Liver X receptor regulates hepatic nuclear O-GlcNAc signaling and carbohydrate responsive element-binding protein activity.

Liver X receptor (LXR)α and LXRβ play key roles in hepatic de novo lipogenesis through their regulation of lipogenic genes, including sterol regulatory element-binding protein (SREBP)-1c and carbohydrate responsive element-binding protein (ChREBP). LXRs activate lipogenic gene transcription in response to feeding, which is believed to be mediated by insulin. We have previously shown that LXRs are targets for glucose-hexosamine-derived O-linked β-N-acetylglucosamine (O-GlcNAc) modification enhancing their ability to regulate SREBP-1c promoter activity in vitro. To elucidate insulin-independent effects of feeding on LXR-mediated lipogenic gene expression in vivo, we subjected control and streptozotocin-treated LXRα/β(+/+) and LXRα/β(-/-) mice to a fasting-refeeding regime. We show that under hyperglycemic and hypoinsulinemic conditions, LXRs maintain their ability to upregulate the expression of glycolytic and lipogenic enzymes, including glucokinase (GK), SREBP-1c, ChREBPα, and the newly identified shorter isoform ChREBPβ. Furthermore, glucose-dependent increases in LXR/retinoid X receptor-regulated luciferase activity driven by the ChREBPα promoter was mediated, at least in part, by O-GlcNAc transferase (OGT) signaling in Huh7 cells. Moreover, we show that LXR and OGT interact and colocalize in the nucleus and that loss of LXRs profoundly reduced nuclear O-GlcNAc signaling and ChREBPα promoter binding activity in vivo. In summary, our study provides evidence that LXRs act as nutrient and glucose metabolic sensors upstream of ChREBP by modulating GK expression, nuclear O-GlcNAc signaling, and ChREBP expression and activity.

expression ( 23 ). Mutations of these residues, however, did not result in a constitutively active ChREBP ( 24 ), which suggests that ChREBP ␣ is regulated by alternative means. In this respect, a recent study reported O-GlcNAcylation of ChREBP in response to hyperglycemia ( 25 ). Moreover, adenoviral overexpression of OGT in liver increased ChREBP O-GlcNAcylation, protein stability, and transactivation of liver pyruvate kinase ( L-pk ) and lipogenic genes ( 25 ). The lipogenic potential of LXRs was not investigated in that study.
Several reports show that LXR is essential for SREBP-1c activity in response to feeding, as SREBP-1c expression and hepatic DNL are drastically reduced in LXR-defi cient mice ( 5,26,27 ). Although LXR clearly stimulates hepatic ChREBP expression after pharmacological activation ( 6 ), whether LXR regulates ChREBP under physiological conditions is less clear and may be dependent on insulin, dietary carbohydrate, and lipid content ( 26,27 ). Notably, Herman et al. ( 28 ) recently identifi ed a novel variant of ChREBP named ChREBP ␤ , which results from an alternative promoter located in exon1b of the ChREB gene. They further showed that glucose metabolism induced transcription of ChREBP ␣ , promoting binding of ChREBP ␣ to a carbohydrate response element (ChORE) identifi ed in exon1b, inducing ChREBP ␤ transcription. Whether LXRs contribute to isoform-specifi c expression of ChREBP is currently not known.
To further explore the role of LXR and ChREBP in hepatic lipogenesis and to investigate insulin-independent effects of feeding on LXR activity, we subjected control and streptozotocin (STZ)-treated LXR +/+ and LXR Ϫ / Ϫ mice to a fasting-refeeding regime. Our results demonstrate that LXR regulates hepatic Gk expression and lipogenic genes also under hyperglycemic conditions independently of insulin. Furthermore, levels of O-GlcNAc-modifi ed LXR were increased upon refeeding in both models, supporting the notion of LXR as a postprandial glucose-O-GlcNAc sensor. In support of our observations in vivo, LXR conferred high glucose responses on the ChREBP ␣ promoter in Huh7 cells, an effect that was signifi cantly inhibited by OGT depletion. We further show that nuclear, but not cytosolic, protein-O-GlcNAc levels are dramatically lower in LXR Ϫ / Ϫ compared with LXR +/+ mice. Accordingly, O-GlcNAcylation of ChREBP and the potential of ChREBP ␣ to induce its target gene L-pk , was lower in refed LXR Ϫ / Ϫ compared with LXR +/+ mice. Combined, our work identifi es LXR as an important integrator of hepatic glucose/ O-GlcNAc metabolism and fatty acid synthesis upstream of ChREBP. of hepatic DNL, glucose also plays a role in this process ( 2 ). The liver X receptors (LXRs) [LXR ␣ (NR1H3) and LXR ␤ (NR1H2)] are oxysterol-activated nuclear receptors with important roles in cholesterol, glucose, and lipid metabolism ( 3,4 ). LXR ␣ is mainly expressed in liver, adipose, intestine, and macrophages, whereas LXR ␤ is ubiquitously expressed. Hepatic LXRs regulate the expression of sterol regulatory element-binding protein (SREBP)-1c ( 5 ) and carbohydrate responsive element-binding protein (ChREBP) ( 6 ), which, together with LXRs, are key regulators of many glycolytic and lipogenic genes, including glucokinase (GK) ( 7 ), acetyl-CoA carboxylase (ACC) ( 8 ), FAS ( 9 ), and stearoyl CoA desaturase (SCD)1 ( 10 ).

