Glucose is a ligand for the oxysterol receptor LXR Supplementary Figure 1 Schematic representation of pathways influencing glucose fate in the liver. Glucose induces insulin secretion, suppressing hepatic gluconeogenesis and through LXR activating SREBP-1c expression and lipogenesis. Glucose can also bind directly to LXR to induce SREBP-1c expression, suppress hepatic glucose output, and increase ChREBP expression. ChREBP activitity is modulated by glucose metabolites, further increasing lipogenesis. For clarity, glycogen metabolism is not included in the diagram. Glucose mm : 0 2 25 HepG2 LXRα β actin 293T LXRα β actin Supplementary Figure 2 Glucose concentration does not influence the levels of LXR protein expressed in transfected cells. HepG2 or 293T cells were transfected with Flag-tagged hlxrα and grown in media with the indicated glucose concentration. Extracts were prepared 48 h post-transfection and analyzed by Western blot. www.nature.com/nature 1
Supplementary Figure 3 Coactivator recruitment dose response curves for D-glucose and LXR agonists. Efficacy is relative to GW3965. Error bars represent s.d. vehicle GW3965 10 µm D-glucose 30 mm PK µg/ml: 0 2 5 0 2 5 0 2 5 38 Kd 28 Kd 17 Kd 14 Kd 7 Kd Supplementary Figure 4 Glucose shields LXRα from protease digestion. Purified histidinetagged hlxrα LBD was incubated with the indicated concentration of proteinase K and either vehicle or ligands for 15 minutes at room temperature. Digestion products were analyzed by SDS PAGE and stained with Comassie blue. Arrows point to the most prominent proteolytic fragments protected by the presence of GW3965 or D-glucose. www.nature.com/nature 2
Supplementary Figure 5 Glucose shifts the melting temperature of LXRβ. Purified hlxrβ LBD alone or incubated with either D-glucose or GW3965 at the indicated concentration was analyzed by differential scanning calorimetry (DSC). This technique is used to determine thermal transition temperatures for samples in solution; ligand binding generally stabilizes the protein and the T m of the protein-ligand complex is higher than that of the protein in the absence of ligand. Both glucose and GW3965 increased the melting point of the protein, indicating that they are binding directly to LXRβ. DSC was performed on a Microcal VP-DSC differential scanning calorimeter, in a temperature range that covered 20-80 C, at a scanning rate equal to 90 C/h. 400 µg of hlxrβ LBD in a total cell volume of 0.4 ml were used in each thermal scan. Subsequent scans containing D-glucose and GW3965 were separated with buffer only samples to avoid cross-contamination within the sample cell. In experiments with added ligand, the reference cell contained an equal concentration of ligand to confirm the absence of any significant contribution to the T m shift. Results were analyzed with Origin 7.0 software. Thermograms display excess heat capacity vs. temperature and melting temperatures were calculated from an intrinsic software feature. Proteins used for protease protection and differential scanning calorimetry were made in-house due to the large amount of protein that these assays require. Briefly, the ligand binding domain (203-461) of human LXRβ was cloned into a vector for expression in baculovirus which contains www.nature.com/nature 3
