and 7-hydroxy-steroids; however the impact of H6PDH on this alternative function of 11β-HSD1 has not been assessed. Methodology We investigated the 11β-HSD1-dependent metabolism of the neurosteroids 7-keto-, 7α-hydroxy- and 7β-hydroxy-dehydroepiandrosterone (DHEA) and 7-keto- and 7β-hydroxy-pregnenolone, respectively, in the absence or presence of H6PDH in intact cells. 3D-structural modeling was applied to study the binding of ligands in 11β-HSD1. Principal Findings We demonstrated that 11β-HSD1 functions in a reversible way and efficiently catalyzed the interconversion of these 7-keto- and 7-hydroxy-neurosteroids in intact cells. In the presence of H6PDH, 11β-HSD1 predominantly converted 7-keto-DHEA and 7-ketopregnenolone into their corresponding 7β-hydroxy metabolites, indicating a role for H6PDH and 11β-HSD1 in the local generation of 7β-hydroxy-neurosteroids. 3D-structural modeling offered an explanation for the preferred formation of 7β-hydroxy-neurosteroids. Conclusions Our results from experiments determining the steady state concentrations of glucocorticoids or 7-oxygenated neurosteroids suggested that the equilibrium between cortisone and cortisol and between 7-keto- and 7-hydroxy-neurosteroids is regulated by 11β-HSD1 and greatly depends on the coexpression with H6PDH. Thus, the impact of H6PDH on 11β-HSD1 activity has to be considered for understanding both glucocorticoid and neurosteroid action in different tissues.

H6PDH colocalizes with 11b-HSD1 at the lumenal surface of the ER-membrane [10], where H6PDH provides cosubstrate NADPH and strongly stimulates 11b-HSD1 reductase activity [10][11][12][13][14]. Thus, a tightly controlled ratio of H6PDH to 11b-HSD1 is crucial since the local regeneration of cortisol and corticosterone in tissues such as liver, skeletal muscle and adipose tissue determines the magnitude of glucocorticoid receptor (GR) activation and subsequent regulation of genes involved in carbohydrate and lipid metabolism. Indeed, chronically elevated local reactivation of glucocorticoids by enhanced 11b-HSD1 activity has been associated with the development of the metabolic syndrome [15]. Although it has not been analyzed so far, modulation by H6PDH might also be relevant for alternative functions of 11b-HSD1, including the detoxification of xenobiotics [16] and the metabolism of 7-keto-and 7-hydroxy-sterols and -steroids [6,[17][18][19].
Recently, it has become clear that 11b-HSD1 also can catalyze the interconversion of the C7-ketone and the C7-hydroxyl on DHEA [19][20][21][22]. DHEA is synthesized in the adrenal glands and serves as a major sex steroid hormone precursor [23]. Besides, another important site for synthesis of DHEA is the brain, where it is further metabolized to its 7a-and 7b-hydroxy-forms. DHEA and its C7-oxygenated metabolites are considered neurosteroids and they play a role in the regulation of emotional responses, memory functions and neuronal excitability [24][25][26]. Pregnenolone and its metabolites also are neurosteroids. 7a-hydroxypregnenolone is the major neurosteroid in amphibians, increasing the neuron locomotor activity in the breeding newts [27].
The major metabolic fate of DHEA and pregnenolone in the brain and other extrahepatic tissues is the CYP7B1-dependent 7ahydroxylation. It was suggested that 7a-hydroxy-DHEA serves as a precursor for other active derivatives [28]. A causal link between declining DHEA levels and age-related loss of cognitive function, including Alzheimer's disease, has been proposed, with evidence for reduced hippocampal CYP7B1 expression [29,30]. Moreover, several studies suggest that 7-keto-and 7-hydroxy-steroids exert neuroprotective, anti-glucocorticoid and immune-modulatory effects [19,29,31,32]. The presence of 11b-HSD1 activity in the brain suggests that 11b-HSD1 may have a role in regulating the neurosteroid actions of the 7-keto-and 7-hydroxy-metabolites of DHEA and pregnenolone. However, details of the role of H6PDH in their metabolism and their in vivo regulation remained unclear, and the reaction direction of the metabolism of these compounds has not been assessed.
