1. Genomics Institute of the Novartis Research Foundation, 10675 John Hopkins Drive, San Diego, California 92121, USA
2. The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, USA

Correspondence to: Enrique Saez^1,2 Correspondence and requests for materials should be addressed to E.S. (Email: esaez@scripps.edu).

Nuclear receptors are ligand-activated transcription factors that coordinate gene expression in response to hormonal and environmental signals. Members of the superfamily that work as heterodimers with the retinoid X receptor (RXR) serve as sensors of dietary components, orchestrating the physiological response to nutrients⁵. LXR- and LXR- (also called NR1H3 and NR1H2, respectively) are RXR partners that recognize oxidized cholesterol and control gene expression linked to cholesterol and fatty acid metabolism^{6, 7, 8}. Activation of LXRs results in decreased atherosclerosis in rodents⁹. LXR ligands have anti-diabetic effects as well, decreasing liver glucose output and increasing peripheral glucose disposal^{10, 11}.

Treatment of rodents with synthetic LXR ligands results in decreased hepatic gluconeogenesis and increased lipogenesis, indicating that LXR serves as a transcription factor that integrates liver carbohydrate and lipid metabolism. Because the rodent diet is virtually devoid of cholesterol, to explore the dietary influences that modulate hepatic LXR function we asked whether LXR activity could be regulated by glucose. Human HepG2 cells were transfected with a Gal4-responsive luciferase reporter and chimaeric constructs of the ligand-binding domain (LBD) of the LXRs fused to the DNA-binding domain (DBD) of yeast Gal4. Transfected cells were maintained in media with various concentrations of glucose and subsequently exposed to increasing amounts of glucose and glycolysis derivatives. Notably, d-glucose elicited a robust 50-fold induction of LXR- and LXR- transactivation activity when cells were grown in no glucose media (Fig. 1a). The ability of the LXRs to respond to glucose and its derivatives is very specific: no effect was seen in other nuclear receptors tested (Fig. 1d).

Figure 1: Glucose induces LXR transcriptional activity.

a, Activation of LXR- and LXR- by glucose and glycolysis intermediates. HepG2 cells transfected with a Gal4–responsive luciferase reporter and expression plasmids encoding Gal4 DBD–LXR-/ LBD chimaeric proteins were plated in no glucose media for 24 h and treated as indicated. d-Fructose-1,6, d-fructose-1,6-bisphosphate; d-fructose-6-P, d-fructose-6-phosphate; d-glucose-6-P, d-glucose-6-phosphate; dl-Gly-3-P, dl-glyceraldehyde-3-phosphate; 22-(R)-HC, 22-(R)-hydroxycholesterol; 22-(S)-HC, 22-(S)-hydroxycholesterol; PEP, phosphoenolpyruvate; d-(-)-3-PG, d-(-)-3-phosphoglycerate. b, Activation profile of natural LXR–RXR heterodimers with glycolysis intermediates (20 mM top dose). Cells were transfected with a 2 LXRE ABCA1-luciferase reporter and expression plasmids for RXR-, and LXR- or LXR-. No effect was seen with pGL3 or TK-luc reporters lacking LXREs. c, Dose response curves on LXR–RXR and LXR–RXR of glucose and glycolytic compounds. Efficacy is relative to GW3965. d, Activation by glucose is LXR specific. HepG2 cells transfected with Gal4 chimaeric constructs for various nuclear receptors were treated with 20 mM glucose for 24 h. e, Activation by glucose requires the LXR LBD. Normalized luciferase values are expressed as fold induction versus DMSO; all error bars indicate s.d. Data are representative of four experiments performed in duplicate.

High resolution image and legend (159K)

To examine whether glucose can regulate the LXR–RXR heterodimers, HepG2 cells were transfected with expression vectors for the LXRs, RXR and a reporter under the control of two copies of the LXR response element (LXRE) of the ABCA1 gene. Glucose and its derivatives stimulated significant LXR–RXR activity in cells grown in no glucose (Fig. 1b). d-Glucose and d-glucose-6-phosphate are more potent on LXR- than LXR- (effector concentration for half-maximum response (EC₅₀) of 308 M versus 3,141 M), and are weaker inducers than known LXR ligands (Fig. 1c and Supplementary Table 1). Activation of the LXR–RXR heterodimer by glucose was dependent on the LXR LBD (Fig. 1e). Glucose concentration did not affect expression of transfected LXR protein (Supplementary Fig. 2).

