Liver X Receptors Downregulate 11β-Hydroxysteroid Dehydrogenase Type 1 Expression and Activity

3T3-L1 preadipocytes were purchased from American Type Culture Collection (Rockville, MD) and propagated in Dulbecco’s modified Eagles medium (DMEM) containing 4.5 g/l glucose, supplemented with 10% bovine calf serum, penicillin/streptomycin (50 units/ml and 50 μg/ml, respectively), and 2 mmol/l glutamine (all from Invitrogen, Carlsbad, CA) at 37°C in humidified atmosphere in presence of 5% CO₂. Two-day postconfluent 3T3-L1 cells were differentiated into adipocytes in six-well plates (Corning-Costar, Bodenheim, Germany) by treatment for 2 days with 0.5 μmol/l dexamethasone, 200 μmol/l 3-isobutyl-1-methylxanthine (both from Sigma), and 1 μg/ml insulin (Roche, Mannheim, Germany) in 3T3-L1 differentiation medium (DMEM supplemented with 10% fetal bovine serum [FBS; Invitrogen], antibiotics, and glutamine) (according to ref. 27, with minor modifications). Differentiation was completed by incubation with 1 μg/ml insulin for additional 3 days in 3T3-L1 differentiation medium. More than 95% of cells were transformed into adipocytes, as evaluated by the detection of lipid droplets by phase contrast microscopy and oil-red O staining (not shown), and this population is referred to as “3T3-L1 adipocytes.” 3T3-L1 adipocytes were carefully washed twice in DMEM, and medium was changed to either 3T3-L1 differentiation medium or a minimal serum-free differentiation medium (modified from ref. 28) (DMEM/F12 supplemented with 10 μg/ml apo-transferrin and antibiotics) 2 days before the addition of drugs in the same medium.

Day 14.5 postconception mouse embryonic fibroblasts (MEFs) were prepared from wild-type, LXRα^−/−, and LXRβ^−/− embryos (for details see the section entitled Animal experiment) by standard methods and propagated in MEF growth medium (DMEM supplemented with 10% FBS, 1 mmol/l pyruvate, 1% [v/v] nonessential amino acids MEM, 2 mmol/l glutamine, penicillin/streptomycin [all from Invitrogen], and 0.1 mmol/l 2-mercaptoethanol). Two-day postconfluent MEFs were treated with 1 μmol/l dexamethasone, 500 μmol/l 3-isobutyl-1-methylxanthine (both from Sigma), 0.5 μmol/l rosiglitazone (BRL49653; generously provided by Steven A. Kliewer, GlaxoSmithKline), and 10 μg/ml insulin (Roche) for 2 days in MEF differentiation medium (AmnioMax basal medium supplemented with 7.5% [vol/vol] AmnioMax complete C-100, 7.5% [vol/vol] FBS [all from Invitrogen], glutamine, and antibiotics) as described (29), followed by 6–7 additional days in MEF differentiation medium including only insulin and rosiglitazone. Depending on the cell line used, 40–95% of cells were finally differentiated into adipocytes, and this proportion did not change by LXR agonist treatment (data not shown). Cells were carefully washed twice, and medium was changed to MEF growth medium for 2 days before drug treatment in the same medium.

Adipocytes were treated for 48 h, unless otherwise indicated, with substances (purchased from Sigma, unless otherwise indicated) added from 1,000-fold stock solutions in the stated solvents as follows: specific LXR agonist T0901317 (1 μmol/l final concentration, dissolved in DMSO; a generous gift of Karobio, Huddinge, Sweden); rosiglitazone (PPARγ agonist, 1 μmol/l, DMSO); Wy14643 (PPARα agonist, 5 μmol/l, DMSO); 9-cis-retinoic acid (RXR agonist, 0.1 μmol/l, ethanol); 22(R)-, 22(S)-, and 20(S)-hydroxycholesterol (all 20 μmol/l, ethanol); and cycloheximide (1.25 μmol/l, DMSO). Parallel samples without drugs included the same solvent(s) (≤0.2% [vol/vol] maximal concentration). To estimate 11β-HSD-1 mRNA half-life, adipocytes were pretreated for 16 h with LXR agonist or vehicle, respectively, before adding actinomycin-d (5 μg/ml, DMSO) and harvesting cells for RNA on multiple time points up to 72 h.