Here, we employed enzyme activity measurements and 3Dstructural modelling to investigate how 7-keto-, 7a-hydroxy-and 7b-hydroxy-DHEA interact with 11b-HSD1. Furthermore, we determined the impact of H6PDH on the 11b-HSD1-dependent metabolism of various 7-keto-and 7-hydroxy-steroids in intact cells by measuring both initial rates of activity and steady state concentrations. We find that when H6PDH supplies sufficient NADPH, this results in 11b-HSD1 having a preference for the reduction of 7-keto-DHEA to 7b-hydroxy-DHEA. Thus, in vivo, depending on the level of H6PDH activity, 11b-HSD1 can regulate the equilibrium between 7-keto-, 7a-hydroxy-and 7bhydroxy-DHEA in brain and other organs.

RESULTS
11b-HSD1 catalyzes the interconversion of 7-keto-and 7-hydroxy-DHEA To determine the ability of 11b-HSD1 to catalyze the interconversion of 7-keto-and 7-hydroxy-DHEA in intact cells, we employed HEK-293 cells that were transfected with either 11b-HSD1 alone or together with H6PDH. This cell line is suitable to assess the effect of H6PDH on 11b-HSD1 function, because of low endogenous H6PDH and lacking 11b-HSD1 and 11b-HSD2 expression, which is indicated by Ct values that were higher than 32 [10].
As shown in Figure 1A, incubation of cells expressing 11b-HSD1 with 1 mM 7a-hydroxy-DHEA led to the formation of both 7b-hydroxy-DHEA and 7-keto-DHEA. Cells co-expressing 11b-HSD1 and H6PDH metabolized less 7a-hydroxy-DHEA, and the accumulation of 7-keto-DHEA was abolished. The fact that 7bhydroxy-DHEA formation could still be observed suggests that the 7-keto-DHEA formed from 7a-hydroxy-DHEA was rapidly further converted to 7b-hydroxy-DHEA. The involvement of other enzymes in the interconversion of 7-keto-and 7-hydroxy-DHEA in HEK cells is unlikely, since the selective 11b-HSD1 inhibitor T0504 [33,34] prevented the formation of 7b-hydroxy-DHEA and 7-keto-DHEA, and no conversion was detected in untransfected cells (data not shown). The recovery of the initially added counts was approximately 75%, which was due to a loss of steroids during sampling and during extraction steps. Similarly, incubation of cells expressing 11b-HSD1 with 1 mM 7b-hydroxy-DHEA led to the formation of almost equal amounts of 7-keto-DHEA and 7a-hydroxy-DHEA, while the formation of 7ahydroxy-DHEA was significantly reduced and that of 7-keto-DHEA almost completely abolished in the presence of H6PDH ( Figure 1B). Inhibition of 11b-HSD1 prevented the formation of 7a-hydroxy-DHEA and 7-keto-DHEA.
We then incubated cells expressing 11b-HSD1 with 1 mM 7-keto-DHEA and observed the formation of almost equal amounts of 7ahydroxy-DHEA and 7b-hydroxy-DHEA ( Figure 1C). In the presence of H6PDH, 11b-HSD1 nearly fully metabolized 7-keto-DHEA and preferentially formed 7b-hydroxy-DHEA. A comparison of the products formed after different time intervals indicated that 11b-HSD1 preferentially catalyzes the conversion of 7-keto-DHEA to 7b-hydroxy-DHEA. The formation of 7a-hydroxy-DHEA is less efficient and its accumulation was only observed after prolonged time of incubation. After 24 h of incubation, a steady state was reached and the ratio of 7-keto-DHEA, 7a-hydroxy-DHEA and 7b-hydroxy-DHEA was independent of the substrate chosen initially ( Figure 1D). Importantly, the presence of H6PDH strongly shifted the equilibrium from the 7-keto-to the 7b-hydroxy-steroid.

