Expression of the Insulin-responsive Glucose Transporter GLUT4 in Adipocytes Is Dependent on Liver X Receptor α*

The insulin-responsive glucose transporter GLUT4 plays a crucial role in insulin-mediated facilitated glucose uptake into adipose tissue and muscle, and impaired expression of GLUT4 has been linked to obesity and diabetes. In this study, we demonstrate that liver X receptors (LXRs) regulate the expression of GLUT4 through direct interaction with a conserved LXR response element in the GLUT4 promoter. The expression of GLUT4 in WAT is induced by a potent LXR agonist in wild type, LXRα-/-, and LXRβ-/- mice but not in LXRα-/-β-/- mice, demonstrating that both LXRs are able to mediate ligand activated transcription of the GLUT4 gene. However, basal and insulin stimulated expression of GLUT4 in epididymal WAT is reduced only in mice carrying ablation of the LXRα isoform. The expression of GLUT4 is furthermore correlated to the induction of LXRα during mouse and human adipocyte differentiation. LXRβ is thus apparently not able to rescue basal expression of GLUT4 in the absence of LXRα. We have previously demonstrated that LXRα is down-regulated in animal models of obesity and diabetes, thus revealing a striking correlation between GLUT4 and LXRα expression in insulin-resistant conditions. This suggests that the LXRα isoform has a unique role in adipose expression of GLUT4 and suggests that alteration of adipose tissue expression of LXRα might be a novel tool to normalize the expression of a gene that is dysregulated in diabetic and insulin-resistant conditions.

The insulin-responsive glucose transporter GLUT4 plays a crucial role in insulin-mediated facilitated glucose uptake into adipose tissue and muscle, and impaired expression of GLUT4 has been linked to obesity and diabetes. In this study, we demonstrate that liver X receptors (LXRs) regulate the expression of GLUT4 through direct interaction with a conserved LXR response element in the GLUT4 promoter. The expression of GLUT4 in WAT is induced by a potent LXR agonist in wild type, LXR␣ ؊/؊ , and LXR␤ ؊/؊ mice but not in LXR␣ ؊/؊ ␤ ؊/؊ mice, demonstrating that both LXRs are able to mediate ligand activated transcription of the GLUT4 gene. However, basal and insulin stimulated expression of GLUT4 in epididymal WAT is reduced only in mice carrying ablation of the LXR␣ isoform. The expression of GLUT4 is furthermore correlated to the induction of LXR␣ during mouse and human adipocyte differentiation. LXR␤ is thus apparently not able to rescue basal expression of GLUT4 in the absence of LXR␣. We have previously demonstrated that LXR␣ is downregulated in animal models of obesity and diabetes, thus revealing a striking correlation between GLUT4 and LXR␣ expression in insulin-resistant conditions. This suggests that the LXR␣ isoform has a unique role in adipose expression of GLUT4 and suggests that alteration of adipose tissue expression of LXR␣ might be a novel tool to normalize the expression of a gene that is dysregulated in diabetic and insulin-resistant conditions.
One important physiological consequence of obesity is a reduced ability to respond to insulin in peripheral tissues. Untreated, this insulin resistance is associated with increased risk for development of cardiovascular disease and, in later stages, type 2 diabetes. Numerous studies have shown that the effect of insulin to increase glucose uptake is directly dependent on recruitment of GLUT4, a facilitative glucose transporter, from an intracellular vesicle pool to the plasma membrane (reviewed in Refs. 1 and 2). GLUT4 is expressed exclusively in tissues exhibiting insulin-stimulated glucose uptake, such as muscle, heart, and adipose tissue (3). In the basal state, GLUT4 is sorted to an intracellular compartment with a low exocytotic rate, which changes to a high exocytotic rate upon insulin stimulation through activation of intracellular phosphorylation cascades (2). Although this insulin-stimulated translocation of GLUT4 is the main factor that increases glucose uptake, it also depends on the amount of stored GLUT4 in such a compartment. The expression of GLUT4 is reduced in rodent models of insulin deficiency (4,5) and in adipose tissue of human obese or type 2 diabetic subjects (6,7), directly linking adipose expression of GLUT4 to insulin resistance. Moreover, recent studies in mice have demonstrated that selective ablation of GLUT4 in adipose tissue leads to decreased whole body glucose tolerance and insulin responsiveness (8), whereas forced overexpression enhances systemic glucose clearance and insulin sensitivity (9). This implies an important role of adipose tissue GLUT4 expression in whole body glucose homeostasis, despite the fact that it is estimated that adipose tissue only accounts for 10% of the insulin-mediated whole body glucose uptake (10).
The differentiation and maintenance of adipose tissue is driven by the transcription factor peroxisome proliferator-activated receptor ␥ (PPAR␥), 1 a member of the nuclear receptor family (11)(12)(13). PPAR␥ heterodimerizes with retinoid X receptor ␣ (RXR␣) and controls transcription of target genes by binding to a direct repeat-1 PPAR␥ response element located in the promoter of target genes (12)(13)(14). Natural high affinity ligands for PPAR␥ have not been identified, but endogenous polyunsaturated fatty acids and 15-deoxy-⌬ 12,14 -prostaglandin J 2  show micromolar affinity for the receptor in line with their serum levels (15,16). Thiazolidinediones (TZDs), a new class of synthetic antidiabetic drugs, have been characterized as high affinity ligands for PPAR␥ (17).
