Reduced de novo lipogenesis in adipose tissue, often observed in obese individuals, is thought to contribute to insulin resistance. Besides trapping excess glucose and providing for triglycerides and energy storage, endogenously synthesized lipids can function as potent signaling molecules. Indeed, several specific lipids and their molecular targets that mediate insulin sensitivity have been recently identified. Here, we report that carbohydrate-response element-binding protein (ChREBP), a transcriptional inducer of glucose use and de novo lipogenesis, controls the activity of the adipogenic master regulator peroxisome proliferator-activated receptor (PPAR)γ. Expression of constitutive-active ChREBP in precursor cells activated endogenous PPARγ and promoted adipocyte differentiation. Intriguingly, ChREBP-constitutive-active ChREBP expression induced PPARγ activity in a fatty acid synthase-dependent manner and by trans-activating the PPARγ ligand-binding domain. Reducing endogenous ChREBP activity by either small interfering RNA-mediated depletion, exposure to low-glucose concentrations, or expressing a dominant-negative ChREBP impaired differentiation. In adipocytes, ChREBP regulated the expression of PPARγ target genes, in particular those involved in thermogenesis, similar to synthetic PPARγ ligands. In summary, our data suggest that ChREBP controls the generation of endogenous fatty acid species that activate PPARγ. Thus, increasing ChREBP activity in adipose tissue by therapeutic interventions may promote insulin sensitivity through PPARγ.
Cellular processes that link blood glucose levels to glucose uptake and metabolism mediate insulin sensitivity. A central coordinator of glucose metabolism is carbohydrate-response element-binding protein (ChREBP) (also Max-like protein X (Mlx)-interacting protein-like or MondoB), a transcription factor activated by glucose metabolites (1–3). Upon activation ChREBP heterodimerizes with Mlx (4) and induces expression of genes primarily involved in glycolysis, de novo lipogenesis, and fatty acid desaturation in organs like liver and adipose tissue (5). Mice with global ChREBP deficiency are intolerant to simple carbohydrates and insulin resistant (6, 7), indicating that ChREBP-controlled glucose use and lipid synthesis are required for physiological glucose homeostasis. ChREBP activation also enhances the desaturation of fatty acids by inducing stearoyl-coenzyme A desaturase 1, alleviating the impairment of AKT (protein kinase B) phosphorylation by saturated fatty acid (8). In contrast, deficiency (9) or liver-specific depletion of ChREBP (10) in ob/ob mice improved hepatic steatosis and glucose tolerance, indicating that nutritional overstimulation of ChREBP also disturbs insulin sensitivity.
Adipose tissue is involved in insulin-dependent glucose disposal via expression and translocation of GLUT4, a glucose transporter that is down-regulated in insulin-resistant states (11). A recent study identified a novel ChREBP isoform that is expressed in adipose tissue and encodes a shortened protein termed ChREBPβ (7). This isoform is regulated by full-length ChREBP (ChREBPα) and lacks most (amino acids 1–177) of the N-terminal domain (amino acids 1–196) that is required for retaining ChREBPα in the cytosol and repressing its activity at low-glucose concentrations, resulting in constitutive activity of the truncated protein (7). Adipose ChREBPβ expression was found to correlate with insulin sensitivity in humans (7, 12, 13). The exact mechanism by which ChREBP mediates insulin sensitivity is currently under investigation. Recent findings suggest that ChREBP-driven de novo lipogenesis has functions far beyond providing fatty acids solely for increased triglyceride accumulation (14) that would increase adiposity and potentially worsen insulin resistance.
We hypothesized that ChREBP activation generates lipids with regulatory/signaling function in adipocytes. Indeed, we found that ChREBP induced peroxisome proliferator-activated receptor (PPAR)γ activity in precursor cells, a ligand-activated nuclear receptor that is pivotal to adipocyte differentiation and whose activation exhibits potent insulin sensitizing properties (15, 16). Accordingly, activation of endogenous ChREBP by high-glucose concentrations or ectopic expression of constitutive-active ChREBP (CA) enhanced adipogenesis, whereas ChREBP depletion or a dominant negative mutant impaired adipogenesis. This suggests that ChREBP in adipose tissue may affect insulin sensitivity by controlling the activity of PPARγ.