Glucose and its derivatives might modulate LXR activity directly, by binding the LXR LBD, or indirectly, via the generation of an endogenous ligand or a post-transcriptional modification. To distinguish these possibilities, we tested the ability of d-glucose and glycolytic intermediates to bind LXR in a cell-free coactivator recruitment assay. In this setting, interaction between the LXR LBD and a peptide from the coactivator SRC-1, measured by fluorescence resonance energy transfer (FRET), provides an indication of ligand binding and receptor activation. Unexpectedly, d-glucose and d-glucose-6-phosphate were both able to induce coactivator recruitment, revealing that they are direct LXR agonists (Fig. 2a). d-Glucose activated the LBD of both LXRs with the same efficacy as 22-(R)-hydroxycholesterol, GW3965 and T0901317, the known LXR ligands. d-Glucose was more potent on LXR- than on LXR- (EC₅₀ of 318 M versus 2,904 M; Supplementary Fig. 3 and Supplementary Table 2). A survey of more than 40 monosaccharides using the FRET assay outlined the specificity of the effect: only d-glucose, l-glucose and d-glucose-6-phosphate brought about coactivator recruitment (Fig. 2b and data not shown).

Figure 2: Glucose is a direct LXR agonist.

a, d-Glucose and glucose-6-phosphate induce SRC-1 recruitment to the LXR- and LXR- LBDs in cell-free FRET assays. Glycolytic intermediates were tested at 20 mM. b, Structure–activity relationship analysis of various monosaccharides (20 mM) on the FRET assay. Values are presented as fold induction versus vehicle (increase in coactivator recruitment measured as a change in 665/615 nm emission relative to vehicle). All error bars indicate s.d.; abbreviations as in Fig. 1. Data are representative of four experiments performed in triplicate.

High resolution image and legend (177K)

To confirm that d-glucose and d-glucose-6-phosphate bind LXR directly, a scintillation proximity assay (SPA) was used to evaluate their ability to displace a potent, labelled synthetic ligand bound to the LBD of the LXRs ([³H]T0901317; dissociation constant (K_d) of 7 nM for LXR- and 22 nM for LXR-). d-Glucose and d-glucose-6-phosphate competed effectively for binding, displacing the labelled synthetic agonist with equivalent efficacy to known ligands (d-glucose inhibition constant (K_i) of 1,801 M for LXR- and 202 M for LXR-; Fig. 3a and Supplementary Table 3). To demonstrate further that d-glucose has the ability to bind directly the LXRs, the SPA assay was run using [³H]d-glucose as the ligand. Labelled glucose bound the LBD of the LXRs and induced scintillation with a K_d of 2,489 M for LXR- and 211 M for LXR- (Fig. 3b). Cold d-glucose and d-glucose-6-phosphate competed for binding and completely displaced [³H]d-glucose from the LBD of the LXRs. Interestingly, the known ligands were only partially able to displace labelled glucose (Fig. 3b), suggesting that glucose can bind LXR in more than one site (see scatchard analysis, inset Fig. 3b), and that glucose and known ligands can bind LXR simultaneously. To verify this finding, an SPA assay was run with a combination of [³H]T0901317 and [³H]d-glucose. The presence of labelled glucose increased scintillation well beyond the level elicited by a maximal dose of labelled T0901317, indicating that these molecules bind LXR concurrently (Fig. 3c). To explore the functional consequences of glucose binding to LXR in the presence of a known ligand, the FRET assay was run with a combination of T0901317 and glucose. Addition of glucose to a maximal dose of a known ligand resulted in increased coactivator recruitment (Fig. 3d). In contrast, addition of GW3965 to a saturating dose of T0901317 (10 M) did not enhance coactivator recruitment. Glucose was also able to protect LXR- from protease digestion (Supplementary Fig. 4), and to increase the melting temperature (T_m) of LXR- in differential scanning calorimetry (Supplementary Fig. 5). These observations in five different biochemical assays demonstrate that d-glucose and d-glucose-6-phosphate are direct agonists of the LXRs that bind more than one site, and can work in combination with a known ligand. The precise binding mode of glucose awaits the resolution of a crystal structure.

Figure 3: Glucose displaces a labelled high-affinity LXR ligand.

a, d-Glucose and glucose-6-phosphate compete for LXR binding and displace [³H]T0901317 (25 nM) in an SPA assay. b, [³H]Glucose binds LXR. A scatchard analysis is shown in the inset. Labelled glucose did not bind the RXR LBD (data not shown). Unlabelled LXR ligands displace bound [³H]glucose (20 mM) but not completely. Values are expressed as percentage binding of labelled compound. Fractional occupancy of the receptor (see Methods) is 95% for LXR- and 98% for LXR-. c.p.m., counts per minute. c, Addition of labelled glucose to a saturating dose of [³H]T0901317 (10 M) increases scintillation in an SPA assay; percentage efficacy is relative to 10 M [³H]T091317. d, Addition of glucose, but not GW3965, to a maximal dose of T0901317 enhances coactivator recruitment. Note the different scales. Values are presented as fold induction versus vehicle (increase in coactivator recruitment measured as a change in 665/615 nm emission relative to vehicle). All error bars indicate s.d.; experiments performed in triplicate.