Materials
In addition to oxysterols and synthetic ligands, our laboratory and others have shown that insulin stimulates the lipogenic activity of LXRs in mouse liver ( 11,12 ). The mechanisms involved are poorly understood, but may include posttranslational modifi cations [i.e., phosphorylation of LXR and/or synthesis of endogenous LXR ligand(s)]. Mitro et al. ( 13 ) reported that glucose bound to and activated LXR; however, these fi ndings are controversial because the LXR ligand binding pocket accommodates small hydrophobic, rather than hydrophilic, compounds ( 14 ). We recently reported that LXRs were posttranslationally modified by O-linked ␤ -N -acetylglucosamine (O-GlcNAc) in response to glucose, which in turn potentiated their transactivation of SREBP-1c ( 15 ). O-GlcNAc is a dynamically regulated posttranslational modifi cation that is catalyzed by O-GlcNAc transferase (OGT). OGT uses UDP-N -acetylglucosamine (GlcNAc) as a substrate to add GlcNAc in O-glycosidic linkage to serine and threonine residues in target proteins, which are removed by O-GlcNAcase (OGA). The intracellular concentration of UDP-GlcNAc is controlled by the nutrient-responsive hexosamine biosynthetic pathway (HBP), a branch of glucose metabolism ( 16 ). Metabolic fl ux via the HBP is controlled by the rate-limiting enzyme glutamine:fructose-6-phosphate amidotransferase (GFAT). Concequently, UDP-GlcNAc pools are reduced when glucose uptake and glucose-6-phosphate (G-6-P) synthesis are limited ( 17 ). Because OGT activity is highly responsive to UDP-GlcNAc concentrations, OGT serves as a unique nutrient sensor ( 16 ). The activity of OGT and OGA are regulated by UDP-GlcNAc availability, protein-protein interactions, and posttranslational modifi cations, but the precise mechanisms involved are poorly understood ( 18,19 ). Of note, the sensitivity of OGT to glucose increases with decreasing insulin signaling ( 20 ), whereas insulin mediates spatio-temporal activation of OGT, regulating substrate specifi city of the enzyme at the plasma membrane ( 21 ).
Several studies have shown that ChREBP is activated by glucose through a variety of glucose metabolites generated downstream of G-6-P, including xylulose-5-phosphate, fructose-6-phosphate, and fructose-2,6-bisphosphate ( 22 ). The mechanisms through which these glucose metabolites modulate ChREBP activity are unclear, but are believed to involve allosteric regulation and dephosphorylation of the canonical isoform ChREBP at Ser196 and Thr666 leading to its nuclear translocation and regulation of target gene

RNA extraction, cDNA synthesis, and quantitative RT-PCR
RNA was isolated with TRIZOL® reagent (Invitrogen) according to the manufacturer's protocol, including high salt precipitation (0.8 M sodium acetate, 1.2 M NaCl) to avoid contaminating polysaccharides to coprecipitate with RNA. Isolated RNA was reverse transcribed into cDNA using a high-capacity cDNA Archive kit (Applied Biosystems). Analysis of mRNA expression was done by quantitative RT-PCR on a 7900HT instrument (Applied Biosystems). Gene expression was normalized against the expression of tata-binding protein (Tbp). Primers were as follows: TaqMan gene expression assays detecting mouse: Acc ␣ (Mm01304257_ m1), Acc ␤ (Mm01204678_m1), elongation of long-chain fatty acid family member 6 ( Elovl6

Generation of LXR antibody
Rat LXR ␣ and LXR ␤ proteins were purifi ed as described previously ( 36 ). Rabbits were immunized at Agisera (Vännäs, Sweden) by a standard immunization program. Briefl y, 0.25 mg of LXR ␣ and LXR ␤ were injected four times and serum collected after 15 weeks. Purifi ed LXR ␣ and LXR ␤ protein (1.4 mg) in 0.2 M NaHCO 3 , 0.5 M NaCl (pH 8.3) buffer were coupled on a HiTrap NHSactivated matrix column (GE Healthcare). The immunized rabbit serum was added to the column and washed according to standard procedure. Elution of anti-LXRs was performed 10 times and eluates were pooled.

Liver extracts and immunoblot analysis
Nuclear and cytoplasmic proteins were prepared using the NE-PER extraction kit (Pierce Biotechnology) with the following inhibitors added to the buffers: 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM ␤ -glycerophosphate, 1 M O-GlcNAcase inhibitor GlcNActhiazoline, and CompleteTM protease inhibitors (Roche Applied Science). Proteins were separated by SDS-PAGE (Bio-Rad) and blotted onto polyvinylidene difl uoride membrane (Millipore ).

Animals and treatment
Male LXR ␣ ␤ -defi cient (LXR Ϫ / Ϫ ) mice and wild-type littermates (LXR +/+ ) were housed in a temperature-controlled (22°C) facility with a strict 12 h light/dark cycle. The mice had free access to food and water at all times. LXR Ϫ / Ϫ mice and corresponding controls had mixed genetic background based on 129/Sv and C57BL/6J strains, backcrossed in C57BL/6J for at least six generations. The generation of the LXR Ϫ / Ϫ mice has been described previously ( 32,33 ). STZ was prepared in a sodium citrate buffer (50 mmol/l, pH 4.5) immediately before injections. STZtreated mice were treated with two intraperitoneal injections of STZ (100 mg/kg) with a 1 day interval. Seven days after the fi rst STZ injection, mice were included in a fasting-refeeding experiment with matched untreated control mice. Mice were fasted for 24 h or fasted for 24 h and refed for 12 h on a diet containing 64% carbohydrates, 31.5% protein, 4.5% fat, and no cholesterol (SDS RM no.1 maintenance, Special Diets Services, UK). The mice were euthanized by cervical dislocation at 8:00 AM and tissues were weighed and snap-frozen in liquid nitrogen and stored at Ϫ 80°C until further analysis. All use of animals was approved and registered by the Norwegian Animal Research authority.