Materials and Methods
Cell culture and differentiation
3T3-L1 and C3H10T1/2 cells (American Type Culture Collection) were grown to confluence in DMEM without pyruvate containing 25mM glucose or Eagle’s Basal medium with 2mM L-glutamine, 1.5-g/L sodium bicarbonate, 5.6mM glucose, and Earle’s Balanced Salt Solution, respectively, both supplemented with 10% fetal bovine serum, and penicillin/streptomycin (Life Technologies). Differing glucose concentrations during differentiation were used as indicated. Cells were induced to differentiate with insulin, dexamethasone and isobutylmethylxanthine as described previously (17). Cells were considered differentiated after 7 days and with at least 90% adipocyte conversion. To induce submaximal differentiation, cells were incubated with 3-fold lower concentrations of the inducers (suboptimal conditions). BOSC23 (18) and HEK293 cells were grown in DMEM containing 10% fetal bovine serum and penicillin/streptomycin. Oil Red-O staining of intracellular lipids in adipocytes was performed as previously described (19).
RNA isolation and quantitative PCR (qPCR)
RNA was isolated by spin column kits (QIAGEN). cDNA was generated using the MMLV-RT (Promega). qPCR was carried out by using SYBR Green PCR Mastermix (Roche) and evaluated using standard curves. mRNA expression was normalized to acidic ribosomal phosphoprotein 70 or hypoxanthine guanine phosphoribosyl transferase. The sequences of primers used in this study are listed in Supplemental Table 1.
Immunoblotting
Whole-cell proteins were prepared by standard methods and separated by SDS-PAGE. Protein concentrations were measured by the bicinchoninic acid method (Thermo Scientific). After incubation with antibodies for PPARγ (sc-7273; Santa Cruz Biotechnology, Inc), adipocyte fatty acid-binding protein 2 (aP2) (2120; Cell Signaling), CCAAT/enhancer-binding protein alpha (C/EBPα) (sc-61; Santa Cruz Biotechnology, Inc), ChREBP (NB400–135, lot J1 or M1 [Novus Biologicals] at 1:2000 dilutions in 0.05% BSA overnight, after blocking the polyvinylidene difluoride membrane with 1% skimmed milk for 1 h), or β-ACTIN (sc-47778; Santa Cruz Biotechnology, Inc), a secondary horseradish-conjugated antibody was added, and a chemiluminescent substrate kit (GE Healthcare) used for detection.
Plasmid cloning and viral infections
Mouse ChREBP cDNA for amino acids 1–864 (ChREBPα), 178–864 (ChREBPβ), 240–864 (ChREBP-CA) (20), or 1–864 with R673A/R674G mutations (ChREBP-DN) (21), kindly provided by Howard C. Towle (University of Minnesota, MN), were inserted downstream of a Kozak sequence into pMSCV (In-Fusion; Clontech) and verified by sequencing. Retroviral particles were generated in BOSC23 cells and used to infect preconfluent 3T3-L1 or C3H10T1/2 cells as previously described (19). Adenoviruses expressing green fluorescent protein (GFP) or a mouse GFP-ChRBEP-CA fusion protein (20) were kindly provided by Susanne Mandrup (University of Southern Denmark, Odense, Denmark), amplified in HEK293 cells, and purified by cesium chloride gradient centrifugation. Titers were determined by the Adeno-X Rapid Titer kit (Clontech) and adipocytes infected as previously described (19) using 1.4 × 108 infectious particles per well of a 12-well plate.
Small interfering RNA (siRNA) knockdown
RNA interference-mediated silencing was performed by electroporating (Nucleofector; Lonza) cells with 3 nmol of siRNA oligonucleotides (Eurogentec) (Supplemental Table 1) as previously described (19).
Lipid synthesis
3T3-L1 adipocytes were incubated with 0.25-μCi/mL [1-14C]-acetic acid and 5mM glucose for 2 hours and harvested in ethanolic KOH. After saponification at 70°C, lipids were extracted by hexane and incorporated radioactivity measured by scintillation.
Reporter assays
3T3-L1 and HEK293 cells were transfected by electroporation or lipofectamine 2000 (Life Technologies) as described previously (19, 22). For determining ChREBP activity, HEK293 cells were transfected overnight with equal amounts of pMSCV plasmids encoding ChREBP isoforms together with an expression plasmid for hemagglutinin-Mlx (4), and a reporter that contained 2 copies of the acetyl-CoA carboxylase (ACC)1 carbohydrate-response element (ChoRE) in its promoter driving firefly luciferase expression (23). The next morning, cells were incubated in media containing 2.5mM glucose for 6 hours and then exposed to either 2.5mM or 25mM glucose for 24 hours before harvesting. ChoRE-, PPARγ-response element (PPRE)-, retinoid acid-response element (RARE)-, and galactose-inducible gene 4 (GAL4)-PPARγ-ligand-binding domain (LBD)/Gal4-DNA-binding domain-firefly luciferase reporter activities were normalized, if not indicated otherwise, to coexpressed Renilla luciferase (Dual Luciferase; Promega).