Total RNA from cell culture wells was isolated by Trizol reagent (Invitrogen) according to the manufacturer’s instructions. To purify RNA from tissue samples, frozen tissues were homogenized in Trizol followed by centrifugation (13,000 rpm for 10 min at 4°C) to remove insoluble material and floating lipids before RNA isolation. RNA was checked for quality by spectrophotometry at 260/280 nm (ratio always >1.9) and 1.5% agarose gel electrophoresis (only occasionally for RNA preparations from cultured cells).

A total of 1 μg RNA was treated with DNase (amplification grade; Invitrogen) before reverse transcription into cDNA by Superscript II (Invitrogen) using random hexamer priming according to the manufacturer’s instructions. Specific mRNA expression was quantitated by Taqman real-time RT-PCR (Applied Biosystems, Foster City, CA). This quantitation method is based on optimal amplification of a short DNA fragment (∼100 bp). A 5′ -dye-3′-quencher-labeled oligonucleotide probe attaches to the middle part of the amplicon and is degraded during amplification by the 5′-nuclease function of the used Taq polymerase, which releases the dye from its quencher, thereby creating fluorescence. Standard curves or a direct comparative method were used for data analysis as applicable. Primer/probe sequences are given in Table 1; the sequence for quantitation of 11β-HSD-1 was derived from reference 30. The commercially available reaction mixture for 18S (Applied Biosystems) was used as a standard in all experiments.

11β-Hydroxysteroid dehydrogenase activity was determined with minor modifications as described (31). Briefly, 3T3-L1 adipocytes in six-well plates treated for 2 days with LXR agonist or vehicle were washed twice before adding 15 nmol/l cortisone in the presence of 100,000 cpm [³H]cortisone as tracer (specific activity 3.1 TBq/mmol) in 1 ml prewarmed 3T3-L1 differentiation medium. Cells were incubated for 15 min at 37°C before removing 900 μl medium. Conversion to cortisol was determined after the addition of excess unlabeled cortisone and cortisol (5 μl of 10 mmol/l solution) and the extraction into ethylacetate, followed by thin layer chromatography (stationary phase, silica plates [Merck, Darmstadt, Germany]; mobile phase, dichloromethane/acetone 4:1 [vol/vol]). Steroids were localized by UV illumination, and the fractional conversion was quantitated by liquid scintillation counting. The background level of radioactive tracer product formation was <1%.

Wild-type (LXRα^+/+β^+/+), LXRα^−/−, LXRβ^−/−, and LXRα^−/−β^−/− mice as well as mice heterozygous for either receptor were generated by gene targeting in our laboratory as described in detail (32,33). The mice used in this study were Sv129/C57BL/6 hybrids finally backcrossed in C57BL/6 mice for three and five generations for the in vivo experiment and MEF preparation, respectively. Mice were housed with a 12-h light/dark cycle in the specific pathogen-free animal unit at the University Hospital at Huddinge. For the experiment, male mice, 10–12 months of age, had free access to water and a low-fat standard rodent diet (R36, Lactamin AB, Vadstena, Sweden) or the same diet supplemented with 0.025% (wt/wt) T0901317 (24) for 7 days. Mice were killed and tissues snap frozen in liquid nitrogen and kept at −80°C until isolation of RNA. The experiment was approved by the local ethics committee for animal experiments, and the guidelines for the use and care of laboratory animals were followed.

Data are presented as means ± SE. If not explicitly stated, data appeared normally distributed and were compared by unpaired Student’s t test, otherwise by Mann-Whitney’s U test. A two-tailed P < 0.05 was considered statistically significant.

