5 (cid:1) -Reduced Glucocorticoids, Novel Endogenous Activators of the Glucocorticoid Receptor*

Metabolism of glucocorticoids to A-ring-reduced dihy-dro- and tetrahydro-derivatives by means of hepatic 5 (cid:1) and 5 (cid:2) -reductases has long been regarded as a pathway of irreversible inactivation. However, 5 (cid:1) -reduced metabolites of other steroids, e.g. testosterone and aldos-terone, have significant biological activity. We investigated whether 5 (cid:1) -reduced metabolites of corticosterone are glucocorticoid receptor (GR) agonists. Corticosterone, 5 (cid:1) -tetrahydrocorticosterone (5 (cid:1) THB), and 5 (cid:1) -dihy-drocorticosterone (5 (cid:1) DHB) were similarly effective in displacing tritiated dexamethasone from binding sites in hepatocytes, whereas 5 (cid:2) -reduced metabolites were less effective in binding. 5 (cid:1) THB had glucocorticoid receptor agonist effects in vitro and in vivo . After transient co-transfection of hGR and a murine mammary tumor virus-luciferase reporter into HeLa cells, 5 (cid:1) THB was active to a comparable extent as corticosterone (28-fold versus 37-fold induction, respectively, at 1 (cid:3) M ) and additive For binding experiments, duplicate incubations of (2 (cid:3) were carried out at 4 °C overnight with 3 H] 4 dexamethasone (1.5 n M final in molybdate buffer (25 (cid:4) 0.1 M , 7.2) and non-radioactive competitor steroid (Steraloids, New-port, molybdate buffer (25 (cid:4) final from 0.63 n M to 200 (cid:4) M The competitors were dexamethasone, corticosterone, 5 (cid:1) -tetrahydrocorticosterone (5 (cid:1) THB), 5 (cid:1) -dihydrocorticosterone (5 (cid:1) DHB), 5 (cid:2) -dihydrocorticosterone (5 (cid:2) DHB), and 5 (cid:2) -tetrahydrocorti-costerone (5 (cid:2) THB). Reactions were terminated by rapid filtration on filtermats pre-soaked in polyethylenimine (0.3%) using a Combi Cell Harvester (Skatron with H 2 Filters were and then encapsulated with Meltilex (PerkinElmer Life and counted in a Microbeta Plus Liquid Scintillation Counter (Wallac). Results tritiated dexamethasone of Glucocorticoid in Lines— maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% (v/v) fetal-calf serum, 100 units/ml penicillin, 100 (cid:4) g/ml strepto-mycin, and 200 m M L -glutamine and at 37 °C and 5% CO Transient transfections were out using the calcium phosphate co-precipitation method. cells at

incubated at 37°C for 30 min. Cell viability was determined by trypan blue exclusion under a light microscope.
Glucocorticoid Receptor Activation in Cell Lines-Cell lines were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal-calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 200 mM L-glutamine and grown at 37°C and 5% CO 2 . Transient transfections were carried out using the calcium phosphate co-precipitation method. HeLa cells were seeded at 3 ϫ 10 5 /60-mm plate in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. After overnight incubation, medium was replaced and cells were transfected with a total of 10 g of DNA comprising 1 g pCH110 (Pharmacia, St. Albans, UK; internal control), 5 g pLTR-Luc (11), 1 g pRShGR␣ (11), and pGEM-3 (Promega, Southampton, UK). After overnight incubation, medium was replaced and steroid was added for 16 h before harvesting and lysis of cells. Luciferase and ␤-galactosidase activities were measured in cell lysates as described previously (12). ␤-Galactosidase activity was assayed using a Tropix Galacto Light Plus kit. All transfections were carried out in triplicate and the mean ratio of luciferase/␤-galactosidase activities was calculated.
Induction of tyrosine aminotransferase (TAT) and phosphoenolpyruvate carboxykinase (PEPCK) mRNA were assessed in H4IIE cells treated overnight with corticosterone, 5␣THB (1 M in ethanol at Ͻ0.1% final volume) or vehicle and/or RU486 (1 M; a GR antagonist). Total RNA was extracted by using TRIzol reagent (Invitrogen), and Northern analysis was carried out by using 32 P-labeled TAT (13), PEPCK (14), and U1 cDNA probes (15). Hybridized probe was quantified by using a Fuji FLA2000 PhosphorImager analyzer.