Blood chemistries and liver lipid analyses
Blood glucose was measured at the time of euthanization with a handheld glucometer (Roche). Serum was separated from blood by centrifugation. Serum insulin was measured using the ultra-sensitive insulin kit from Mercodia (Mercodia AB, Uppsala, Sweden) according to the manufacturer's instructions. Serum lipid quantifi cation, using enzymatic in vitro tests, was done on a Roche Hitachi 917. TGs were determined with a TG GPO-PAP kit (11730711216; Roche). Liver samples were homogenized in cold sucrose buffer [0.25 M sucrose, 10 mM HEPES, 1 mM Na-EDTA (pH 7.4)] and centrifuged at 600 g for 10 min at 4°C and hepatic TGs were measured in the supernatant using the TG GPO-PAP kit according to the manufacturer's protocol.

Succinylated wheat germ agglutinin pulldown
Nuclear extracts (100 g) were incubated with protein A/Gagarose beads (sc-2003; Santa Cruz Biotechnology) for 1 h at 4°C. Cleared extracts were transferred to new tubes and incubated with 30 l of succinylated wheat germ agglutinin (sWGA)-agarose (Vector Lab) with or without 0.5 M GlcNAc overnight at 4°C. After four washes (PBS, 0.2% NP-40), proteins were eluted from the beads in 2× Laemmli buffer and separated by SDS-PAGE. The captured proteins were analyzed by immunoblotting.

Statistical analysis
Statistical analyses were performed using SPSS 22 (IBM Corp.). All data are presented as means and SEM, and error bars for all results are derived from biological replicates rather than technical replicates. The homogeneity of variances between groups was evaluated using Levene's test. If signifi cant, a Welch test was applied to assess overall differences combined with a Dunnett's T3 post hoc test to establish intergroup comparisons. If Levene's test was not signifi cant and homogeneity of variances could be assumed, statistical differences between groups were determined by one-way ANOVA followed by Tukey's multiple comparison tests. P < 0.05 was considered statistically signifi cant.

Chromatin immunoprecipitation
Chromatin immunoprecipitation (ChIP) experiments were performed as described previously ( 41 ). Briefl y, liver tissue was cross-linked with 1% formaldehyde for 10 min at room temperature. Cross-linking was stopped by 10 min incubation with 12. . All washing steps were done for 5 min at 4°C. DNAprotein complexes were eluted with 1% SDS and 0.1 M NaHCO 3 and reverse cross-linked by adding 0.2 M NaCl and overnight incubation at 65°C. DNA was purifi ed by phenol-chloroform extraction, precipitated in ethanol with sodium acetate, and dissolved in water. DNA enrichment was quantifi ed by quantitative RT-PCR. ChIP PCR primers spanning mouse LXR response elements (LXREs) were designed to amplify known LXR-enriched regions ( 41 ) and were as follows: . The binding of ChREBP to the ChORE in the promoter-proximal enhancer of L-pk has been described previously ( 25 ) and was as follows: L-pk ChORE (5 ′ -GTCCCACACTT-TGGAAGCAT, 5 ′ -CCCAACACTGATTCTACCC).

Recombinant proteins and GST pulldown
pET-24b-His-ncOGT, pGEX-4T-1-hLXR ␣ , and pGEX-4T-1-hLXR ␤ were expressed in Escherichia coli BL21 (DE3) Star (Invitrogen). Human His-ncOGT was grown and purifi ed as previously described ( 30 ). GST fusion proteins were expressed as described different between genotypes in refed control mice, but were almost abolished (70% lower in cytosolic, 90% lower in nuclear) in refed LXR Ϫ / Ϫ STZ-treated hyperglycemic mice compared with refed LXR +/+ mice treated with STZ ( Fig.  1C ; band intensity quantifi cations relative to loading control using Image J software). This suggests that transcriptional and posttranscriptional regulation of ChREBP ␣ is dependent on LXR under hyperglycemic conditions lacking the postprandial insulin signal. ChREBP ␣ is believed to be retained inactive in the cytoplasm during fasting ( 23 ). Our observed nuclear localization of ChREBP in the fasted state may refl ect levels of the constitutively nuclear localized ChREBP ␤ ( 25 ). However, we were not able to distinguish between the two isoforms in Western blotting analysis (ChREBP ␣ , 93 kDa; ChREBP ␤ , 90 kDa). Furthermore, we did not observe a signifi cant fasting-refeeding shuttling of ChREBP between the cytosolic and nuclear compartments. Rather, cytosolic and nuclear ChREBP protein levels were comparable under each condition. ChIP assays revealed increased LXR recruitment to an LXRE in the ChREBP ␣ promoter in refed control and STZtreated mice. LXR recruitment to an LXRE in the SREBP-1 promoter was evident in refed control mice, but only a weak nonsignifi cant increase in LXR binding was observed in STZ-treated mice ( Fig. 1D ), which did not refl ect the strong induction of Srebp-1c mRNA levels in these mice ( Fig. 1B ). This may be due to increased coactivator recruitment to this LXRE and/or increased SREBP-1c autoregulation under hyperglycemic conditions. Taken together, these data suggest that LXRs are essential regulators of SREBP-1 and ChREBP expression also in the absence of postprandial insulin.