High resolution image and legend (128K)

To evaluate the physiological relevance of glucose as an LXR agonist, we compared the pattern of gene expression induced in HepG2 cells by d-glucose and GW3965. In cells grown in the absence of glucose or in low glucose conditions, overnight treatment with either compound (1 M GW3965 or 20 mM d-glucose) stimulated expression of genes involved in fatty acid synthesis and repressed expression of gluconeogenic genes (Fig. 4a). Glucose and GW3965 inhibited these genes under conditions of maximal induction (cAMP, dexamethasone) to a degree comparable to that seen with insulin, which seems to work with GW3965 and d-glucose to suppress expression of gluconeogenic genes (Supplementary Fig. 6). Interestingly, d-glucose was also able to significantly induce expression of LXR target genes involved in cholesterol homeostasis (ABCA1, ABCG5, ABCG8, ABCG1, CETP) that are not insulin regulated and whose expression was heretofore not associated with glucose levels (Fig. 4a). The efficacy of known LXR ligands was potentiated with increasing glucose concentration, indicating that glucose can work together with established LXR ligands. Expression of multiple control genes was not positively affected by glucose concentration, showing that health of the cells was not affected by glucose concentration in the experimental time course used. To verify that d-glucose-mediated stimulation of LXR target genes requires LXR, the experiment was repeated in HepG2 cells transfected with a short-interfering (si)RNA pool against human LXR-, the dominant subtype in these cells. Induction of LXR-dependent target genes by d-glucose and other ligands was blocked in cells transfected with siRNA against LXR-, but not in unperturbed cells or cells transfected with control siRNA (Supplementary Fig. 7). d-Glucose failed to upregulate LXR target genes when cells were grown in the presence of the glucose uptake inhibitor cytochalasin B, whereas GW3965 remained active, but to a diminished degree (Supplementary Fig. 8). These results establish that d-glucose can activate endogenous LXRs and regulate genuine LXR targets in liver cells, and that it may do so together with other LXR ligands.

Figure 4: Glucose regulates direct LXR target genes in vivo.

a, HepG2 cells cultured in 0 mM (white bars), 2 mM (grey bars) or 25 mM (black bars) glucose medium were treated overnight with GW3965 (1 M), 22-(R)-hydroxycholesterol (5 M), or d-glucose (20 mM) and gene expression was analysed using qRT–PCR. Glucose stimulates expression of direct LXR cholesterol homeostasis target genes. Note that efficacy of known LXR ligands increases with increasing glucose concentration. b, Glucose induces LXR target genes in mouse liver. Mice fasted overnight were challenged orally with GW3965 (50 mg kg^-1), or re-fed with a glucose or sucrose diet and killed 6 h later. d-Glucose and GW3965 regulate the same direct LXR targets (genes involved in cholesterol and fatty acid metabolism) as well as indirect carbohydrate metabolism targets. All error bars represent s.d., n = 5–6 mice per group. Asterisk, P < 0.05; double asterisk, P < 0.001 treatment versus fasted.

High resolution image and legend (217K)

To extend these observations to animals, fasted mice were challenged with GW3965 or diets where the source of carbohydrate was exclusively sucrose or d-glucose. All diets were devoid of cholesterol to minimize endogenous generation of oxysterols. d-Glucose and GW3965 induced similar changes in hepatic gene expression, triggering a pattern expected to limit hepatic glucose output and increase fatty acid synthesis (Fig. 4b). d-Glucose also strongly stimulated expression of genes that require LXR for induction (for example, Abca1, Abcg1, Sp). Induction of insulin-independent LXR target genes was also observed with sucrose, the glucose:fructose disaccharide present in standard chow. These findings indicate that LXR functions as a glucose sensor in vivo that responds to increasing liver glucose uptake. To examine the effect of insulin on glucose-stimulated, LXR-related hepatic gene expression, animals rendered insulin-deficient via streptozotocin (STZ) injection were treated in an identical manner. In the absence of an increase in insulin, glucose and GW3965 were still able to induce expression of LXR-dependent genes, repress gluconeogenesis genes, and upregulate fatty acid synthesis genes (Supplementary Fig. 9). Moreover, glucose was also able to induce upregulation of bona fide LXR target genes in the intestine of wild-type and STZ-treated mice, confirming the role of glucose as a physiological LXR ligand in another tissue that faces significant glucose influx and in which the role of insulin is not as prominent (Supplementary Fig. 10 and data not shown).