Recently, we suggested that the liver X receptors (LXRs) are involved in the regulation of adipose tissue fatty acid metabolism and triglyceride (TG) storage (18). LXR␣ (19,20) and LXR␤ (21,22) heterodimerize with RXR␣ and control transcription by binding to a direct repeat type 4 LXR response element (LXRE) located in the promoter of their target genes (20,23). The LXR␤ isoform is ubiquitously expressed in adults (21), whereas the expression of LXR␣ is predominantly restricted to tissues known to play important roles in lipid metabolism, such as liver, skeletal muscle, adipose tissue, kidney, and small intestine, but a lower expression level is also seen in spleen and pituitary and adrenal glands (19,20,24). Naturally occurring oxysterols, mainly derivatives of cholesterol, function as high affinity ligands for LXRs (25,26), and recently, a synthetic nonsteroidal LXR agonist was described (27).
Several studies have demonstrated the importance of LXRs in cholesterol homeostasis (i.e. by regulating ATP-binding cassette transporters (28,29), human cholesterol ester transfer protein (30), apolipoprotein E (31), and the rodent cholesterol-7␣ hydroxylase gene (26,32)). In addition, LXRs have been linked to lipid metabolism by their direct regulation of the sterol regulatory element-binding protein 1c (SREBP-1c) (33,34) and fatty acid synthase (FAS) (35) promoters as well as the observation that insulin-mediated regulation of these lipogenic genes in liver is dependent on LXRs (36). This suggests that the LXRs are important regulators of lipid metabolism, in addition to their roles as regulators of cholesterol homeostasis.
Earlier studies have shown that the transcription of GLUT4 in adipocytes is induced by ligands that activate PPAR␥ (37) and that the reduced levels of GLUT4 in adipose tissue in animal models of diabetes is normalized by treatment with PPAR␥ activators (38,39). The exact molecular mechanism responsible for this regulation is, however, not known. We demonstrated recently that the expression of LXR␣ is regulated similarly to GLUT4 in adipose tissue by being repressed in a model of obesity, which was normalized by treatment with a potent PPAR␥ activator (18). Furthermore, activation of LXRs in adipose tissue increases basal glucose uptake (40) and incorporation of TGs into lipid droplets (18). Since all of these factors pointed to a role of LXRs in the regulation of glucose metabolism, we initiated a study to determine whether genes involved in this pathway are regulated by LXRs.
Here we demonstrate that the adipose tissue expression of GLUT4 is directly regulated by both LXR␣ and LXR␤ upon ligand stimulation but that the basal expression of GLUT4 is selectively dependent on the LXR␣ isoform. In view of the recently proposed antidiabetic effects of LXRs (41), this work supports such a role for LXRs, since we now demonstrate a direct link between LXR␣ and a gene that is down-regulated in diabetic and insulin-resistant conditions.
3T3-L1 cells were differentiated into adipocytes by the addition of adipogenic factors (23) with some adjustments. Preadipocytes were seeded at passage 6 -8, grown 1-2 days postconfluence, and exposed to adipogenic medium (0.5 mM isobutylmethylxanthine, 0.1 M dexamethazone, and 1 g/ml insulin) for 3 days and then medium containing insulin (1 g/ml) for 3 days, followed by an additional 7 days with regular medium to obtain mature adipocytes (day 13).
Cloning of the GLUT4 cDNA and Probe-The mouse GLUT4 cDNA and the human GLUT4 probe were generated by reverse transcriptase polymerase chain reaction (RT-PCR) from total RNA isolated from differentiated 3T3-L1 and SGBS cells, respectively, using the ImProm-II™ reverse transcription system (Promega) with an oligo(dT) 16 primer, followed by a PCR (30 cycles) with Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA). The obtained PCR fragments were cloned into the pPCR-Script vector using the PCR-Script TM Amp cloning kit (Stratagene). PCR fragments obtained by a second PCR amplification from these vectors were used for labeling reactions in Northern analysis.
Cloning and Mutagenesis of the GLUT4 Promoter-A previously described nucleotide sequence for the human GLUT4 promoter (accession number M61126) was used as a bait in a BLAST search against htgs sequences at NCBI to identify a longer 5Ј-upstream sequence (annealed to accession number AC003688). The mouse GLUT4 promoter was identified with the full-length mouse GLUT4 mRNA as bait (accession number BC014282 annealed to accession number AL596185.8). For both promoters, the promoter sequence spanning 10,000 bp up-and downstream from the transcription start site was extracted and analyzed using a consensus LXRE (DGGTYA HWHW MGKKCA) generated by the GCG program (Wisconsin Package version 10.0, Genetics Computer Group (GCG), Madison, WI) to localize potential LXR response elements.