Animal studies
Animal procedures were in accordance with institutional guidelines and approved by the corresponding authorities. Male C57BL/6J mice were killed at an age of 3 months for analyzing tissue gene expression.
Ex vivo fat pads
A protocol for the isolation and cultivation of human adipose tissue explants (24) was modified for the preparation of murine adipose tissue as follows. Male C57BL/6J mice were killed, epididymal (epi) fat pads isolated, cut into small pieces of approximately 8 mm3, and kept in DMEM (5mM glucose) containing 0.5% BSA and penicillin/streptomycin. Fat pad pieces were infected in 230-μL DMEM with equal titers (3.8 × 108 infectious particles per fat pad) of adenoviruses expressing GFP or a CA-ChREBP-GFP fusion protein overnight. The next morning, fat pads were incubated in DMEM (5mM glucose), 10% fetal bovine serum, and penicillin/streptomycin; 72 hours after the infection, fat pads were subjected to fluorescence microscopy and mRNA isolation.
Statistical analysis
Representative results of at least 3 independent cell culture experiments are shown and results presented as mean ± SD. Significance was determined by the 2-tailed Student’s t test or ANOVA, as appropriate, and P < .05 was deemed significant (*, P < .05).
Results
Expression of ChREBP isoforms during in vitro adipocyte differentiation
We first analyzed ChREBP transcript levels of both isoforms in selected tissues of ad libitum-fed mice. As expected, total ChREBP mRNA levels were highest in liver, followed by sc and epi white adipose tissue (WAT). ChREBPβ mRNA was determined by amplifying part of the specific 5′-untranslated region by qPCR as previously described (7). It comprised approximately 12%, 33%, and 9% of the total tissue ChREBP mRNA, respectively (Figure 1A, estimated by cycle threshold values after confirming comparable amplification efficiencies using standard curves), showing that despite the highest absolute expression of ChREBPβ in liver, sc WAT exhibited the biggest contribution of this isoform to total ChREBP mRNA. To which extent ChREBPβ contributes to total ChREBP protein levels could not be determined, because we were unable to detect a distinguishable band for ChREBPβ protein by immunoblotting in either of these tissues. We then investigated ChREBP expression in adipocyte lineage-committed 3T3-L1 cells, a widely used model for adipocyte differentiation (25). Similar to the induction of genes like Pparγ2 and aP2 (26), ChREBPα mRNA increased more than 10-fold after 7–8 days of differentiation to adipocytes, which is consistent with previous reports (27, 28) and the observed induction during differentiation of human precursor cells (29), whereas ChREBPβ mRNA remained at low levels (Figure 1B). Consistently, the used ChREBP antibody showed only a band corresponding to ChREBPα protein in differentiated 3T3-L1 adipocytes (Figure 1C), migrating slower than expected for its calculated molecular weight. Compared with sc WAT, qPCR cycle threshold values for ChREBPα in mature 3T3-L1 adipocytes were 2–3 cycles higher, indicating that its expression was 4–8 times lower than in sc WAT. ChREBP’s heterodimeric partner Mlx is also present in 3T3-L1 cells and increases its expression during differentiation (microarray data in Ref. 30). Interestingly, pluripotent C3H10T1/2 cells that in contrast to 3T3-L1 cells are routinely cultured in Eagle’s Basal medium at low-glucose concentrations (5.6mM vs 25mM), differentiated to adipocyte without a major increase in ChREBPα mRNA expression. Switching to 25mM glucose during the 7–8 days of differentiation robustly increased ChREBPα expression in adipocytes (Figure 1D), whereas ChREBPβ expression remained very low in these cells (data not shown). In summary, ChREBPα is induced during adipocyte differentiation of 3T3-L1 cells and in C3H10T1/2 cells if differentiated in the presence of high-glucose concentrations, suggesting a regulatory function of ChREBP during differentiation.
Expression of ChREBP isoforms during adipocyte differentiation. A, Contribution of α- and β-isoforms to total ChREBP mRNA expression in liver, sc, and epi WAT of ad libitum-fed C57BL/6J mice. B and C, Undifferentiated and differentiated 3T3-L1 cells were analyzed for mRNA and protein levels of indicated genes. D, C3H10T1/2 cells were differentiated in cell culture media containing glucose concentrations as indicated. mRNA expression of ChREBPα before and after differentiation was determined by qPCR.