11β-HSD-1 and LXRs, particularly LXRα, are highly expressed in adipocytes (14,34). Treatment of adipocytes differentiated from 3T3-L1 preadipocytes and mouse embryonic fibroblasts, respectively, with a synthetic LXR agonist (T0901317) (18) considerably downregulated 11β-HSD-1 mRNA (Fig. 1A), similar to treatment with a PPARγ ligand (rosiglitazone, Fig. 1) (33), whereas the PPARα ligand (WY14643) was ineffective. Moreover, 9-cis-retinoic acid, a ligand for the obligate heterodimer partner of these receptors, RXR (35) (but also of the retinoic acid receptor) exhibited an even stronger inhibitory effect. Treatment of differentiated adipocytes from embryonic fibroblasts derived from LXRα^−/− and LXRβ^−/− single knockout embryos revealed that each of the LXRs can mediate this effect (data not shown). In contrast to 11β-HSD-1, mRNA levels of two adipocyte marker genes, adipocyte-specific fatty acid binding protein and glycerol-3-phosphate dehydrogenase, did not change or were enhanced by treatment with the LXR agonist, showing that downregulation of 11β-HSD-1 mRNA was not due to a nonspecific toxic effect (Fig. 1B).

The effect of the LXR agonist was concentration-dependent, with an EC₅₀ of 52 nmol/l (Fig. 2) (95% CI 27–102 nmol/l). The concentration is in the specific range of the drug and strikingly conforms to the published EC₅₀ of ∼50 nmol/l derived from cell reporter assays (18). Thus, the LXR agonist most likely provokes 11β-HSD-1 downregulation in adipocytes by specifically activating LXRs.

To evaluate potential regulation by natural LXR ligands, which are chemically unrelated to the synthetic ligand, 22(R)-hydroxycholesterol and 20(S)-hydroxycholesterol were tested for their effect on 11β-HSD-1 mRNA levels (Fig. 3). Both 22(R)-hydroxycholesterol and 20(S)-hydroxycholesterol decreased 11β-HSD-1 mRNA levels in 3T3-L1 adipocytes, whereas the inactive stereoisomer 22(S)-hydroxycholesterol (16) was ineffective. Thus, endogenous LXR ligands could participate in the regulation of 11β-HSD-1 gene expression.

A reduction of 11β-HSD-1 mRNA by treatment with the specific LXR agonist was evident after a lag period of at least 8 h and achieved its maximum after ∼48 h (Fig. 4). Therefore, 48 h drug treatment was generally applied. The half-life of 11β-HSD-1 mRNA was almost identical to that of 18S mRNA (∼24 h) in 3T3-L1 adipocytes (data not shown), which could delay the detection of subtle differences during the first hours following treatment initiation. The apparent mRNA half-life was not changed by treatment with the LXR agonist (data not shown), indicating that altered gene expression, but not enhanced mRNA degradation, causes the decrease in 11β-HSD-1 mRNA level.

On the other hand, the delay in decreasing 11β-HSD-1 gene expression by LXRs could point to an indirect mechanism. Experiments in presence of cycloheximide revealed that ongoing protein synthesis was required for the effect of LXR agonist on 11β-HSD-1 mRNA (Fig. 5). In contrast, upregulation of sterol regulatory element-binding protein (SREBP) 1c mRNA, a known LXR target gene (24), was not inhibited by cycloheximide.

11β-HSD-1 regulates local glucocorticoid concentrations due to its enzymatic activity. Therefore, we assessed whether LXR not only downregulates 11β-HSD-1 mRNA expression but also enzymatic activity to convert exogenous cortisone to cortisol. Parallel to its downregulation on the mRNA level, treatment of adipocytes with LXR agonist significantly decreased 11β-HSD-1 enzymatic activity to 78% of control values after 2 days (Fig. 6).