The full-length GLUT4 promoter was amplified by PCR with Pfu Turbo DNA polymerase (Stratagene) from human genomic DNA (6550-1; Clontech) with primers selected by the Primer3 program (51) (primers were 5Ј-h-GLUT4-promoter (TAAGATCTCTGTGCTGGAACT-CAGGGATCA) and 3Ј-h-GLUT4-promoter (TAAGATCTTCGGAGC-CTATCTGTTGGAAGC)). The promoter was amplified in a 100-l PCR mixture (100 -500 ng of template, 1ϫ cloned Pfu buffer, 200 M each dNTP, 2.5 units of Pfu Turbo polymerase, and 20 pmol of each primer) in a long cycled PCR (2 min at 95°C; 1 min at 95°C, 45 sec at 60°C (2 min/1000 bp) at 72°C for 30 cycles; 10 min at 72°C). The amplified promoter fragment was inserted into the pPCR-Script vector (Stratagene) prior to insertion into the BglII site in the pGL3-Basic luciferase reporter vector (Promega). The GLUT4 SmaI deletion reporter was made by restriction cutting of the full-length reporter with SmaI followed by religation of the vector.
The mutation of the LXRE element in the human GLUT4 reporter was introduced by employing the mutated oligonucleotides used in the EMSA (Fig. 5A) as mutation-targeting primers. 10 ng of the full-length GLUT4 reporter was amplified in a 50-l PCR mixture (1ϫ cloned Pfu buffer, 200 M each dNTP, 2.5 units of Pfu Turbo polymerase, and 15 pmol/each mutation oligonucleotide) in a long cycled PCR reaction (2 min at 95°C; 45 s at 95°C, 30 s at 55°C, 15 min at 68°C for 20 cycles). Following PCR, the PCR mixture was treated with DpnI (1 unit) at 37°C for 1 h and transformed into supercompetent cells. Positive clones were identified by restriction analysis and verified by sequencing.
Transfection, Luciferase Assay, and Electrophoretic Mobility Shift Assay-For reporter gene assays, COS-1 cells were transiently transfected in six-well dishes with luciferase reporters (5 g) and co-transfected with pCMX-RXR␣, pCMX-LXR␣ (1 g each), and pSV-␤-galactosidase (3 g) expression vectors with calcium phosphate precipitation (42). Total DNA concentration was adjusted to 12 g with corresponding empty expression vectors and pGL3-basic vector. Differentiated human SGBS cells were transfected (12 wells) with reporters (800 ng) and pTK Renilla luciferase (80 ng) with LipofectAMINE Plus reagent (Invitrogen). Total DNA concentration was adjusted to 1.0 g by the addition of pGL3-basic vector. After 3 h of transfection, medium containing appropriate reagents was added for 48 h. Cells were harvested in 100 l lysis buffer, and luciferase activities were measured using the dual luciferase assay kit (Promega).
For preparation of nuclear extract, COS-1 cells were transfected in 10-cm dishes with 20 g of vector mixed with 70 l of LipofectAMINE (Invitrogen) in 10 ml of serum and antibiotic-free medium for 6 h, followed by 48-h incubation in 20 ml of medium containing serum and antibiotics. Nuclear extracts from 3T3-L1 and transfected cells were isolated according to Ref. 44, except that the proteinase inhibitor phenylmethylsulfonyl fluoride was replaced with 2ϫ Complete (Roche Applied Science). Protein concentrations were measured by a BC assay (Interchim, Montlucon, France).
In the 24-h feeding experiment, mice were gavage-fed twice (24 and 8 h before the mice were sacrificed) using 50 mg/kg T0901317 or 1 mg/kg darglitazone in a vehicle containing 1% carboxymethyl-cellulose (Sigma). Control mice received vehicle only. In the experiments including insulin injections, the mice were first fed twice with T0901317 (30 mg/kg) as described above and then injected with PBS (as a control) or insulin as a single 0.2-unit injection (intraperitoneally) of Actrapid insulin (Novo Nordisk, Bagsvaerd, Denmark) 3 h before they were sacrificed. For the 1-week feeding experiment, the mice were fed orally once a day with darglitazone (1 mg/kg) or T0901317 (30 mg/kg). For all in vivo experiments, tissues were rapidly frozen in liquid nitrogen and stored at Ϫ70°C until isolation of total RNA.
Quantitative RT-PCR Analysis-Total RNA was isolated using TRIZOL® Reagent (Invitrogen). DNA was removed from RNA preparations (DNA-free kit; Intermedica AB, Stockholm, Sweden), and first strand synthesis was performed using random primers with the Superscript first strand synthesis system for RT-PCR (Invitrogen). Quantitation of mRNA was performed with a quantitative, real time PCR approach using the ready-to-use assay SYBR green PCR core reagents kit based on the Taqman technology (Applied Biosystems, Stockholm, Sweden). The threshold cycles (Ct) for the GLUT4 and the 36B4 control signals were determined in triplicate experiments, and the relative RNA quantitation was calculated using the comparative Ct method, where ⌬Ct is Ct GLUT4 Ϫ Ct 36B4 . The ⌬Ct values were used to calculate 2 Ϫ⌬Ct . GLUT4 and 36B4 control were amplified with similar PCR efficiency using the following primers: GLUT4 fw, 5Ј-TCATTGTCGGCAT-GGGTTT-3; GLUT4 rev, 5Ј-CGGCAAATAGAAGGAAGACGTA-3Ј; 36B4 fw, 5Ј-GAGGAATCAGATGAGGATATGGGA-3Ј; and 36B4 rev, 5Ј-AAGCAGGCTGACTTGGTTGC-3Ј.