Expression of hepatic de novo lipogenic genes and production of plasma TGs through LXR are not dependent on insulin signaling
We next investigated the expression of genes downstream of the LXR lipogenic axis to further delineate the effects of hyperglycemic conditions on LXR activity. A schematic representation of DNL with the proteins involved is provided in Fig. 2A . Gk and L-pk mRNA levels were induced approximately 50-fold and 6-fold in refed control mice, respectively ( Fig. 2B ). These genes were also signifi cantly upregulated in refed STZ-treated mice; however, the level of induction was less than that observed in control mice (10-fold for Gk , 3-fold for L-pk ). Of note, expression of gluconeogenic genes, glucose-6-phosphatase and phosphoenolpyruvate carbokinase ( PEPCK ), and the gluconeogenic coactivator peroxisome proliferator-activated receptor ␥ coactivator 1-␣ ( PGC1 ␣ ) were strongly downregulated in both control and STZ-treated mice (data not shown), supporting increased G-6-P synthesis above fasting levels in hyperglycemic STZ-treated mice. In support of our observations, others have shown downregulation of glucose-6-phosphatase and PEPCK in response to glucose and sucrose refeeding in control and STZ-treated mice ( 13,43 ). Moreover, GLUT2 expression is increased in refed STZadministered mice, ensuring increased hepatic glucose uptake in hyperglycemia ( 43 ). The mRNA levels of Fas , Elovl6 ,

Feeding induces LXR ␣ protein levels
To examine the effects of insulin on LXR-regulated hepatic lipogenic target gene expression, LXR +/+ and LXR Ϫ / Ϫ control and STZ-treated mice were placed on a fastingrefeeding regime. STZ was used to destroy pancreatic ␤cells and insulin production. Using this model, the effect of insulin could be distinguished from those of hyperglycemia in refed STZ-treated mice (low insulin, high glucose) compared with untreated control mice (high insulin, high glucose). In hepatocytes expressing the insulin-independent low-affi nity glucose transporter, GLUT2, glycolysis is favored in the fed state when glucose is abundant ( 42,43 ). LXR +/+ and LXR Ϫ / Ϫ refed control mice showed increased serum insulin and blood glucose levels, whereas refed STZtreated mice were hyperglycemic with fasting insulin levels ( Fig. 1A ). Insulin levels were slightly, but not signifi cantly, higher in fed LXR +/+ compared with LXR Ϫ / Ϫ control mice, supporting the known role of LXR in glucose homeostasis, insulin secretion, and pancreatic function ( 44,45 ). No signifi cant differences in body weight or liver weight were found between treatment-matched LXR +/+ and LXR Ϫ / Ϫ mice (supplementary Table 1). Lxr ␣ and Lxr ␤ mRNA expression levels were unaffected in refed control and STZ-treated LXR +/+ mice when compared with treatmentmatched fasted mice ( Fig. 1B ). Nuclear LXR ␣ protein levels were 2.1-to 2.5-fold increased in refed control and STZtreated mice ( Fig. 1C ; band intensity quantifi cations relative to loading control using Image J software), whereas LXR ␤ protein levels remained unchanged. These data suggest that feeding increases LXR ␣ protein expression via posttranscriptional or posttranslational mechanisms in vivo.

LXR confers differential regulation of hepatic ChREBP isoform expression
The expression and activity of SREBP-1c and ChREBP are regulated by insulin and glucose ( 43,46 ). Because SREBP-1c and ChREBP can work in concert with LXRs to regulate hepatic lipogenic genes ( 5, 6, 47 ), we investigated the expression of SREBP-1 and ChREBP upon refeeding under normal and hyperglycemic insulin-independent conditions. In agreement with previous reports, SREBP-1c mRNA and protein levels were strongly induced in refed control and STZ-treated LXR +/+ mice ( Fig. 1B, C )  Ϫ / Ϫ mice) ( Fig. 1B ). Expression levels of the newly identifi ed shorter isoform, Chrebp ␤ ( 28 ), were strongly induced by refeeding in both control and STZ-treated mice (6-fold), and almost abolished in LXR Ϫ / Ϫ mice. Interestingly, ChREBP cytosolic and nuclear protein levels, detected by immunoblotting with an antibody recognizing ChREBP ␣ as well as ChREBP ␤ (our own observations in ChREBP ␤ overexpressing cells), were not statistically  LXR ␣ /RXR ␣ transfected cells after high glucose (25 mM) compared with low glucose (1 mM) treatment for 24 h ( Fig. 3B ). To determine whether OGT-mediated O-GlcNAc signaling was conferring the glucose effect on ChREBP ␣ expression through LXR ␣ , Huh7 cells, cultured in physiological (5 mM) and high glucose, were transfected with scrambled siRNA or siRNA against OGT. One to two days after siRNA transfection, cells were transfected with LXR ␣ /RXR ␣ and ChREBP ␣ promoter construct in siRNAcontaining medium and cells were analyzed for luciferase activity the following day. The stimulating effect of high glucose through LXR ␣ was approximately 30% reduced in siOGT-transfected cells, whereas no signifi cant reduction was observed under normoglycemic (5 mM) conditions ( Fig. 3C ), arguing for a role for OGT signaling in high glucose/LXR-mediated transactivation of the ChREBP ␣ promoter.