Metabolic Transformation of Steroids-Medium was retained at Ϫ20°C from all cell culture experiments. Steroids in medium were recovered by solid-phase extraction. Sep-pak C18 cartridges (Waters, Herts, UK) were activated using methanol (5 ml) followed by water (5 ml). Medium was loaded on to the column, and the eluant was discarded. The column was washed with water (5 ml) and steroids were eluted in methanol (2 ml). The eluant was dried under oxygen-free nitrogen at 60°C and resuspended in ethyl acetate (2 ml) and water (200 l). The organic layer was separated and dried before derivatization to form methoxime-trimethylsilyl derivatives as described (16). Gas chromatographic mass spectrometric analysis was performed in electron impact mode (70 eV) using a Polaris Q ion-trap gas chromatography mass spectrometer (GCMS; Thermofinnigan, UK) (17). The identities of steroids were confirmed by using analytical standards obtained from Steraloids. The level of detection for steroids was Ͻ1 nM. All solvents were high pressure liquid chromatography glass distilled grade (Rathburn, Walkerburn, UK), and reagents were obtained from Sigma (Poole, UK).
In Vivo Glucocorticoid Activity and Suppression of ACTH-Adult lean male Zucker rats (6 -8 weeks old, Harlan Olac Ltd, n ϭ 6 per group) were fed ad libitum and maintained on a 12-h dark/light cycle. Rats were bilaterally adrenalectomized under halothane anesthesia and given 0.9% saline drinking water for 1 week before steroid treatment. Corticosterone, 5␣THB (both 5 mg/kg body weight) or vehicle (saline with 10% ethanol and 10% dimethyl sulfoxide) were injected intraperitoneally at 0900 h. Blood samples were taken at 0, 60, and 120 min after injection, and plasma was stored at Ϫ80°C.
Plasma ACTH concentrations were determined by radioimmunoassay (Eurodiagnostica, Arnhem, The Netherlands). The inter-and intraassay coefficients of variation were Ͻ 4%. Plasma corticosterone levels were measured using an in-house radioimmunoassay (18). The interand intra-assay coefficients of variation were Ͻ10%.
Tissue Concentrations of Steroids-Livers (ϳ500 mg) from adult lean male Zucker rats (6 -8 weeks old, n ϭ 3) were homogenized in ice-cold potassium phosphate buffer (pH 7, 5 vol, 0.05 M) containing epi-tetrahydrocorticosterone and epi-corticosterone (250 ng) (Steraloids). The homogenate was dripped slowly into pre-chilled ethanol (95%) containing glacial acetic acid (3% v/v). The sample was stored overnight at Ϫ80°C, allowed to return to 4°C, and sonicated chilled (1-2 min). The sample was left on ice for 20 -30 min and centrifuged (20,000 ϫ g, 20 mins, 4°C). The supernatant was reduced to dryness under a stream of nitrogen at 60°C, and the dried residue was dissolved in methanol (0.5 ml, 80% v/v), sonicated (1 min), and additional water (0.5 ml) was added. This solution was passed through an activated Sep-pak column (C18, Waters). The steroids were eluted from the column in methanol (2 ml) after washes with water and methanol (40% v/v). The solvent was removed under a stream of nitrogen. Conjugates were hydrolyzed, and the resulting steroids were derivatized and analyzed by GCMS as described above and previously (17). Calibration and blank standards were prepared concomitantly in the absence of tissue.
Statistics-All data are expressed as means Ϯ S.E. Data were analyzed by analysis of variance followed by least squares difference (Fisher's LSD) post hoc tests.

5␣-Reduced Glucocorticoids Activate Glucocorticoid Receptors
ethasone to a minimal degree, having K d values in the M range.