Analysis of substrate binding by 3D-modelling
To better understand how 7-keto-, 7a-hydroxy-and 7b-hydroxy-DHEA interact with 11b-HSD1, we inserted these steroids into the active site of the recently solved 3D-structure of 11b-HSD1 [35]. In constructing models of DHEA analogs with 11b-HSD1, there are two different orientations in which DHEA analogs can fit in the substrate binding site of 11b-HSD1. In one configuration, the D ring of DHEA is oriented towards the interior of 11b-HSD1. In the other, the A ring is oriented towards the interior.
To obtain the first orientation for analysis, we first extracted E. coli 7a-HSD from the PDB (file: 1FMC). 7a-HSD contains chenodeoxycholic acid, which has a 7a-hydroxyl group. The D ring of chenodeoxycholic acid is oriented towards the interior of 7a-HSD [36]. We superimposed 1FMC with 1Y5R. As shown in Figure 2, the 7a-hydroxyl in chenodeoxycholic acid and the catalytic tyrosine in 7a-HSD superimpose nicely on the 11bhydroxyl in corticosterone and the catalytic tyrosine in 11b-HSD1. Initial models of 7-keto-DHEA (with NADPH) and 7b-hydroxy-DHEA (with NADP + ) with the D ring oriented towards the interior of 11b-HSD1 were constructed by using the Biopolymer option in Insight II for conversion of the 7a-hydroxyl to a ketone and to a 7b-hydroxyl, respectively.
The second orientation was obtained by overlaying 7b-hydroxy-DHEA with the 11b-hydroxyl on corticosterone in 11b-HSD1 with NADP + (Figure 3). To obtain initial models of 7-keto-DHEA (with NADPH) and 7a-hydroxy-DHEA (with NADP + ) with the A ring oriented towards the interior of 11b-HSD1, we used the Biopolymer option in Insight II for conversion of the 7b-hydroxyl to a ketone and to a 7a-hydroxyl, respectively. Thus, we had six tertiary complexes of mouse 11b-HSD1 with 7-keto-, 7a-hydroxy-or 7b-hydroxy-DHEA ( Figure 4) for refinement by energy minimization using Discover 3. Figure 4A shows the minimized 3D models of 7-keto-, 7ahydroxy-or 7b-hydroxy-DHEA with the D ring oriented towards the interior of 11b-HSD1. All of these C7-DHEA analogs have either the hydroxyl or ketone in a favorable position for catalysis by 11b-HSD1. For example, the distance between Tyr-183 and 7a-hydroxyl, 7-keto and 7b-hydroxyl is 2.9 Å , 2.8 Å and 3.1 Å , respectively. The distance between the nicotinamide C4 and 7ahydroxyl, 7-keto and 7b-hydroxyl is 3.9 Å , 3.5 Å and 3.0 Å , respectively. Figure 4B shows the minimized structures of 7-keto-, 7a-hydroxy-or 7b-hydroxy-DHEA with the A ring oriented towards the interior of 11b-HSD1. 7-keto-DHEA and 7bhydroxy-DHEA have the ketone and hydroxyl, respectively, in a favorable position for catalysis by 11b-HSD1. In contrast, the 7a-hydroxyl on DHEA is 4.8 Å from the nicotinamide C4.
The models show that 7b-hydroxy-DHEA is favored for reduction when 7-keto-DHEA is in two orientations in 11b-HSD1, while 7a-hydroxy-DHEA is favored in one orientation. Thus, our models can explain the preference for formation of 7bhydroxy-DHEA by 11b-HSD1.

Metabolism of other 7-keto steroids by 11b-HSD1
We next investigated whether 11b-HSD1 acts on other 7-ketosteroids and demonstrate for the first time that 11b-HSD1 metabolizes 7-ketopregnenolone and 5a-androstane-3b-ol-7,17dione (5-ADION)( Figure 5) and that metabolism of both 7-ketosubstrates to 7b-hydroxy-steroids was strongly stimulated by H6PDH. For 7-keto-pregnenolone, the equilibrium was shifted from the 7-keto to the 7b-hydroxy form, comparable with the metabolism of 7-keto-DHEA. Since 7a-hydroxypregnenolone and 5a-androstane-3b-ol-7-hydroxy,17-one were not available, the formation of these products could not be analyzed. Peaks in the GC-MS analyses that may correspond to these products were observed and suggest preferential formation of the 7b-hydroxy metabolites (data not shown). The 11b-HSD1-dependent metabolism of 7-ketopregnenolone and 5a-androstane-3b-ol-7,17-dione was almost completely abolished in the presence of inhibitor T0504 ( Figure 5), and cells transfected with pcDNA3.1 control plasmid instead of the plasmid for 11b-HSD1 did not show any 7keto-reductase activity (data not shown).