To analyze whether downregulation of 11β-HSD-1 gene expression by LXR also occurs in vivo, we treated wild-type and LXRα^−/−β^−/− mice, previously generated in our lab (32,33), with the synthetic LXR agonist T0901317 or a control diet for 7 days. 11β-HSD-1 mRNA expression was evaluated in epididymal and brown adipose tissue (BAT) as well as liver (Fig. 7A) along with the known LXR target gene SREBP-1c (Fig. 7B). Regulation of SREBP-1c was significantly increased by the treatment in all analyzed tissues of wild-type but not LXRα^−/−β ^−/− animals, revealing that the treatment was sufficient to induce LXR-specific responses (Fig. 7B). 11β-HSD-1 gene expression showed large interindividual variation in epididymal adipose tissue, thereby prohibiting detection of regulation in the given number of animals (Fig. 7A). 11β-HSD-1 mRNA was decreased to 41 ± 9% in BAT of LXR agonist-treated mice. This decrease was statistically different at P = 0.06 (U test) despite large variation among control mice. Notably, LXR agonist treatment significantly diminished 11β-HSD-1 mRNA expression in liver to 52 ± 5% of control values (P < 0.01). Moreover, the specificity of the regulation was underlined by the fact that the LXR agonist did not alter 11β-HSD-1 gene expression in LXRα^−/−β ^−/− animals in either tissue. Liver PEPCK is a typical glucocorticoid-responsive gene and a key enzyme in gluconeogenesis (36). In parallel with the decrease in hepatic 11β-HSD-1 gene expression, PEPCK mRNA expression was markedly decreased in liver to 38% of control values in LXR agonist-treated wild-type but not LXRα^−/−β^−/− mice (Fig. 7C).

11β-HSD-1 activates circulating inactive 11-oxo corticosteroids, thereby regulating local glucocorticoid concentrations in peripheral tissues, particularly adipose tissue and liver (37–39). Here, we show that 11β-HSD-1 is downregulated by the nuclear oxysterol receptors LXRα/β in adipocytes and liver, suggesting a novel role of these receptors in endocrine function with the potential to exert positive effects on insulin sensitivity.

The data presented here clearly reveal that activation of LXRs downregulate 11β-HSD-1 gene expression in adipocytes. In particular, we demonstrate similar effects of chemically unrelated LXR agonists including bona fide natural ligands [22(R)-, 20(S)-hydroxycholesterol] and the lack of effect of an inactive stereoisomer [22(S)-hydroxycholesterol]. Moreover, an EC₅₀ equal to that established for LXR-specific effects of the used synthetic ligand (18) and the lack of effect in LXRα^−/−β^−/− mice underline the specificity of the LXR-mediated downregulation. No significant differences in 11β-HSD-1 mRNA half-life could be found, emphasizing that the decrease in mRNA levels was due to downregulation of gene transcription. Thus, activation of LXRs by endogenous and synthetic ligands causes downregulation of 11β-HSD-1 gene expression in vitro and in vivo.

The observed downregulation of adipocyte 11β-HSD-1 expression was more pronounced than that of 11β-HSD-1 activity levels (∼50 vs. 78%). This difference could be due to disequilibria at the treatment time of 48 h between RNA and protein half-lives. On the other hand, 11β-HSD-1 activity is also influenced by the availability of NADPH, whose provision is controlled by glucose-6-phosphate dehydrogenase. Because glucose-6-phosphate dehydrogenase is induced by SREBP-1c (40), it could be speculated that increased expression of SREBP-1c by LXR agonists induces glucose-6-phosphate dehydrogenase, thereby increasing NADPH levels.

11β-HSD-1 is the first published example showing that LXRs can mediate the downregulation of a specific gene. Except the notion that ongoing protein synthesis is required (Fig. 5), no information is currently available on the underlying mechanism. The downregulatory effects found with LXR ligands parallel those published previously about adipocytes for PPARγ ligands (34). According to the tissue distribution of PPARs (41) and the known tissue-specific activation of other PPAR response elements (42), the effect of PPARα agonists to downregulate 11β-HSD-1 expression appears restricted to liver (43) and did not occur in our experiments with cultured adipocytes (Fig. 1A). Due to their downregulatory role on 11β-HSD-1 expression in adipocytes, thereby decreasing production of active glucocorticoids, some of the insulin-sensitizing effects of thiazolidinediones (PPARγ agonists) might be mediated by 11β -HSD-1 (34). Interestingly, LXRα, the major LXR in adipose tissue and liver (14), is a direct target gene of PPARγ as well as PPARα (44–46). Thus, the effects of PPARs could be mediated by increasing LXRα expression, though a delay in downregulation by PPARγ compared with LXR activation could not be observed (Fig. 4). PPARγ strongly promotes adipocyte differentiation (47,48), and differentiation into adipocytes induces expression of 11β-HSD-1 in precursor cells (11,34). Accordingly, PPARγ agonist treatment even increased 11β-HSD-1 activity in human adipose stromal cells, perhaps by inducing adipogenesis (49). In contrast, LXRs are not required for adipocyte differentiation, and in vitro treatment of preadipocytes with an LXR agonist increases adipocyte triglyceride content only moderately (H. Nebb, personal communication). Thus, LXR agonist treatment could be particularly useful for the reduction of local glucocorticoid reactivation by diminishing 11β-HSD-1 expression without negative side effects through worsening obesity.