Expression of GLUT4 during Differentiation of Fibroblasts
into Adipocytes-In our previous studies on LXR␣ in adipocytes, we have noticed that the induction of this nuclear receptor is delayed during differentiation, although we and others have established this gene as directly regulated by PPAR␥ in mature adipocytes (18) and macrophages (47). Interestingly, the GLUT4 transcript has previously been described as a late adipocyte differentiation marker (i.e. induced later than the other characterized PPAR␥ target genes) and moreover to be regulated by PPAR␥ activators in mature adipocytes (37). Thus, GLUT4 was a proven candidate for further investigation.
To test whether the induction of GLUT4 and LXR␣ overlaps during adipocyte differentiation, we cultured mouse 3T3-L1 cells 2 days postconfluence and let them differentiate into adipocytes. Samples were withdrawn each day during the differentiation process (from day Ϫ1 until day 7). As described earlier, using this differentiation protocol (18), the PPAR␥ transcript was strongly induced from day 2 to 3 during differentiation (Fig. 1A), with an overlapping induction of a set of characterized PPAR␥ target genes (i.e. aFABP) (14). However, both the LXR␣ and GLUT4 transcripts were induced 1 day later, accompanied by induction of the LXR target gene SREBP-1, as demonstrated previously (18).
To test whether this induction sequence also occurred during human adipocyte differentiation, we examined the expression pattern of relevant genes during differentiation of a recently characterized human preadipocyte cell strain (43). This cell strain (termed SGBS) is derived from the stromal cell fraction of subcutaneous adipose tissue of an infant with Simpson-Golabi-Behmel syndrome. Although the biological mechanism behind the prolonged differentiation capacity is still unknown, SGBS adipocytes are morphologically, biochemically, and functionally identical to in vitro differentiated adipocytes from healthy subjects and are therefore an ideal human adipocyte in vitro model system. SGBS cells were grown confluent prior to exposure to adipogenic medium, and samples were withdrawn each day during the differentiation program (from day Ϫ1 until day 15) and used for Northern analysis. A low expression level of the PPAR␥ transcript was detected already at day Ϫ1, but a robust induction of the PPAR␥ transcript and the PPAR␥ target gene aFABP was observed later around day 3 to 4 ( Fig. 1B  and data not shown). The expression of LXR␣ was also induced at this time point, whereas the GLUT4 transcript was induced much later during the differentiation process. Weak GLUT4 mRNA signals could be detected on day 7, with a stronger induction around day 9. Interestingly, three different characterized LXR target genes, FAS (35), SREBP-1 (33,34), and LXR␣ itself (48) were also induced further at this later time point.
Regulation of GLUT4 mRNA by LXR Activation in Vitro-To test whether activation of LXRs affects the expression of GLUT4 in mature adipocytes, mouse 3T3-L1 cells were differentiated for 13 days and stimulated with the synthetic LXR ligand T0901317 (1 M) for 24 h, and samples were examined by Northern analysis. Exposure of differentiated 3T3-L1 cells to the LXR ligand induced the level of the GLUT4 mRNA transcript 2.3-fold (Fig. 1, C and D). To examine whether a similar regulation also applies to humans, differentiated SGBS cells were stimulated with T0901317 (0.1 M) for 24 h. A 1.6-fold induction of GLUT4 expression was observed.
Characterization of the GLUT4 Promoter-To identify potential LXR response elements in the GLUT4 promoter, a previously described part of the human GLUT4 promoter was used as bait to identify a longer form of the GLUT4 promoter by BLAST against nr and htgs sequences at NCBI (accession numbers used and identified are listed under "Experimental Procedures"). The promoter sequence spanning 10,000 bp upand downstream from the transcription start site was extracted and analyzed by a consensus LXRE (DGGTYA HWHW MGKKCA). By this method, several high score candidate LXREs were identified in the human promoter ( Fig. 2A; only the most promising element in the cloned promoter is shown). We next extracted and performed the same analysis of the mouse GLUT4 promoter. Finally, we aligned the human and mouse promoters and identified one of the promising LXREs as evolutionarily conserved (Fig. 2, A and B). Interestingly, this element is located within a region previously determined to be very important for expression of the GLUT4 gene in adipose, skeletal, and heart tissues (49).
Transfection with the Human GLUT4 Promoter-To examine the validity of the particular LXRE found in the human GLUT4 promoter, we cloned this promoter (sequence spanning from Ϫ2142 to ϩ223) and inserted it into the pGL3-basic reporter vector. Transient transfection of this luciferase reporter into COS-1 cells in combination with RXR␣ and LXR␣ expression vectors and treatment with the RXR ligand 9-cis-retinoic acid (9-cis-RA; 1 M) and the LXR ligand T0901317 (1 M), induced the reporter activity (Fig. 3A). A maximal 12-fold increase in reporter activity was observed following the addition of both receptor expression vectors and ligands. Furthermore, transfection experiments with increasing concentrations of T0901317 induced the reporter activity in a dose-dependent manner (Fig. 3B) with an estimated EC 50 value well in agreement with previous reports (27).