Assessment of 7-oxygenated neurosteroids as substrates of 11b-HSD1 by determining competition with glucocorticoids
The modeling analyses indicated that the 7-keto and 7-hydroxy metabolites of DHEA occupy the same binding site in 11b-HSD1 as cortisone and cortisol, respectively. If the 7-keto and 7-hydroxy metabolites of DHEA and pregnenolone are substrates for either the reductase or the dehydrogenase reaction, then they should compete with the reduction of cortisone and the dehydrogenation of cortisol, respectively. Indeed, as shown in Table 1, all three 7keto-steroids tested preferentially inhibited 11b-HSD1-dependent reduction of cortisone to cortisol with much weaker effects on the reverse reaction. In contrast, 7-hydroxy-steroids preferentially inhibited the dehydrogenase reaction, whereby 7b-hydroxy-DHEA was more potent than 7a-hydroxy-DHEA, in line with the observation that 7b-hydroxy-DHEA is a better substrate for 11b-HSD1 than 7a-hydroxy-DHEA.
H6PDH stimulates 11b-HSD1 reductase activity and alters the steady state ratio of cortisone to cortisol Recent studies showed that H6PDH affects 11b-HSD1 function by strongly stimulating the reduction of cortisone and abolishing the oxidation of cortisol [10,13]. These studies determined only initial rates of conversion. Here, we determined the steady state ratio of cortisone to cortisol in HEK-293 cells transfected with 11b-HSD1 alone or cotransfected with 11b-HSD1 and H6PDH. As shown in Table 2, a steady state ratio of 70-75% cortisol to 25-30% cortisone was reached in cells expressing 11b-HSD1, independent of whether cortisone or cortisol were supplied initially. Co-expression with H6PDH shifted the steady state ratio to 90-95% cortisol and 5-10% cortisone, respectively. Interestingly, the two structurally distinct 11b-HSD1 inhibitors flavanone [34] and BNW7 [37] both had little or no effect on the steady state ratio of glucocorticoids, suggesting that the presence of inhibitors only lowers the time until the equilibrium is reached and that the NADPH availability is the key determinant for the control of the intracellular ratio of cortisone to cortisol. A similar shift from ketoto hydroxy-steroids was observed with 7-substituted steroids (Figure 1 and 5).

DISCUSSION
The cloning of 11b-HSD1 has stimulated important advances in understanding its molecular properties and physiological actions. As a result, recombinant 11b-HSD1 in lysates of cells or upon purification became available for determining the kinetic constants for reduction of cortisone and oxidation of cortisol, which revealed that 11b-HSD1 only had a slight preference for reduction of the keto group at C11 on glucocorticoids [2,3,38]. In contrast, studies of 11b-HSD1 in whole cells, showed that 11b-HSD1 preferentially was a reductase [4][5][6][7]. The reaction direction of 11b-HSD1 is crucial with respect to its potential as a target for treatment of metabolic diseases, where a decrease of glucocorticoid reactivation, especially in adipose tissue, is thought to be beneficial [15,39]. Whether 11b-HSD1 acts as a reductase or dehydrogenase in a given tissue also is important regarding its reported alternative  functions. Several studies showed 11b-HSD1-dependent carbonyl reduction of a variety of chemicals, indicating that 11b-HSD1 has a role in detoxification of xenobiotics [16] and oxidized cholesterol [6,17,18]. Importantly, it was recently shown that 11b-HSD1 is closely localized to H6PDH in the endoplasmic reticulum [10]. H6PDH provides NADPH for reductases, indicating that the mechanism by which 11b-HSD1 regulates cortisol levels was more complex than previously thought [10,12,13].