an N-terminal hexa-his tag with a TEV protease cleavage site. Lysis of insect cells was achieved by sonication. After removal of cell debris by centrifugation, the resulting lysate containing soluble LXRβ LBD was purified by Ni-NTA chromatography followed S200 gel filtration chromatography. LXRβ-containing fractions were pooled and the protein was concentrated to ~10 mg/ml. The purity of the protein was evaluated to be higher than 95% based on SDS PAGE. Human LXRα(183-447) was cloned into the same vector and processed identically. Identity of purified proteins was checked by standard LC MS/MS methods after trypsin digestion, and their functional activity evaluated by a SPA assay with a reference compound. Supplementary Figure 6 Glucose and other LXR ligands suppress gluconeogenic genes under conditions of maximal induction. HepG2 cells were pre-treated for 8 h with 1 mm 8- Br-cAMP and 1 µm dexamethasone prior to addition of ligands or insulin (100 nm). Insulin has no effect on the expression of ABCG1, but appears to work together with LXR ligands to suppress gluconeogenesis. Error bars represent s.d. www.nature.com/nature 4
Supplementary Figure 7 LXR is required for glucose modulation of cholesterol efflux genes. In cells pretreated for 48 h with sirna against hlxrα, neither D-glucose nor GW3965 stimulate expression of ABCA1, ABCG1, or CETP (see Methods). Error bars represent s.d. www.nature.com/nature 5
Supplementary Figure 8 Inhibition of glucose uptake suppresses induction of LXR targets by glucose and diminishes the response to GW3965. Cells were pre-treated for 16 h with cytochalasin B (CcB, 5µM) prior to ligand addition (see Methods). Error bars represent s.d. www.nature.com/nature 6
Supplementary Figure 9 Glucose can regulate LXR targets in the absence of insulin secretion. Insulin-depleted STZ-injected mice fasted overnight were challenged orally with GW3965 (50 mpk), or refed with a glucose or sucrose diet and sacrificed 6 h later. Hepatic gene expression was assessed by qrt-pcr. Note that expression of direct LXR targets (e.g. SPα, SREBP-1c) correlates with circulating glucose, not insulin. Error bars represent s.d. *p<0.05, **p<0.001 treatment vs. fasted. www.nature.com/nature 7
Supplementary Figure 10 Glucose induces expression of insulin-independent LXR targets and SREBP-1c in intestine. Sucrose has a milder effect. Wild-type mice fasted overnight were challenged orally with GW3965 (50 mpk), or refed with a glucose or sucrose diet and sacrificed 6 h later. Hepatic gene expression was assessed by qrt-pcr. Error bars represent s.d. *p<0.05, **p<0.001 treatment vs. fasted. www.nature.com/nature 8
Materials and Methods Constructs Expression constructs for chimeric NR LBDs were generated by fusing the LBDs to the DBD of GAL4 and inserting them into pcdna3. Assays were validated using known ligands. The ptk-2xlxre-abca1-luc reporter and pcdna3 expression plasmids for LXRα/β have been described 22. Cell culture and treatments HepG2 cells were cultured in DMEM with 10% FBS. For transcriptional assays, cells were plated in 384 well plates in DMEM with varying glucose concentration (0, 2, 25 mm) with 10% fetal bovine lipoprotein deficient serum (FBLDS, Intracel) in the presence of 7.5 μm lovastatin and 100 μm mevalonic acid. For gene expression analysis, cells were plated in 48 well plates in the above conditions, and treated for 16 h with DMSO, 1 μm GW3965, 5 μm 22-R-HC, or 20 mm D-glucose. Induction of gluconeogenic genes was achieved using 8-Br-cMAP (1 mm) and dexamethasone (1 µm) for 8 h prior to test compound or insulin (100 nm) addition; cells were harvested 6 h after ligand addition. For sirna experiments, HepG2 cells were transfected with an hlxrα (Dharmacon, SMART pool M-003413-01-0005) or control sirna pool (Dharmacon, sicontrol non-targeting sirna pool, D-001206-13-05, designed to have >4 mismatches with all known human and mouse genes) using DharmaFECT 4; 48 h later, ligands were added, and cells were harvested 24 h later. To check for knockdown, nuclear extracts were analyzed by Western blot using an antibody against LXR (Calbiochem). In glucose uptake inhibition experiments, cells were pre-treated with 5 µm cytochalasin B for 16 h before addition of ligands for 8 h. Transcriptional assays HepG2 cells were transfected using FuGENE 6 (Roche). 