A deletion construct of the promoter lacking the LXRE (GLUT4-SmaI-del), and a full-length construct containing a mutated version of the LXRE (GLUT4-LXRE-mut) (Fig. 3C) were transfected into COS-1 cells in the presence of RXR␣ and LXR␣ expression vectors and the LXR ligand T0901317 (1 M) as indicated. In both the GLUT4-SmaI-del and the LXRE-mut deletion reporters, the ability of the RXR␣/LXR␣ heterodimer complex to induce transcription of the GLUT4 promoter was significantly reduced (Fig. 3D).
To test the functionality of the identified LXRE in a physiologically more relevant cell strain, SGBS cells were transiently transfected with full-length and LXRE-mut reporters (Fig. 3E) and stimulated with vehicle (Me 2 SO) or T0901317 (1 M). As expected, only the wild type reporter responded to the LXR␣ agonist (Fig. 3F).
The Human and Mouse LXREs in the GLUT4 Promoter Bind LXR␣-To demonstrate in vitro binding of LXR␣ to the identified LXRE, we performed an EMSA with double-stranded oligonucleotides containing the human and mouse LXREs (Fig.  4A, oligonucleotide sequences). A specific protein-DNA complex with the human (Fig. 4B) and mouse (Fig. 4C) GLUT4 LXREs was observed in the presence of in vitro translated RXR␣ and LXR␣ proteins. These complexes were specific, since the binding was eliminated by excess unlabeled wild type but not mutated oligonucleotides. Similarly, a strong specific binding was observed when nuclear extracts from COS-1 cells transfected with either RXR␣ or LXR␣ expression vectors were combined (Fig. 4D). A weaker band was observed when only nuclear extract from cells transfected with the LXR␣ expression vector was used, suggesting a complex formation between endogenously expressed RXR␣ and the exogenous LXR␣. Nuclear extracts from differentiated, but not from undifferentiated, 3T3-L1 cells formed a complex with the GLUT4 LXRE, with identical molecular weight to the heterodimer complex formed by in vitro translated RXR␣ and LXR␣ proteins ( Fig. 4E; only binding to the human element is shown). This suggests that the complex formed by endogenous proteins expressed in differentiated 3T3-L1 cells consists of a heterodimer complex formed by RXR␣ and LXR␣.
GLUT4 Expression Is Induced in Vivo by Ligand Activation of LXRs-To examine the role of LXR activation on the expression of GLUT4 in vivo, male mice (10 weeks; 25-30 g) were fed twice (24 and 8 h before they were sacrificed) with the synthetic PPAR␥ activator darglitazone (1 mg/kg) and/or the synthetic LXR agonist T0901317 (50 mg/kg). RNA isolated from white adipose tissue (WAT) and skeletal muscle was examined by Northern analysis.
In epididymal WAT, the expression of GLUT4 was induced ϳ3-fold by T0901317 treatment and 4-fold by combined T0901317 and darglitazone treatment (Fig. 5, A and B). Interestingly, no effect on GLUT4 mRNA was observed by administration of darglitazone alone to these healthy animals, although the well characterized PPAR␥ target gene aFABP was up-regulated (data not shown). This result differs from previous observations in vitro, where transcription of GLUT4 in adipocytes was induced by ligands that activated PPAR␥ (37). However, since PPAR␥ also regulates the expression of LXR␣ (18,42), it could be that this in vitro effect is secondarily mediated through increased LXR␣ expression and thus increased potential for activation through LXR␣.
In muscle, only a minor, but statistically significant, 1.5-and 1.6-fold induction of GLUT4 transcript was observed following darglitazone and T0901317 treatment, respectively (Fig. 5, C  and D).
Basal and Insulin Regulation of GLUT4 Expression Is Dependent on LXR␣-We recently demonstrated that insulin regulation of several key lipogenic enzymes in liver is dependent on LXRs (36). Since glucose transported through GLUT4 supplies the adipose cell with substrate for fatty acid synthesis, we next investigated whether GLUT4 is regulated by insulin and FIG. 4. The LXRE in the human and mouse GLUT4 promoters recruits the LXR␣/RXR␣ heterodimer. A, the nucleotide sequence for the different oligonucleotides used. Only the upper primer is shown. Each half-site in the DR4 element is indicated in boldface type, and the base substitution in the mutated oligonucleotides is given in the Ϫmut oligonucleotides. B, direct and specific binding of the RXR␣/LXR␣ heterodimer to the LXRE in the human GLUT4 promoter. The EMSA was performed with annealed and 32 P-labeled h-GLUT4-LXRE oligonucleotides and incubated in the presence of in vitro translated RXR␣ and/or LXR␣ proteins as indicated. C, direct and specific binding of the RXR␣/ LXR␣ heterodimer complex to the LXRE in the mouse GLUT4 promoter. D, nuclear extracts (0.5 g) isolated from COS-1 cells transfected with either RXR␣ or LXR␣ expression vectors form a strong specific complex with the GLUT4 LXRE only when these extracts are combined. E, nuclear extract (8 g) isolated from differentiated (day 13) but not from undifferentiated 3T3-L1 cells (day 0) forms a specific complex with the human GLUT4 LXRE with similar molecular weight to the complex formed by in vitro translated RXR␣ and LXR␣ proteins. For all experiments, the competition was performed using unlabeled oligonucleotides (LXREs and LXRE-mut oligonucleotides) as competitors in 5-, 10-, or 25-fold molar excess.
whether this regulation is dependent on LXRs. Wild type, LXR␣ Ϫ/Ϫ , LXR␤ Ϫ/Ϫ , and LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice were fed twice (24 and 8 h before they were sacrificed) with vehicle or T0901317 (30 mg/kg) and injected with PBS or insulin (0.2 units of Actrapid in PBS) 3 h before they were sacrificed. RNA isolated from epididymal WAT was examined by quantitative RT-PCR analysis.