LXR regulates nuclear O-GlcNAc signaling
Having established a role for LXR in mediating glucose responses on the promoters of ChREBP ␣ and SREBP-1c ( 15 ) via HBP/O-GlcNAc signaling, we next investigated the impact of feeding and LXR on expression of central genes in this pathway ( Fig. 4A ). Expression levels of GFAT1, the rate limiting enzyme of HBP, was upregulated by feeding in the absence and presence of insulin, whereas the less characterized isoform, GFAT2, was only upregulated in STZtreated mice (3-fold) and more so in LXR Ϫ / Ϫ mice (6-fold), suggesting a possible negative feedback by LXRs on GFAT2 expression under hyperglycemic conditions. GFAT1, but not GFAT2, was localized in both cytoplasmic and nuclear fractions in LXR +/+ and LXR Ϫ / Ϫ mice, suggesting a possible nuclear production of glucosamine-6-phosphate, an intermediary product in the HBP generating UDP-GlcNAc, the substrate for OGT ( Fig.4B ). Hepatic UDP-GlcNAc concentrations have been previously shown to be increased by approximately 60% in ad libitum-fed STZ-treated rats compared with controls ( 49 ). OGT and OGA expression were not regulated by feeding or LXR ( Fig. 4A, B ). Surprisingly, nucleocytoplasmic OGT protein expression was unaffected ( Fig. 4B ), but nuclear protein O-GlcNAcylation was signifi cantly lower in refed LXR Ϫ / Ϫ mice under both control and hyperglycemic conditions ( Fig. 4C ), suggesting that LXR, by direct or indirect mechanisms, regulates nuclear OGT activity. In line with our observations in vivo, LXR potentiated nuclear, but not cytosolic, protein O-GlcNAcylation in LXR-tranfected Huh7 cells compared with empty vector controls (data not shown). Cytosolic protein O-GlcNAcylation was not affected by LXR defi ciency as revealed by Western blotting analysis using a pan-O-GlcNAcspecifi c antibody (CTD 110.6), and total nuclear and cytosolic protein O-GlcNAcylation levels were similar in control and STZ-treated mice ( Fig. 4C ).

LXRs regulate O-GlcNAcylation of ChREBP and binding of ChREBP ␣ to the L-pk promoter
Recently, O-GlcNAcylation of ChREBP was shown to potentiate its activity, stability, and binding to the promoterproximal enhancer of L-pk ( 25 ). Based on that study and Acc ␤ , and Gpat were elevated to a similar degree in both refed control and STZ-treated mice. Notably, the mRNA levels of Acc ␣ and Scd1 were highly induced in LXR +/+ STZtreated mice, where SREBP-1c may be important ( 43 ). Interestingly, GK and L-PK protein levels did not mimic the induced mRNA expression levels in refed LXR +/+ mice, there were only small insignifi cant changes, which could be due to posttranslational regulation of the active proteins ( Fig. 2C ). GK protein levels were 1.4 times higher in refed control compared with fasted control LXR +/+ mice and 1.6 times higher in refed STZ compared with fasted STZ LXR +/+ mice. Fasting GK protein levels were 2.8 times higher in fasted control compared with fasted STZ mice and GK and L-PK levels were signifi cantly lower in refed LXR Ϫ / Ϫ mice compared with LXR +/+ mice in both control (GK 50% lower, L-PK 92% lower) and STZ-treated mice (GK 65% lower, L-PK 77% lower) ( Fig. 2C ; band intensity quantifi cations relative to loading control using Image J software). LXR binding to the LXREs in the promoters of Gk , Fas , and Scd1 was signifi cantly induced in refed control mice, but LXR binding was only signifi cantly induced to the Scd1 LXRE in STZ-treated mice ( Fig. 2D ). Differences may be due to alternative LXREs in promoters or enhancers of these genes, and/or coregulator interactions. The induction of lipogenic genes was followed by increased plasma TGs in refed control and STZ-treated LXR , mice ( Fig. 2E ). Plasma nonesterifi ed fatty acids and cholesterol were not statistically different between LXR +/+ and LXR Ϫ / Ϫ mice (data not shown). Similar to previous reports, LXR Ϫ / Ϫ mice accumulated fewer hepatic TGs compared with LXR +/+ littermates, and in STZ-treated compared with control mice ( 43,48 ) ( Fig. 2E ). No differences in expression of hepatic fatty acid oxidation genes ( Cpt1 and Acox1 ) or genes involved in endoplasmic reticulum stress ( Hspa5 , Hsp90b1 , and Calr ) were observed, suggesting that the lower hepatic TG content was independent of these pathways (data not shown). Taken together, our fi ndings suggest that LXR regulates hepatic lipogenic genes and TG content under hyperglycemic conditions independently of insulin, suggestive of a role for glucose in regulating LXR activity as previously suggested ( 13,15 ).