24 h later, ligands were added as 12- point dose response curves starting at 10 μm for GW3965 and T0901317, 30 μm for 22-R-HC and 22-S-HC, and 20 mm for carbohydrates. 16 h later, luciferase activity was read and normalized to an internal pcmv-gfp control. Fluorescence-resonance energy transfer assay The purity and identity of commercial proteins used in binding assays was verified by mass fingerprinting using trypsin and MALDI-TOF mass spectrometry. Purity was greater than 85% as evaluated by SDS PAGE, with no detectable protease, DNase, or RNase activity. Functional activity of all protein batches received was tested using coactivator recruitment and SPA assays with reference compounds to insure that results with standards were within published values. In all cases, fractional occupancy using a 12-point dose response to labeled T0901317 was calculated to verify protein quality prior to its use in FRET or SPA assays. The top dose of labeled T0901317 used was greater than 400 times its K d, ensuring receptor saturation. FRET assay was performed in 384 well plates in 20 μl of volume. A mix of 5 nm His-LXRαLBD or His-LXRβLBD (Roche), 5 nm biotinylated SRC-1 peptide (Biotin- CPSSHSSLTERHKILHRLLQEGSPSC-OH), Europium-labeled anti-his antibody (Perkin Elmer) and allophycocyanin-labeled streptavidin (Prozyme) was prepared. Ligands were tested in 12-point dose response curves starting at 10 μm for GW3965 and T0901317, 30 μm for 22-R- HC and 22-S-HC, and 20 mm for glycolytic intermediates. Plates were incubated for 3 h at rt www.nature.com/nature 9
and FRET read on an AnalystGT (Molecular Devices). In combination experiments, compounds were added simultaneously. Scintillation proximity assay SPA was performed as in Janowski et al 23. Scintillant-filled beads (Amersham) were diluted in SPA buffer (10 mm K 2 HPO 4, 10 mm KH 2 PO 4, 2 mm EDTA, 50 mm NaCl, 1 mm DTT, 2 mm CHAPS, 10% glycerol) to a concentration of 5 mg/ml. Assay was performed in 384 well plates (Packard) in 20 μl containing beads (0.2 mg per well) and His-LXRαLBD (120 ng per well) or His-LXRβLBD (50 ng per well) (Roche). Amount of protein did not deplete ligand concentration. Binding of [ 3 H]-glucose was tested as a 12-point dilution curve starting at 70 mm. Displacement assays used 25 nm [ 3 H]-T0901317 (Amersham) or 20 mm [ 3 H]-glucose (Amersham). Ligands were tested in 12-point dose response curves starting at 12 μm for GW3965 and T0901317, 35 μm for 22-R-HC and 22-S-HC, and 70 mm for carbohydrates. In combination experiments, ligands were added simultaneously. Plates were shaken for 3 h at room temperature and read in a Packard Topcount at 1 min per well. Wells devoid of competitor represented 100% binding. Non-specific binding was measured by leaving LXR protein out of the SPA reaction. To calculate the fraction of LXR protein that bound ligand, we used the following formula from the law of mass action: Fractional occupancy = [Receptor- Agonist]/[Receptor], where [Receptor-Agonist] is the concentration of the receptor-agonist complex, and [Receptor] is the total concentration of the receptor. In our assays, the fraction of LXRs that bind the ligand is 95-99%. Generation of K d and K i values Dose response and competition curves were generated by nonlinear regression analysis using GraphPad Prism and the apparent equilibrium association/dissociation constants (K d and K i values) were determined using the method described by DeBlasi and colleagues 24. Animal experiments 10-week-old male C57BL/6 mice were maintained on standard chow diet. Mice fasted for 24 h were refed with