5␣THB Activates GR in a Transient Transfection Assay-HeLa cells were transiently co-transfected with an expression plasmid encoding human GR and LTR-luciferin (in which luciferin expression is driven by the glucocorticoid-responsive murine mammary tumor virus (MMTV) promoter). Both corticosterone and 5␣-reduced corticosterone metabolites stimulated luciferase activity (Fig. 3A) was dependent upon co-transfection of glucocorticoid receptor. To test whether 5␣THB was a full or partial agonist at GR, 5␣THB was added to the incubation medium together with corticosterone. The effects upon luciferase activity were additive (Fig. 3B), suggesting that 5␣THB is a full agonist. It was not possible to measure maximal activation because high concentrations (Ͼ2 M) of corticosterone and 5␣THB caused cell death. To eliminate the possibility that 5␣THB exerts its effects through conversion to corticosterone, the medium from cell experiments was analyzed by GCMS. Corticosterone was not detected in the medium from HeLa cells to which 5␣THB had been added and likewise 5␣THB was not detected in the medium to which corticosterone had been added (data not shown).
5␣THB Increases TAT and PEPCK mRNA in H4IIE Cells-The experiments above relied upon GR added exogenously. To test whether 5␣THB could activate endogenous GR, the ability of 5␣THB to increase TAT and PEPCK mRNA expression was assessed in the glucocorticoid-responsive cell line, H4IIE. After 16 h of incubation with 5␣THB, TAT mRNA expression was induced, albeit to a lesser extent than by corticosterone (Fig.  4A). In the case of PEPCK, corticosterone and 5␣THB induced the transcription of mRNA of PEPCK to the same degree (Fig.  4B). Incubation of H4IIE cells with the GR antagonist RU486 had no effect alone and prevented induction of transcription of TAT and PEPCK mRNAs by 5␣THB.
5␣THB Has Glucocorticoid Effects in Vivo-To test whether 5␣THB is a GR agonist in vivo, ACTH suppression in response to a bolus of steroid was measured in adrenalectomized animals. Basal plasma ACTH levels were high, consistent with prior adrenalectomy. 1 h after steroid injection, plasma ACTH levels were suppressed in corticosterone-treated animals compared with vehicle-treated animals. Suppression of ACTH levels was also seen with 5␣-THB, although the effect was smaller than that with corticosterone (Fig. 5A). Plasma corticosterone levels were increased significantly in the corticosterone-treated animals by 1 h post-injection and remained elevated at 2 h. In contrast, plasma corticosterone levels in vehicle and 5␣THBtreated animals remained low throughout the experiment (Fig. 5B).
Endogenous Concentrations of 5␣-Glucocorticoid Metabolites-Similar concentrations of corticosterone (range 0.42-0.53 nmol/g tissue) and 5␣-THB (range 0.14 -1.2 nmol/g tissue) were measured by gas chromatography/mass spectrometry in extracts from rat liver homogenates (Fig. 6). DISCUSSION Our studies demonstrate that 5␣-reduced glucocorticoids are present at significant concentrations in vivo and are able to compete with dexamethasone for its binding sites in hepatocytes with affinity similar to that of corticosterone. We have shown in two cell models that this binding is associated with increased transcription by GR. Further, 5␣THB has glucocorticoid activity in vivo, as judged by suppression of the hypothalamic-pituitary adrenal axis.
Binding sites for glucocorticoids other than the cytosolic GR exist in rat liver. Low affinity glucocorticoid binding sites (LAGS) have been demonstrated in the microsomal fraction (19) and nuclear envelope (20). However, the nature and function of LAGS is uncertain. Melville et al. (21) have shown that LAGS have a preference for 5␣rather than 5␤-reduced steroids. It is possible that LAGS contributes to total binding in

5␣-Reduced Glucocorticoids Activate Glucocorticoid Receptors
hepatocytes presented here, so that observed K d values may be higher than we would find using purified GR.
In isolated hepatocytes, 5␤-reduced metabolites of corticosterone show substantially less binding to GR than corticosterone or 5␣-reduced glucocorticoids. The ability of 5␣-but not 5␤-reduced metabolites to bind to GR is probably due to differences in stereochemistry around the A/B ring junction. Previous studies of dexamethasone binding to cytosolic GR indicate that binding is sensitive to the configuration at the C5 position (20). The A/B ring junction of the 5␤ stereoisomers is "cis" and therefore skewed relative to the plane through rings B, C, and D, whereas the 5␣ isozyme is "trans" and, hence, has a structure akin to corticosterone. Similar stereospecificity in favor of 5␣-rather than 5␤-metabolites has been observed for binding to the androgen receptor (22).