In wild type mice, the expression of GLUT4 was unchanged by insulin injection alone (Fig. 6A). In contrast, the insulinresponsive transcription factor SREBP-1, was induced severalfold (data not shown), suggesting that GLUT4 is not normally transcriptionally regulated by insulin. Still, a synergistic induction of GLUT4 was observed with combined insulin and T0901317 treatment compared with T0901317 treatment alone. In both LXR␣ Ϫ/Ϫ and LXR␤ Ϫ/Ϫ mice, the expression of GLUT4 was induced by T0901317 treatment (Fig. 6, B and C), with no additional effect of insulin injections. As expected, T0901317 treatment had no effect on GLUT4 expression in LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (Fig. 6D), demonstrating that regulation by the LXR activator is dependent on the presence of at least one LXR isoform. The basal GLUT4 expression was slightly lower in LXR␣ Ϫ/Ϫ mice compared with the other animal groups, and the slightly increased GLUT4 expression after insulin treatment was clearly absent in the LXR␣ Ϫ/Ϫ mice compared with the other animal groups (Fig. 6E). This indicates that the LXR␣ isoform, but not the LXR␤ isoform, plays a unique role for basal and insulin-regulated expression of GLUT4 in epididymal WAT.
PPAR␥ Regulation of GLUT4 in WAT Is Only Visible in the Absence of LXRs-To test whether PPAR␥-mediated regulation of GLUT4 expression is dependent on LXRs, wild type and LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice were fed with darglitazone (1 mg/kg) or T0901317 (30 mg/kg) for 7 days, and subcutaneous WAT was examined by Northern analysis. In wild type mice, the GLUT4 expression was not altered by darglitazone treatment, whereas GLUT4 was induced 2-fold by darglitazone treatment in LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (Fig. 7A). A similar 2-fold stimulation with darglitazone treatment also occurred in epididymal WAT in LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (results not shown). In contrast, expression of the well characterized PPAR␥ target gene aFABP was increased ϳ2-fold by darglitazone treatment in both groups (Fig.  7B). The fact that PPAR␥ does not regulate GLUT4 expression in normal mice, although aFABP is regulated, argues against FIG. 5. GLUT4 is regulated in vivo by ligand activation of LXRs. Mice were gavage-fed with twice vehicle, darglitazone (Darg; 1 mg/kg), T0901317 (50 mg/kg), or a combination of darglitazone and T0901317, 24 and 8 h prior to being sacrificed. WAT (epididymal) and skeletal muscle (gastrocnemius) were rapidly taken out and subjected to Northern blot analysis. A, expression of GLUT4 mRNA in epididymal WAT. B, the expression of GLUT4 in epididymal WAT correlated against L27 as internal standard. Shown is relative expression of GLUT4 after treatment with darglitazone (n ϭ 5), T0901317 (n ϭ 3), or T0901317 and darglitazone (n ϭ 3) compared with controls (n ϭ 5). C, expression of GLUT4 mRNA in muscle. D, the expression of GLUT4 in muscle correlated against L27 as an internal standard. Shown is relative expression of GLUT4 after treatment with darglitazone (n ϭ 3) or T0901317 (n ϭ 3) compared with controls (n ϭ 3). The result is given as mean Ϯ S.E. p values were determined by one-way analysis of variance; *, p Ͻ 0.05; **, p Ͻ 0.01.

FIG. 6. Insulin and LXR activators synergistically regulate GLUT4 expression in vivo.
Mice were gavage-fed twice with either vehicle or T0901317 (30 mg/kg) prior to injection (intraperitoneally) of PBS or Actrapid (insulin; Ins) and killed 3 h after insulin/PBS injections. Epididymal WAT was rapidly taken out and used for quantitative RT-PCR analysis. The expression of GLUT4 was normalized against 36B4 to determine the relative GLUT4 expression. A, GLUT4 expression in wild-type controls; B, GLUT4 expression in LXR␣ Ϫ/Ϫ ␤ ϩ/ϩ mice; C, GLUT4 expression in LXR␣ ϩ/ϩ ␤ Ϫ/Ϫ mice; D, GLUT4 expression in LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice. Bars represent 2-⌬Ct values (see "Experimental Procedures") Ϯ S.E. (n ϭ 4 -6). p values were determined by one-way analysis of variance (p Ͻ 0.05). Expression different from control is marked with an asterisk. Expression of GLUT4 after combined insulin and T0901317 treatment of wild type mice was significantly higher compared with identical treatment of the three other groups (a; p Ͻ 0.05). E, basal GLUT4 expression and expression after insulin stimulation was lower in LXR␣ Ϫ/Ϫ ␤ ϩ/ϩ mice compared with the other groups (b; p Ͻ 0.01).