High glucose regulates LXR transactivation of the ChREBP ␣ promoter via O-GlcNAc
We have previously reported increased O-GlcNAcylation of LXR in vivo and in vitro in response to refeeding and high glucose treatment contributing to increased SREBP-1c promoter activation ( 15 ). To examine whether high glucose can regulate the ChREBP ␣ promoter through LXRs, Huh7 cells were cotransfected with LXR ␣ or LXR ␤ and/or RXR ␣ expression vectors together with a luciferase reporter construct containing 3.2 kb of the 5 ′ -fl anking region of the mouse ChREBP ␣ promoter ( Fig. 3A, B ). Cotransfected with RXR ␣ , LXR ␣ conferred an approximately 30-fold induction of ChREBP ␣ promoter-luciferase activity as compared with 8-fold by LXR ␤ in Huh7 cells adapted to high glucose (25 mM) ( Fig. 3A ). Moreover, in cells adapted to physiological glucose levels (5 mM), we observed a 30% increase in ChREBP ␣ promoter-luciferase activity in the reduced L-PK expression and protein O-GlcNAc levels observed in LXR Ϫ / Ϫ mice, we investigated whether LXR could regulate ChREBP activity through modulating O-GlcNAcylation of ChREBP. Using lectin (sWGA) pulldown experiments, we observed lower levels of O-GlcNAcylated proteins precipitated from nuclear lysates isolated from refed LXR Ϫ / Ϫ mice compared with LXR +/+ controls, which was in agreement with reduced total protein O-GlcNAc ( Fig. 5A ). Specifically, ChREBP O-GlcNAcylation was strongly reduced in refed LXR Ϫ / Ϫ mice compared with LXR +/+ mice as determined by 83 and 73% reduced recovery of ChREBP protein after sWGA pulldown in control and STZ-mice, respectively ( Fig. 5A ; band intensity quantifi cation using Image J software). As expected and in line with our previous observations ( 15 ), LXR ␣ was strongly recovered on sWGA beads in refed liver, whereas LXR ␤ was recovered under both fasted and refed conditions, suggestive of postprandial glucose-O-GlcNAc sensing via LXR ␣ . SREBP-1 was not recovered on sWGA beads, suggesting that SREBP-1 is not a target for O-GlcNAc modifi cation (data not shown). In ChIP assays, recruitment of ChREBP to its cognate DNA-binding site in the L-pk promoter was signifi cantly higher in refed control compared with fasted mice ( Fig. 5B ), consistent with hepatic L-pk mRNA expression ( Fig. 2B ). Interestingly, ChREBP was not recruited to L-pk in control livers of LXR Ϫ / Ϫ mice despite high ChREBP ␣ protein expression, suggesting that ChREBP ␤ is essential for ChREBP-dependent target gene activity in response to feeding in the presence of postprandial insulin signaling. Although ChREBP binding to the L-pk promoter was increased in the liver of refed STZ-treated mice ( Fig.  5B ), L-pk mRNA was only modestly increased ( Fig. 2B ), suggesting that the insulin signal is potentiating ChREBP ␣ and/or ChREBP ␤ transactivation on the L-pk promoter.

LXR interacts and colocalizes with OGT in vitro and in vivo
Prompted by our observations that LXRs are targets for O-GlcNAc modifi cation ( 15 ) and appear to potentiate nuclear O-GlcNAc signaling without affecting expression of GFAT1/2 , OGT , or OGA , we performed immunofl uorescence and coimmunoprecipitation experiments to determine a possible colocalization and/or interaction of LXRs and OGT. Ectopically expressed cherry-OGT and EGFP-LXR ␣ or EGFP-LXR ␤ colocalized in the nucleus of Huh7 cells treated with DMSO or GW3965, a synthetic LXR and RXR ␣ expression vector and/or LXR ␣ or LXR ␤ expression vectors or empty vector (pcDNA3.1) (n = 3). Dual luciferase assay was performed and lysates were immunoblotted with antibodies against FLAG, RXR ␣ , and ␤ -actin as loading control. A schematic fi gure of the ChREBP ␣ reporter construct (ChREBP ␣ -prom) including the two LXREs is shown. B: Huh7 cells maintained in 5 mM glucose were transfected with ChREBP ␣ -prom or pGL3-Basic with empty expression vector or RXR ␣ and LXR ␣ expression vectors. All transfections contained pRL-CMV. After 6 h, the cells were treated with 1 mM or 25 mM glucose for 24 h (n = 3). Dual luciferase assay was performed and lysates were subjected to SDS-PAGE and immunoblotted with an antibody detecting O-GlcNAc epitopes on proteins (CTD 110.6) with ␤ -actin as loading control. One representative Western blot is shown (n = 3). C: Huh7 cells maintained in 5 mM glucose (upper panel) and 25 mM glucose (lower panel) were transfected with 40 nM siRNA, nontargeting pool (siSrc), or human OGT (siOGT). After 24 h (25 mM glucose) and 48 h (5 mM glucose) the cells were further transfected with ChREBP ␣prom, pRL-CMV with empty vector, or RXR ␣ and LXR ␣ expression vectors. Dual luciferase assay was performed after another 24 h. Lysates were subjected to SDS-PAGE and immunoblotted with antibodies against OGT, O-GlcNAc (CTD 110.6), LXR ␣ , and RXR ␣ and ␤ -actin as loading control (n = 3). All values are given as mean ± SEM. Statistical differences are shown as * P < 0.05 or *** P < 0.001 between respective controls or as ### P < 0.001 between indicated groups .