Expression and activity of ectopically expressed ChREBP isoforms
We generated pMSCV vectors encoding ChREBPα, ChREBPβ, and a previously described constitutive-active (ChREBP-CA) isoform that lacks the complete N-terminal low-glucose inhibitory domain (Figure 2A) (20). Equal amount of plasmid DNA transfected in HEK293 cells yielded comparable levels of total ChREBP mRNA, whereas ChREBPβ and ChREBP-CA protein levels were much lower than that of ChREBPα (Supplemental Figure 1, A and B, respectively). In order to validate functionality of the expressed isoforms, we determined ChREBP activity by coexpressing the heterodimeric partner Mlx and a luciferase reporter that is driven by ACC1 ChoREs (23). Ectopic ChREBPα expression conferred robust glucose sensing with an approximately 9-fold increase of luciferase activity upon exposure to 25mM glucose (Figure 2B). ChREBPβ was constitutively active, inducing luciferase activity even at 2.5mM glucose. ChREBP-CA showed the strongest constitutive activity, more than 10-fold higher than that of ChREBPβ (Figure 2B). Residual glucose responsiveness in ChREBPβ and ChREBP-CA-expressing cells may be due to the presence of endogenous sensing proteins in HEK293 cells. The rather limited maximal activity of ChREBPβ was surprising considering previous results (7). Experimental differences, such as the expression of untagged vs N-terminally flag-tagged ChREBP, may be responsible. However, normalizing luciferase activity to ChREBP protein levels instead of the activity of coexpressed Renilla luciferase showed indeed a much higher activity of ChREBPβ compared with ChREBPα (Supplemental Figure 1C). A reason for the lower maximal activity of ChREBPβ could be that the N terminus of ChREBPβ, but not that of ChREBP-CA, contains 19 amino acids that are part of the Mondo conserved region IV (amino acids 178–196, compare Figure 2A). Reduced protein levels of ectopically expressed β vs α have been reported earlier (7) and, although expressing similar mRNA levels, were also observed in 3T3-L1 preadipocytes that were infected by retroviral vectors prepared from these pMSCV plasmids (Figure 2C, left and right panels), suggesting decreased stability of the shorter isoforms. Indeed, both ChREBPβ and ChREBP-CA lack the N-terminal domain that allows interaction with 14–3-3 proteins, which is thought to control cytoplasmic sequestration and protein stability (31, 32). We then determined mRNA expression of several known ChREBP target genes (33) and found a significant induction only in 3T3-L1 cells expressing ChREBP-CA (Figure 2D). Taken together, we show that N-terminally truncated ChREBP isoforms express at lower protein levels, likely due to reduced protein stability, and that both ChREBPβ and ChREBP-CA are constitutively active ChREBP isoforms that differ in their maximal activity. However, only ectopic ChREBP-CA was potent enough to induce known ChREBP target genes in 3T3-L1 precursor cells under standard culture conditions.
Expression and activity of ectopically expressed ChREBP isoforms. A, Depicted protein structure of ChREBPα, ChREBPβ, or a known CA. B, pMSCV plasmids encoding these isoforms, together with a plasmid expressing Mlx and a ACC1-ChoRE luciferase reporter were transfected in HEK293 cells and 16 hours later incubated with cell culture media containing 2.5mM glucose for 6 hours before exposing cells to either 2.5mM or 25mM glucose for 24 hours as indicated. Cells were harvested and luciferase activity determined. C, pMSCV plasmid-derived retroviral particles were used to infect 3T3-L1 preadipocytes. Two days after reaching confluency (d 0), mRNA (left panel) and protein (right panel) expression of ChREBP isoforms was determined by qPCR and immunoblotting, respectively. D, mRNA expression of known ChREBP target genes was determined in cells described in C by qPCR. *, P < .05; #, P < .05 vs ChREBPα activity at low glucose.