Here, we provide evidence that in concert with H6PDH, 11b-HSD1 also regulates the levels of 7a-hydroxy-, 7b-hydroxy-and 7keto-neurosteroids derived from DHEA and pregnenolone (Figure 1 and 5). We show that regulation of cosubstrate levels by H6PDH is crucial for determination of the directionality of the synthesis of neurosteroids and glucocorticoids by 11b-HSD1, expanding the physiological functions of 11b-HSD1. We tested various 7-keto-and 7-hydroxy-steroids as substrates of 11b-HSD1 and investigated the influence of H6PDH on their metabolism in intact cells. In line with the results from Muller et al. using microsomal fractions from yeast expressing human 11b-HSD1 [22], we detected efficient interconversion of 7-keto-and 7hydroxy-steroids in an intact human cell system. Importantly, we demonstrate that cells co-expressing 11b-HSD1 and H6PDH predominantly metabolize 7-keto-DHEA, 7-ketopregnenolone and 5a-androstane-3b-ol-7,17-dione into their corresponding 7bhydroxy metabolites.
The present study provides evidence that 11b-HSD1 plays an important role in regulating the intracellular equilibrium between C7-and C11-keto and their -hydroxy metabolites, respectively. Depending on the level of H6PDH activity, the steady state ratio between keto-and hydroxy-steroids can be controlled. That cosubstrate levels influence how 11b-HSD1 regulates an equilibrium between active versus inactive or less potent steroid hormone metabolites resembles 17b-HSD enzymes. Auchus and coworkers recently showed that at equilibrium 17b-HSD1, 2 and 3 catalyze both reductase and dehydrogenase reactions; thus, these enzymes do not just drive steroid flux in one direction [40,41].
Our structural modeling provides an explanation for the preferential formation of 7b-hydroxy-DHEA from 7-keto-DHEA. First, the models show that when 7-keto-DHEA, 7a-hydroxy-DHEA and 7b-hydroxy-DHEA have their D ring pointing into the interior of the enzyme, they are in a favorable position to interact with the catalytic tyrosine and the nicotinamide ring ( Figure 4A). A similar orientation of the D ring is found for the bile acid chenodeoxycholic acid crystallized in 7a-HSD and corticosterone in 11b-HSD1 (Figure 2). The close superposition of the 7ahydroxyl on chenodeoxycholic acid and 11b-hydroxyl on corticosterone supports an earlier hypothesis by Lathe, who recognized the rotational symmetry between the 11b-and the 7aposition of the steroid backbone and suggested that some binding sites may recognize both 7a-and 11b-modified steroids [42]. The important discovery from the 3D models is that there are favorable interactions between the C7-hydroxyl and the catalytic site of 11b-HSD1 when the A ring on 7b-hydroxy-DHEA is oriented towards the interior of 11b-HSD1 ( Figure 4B). Thus, there are two orientations of 7-keto-DHEA in 11b-HSD1 that can lead to 7b-hydroxy-DHEA in the presence of NADPH, but only one configuration of 7-keto-DHEA is favorable for reduction to 7a-hydroxy-DHEA. This analysis reveals an important role for H6PDH in regulating the specificity for formation of 7b-hydroxy-DHEA. By providing sufficient NADPH, H6PDH can increase cellular levels of 7b-hydroxy-DHEA.