DISCUSSION
LXRs regulate genes involved in hepatic fatty acid synthesis in response to synthetic and endogenous ligands and insulin ( 50 ). In addition to being a substrate for lipogenesis, glucose has emerged as a lipogenic signal ( 2 ) that activates LXR ( 13,15 ), ChREBP ( 22 ), and SREBP-1c ( 43 ). In contrast to SREBP-1c, ChREBP expression and activity is seemingly unaffected by LXR defi ciency in refed mice when both postprandial insulin and glucose signals are present ( 26 ). Pharmacological activation of LXR induces hepatic ChREBP expression ( 6,41 ); however, whether LXR regulates ChREBP and de novo lipogenic gene expression in vivo under physiological conditions has not been fi rmly established ( 26,27 ). agonist ( Fig. 6A ). However, colocalization of LXRs and OGT was independent of ligand treatment (GW3965) or glucose concentration in the culture medium (5 mM vs. 25 mM) (data not shown). By GST pulldown experiments, we show that OGT and LXRs also interacted in a bacterial GST-LXR ␣ / ␤ /His-OGT coexpressed system, suggesting a direct interaction between OGT and LXRs ( Fig. 6B ). Finally, using a reciprocal coimmunoprecipitation strategy, we observed coimmunoprecipitation of OGT and LXRs in Huh7 cells cotransfected with HA-OGT and FLAG-LXR ␣ or FLAG-LXR ␤ ( Fig. 6C ) and in the liver of fed wild-type mice ( Fig. 6D ). The interaction between OGT and LXRs in Huh7 cells was not affected by GW3965 (data not shown), suggesting that LXR interacts with OGT, indirectly or directly, via its N-terminal domain. compartments were immunoblotted with antibodies against OGT, GFAT1/2, and GFAT2. ␣ -Tubulin and lamin A were used as controls for Cyt and NE fractions, respectively. Each lane represents lysate from independent mice from each experimental group. A representative Western blot is shown (n = 2). C: Cytosolic and nuclear liver lysates were immunoblotted with anti-O-GlcNAc antibody (CTD110.6). ␣ -Tubulin and lamin A were used as controls for Cyt and NE fractions, respectively. One representative Western blot is shown (n = 2). Quantifi cation of total cytosolic and nuclear O-GlcNAc modifi ed proteins was analyzed by Image J. All values are given as mean ± SEM (n = 4-5). * P < 0.05, ** P < 0.01, *** P < 0.001 compared with fasted within experimental group. ## P < 0.01, ### P < 0.001 compared with refed between indicated groups. ns, not signifi cant.
to be mediated by SREBP-1, because its expression is dependent on LXR ( 5,47 ). Furthermore, we performed a liftover of rat ChREBP ChIP-Sequencing profi les to mouse DNA and designed ChIP primers corresponding to ChREBP binding peaks in rat Fas , Scd1 , and Acc promoters, but we were unable to detect ChREBP recruitment to the corresponding genomic regions in mouse liver extracts from control or STZ-treated wild-type or LXR Ϫ / Ϫ mice (Mandrup et al., unpublished observations ). Future studies designed to delineate the relative role of LXR, SREBP-1c, and ChREBP in response to different diets and glucose levels will be important to further address their cooperative or independent regulation of lipogenic genes.
In the present study, we show that Chrebp ␣ expression was blunted in STZ-treated LXR Ϫ / Ϫ mice, which is consistent with Chrebp ␣ being an LXR target gene in liver ( 6,13,27 ). However, ChREBP ␣ mRNA and protein levels were minimally affected by LXR defi ciency in refed control mice, suggesting that ChREBP ␣ expression becomes highly dependent on LXR in the absence of insulin. In line with these observations, we show that high glucose induced ChREBP ␣ promoter activity through LXR in vitro, an effect that was inhibited by OGT depletion. Guinez et al. ( 25 ), reported increased ChREBP protein expression , but To discriminate between insulin and dietary activation of LXR in vivo, we subjected LXR +/+ and LXR Ϫ / Ϫ control and STZ-treated mice to a fasting-refeeding regime. The mice were fed a diet containing no cholesterol, omitting effects of dietary oxysterols on LXR activity. We show that LXRs regulate essential genes in the glycolytic and fatty acid synthesis pathway; Gk , L-pk Srebp-1 , Chrebp ␣ / ␤ , Fas , Acc ␣ , Acc ␤ , Scd1 , and Gpat in response to feeding also in the absence of insulin. The refeeding-induction of these genes was blunted in LXR Ϫ / Ϫ control and hyperglycemic mice and correlated with lower protein levels of GK, L-PK, FAS, and SCD1 and serum TGs in LXR Ϫ / Ϫ mice, supporting the notion that LXRs are important glucose sensors of hepatic lipogenesis ( 13,15 ), as glycolysis is dominant in the fed state when glucose is abundant ( 42 ). Recent ChIP-Sequencing analyses of LXR and SREBP-1 binding profi les in mouse liver and ChREBP binding in human HepG2 cells have revealed that their chromatin binding occurs at many genes involved in lipid metabolism ( 39,41,51 ). Moreover, mice defi cient in LXR ( 5,52 ), ChREBP ( 53 ), or SREBP-1c ( 54 ) all exhibit reduced expression of lipogenic enzymes, indicating that they each have prominent roles in the lipogenic pathway. The expression levels of lipogenic genes observed in LXR Ϫ / Ϫ mice are not expected dependent on LXRs in both control and STZ-mice, which suggests a role for LXRs in regulating ChREBP ␣ DNA binding and/or its transactivating activity on the ChREBP ␤ promoter under both physiological and hyperglycemic conditions because ChREBP ␣ has been shown to regulate the ChREBP ␤ promoter in response to postprandial glucose levels ( 28 ).
ChREBP activity has been reported to be dependent on glucose metabolites downstream of GK ( 22 ). In our study, feeding induced ChREBP ␣ activity (increased ChREBP ␣specifi c target gene expression of Chrebp ␤ and L -pk and increased ChREBP promoter binding to the L-Pk promoter) in both control and STZ-treated mice, supporting increased GK-mediated G-6-P synthesis above fasting levels no increase in total Chrebp mRNA expression in liver and primary hepatocytes isolated from mice with adenoviral overexpression of OGT, arguing against a role for LXR in mediating a transcriptional glucose-response via O-GlcNAc on ChREBP ␣ . However, these experiments were performed in the presence of insulin, and relative expression of ChREBP ␣ and ChREBP ␤ isoforms were not investigated.