The concentration of glucose required to activate LXR is within physiological range, as blood glucose concentration in healthy individuals can vary from 3 to 9 mM during the day¹. In tissues that express the high-capacity glucose transporter GLUT2 (liver, pancreas, kidney; Michaelis constant (K_m) of 20–40 mM), glucose enters the cell readily and can diffuse freely to the nucleus, for the nuclear pore displays passive permeability for molecules <500 Da¹². Direct measurement of nuclear glucose concentration using live imaging of FRET-based glucose nanosensors in kidney cells indicates that it is approximately 50% that of the external medium¹³.

Insulin promotes hepatic lipogenesis through induction of SREBP-1c^{14, 15}. Deletion of this SREBP isoform results in impaired fatty acid and triglyceride synthesis¹⁶. SREBP-1c expression is abolished in LXR-/ null animals, showing that the LXRs are primary transcriptional regulators of this gene¹⁷. Recent work has shown that insulin-mediated activation of SREBP-1c expression requires LXR, leading to the hypothesis that insulin may regulate SREBP-1c expression through the production of an LXR ligand¹⁸. We would like to suggest that glucose itself is the LXR ligand involved in this response (Supplementary Fig. 1). LXR could also influence hepatic glucose fate via its ability to induce expression of ChREBP (Fig. 4a, b), although activity of this transcription factor is regulated at the post-transcriptional level. Deletion of ChREBP reduces but does not ablate lipogenesis¹⁹, indicating that concerted action of SREBP-1c and ChREBP is necessary for normal fatty acid and triglyceride synthesis in vivo.

The integration of glucose sensing and control of lipogenesis in the same protein may provide an explanation for the observation that low-fat, high-carbohydrate diets induce hypertriglyceridaemia²⁰: LXR can sense surplus glucose, induce fatty acid synthesis, and prompt hepatic export of very low density lipoprotein (VLDL)²¹. Because LXR acts as the body's sensor of pathogenic cholesterol build-up, its ability to bind both glucose and oxysterols also suggests that LXR may be the link between hyperglycaemia and atherosclerosis. However, it remains to be established whether enough intracellular d-glucose or d-glucose-6-phosphate is present to activate LXR in tissues other than liver and intestine.