11β-HSD-1 expression was found to be downregulated to ∼50% by long-term LXR agonist treatment in BAT and liver of wild-type but not LXRα ^−/−β^−/− knockout mice (Fig. 7A). Importantly, already a two- to threefold overexpression of 11β-HSD-1 in omental adipocytes was sufficient to provoke a phenotype reflecting the human metabolic syndrome, including hyperlipidemia and decreased insulin sensitivity (6). Accordingly, it can be expected that a decrease in 11β-HSD-1 expression and activity as observed in our experiments will significantly ameliorate 11β-HSD-1-mediated effects on metabolism and has the potential to increase insulin sensitivity. Notably, the hepatic decrease of 11β-HSD-1 expression was associated with a strong downregulation of PEPCK expression. Because PEPCK is a glucocorticoid-responsive gene in liver (36), the parallel regulation with 11β-HSD-1 could point to a physiological importance of diminished local glucocorticoid production by LXR-mediated reduction of 11β-HSD-1. Since PEPCK catalyzes the rate-limiting step in gluconeogenesis, its downregulation by LXR agoinst treatment emphasizes a role of LXRs in regulating hepatic glucose homeostasis with probable beneficial effects of LXR agonists in diabetic patients.

Due to large variation in gene expression, we could not detect LXR-mediated regulation of 11β-HSD-1 in epididymal WAT, although gene regulation in WAT generally parallels that in BAT, except for a small number of BAT-specific genes such as UCP 1 (50). However, expression of 11β-HSD-1 mRNA in epididymal WAT was only half of that in BAT and approximately one-third of that observed in liver of untreated control animals (Fig. 7, legend). Hence, LXR-driven downregulation appears particularly effective in tissues with high basal amounts of 11β-HSD-1 mRNA. Nonobese subjects and animals, as those analyzed in this study, express considerably lower levels of 11β-HSD-1 in visceral adipose tissue compared with obese subjects (4,5, 51). It could be speculated that LXR agonists particularly downregulate the high 11β-HSD-1 expression as it occurs in visceral adipose tissue of obese subjects.

Beyond discovering an involvement of LXRs in regulation of endocrine function, these data indicate that a downregulation of 11β-HSD-1 expression is feasible by LXR agonist treatment. Of course, the physiological consequences of such a treatment in humans have yet to be explored. Due to the pronounced effect of LXRs on cholesterol efflux, particularly from macrophages (19–22), implicating a strong anti-atherogenic effect, such substances are currently developed for clinical application. Though further analyses on the pharmacological effects of such drugs must be awaited, a decrease in 11β-HSD-1 expression by LXR agonists could have beneficial effects on the metabolic control in patients with type 2 diabetes who are at high risk for developing cardiovascular disease.

This research has been supported by a Marie Curie Fellowship of the European Community program Human Potential under contract number HPMF-CT-2000-00898 (to T.M.S.). The European Commission is not responsible for any views or results expressed. This study was further supported by grants from the Novo Nordisk Foundation, the Swedish Union of Physicians (to U.O.), the Norwegian Cancer Society (Grant no. A97030/002, to K.R.S.), the Tore Nilsson Foundation, the Swedish Cancer Society (to G.U.S.), the Swedish Scientific Council, and KaroBio AB, Sweden (to J.-Å.G.).

We are grateful to Jacob B. Hansen for sharing his great experience in adipocyte differentiation.