sucrose and D-glucose diets lacking cholesterol (Research Diets Inc., D05121205, D05121203) or gavaged with GW3965 50 mpk, 6 h before sacrifice. Age-matched mice were rendered insulin-deficient via injection of streptozotocin (100 µg/kg IP) on 3 consecutive days. STZ-treated mice were used once hyperglycemia (glucose > 350 mg/dl) became evident (6 days). All experiments were carried out with strict awareness of the light cycle in place to avoid variation in gene expression arising from potential circadian regulation. Experiments were approved by GNF s IACUC. Gene expression analysis RNA was analyzed by TaqMan qrt-pcr using the one-step Superscript III platinum reagent (Invitrogen). Samples were run in triplicate as multiplexed reactions with a normalizing internal control (36B4). Probe and primer sequences are available on request. Statistical analysis was performed by using a two-tailed Student s t test. www.nature.com/nature 10
Supplementary Table 1 Structure-activity relationship of glycolysis derivatives on LXR/RXR transactivation assay LXRα LXRβ Compound K d, μm EC 50, μm Efficacy K d, μm EC 50, μm Efficacy GW3965 0.147 ± 0.022 0.188 1 0.036 ± 0.006 0.051 1 T0901317 0.021 ± 0.004 0.033 1 0.005 ± 0.001 0.007 1 22-(R)-HC 4.05 ± 1.05 4.99 0.79 2.69 ± 0.61 2.97 0.90 22-(S)-HC - - - - - - D-(+)-Glucose 1770 ± 99 3141 0.55 191 ± 37 308 0.61 D-Glucose-6-P - No fit 0.44 - No fit 0. 39 D-(-)-Fructose - No fit 0.20 - No fit 0.10 D-Fructose-6-P - No fit 0.12 - No fit 0. 13 D-Fructose-1,6-P - - - - - - DL-Glyceraldehyde-3-P - - - - - - D-(-)-3-Phosphoglycerate - - - - - - Phosphoenolpyruvate - - - - - - Pyruvate - - - - - - K d values are presented ± S.D. EC 50, effective concentration for 50% maximal activation; Efficacy, maximal fold activation relative to GW3965; -, below detection; No fit, no plateau at 20 mm. www.nature.com/nature 11
Supplementary Table 2 Structure-activity relationship of glycolysis derivatives on coactivator recruitment assay LXRα LXRβ Compound K d, μm EC 50, μm Efficacy K d, μm EC 50, μm Efficacy GW3965 0.065 ± 0.02 0.186 1 0.027 ± 0.005 0.049 1 T0901317 0.008 ± 0.002 0.025 1 0.025 ± 0.004 0.05 1 22-(R)-HC 1.12 ± 0.5 4.72 0.82 1.38 ± 0.39 3 0.94 22-(S)-HC - - - - - - D-(+)-Glucose 1931 ± 401 2904 0.86 103.5 ± 29 318 0.98 L-Glucose 1410 ± 528 2932 0.86 121.7 ± 54 380 0.99 D-Glucose-6-P 1765 ± 695 5000 0.74 1950 ± 86.7 1366 0.43 D-(-)-Fructose - - - - - - D-Fructose-6-P - - - - - - D-Fructose-1,6-P - - - - - - DL-Glyceraldehyde-3-P - - - - - - D-(-)-3-Phosphoglycerate - - - - - - Phosphoenolpyruvate - - - - - - Pyruvate - - - - - - K d values are presented ± S.D. EC 50, effective concentration for 50% maximal activation; Efficacy, maximal fold activation relative to GW3965; -, below detection. www.nature.com/nature 12
Supplementary Table 3 Structure-activity relationship of glycolysis derivatives on SPA displacement assay LXRα LXRβ Compound K i, μm EC 50, μm Efficacy K i, μm EC 50, μm Efficacy GW3965 0.074 ± 0.003 0.121 1 0.028 ± 0.007 0.040 1 T0901317 0.014 ± 0.007 0.021 1 0.022 ± 0.004 0.034 1 22-(R)-HC 0.350 ± 0.024 5.3 0.80 0.169 ± 0.027 2.76 0.82 22-(S)-HC 0.180 ± 0.038 2.7 0.72 0.184 ± 0.010 2.51 0.84 D-(+)-Glucose 1801 ± 102 2701 0.79 202 ± 12 303 0.66 L-Glucose 1745 ± 305 3050 0.80 178 ± 45 299 0.65 D-Glucose 6-P 2967 ± 721 4475 0.78 744 ± 46 1116 0.73 D-(-)-Fructose - - - - - - D-Fructose 6-P - - - - - - D-Fructose 1,6-P - - - - - - DL-Glyceraldehyde 3-P - - - - - - D-(-)-3-Phosphoglycerate - - - - - - Phosphoenolpyruvate - - - - - - Pyruvate - - - - - - K i values are presented ± S.D. EC 50, effective concentration for 50% maximal displacement of [ H]-T0901317; Efficacy, maximal activity relative to GW3965; -, below detection. 3 www.nature.com/nature 13