GLUT4 being a direct PPAR␥ target gene. Still, the PPAR␥ activator is able to regulate GLUT4 expression in the absence of LXRs, suggesting that ablation of the LXRs results in an altered phenotype with disruption of a mechanism that normally prevents PPAR␥ activation of the GLUT4 promoter in mice with normal insulin levels. Therefore, the LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mouse is not a suitable tool to determine whether the PPAR␥-LXR␣ pathway is important for the PPAR␥ ligand-mediated regulation of adipose GLUT4 expression.
Treatment with the LXR activator T0901317 for 7 days had no effect on subcutaneous GLUT4 expression in wild type mice (Fig. 6A). The induction of GLUT4 expression was also much lower in epididymal WAT after prolonged treatment compared with a shorter 24-h feeding period (Figs. 5 and 6 and data not shown). The LXR target gene SREBP-1 was only induced 2-fold in wild type mice, and, as expected, no regulation was observed in the LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (Fig. 7C). Thus, the initial induction of GLUT4 expression after 24-h activation of LXRs seems to be transient and is no longer observed after prolonged treatment with a potent LXR activator. DISCUSSION Both the reduced amount of GLUT4 in adipose tissue and reduced insulin-mediated recruitment of GLUT4 from intracellular vesicles to the plasma membrane are strongly associated with obesity and type 2 diabetes (4,50). Accordingly, modulation of pathways involved in this dysregulation may improve insulin sensitivity and reduce the hyperglycemia in type 2 diabetes. In this study, we demonstrate that GLUT4 expression in adipose tissue, and to a lesser extent also in skeletal muscle, is increased by ligand activation of LXRs. We furthermore demonstrate that this regulation is directly mediated by binding of an RXR/LXR heterodimer complex to a conserved DR-4 type LXRE in the GLUT4 promoter. Despite the fact that both LXRs are able to mediate ligand-activated transcription of the GLUT4 gene in vivo, several observations suggest that LXR␣ plays a unique role in the basal regulation of adipose GLUT4 expression. Adipose-specific activation of LXR␣ might therefore represent a novel tool for increasing insulin-stimulated glucose uptake.
We have recently demonstrated that activation of LXRs increases TG accumulation in adipocytes (18), presumably by direct regulation of lipogenic genes as SREBP-1c (33,34) and FAS (35). That fits with the finding that LXRs also regulate GLUT4, since increased glucose uptake through GLUT4 increases the substrate availability for TG synthesis. We found that the GLUT4 transcript was absent prior to expression of LXR␣ during both mouse and human adipocyte differentiation, suggesting that LXR␣ expression is important for the expres-sion of GLUT4 (Fig. 1). Still, adipose expression of GLUT4 is not dependent on LXR␣, since GLUT4 is expressed also in LXR␣ Ϫ/Ϫ ␤ ϩ/ϩ and LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (Figs. 6 and 7). Rather, it seems that LXR␣ functions to increase the basal transcription level of the GLUT4 gene. This transcription is mediated through a typical DR-4 type LXRE, with similar nucleotide sequence and distance from the transcription start site as those found in the FAS and SREBP-1 promoters (33)(34)(35). This and the fact that the GLUT4 LXRE is located within a region that is highly conserved between human and mouse GLUT4 promoters (Fig. 2) suggest that regulation through this element is important. Transfection experiments in COS-1 cells and human SGBS adipocytes indeed demonstrate that this element is functional and necessary for LXRs to be able to activate the GLUT4 promoter (Fig. 3). The GLUT4 LXRE also specifically recruits the RXR/LXR heterodimer, as determined by in vitro binding experiments (Fig. 4). The induced GLUT4 expression observed after stimulation with a potent LXR activator, both in vitro in mouse and human adipocytes (Fig. 1) and in whole animals (Figs. 5 and 7), is therefore likely to be mediated through this particular LXRE. Thus, these analyses establish GLUT4 as a novel direct LXR target gene.
We recently demonstrated that insulin regulation of several important lipogenic enzymes in liver is dependent on LXRs, since insulin regulation of several of these genes was impaired in the LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice (36). The experiment performed here does not indicate that insulin stimulation alone regulates GLUT4 expression at the transcriptional level in normal mice (Fig. 7A). However, after activation of the LXRs with a potent LXR activator, GLUT4 expression was synergistically induced by insulin. This regulation was abolished in both single knockout mice, but in the absence of both LXRs, GLUT4 transcription was unexpectedly also increased by insulin alone (Fig. 7D). Since the LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice, in contrast to wild type mice, also responded to the insulin-sensitizing PPAR␥ activator after 1 week of treatment (Fig. 7), the insulin sensitivity of the adipose tissue is apparently altered in the LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice. It is therefore difficult to evaluate whether GLUT4 expression is stimulated by insulin. On the other hand, the different effects obtained by removal of the LXRs, suggest a complex regulation of GLUT4 expression, where factors such as insulin, LXRs, and PPAR␥ seem to be able to compensate for each other.