The metabolism of C7-keto and -hydroxy-DHEA and C7-keto and -hydroxy-pregnenolone by 11b-HSD1 and H6PDH can regulate the activities of these neurosteroids in the brain. CYP7B1 is responsible for the production of 7a-hydroxy-DHEA, mainly in the hippocampus [43]. CYP7B1 knock-out mice are unable to produce 7a-hydroxy-DHEA and 7b-hydroxy-DHEA [44], suggesting that 7a-hydroxy-DHEA serves as a precursor for other derivatives including 7b-hydroxy-DHEA and 7-keto-DHEA [28].  The ratio of cortisone to cortisol was measured after incubating HEK-293 cells expressing 11b-HSD1 or coexpressing 11b-HSD1 and H6PDH for 16   The mechanism, however, for the formation of 7b-hydroxy-DHEA and 7-keto-DHEA remained unclear. The present findings indicate that, depending on the activity of H6PDH, 11b-HSD1 can regulate the availability of 7a-hydroxy-7b-hydroxy-and 7keto-metabolites of DHEA and pregnenolone in tissues expressing CYP7B1, which can control local concentrations of neurosteroids. Since cerebrospinal fluid from patients with vascular dementia contained higher levels of 7b-hydroxy-DHEA than that from patients with Alzheimer's disease, Kim et al. suggested that 7bhydroxy-DHEA may be more neuroprotective than 7a-hydroxy-DHEA and that the ratio of 7b-hydroxy-DHEA to 7a-hydroxy-DHEA could be a marker to distinguish between vascular dementia and Alzheimer's disease [45]. Moreover, other 7b-hydroxy-steroids appear to have important protective actions in the brain. For example, Pringle et al. reported more efficient reduction of ischemiainduced neuronal damage by 7b-hydroxy-epiandrosterone compared with its 7a-epimer, both in vivo and in vitro [46].
Interestingly, a single-nucleotide polymorphism associated with a six-fold increased risk for sporadic Alzheimer's disease was recently identified in the promoter region of the HSD11B1 gene. The polymorphism reduced promoter activity, and the authors suggested that 11b-HSD1 might act as a dehydrogenase in hippocampus and that the polymorphism may result in insufficient inactivation of glucocorticoids with increased neuronal damage [47]. Our data indicate that this polymorphism also will alter levels of 7a-hydroxy-and 7b-hydroxy-DHEA and -pregnenolone, respectively, in the brain, which could contribute to the increased incidence of Alzheimer's disease. A previous study on the distribution of H6PDH activity indicated a ten-fold lower expression in the brain compared with the liver, but the exact localization has not been determined [48]. It will be important to investigate the distribution and activity of H6PDH in specific regions of the brain, with respect to the metabolism of both glucocorticoids and C7-oxygenated steroids by 11b-HSD1.
In conclusion, we provide evidence for a role of 11b-HSD1 in the regulation of the relative ratios of 7a-hydroxy-, 7b-hydroxyand 7-keto-metabolites of DHEA and pregnenolone, respectively. 11b-HSD1 acts as a reversible enzyme, whereby H6PDH by providing cosubstrate NADPH mediates a shift in the steady state ratio from inactive to active glucocorticoids and from 7a-hydroxyand 7-keto-steroids to their 7b-hydroxy-forms, respectively. The role of 11b-HSD1 and H6PDH in the metabolism of neurosteroids should be kept in mind when using 11b-HSD1 inhibitors for treatment of patients with the metabolic syndrome.

Cell culture and transient transfection
HEK-293 cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal calf serum (FCS), 50 units/ml penicillin, 50 mg/ml streptomycin and 2 mM glutamine. Cells were transfected with expression plasmids for FLAG-tagged 11b-HSD1 and myc-tagged H6PDH or pcDNA3.1 control using the Ca 2+ -phosphate precipitation method. The epitope tags had no effect on enzymatic activities or expression levels of 11b-HSD1 and H6PDH [10,49]. A total amount of 8 mg DNA was used per 10 cm culture dish or 2 mg DNA per well of a six-well culture plate.