In agreement with earlier observations ( 28,55 ), feeding in both mouse models strongly induced mRNA expression of Chrebp ␤ , which is believed to have higher activity than ChREBP ␣ due to a lack of the N-terminal low glucose inhibitory domain and constitutive nuclear localization ( 28 ). Feeding-regulation of Chrebp ␤ mRNA expression was . Twenty-four hours after transfection, cells were treated for 2 h with DMSO or 1 M of GW3965. B: GST pulldown of bacterial lysates coexpressing His-OGT and GST, GST-LXR ␣ or GST-LXR ␤ . Precipitated proteins were subjected to SDS-PAGE and immunoblotted with antibodies against GST or OGT (AL28). C: Huh7 cells were cotransfected with HA-OGT and FLAG-LXR ␣ or FLAG-LXR ␤ plasmids and incubated for 24 h under high glucose conditions. Immunoprecipitations (IP) of OGT and FLAG-LXRs were blotted with anti-FLAG and anti-OGT antibodies. Mouse IgG IP was performed as control. Input lysates (Inp, 5%) and immunoprecipitated proteins were subjected to SDS-PAGE and blotted with anti-FLAG and anti-OGT antibodies. D: Liver lysates pooled from three fed wild-type (WT) and LXR Ϫ / Ϫ male mice were immunoprecipitated with anti-OGT or rabbit IgG antibody. Immunoprecipitated proteins were subject to SDS-PAGE and blotted with anti-OGT and anti-LXR antibodies. mice, cytosolic protein O-GlcNAc levels were not reduced, suggesting that the effect of LXR on nuclear O-GlcNAc levels is not due to decreased fl ux through the HBP. Instead, we believe this is due to LXR-dependent expression of a protein(s) modulating HBP fl ux or OGT activity and/ or that LXR may infl uence nuclear substrate specifi city of OGT, which has been suggested for other OGT-interacting proteins ( 57 ). In support of the latter hypothesis, we observed that OGT and LXR interact and colocalize in the nucleus. Because OGT has been shown to be an important coactivator of nutrient-responsive gene transcription ( 59-61 ), we hypothesize a role for LXR in regulating substrate specifi city of OGT on specifi c target gene promoters. OGT has previously been shown to target and activate ChREBP ␣ , leading to increased ChREBP protein stabilization and binding to the L-pk promoter concomitant with increased DNL gene expression ( 25 ). This model is in agreement with our results showing reduced nuclear O-GlcNAc modifi ed ChREBP, L-PK expression, and binding of ChREBP to the L-pk promoter in LXR Ϫ / Ϫ mice, suggesting that LXRs potentiate ChREBP activity via regulation of Gk expression and nuclear O-GlcNAc signaling.
In summary, we have found that LXRs act as postprandial nutrient and glucose metabolic sensors upstream of ChREBP in vivo based on the following observations: 1 ) LXRs do not require insulin for activation of Gk and lipogenic gene expression in response to refeeding, specifically, both SREBP-1c and ChREBP ␤ expression is completely dependent on LXRs under hyperglycemic insulin-independent conditions. 2 ) LXRs interact with and modify the activity of nuclear OGT, a sensor of glucose metabolic fl ux via the HBP. 3 ) Transactivation of the L-PK promoter by glucose sensing ChREBP ␣ is dependent on LXR. 4 ) In the absence of insulin, feeding-induced ChREBP ␣ expression is dependent on LXR (in vivo), in part via glucose-O-GlcNAc-dependent signaling (in vitro) under hyperglycemic and hypoinsulinemic conditions. Furthermore, feeding-induced ChREBP activity appeared to be dependent on LXRs, as refl ected by decreased mRNA expression of L-pk and Chrebp ␤ and no induction of ChREBP L-pk promoter binding in refed LXR Ϫ / Ϫ control mice despite high ChREBP ␣ protein expression (low Chrebp ␤ expression). These fi ndings are supported by reduced RNA polymerase II recruitment to L-pk in LXR Ϫ / Ϫ compared with LXR +/+ mice treated with pharmacological LXR agonists ( 41 ). In contrast to our results, one study reported that LXR is dispensable for ChREBP-activated DNL gene expression in high carbohydrate refed mice ( 26 ). However, in that study, the transcriptional level of Gk was unaffected by the loss of LXR, suggesting different responses due to different diets and/or LXR mouse models. We observed lower GK mRNA and protein levels in LXR Ϫ / Ϫ mice. The role of LXR in regulating GK is supported by reduced GK expression in LXR Ϫ / Ϫ mice ( 27,45 ), reduced glucose phosphorylation to G-6-P by GK in LXR ␣ Ϫ / Ϫ mice ( 49 ), elevated GK transcription in mice treated with pharmacological LXR agonists ( 56 ), and LXR binding to an LXRE in Gk promoter in rat liver extracts ( 7 ). Moreover, we report induced binding of LXR to a mouse LXRE in the GK promoter upon refeeding, with a similar trend in STZtreated mice. Elevated Gk transcription in STZ-treated mice may be mediated by SREBP-1c ( 43 ), and possibly other transcription factors or coactivator complexes recruited to LXR at the GK promoter. We have previously shown that binding of LXR to the LXRE in the SREBP-1c promoter was unaffected by LXR agonist treatment, whereas RNA polymerase II was recruited to the Srebp-1c gene upon LXR activation, supporting that LXR activation of target genes may be mediated by coregulator recruitment to some genes ( 41 ).
Surprisingly, we observed a strong reduction in nuclear O-GlcNAc signaling in the liver of control and STZ-treated refed LXR Ϫ / Ϫ mice. Despite lower GK expression in LXR Ϫ / Ϫ Fig. 7. LXR regulates ChREBP ␣ activity in response to feeding. LXR regulates the hepatic expression of lipogenic enzymes directly and through SREBP-1c expression and ChREBP ␣ activity in response to elevated postprandial glucose and insulin levels (Control). LXR maintains its ability to induce the expression of lipogenic target genes under hyperglycemic and hypoinsulinemic conditions (STZ), by regulating the expression of SREBP-1c and expression and activity of ChREBP ␣ .