Despite the fact that both LXR isoforms generally are believed to compensate for each other and activate transcription of the same set of LXR target genes, the results from the use of single knock-out mice in this study suggest that there are distinct differences between the LXR␣ and LXR␤ isoforms regarding GLUT4 expression. LXR␤ apparently binds to the FIG. 7. Prolonged treatment with a PPAR␥ activator induces adipose expression of GLUT4 in mice lacking LXRs. Wild-type controls (LXR␣ ϩ/ϩ ␤ ϩ/ϩ ) and LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice were gavage-fed with vehicle, darglitazone (Darg; 1 mg/kg), or T0901317 (30 mg/kg) for 7 days and killed 8 h after the last administration of ligands. Subcutaneous WAT was subjected to Northern blot analysis, and equal amounts of total RNA from each animal (n ϭ 4 -6) were pooled and applied as two individual samples on the Northern gel. The expression of GLUT4 was normalized against the 36B4 signal. A, relative expression of GLUT4. B, relative expression of aFABP. C, relative expression of SREBP-1.
GLUT4 promoter and modulates transcription of the GLUT4 gene in the presence of a potent LXR activator (Fig. 6). However, LXR␤ is not able to rescue the lower basal expression of GLUT4 seen in epididymal WAT in the absence of LXR␣. During adipocyte differentiation, neither GLUT4 nor the LXR target genes SREBP-1 and FAS are induced before the expression of LXR␣ is turned on, whereas LXR␤ is expressed also in undifferentiated fibroblast like cells ( Fig. 1 and Ref. 18). Furthermore, only nuclear extracts from differentiated 3T3-L1 cells, where both RXR␣ and LXR␣ are expressed, are able to form a complex with the GLUT4 LXRE. Similar specific requirement for the LXR␣ isoform has also been demonstrated for induction of the rodent cholesterol-7␣ hydroxylase gene in liver (32,45). However, since LXR␤ fails to activate the LXRE in the murine cholesterol-7␣ hydroxylase promoter (26), the selective effects of the LXR isoforms on the cholesterol-7␣ hydroxylase promoter are different from what we observe for the basal regulation of the GLUT4 promoter in epididymal WAT.
After 1 week of LXR ligand feeding to wild type mice, GLUT4 expression in either epididymal or subcutaneous WAT was not different from control feed animals ( Fig. 7 and results not shown). This suggests that a mechanism exists that prevents prolonged induction of GLUT4 through LXR activation. A similar regulation has also recently been demonstrated for lipogenic genes as FAS and SREBP-1 in liver, which decline to almost normal expression levels after prolonged treatment with LXR activators (35). Interestingly, the expression of LXR␣ and SREBP-1 is similarly regulated during prolonged treatment with the LXR activator in adipose tissue, 2 directly linking the expression level of LXR␣ to the induction level of GLUT4 and SREBP-1 in adipose tissue. Along with our observation of a reduced basal expression of GLUT4 in LXR␣ Ϫ/Ϫ mice, but not in LXR␣ Ϫ/Ϫ ␤ Ϫ/Ϫ mice, LXR␤ might prevent prolonged LXR␣mediated activation, perhaps by binding to and occupying the LXRE elements in the promoter region of LXR␣ target genes.
The expression of GLUT4 is reduced in rodent models of insulin deficiency (4,5) and in adipose tissue of human obese or type 2 diabetic subjects (6, 7), directly linking adipose expression of GLUT4 to insulin resistance. We have previously demonstrated that LXR␣ also is down-regulated under such conditions (18), thus revealing a striking correlation between GLUT4 and LXR␣ expression in insulin-resistant conditions. Treatment with anti-diabetic TZDs, which are thought to mediate their effect as high affinity ligands for PPAR␥ (17), furthermore normalizes the reduced adipose expression of both GLUT4 (38,39) and LXR␣ (18). Since LXR␣ is a downstream target gene for PPAR␥ (18,47), the beneficial normalization of GLUT4 expression by TZD treatment might therefore actually be mediated through increased expression and activation of LXR␣. Unfortunately, the altered phenotype obtained by the removal of LXRs makes it difficult to prove this experimentally with the knock-out mice. However, we found no regulation of GLUT4 in subcutaneous WAT after 1 week of treatment with darglitazone, a potent TZD, although the well characterized PPAR␥ target gene, aFABP, was induced (Fig. 6). This argues against the notion that GLUT4 is a typical direct PPAR␥ target gene and rather suggests that the TZDs modulates GLUT4 expression through a secondary mechanism. For the moment it is unclear whether this regulation mechanism is mediated exclusively through LXRs or whether PPAR␥ also regulates GLUT4 expression through another uncharacterized mechanism.
In view of these considerations, LXR␣ is an interesting tar-get for a new generation of antidiabetic drugs. Unfortunately, administration of currently available LXR activators increases plasma and liver TG levels in normal as well as diabetic animals, with complications such as hypertriglyceridemia and liver steatosis (27,41). Nevertheless, our analysis suggests that specific activation of LXR␣ selectively in adipose tissue could be a complement or replacement to currently available treatment modalities for reduction of hyperinsulinemia and hyperglycemia in type 2 diabetic subjects.