11b-HSD activity assays
The enzyme activity in intact cells was measured as described [10]. Briefly, HEK-293 cells, grown in 10 cm culture dishes and transfected with FLAG-tagged 11b-HSD1 alone or co-transfected with myc-tagged H6PDH, were detached 24 h post-transfection and distributed in 96-well plates at a density of 309000-409000 cells per well. After 16 h, cells were incubated in serum-and steroid-free medium and the conversion of radiolabeled cortisone to cortisol (or the reverse activity) was determined upon incubation for 1 h at 37uC in a total volume of 50 ml containing 200 nM cortisone (or cortisol). The reaction was stopped by adding methanol containing 2 mM unlabeled cortisone and cortisol, followed by separation of steroids by TLC and scintillation counting. Measurements with freshly prepared cell lysates were carried out in the presence of [1,2,6,7-3 H]-cortisone and cofactor NADPH at final concentrations of 200 nM and 500 mM for the reductase reaction and with 50 nM [1,2,6,7-3 H]-cortisol and 500 mM cofactor NAD + for the dehydrogenase reaction, respectively, in the absence or presence of various concentrations of competing 7-oxygenated neurosteroid ligands. The concentration of dimethylsulfoxide or methanol from the solvent was 0.1% in all reactions and had no effect on enzyme activity. Data (mean6SD) were obtained from four independent experiments and were calculated using the Data Analysis Toolbox (Elsevier MDL, Allschwil, Switzerland). Measurement of 7-keto-steroid reductase and 7-hydroxy-steroid dehydrogenase activity in intact cells HEK-293 cells were seeded in six-well plates at a density of 9009000 cells per well. After 24 h, cells were transfected with expression vector for 11b-HSD1 or co-transfected with vectors for 11b-HSD1 and H6PDH. An empty pcDNA3.1 plasmid was used in transfections lacking H6PDH to adjust total DNA amounts in all samples. The medium was replaced 18 h post-transfection by steroid-free medium followed by addition of 1 mM of the steroid substrate to be tested. Compound T0504 at 1 mM was used in control reactions to achieve almost complete inhibition of 11b-HSD1 [33,34]. After incubation for various time intervals, reactions were terminated by addition of an equal volume of dichloromethane, containing 500 ng medroxyprogesterone as an internal standard. Since radiolabeled C7-oxygenated metabolites were not commercially available, we applied GC-MS for detection. Steroids were extracted and the organic phase separated after centrifugation for 5 min at 20006g. The organic solvent was evaporated under nitrogen flow, 500 ng of stigmasterol added as an external standard, and the solvent evaporated again. Derivatization was carried out by adding 100 ml of 2% methoxyamine-HCl in pyridine, followed by incubation at 60uC for 1 h. After evaporation of the solvent, 100 ml of trimethylsilylimidazole was added and the samples were incubated for 16 h at 100uC. After cooling to 25uC, the samples were dissolved in 500 ml of cyclohexane:pyridine:hexamethyldisilazane (98:1:1), and the steroids were purified using Lipidex-500 columns. Finally, steroids were dissolved in 200 ml of cyclohexane, sonicated for 2 min and subjected to GC-MS analysis on a Hewlett Packard gas chromatograph 6890 equipped with a mass-selective detector 5973 by selected ion monitoring.

3D-Models of 11b-HSD1
Mouse 11b-HSD1 (PDB ID:1Y5R) was extracted from the Protein Data Bank (PDB) for use as a template to investigate the interactions of 7-keto-, 7a-hydroxy-and 7b-hydroxy-DHEA with 11b-HSD1. We used 1Y5R because it contains both corticosterone and NADP + , which allows us to superimpose C7-DHEA analogs on corticosterone. Human 11b-HSD1 (PDB ID: 1XU7) and guinea pig 11b-HSD1 (PDB ID: 1XSE) were co-crystallized with NADP + , but not with a glucocorticoid. The close similarity of human and mouse 11b-HSD1, which have 81% sequence identity, allows mouse 11b-HSD1 to be a good model for the interaction of compounds with human 11b-HSD1 [35].
To investigate an alternative, inverted binding mode of 7-ketoand 7-hydroxy-DHEA in 11b-HSD1, we used chenodeoxycholic acid from E. coli 7a-hydroxysteroid dehydrogenase (7a-HSD). Chenodeoxycholic acid is an excellent template because it has a 7a-hydroxyl. We superimposed the PDB file 1FMC (E. coli 7a-HSD with chenodeoxycholic acid) with 1Y5R and extracted chenodeoxycholic acid, which was merged into 11b-HSD1. For conversion of the 7a-hydroxyl to a ketone or a 7b-hydroxyl we used the Biopolymer option in Insight II. The energy of each model of 11b-HSD1 with C7-DHEA analogs was minimized using Discover 3 (Accelrys Inc., San Diego, CA, USA), which was run for 10,000 iterations, using a distant dependent dielectric constant of 2.