References

1. Zierler, K. Whole body glucose metabolism. Am. J. Physiol. 276, E409–E426 (1999) | PubMed | ChemPort |
2. Moore, M. C., Cherrington, A. D. & Wasserman, D. H. Regulation of hepatic and peripheral glucose disposal. Best Pract. Res. Clin. Endocrinol. Metab. 17, 343–364 (2003) | Article | PubMed | ISI | ChemPort |
3. Cherrington, A. D. The role of hepatic insulin receptors in the regulation of glucose production. J. Clin. Invest. 115, 1136–1139 (2005) | Article | PubMed | ChemPort |
4. Yamashita, H. et al. A glucose-responsive transcription factor that regulates carbohydrate metabolism in the liver. Proc. Natl Acad. Sci. USA 98, 9116–9121 (2001) | Article | PubMed | ChemPort |
5. Shulman, A. I. & Mangelsdorf, D. J. Retinoid x receptor heterodimers in the metabolic syndrome. N. Engl. J. Med. 353, 604–615 (2005) | Article | PubMed | ChemPort |
6. Schultz, J. R. et al. Role of LXRs in control of lipogenesis. Genes Dev. 14, 2831–2838 (2000) | Article | PubMed | ISI | ChemPort |
7. Peet, D. J. et al. Cholesterol and bile acid metabolism are impaired in mice lacking the nuclear oxysterol receptor LXR. Cell 93, 693–704 (1998) | Article | PubMed | ISI | ChemPort |
8. Janowski, B. A., Willy, P. J., Devi, T. R., Falck, J. R. & Mangelsdorf, D. J. An oxysterol signalling pathway mediated by the nuclear receptor LXR. Nature 383, 728–731 (1996) | Article | PubMed | ISI | ChemPort |
9. Joseph, S. B. et al. Synthetic LXR ligand inhibits the development of atherosclerosis in mice. Proc. Natl Acad. Sci. USA 99, 7604–7609 (2002) | Article | PubMed | ChemPort |
10. Laffitte, B. A. et al. Activation of liver X receptor improves glucose tolerance through coordinate regulation of glucose metabolism in liver and adipose tissue. Proc. Natl Acad. Sci. USA 100, 5419–5424 (2003) | Article | PubMed | ChemPort |
11. Cao, G. et al. Antidiabetic action of a liver x receptor agonist mediated by inhibition of hepatic gluconeogenesis. J. Biol. Chem. 278, 1131–1136 (2003) | Article | PubMed | ChemPort |
12. Stehno-Bittel, L., Perez-Terzic, C. & Clapham, D. E. Diffusion across the nuclear envelope inhibited by depletion of the nuclear Ca²⁺ store. Science 270, 1835–1838 (1995) | Article | PubMed | ChemPort |
13. Fehr, M., Lalonde, S., Ehrhardt, D. W. & Frommer, W. B. Live imaging of glucose homeostasis in nuclei of COS-7 cells. J. Fluoresc. 14, 603–609 (2004) | Article | PubMed | ChemPort |
14. Shimomura, I. et al. Insulin selectively increases SREBP-1c mRNA in the livers of rats with streptozotocin-induced diabetes. Proc. Natl Acad. Sci. USA 96, 13656–13661 (1999) | Article | PubMed | ChemPort |
15. Eberle, D., Hegarty, B., Bossard, P., Ferre, P. & Foufelle, F. SREBP transcription factors: master regulators of lipid homeostasis. Biochimie 86, 839–848 (2004) | Article | PubMed | ISI | ChemPort |
16. Liang, G. et al. Diminished hepatic response to fasting/refeeding and liver X receptor agonists in mice with selective deficiency of sterol regulatory element-binding protein-1c. J. Biol. Chem. 277, 9520–9528 (2002) | Article | PubMed | ISI | ChemPort |
17. Repa, J. J. et al. Regulation of mouse sterol regulatory element-binding protein-1c gene (SREBP-1c) by oxysterol receptors, LXR and LXR. Genes Dev. 14, 2819–2830 (2000) | Article | PubMed | ISI | ChemPort |
18. Chen, G., Liang, G., Ou, J., Goldstein, J. L. & Brown, M. S. Central role for liver X receptor in insulin-mediated activation of Srebp-1c transcription and stimulation of fatty acid synthesis in liver. Proc. Natl Acad. Sci. USA 101, 11245–11250 (2004) | Article | PubMed | ChemPort |
19. Iizuka, K., Bruick, R. K., Liang, G., Horton, J. D. & Uyeda, K. Deficiency of carbohydrate response element-binding protein (ChREBP) reduces lipogenesis as well as glycolysis. Proc. Natl Acad. Sci. USA 101, 7281–7286 (2004) | Article | PubMed | ChemPort |
20. Parks, E. J. Changes in fat synthesis influenced by dietary macronutrient content. Proc. Nutr. Soc. 61, 281–286 (2002) | Article | PubMed | ChemPort |
21. Lin, J. et al. Hyperlipidemic effects of dietary saturated fats mediated through PGC-1 coactivation of SREBP. Cell 120, 261–273 (2005) | Article | PubMed | ISI | ChemPort |
22. Joseph, S. B. et al. LXR-dependent gene expression is important for macrophage survival and the innate immune response. Cell 119, 299–309 (2004) | Article | PubMed | ISI | ChemPort |
23. Janowski, B. A. et al. Structural requirements of ligands for the oxysterol liver X receptors LXR and LXR. Proc. Natl Acad. Sci. USA 96, 266–271 (1999) | Article | PubMed | ChemPort |
24. DeBlasi, A., O'Reilly, K. & Motulsky, H. J. Calculating receptor number from binding experiments using same compound as radioligand and competitor. Trends Pharmacol. Sci. 10, 227–229 (1989) | Article | PubMed | ChemPort |

Supplementary Information

Supplementary information accompanies this paper.

Acknowledgements

We thank S. Bohan, R. Romeo, B. Geierstanger, M. Chalmers, P. Griffin, N. Gekakis, M. Crestani, H. Shimano, T. Matsuzaka, P. Tontonoz, B. Laffitte, S. Joseph, A. Brock, E. Peters and K. Nettles for technical help and/or useful comments. C.G. was a visiting scientist supported by a fellowship from the Department of Pharmacological Sciences, University of Milano, Italy.

Author Contributions L.V., C.G., E.H. and A.K. performed experiments; P.A.M. and V.M. designed and performed experiments and analysed data; and N.M. and E.S. designed and performed experiments, analysed data, and wrote the manuscript.

Competing interests statement

The authors declare no competing financial interests.

MORE ARTICLES LIKE THIS

These links to content published by NPG are automatically generated.