To address whether 5␣THB was not only bound to but also activated GR, we investigated GR-dependent responses in cells. We have shown in two different cell lines that 5␣-reduced glucocorticoids induced GR-dependent gene transcription. First, in a transient transfection system, luciferase expression was placed under control of the MMTV promoter linked to a glucocorticoid response element (GRE). To induce translation of luciferase, ligand-activated GR must dimerize and associate with the GRE. The results indicate that the 5␣-reduced metabolites were able to induce transcriptionally a glucocorticoidsensitive reporter and that this response was dependent completely upon co-transfected GR. The effects of 5␣THB upon GR activation in this system were additive with those of corticos- Mass chromatogram (m/z 564) shows methoxime-trimethylsilyl derivative of epi-THB (internal standard), 5␤-THB and 5␣THB, recovered from mixture of analytical standards exposed to the procedure, allowing extraction of glucocorticoid metabolites from liver (A) and liver (500 mg) of a lean Zucker rat spiked with internal standard (B).

5␣-Reduced Glucocorticoids Activate Glucocorticoid Receptors
terone, suggesting this metabolite is a full agonist. However, it was not possible to compare maximal activation of 5␣THB to that of corticosterone in this system, because cell viability was compromised at the higher concentrations of glucocorticoids that would have been necessary (Ͼ2 M).
In a second model, H4IIE cells, activation of endogenous GR was measured by induction of transcription of endogenous TAT and PEPCK mRNAs. Expression of TAT and PEPCK mRNAs are controlled by numerous factors, including an upstream positive GRE (23,24). Transcriptional up-regulation of both genes was observed in the presence of 5␣THB, and this was prevented by the addition of the GR antagonist, RU486.
Finally, we showed that 5␣THB administration in vivo induced a classical negative feedback effect upon the hypothalamic-pituitary-adrenal axis. In suppressing ACTH, 5␣THB had a slower onset of effect than corticosterone. This could relate in part to different central nervous system availability of 5␣THB compared with corticosterone. Although 5␣THB is bound minimally to plasma proteins (25), its access across the blood-brain barrier and its susceptibility to export from the central nervous system by multidrug resistance P-glycoproteins are unknown. Moreover, corticosterone acts through both mineralocorticoid and glucocorticoid receptors. The MR contributes more to early "shut-off" after stress and may account for the rapid effect of corticosterone (26). It is not known whether 5␣THB can bind to MR; however, 5␣DHB has no mineralocorticoid activity (27).
Importantly, we found no evidence in any of our model systems of the effects of 5␣-reduced corticosterone being dependent upon metabolism to corticosterone. Using GCMS analysis of cell culture medium, or radioimmunoassay of plasma, corticosterone was not detected after administration of 5␣THB. The concentrations of corticosterone and 5␣THB were similar in extracts of liver from lean rats, suggesting that sufficient 5␣reduced metabolites of corticosterone are present to influence physiological processes. The assay did not reveal the presence of 5␤-reduced glucocorticoid metabolites, in keeping with data from urinary analyses reported previously, showing that the 5␣-reduced glucocorticoids are the predominant A-ring-reduced metabolite in the Zucker rat (16). This result is unlike that from human urine, where equivalent 5␣-and 5␤-reduced metabolites of cortisol are detected.
The implication of these findings is that, when present at physiological concentrations, 5␣-reduced glucocorticoids contribute to GR activation. Potentially, this will occur wherever 5␣-reductases are expressed, including glucocorticoid target tissues such as liver and adipose tissue. This provides another intracrine mechanism, over and above the influence of 11␤hydroxysteroid dehydrogenases (28), whereby local concentrations of active glucocorticoids can be modulated independently of circulating corticosterone levels. These findings are of sig-nificant clinical importance because our group has demonstrated previously that 5␣-reduction of glucocorticoids is increased (often at least 2-fold) in human obesity and in the Zucker rat model of obesity (16,29). 5␣-Reduction of glucocorticoids is also increased in women with polycystic ovary syndrome (30). Whether 5␣-reductases contribute to GR activation and the adverse metabolic phenotype in these subjects is now a priority for further investigation.