ChREBP activation is proadipogenic
We next investigated whether ChREBP is functionally involved in adipocyte differentiation. We chose to study cells transduced with ectopic ChREBPα as the endogenously expressed isoform, and ChREBP-CA due to its potent activity in undifferentiated 3T3-L1 cells. Induction of differentiation by suboptimal conditions led to low differentiation in control cells, assessed by Oil Red-O staining and phase-contrast microscopy. Cells expressing ChREBPα appeared to differentiate slightly better, whereas cells expressing ChREBP-CA exhibited much enhanced adipocyte conversion (Figure 3A). Because enhanced lipid accumulation suggests increased differentiation, we analyzed typical marker genes for differentiation. Only cells induced with ChREBP-CA showed elevated expression of aP2 and Pparγ2 at different days of differentiation (Figure 3, B and C, left panel) and reduced expression of preadipocyte-specific preadipocyte factor 1 (Pref-1) (34) and Gata2 (35) (Figure 3C, right panel). This indicates that ectopic ChREBPα increases lipid accumulation without affecting differentiation, whereas ChREBP-CA enhances lipid accumulation as well as the expression of adipocyte-specific marker genes.
ChREBP activation enhances adipocyte differentiation. A, Preconfluent 3T3-L1 cells, infected with retroviruses expressing the indicated ChREBP isoforms, were induced to differentiate. Adipocytic conversion was assessed by Oil Red-O staining, phase-contrast microscopy at day 8, and (B) expression of adipocyte-specific aP2 and (C) Pparγ2 at day 2 and day 8 (left panel), and preadipocyte-specific Pref-1 and Gata2 at day 8 (right panel) determined by qPCR. D, 3T3-L1 cells were induced to differentiate in the presence of low- or high-glucose concentrations and analyzed as described above. E, Preconfluent C3H10T1/2 cells were infected with indicated retroviruses and induced to differentiate. mRNA expression of aP2, Pparγ2, Lpl, and Cd36 was determined at day 8. F, C3H10T1/2 cells were induced to differentiate in the presence of low- or high-glucose concentrations and analyzed as described above. *, P < .05.
Because ChREBPα is activated by glucose metabolites, we investigated whether glucose itself regulates adipogenesis. We found that differentiation of 3T3-L1 cells at 5mM glucose was low and improved tremendously at 25mM when determined by Oil Red-O staining, phase-contrast microscopy, and the expression of differentiation markers (Figure 3D), suggesting that glucose-dependent activation of ChREBP in combination with ChREBP-independent effects of high glucose, such as its function as an energetic/lipogenic substrate, strongly promote differentiation. The proadipogenic effects of both ChREBP-CA and high-glucose levels were reproducible in C3H10T1/2 cells (Figure 3, E and F, respectively). Interestingly, ectopic ChREBP-CA was more potent in enhancing differentiation of C3H10T1/2 cells than of 3T3-L1 cells, consistent with the low basal ChREBP expression and/or activity in these cells cultured at 5.6mM glucose. In summary, we found that expressing ChREBP-CA in precursor cells promoted adipogenesis in vitro in 2 established cellular models of adipocyte differentiation. Furthermore, activation of endogenous ChREBP may, at least in part, contribute to the proadipogenic effects of high-glucose levels.
ChREBP-CA induces PPARγ activity
DNL driven by the transcriptional regulator sterol-regulatory element-binding protein (Srebp) and its target genes fatty acid synthase (Fasn) has been linked to the generation of PPARγ-activating lipids (36, 37). We therefore tested whether ectopic ChREBPα and ChREBP-CA expression activate PPARγ. Indeed, PPRE-driven luciferase activity was increased in 3T3-L1 precursor cells that express ChREBP-CA, even without the addition of the differentiation cocktail (d 0) (Figure 4A). Inducing differentiation for 2 days by insulin, dexamethasone, and isobutylmethylxanthine amplified this increase and demonstrated PPARγ activation also for cells overexpressing ChREBPα (Figure 4A). The increase in PPARγ activity was unlikely due to nonspecific effects of ectopic ChREBP, because a similarly transfected RARE-driven luciferase reporter (17), responsive to all-trans retinoic acid (Supplemental Figure 2A), was not activated and rather reduced (Supplemental Figure 2B), likely due to the enhanced differentiation of ChREBP-CA-expressing cells and the therewith accompanied down-regulation of retinoic acid receptor expression (38). Interestingly, the increase in PPARγ activity upon ChREBPα expression (Figure 4A) was not sufficient for a significant enhancement of differentiation, as demonstrated in Figure 3, A–C. Taken together, these results show that expressing ChREBP-CA induces PPARγ activity, which may be responsible for its proadipogenic effects.
ChREBP activation induces PPARγ activity and ChREBP inhibition impairs differentiation. A, Preconfluent 3T3-L1 cells infected with retroviruses expressing indicated ChREBP isoforms were transfected with a PPRE-driven luciferase reporter before or 2 days after initiating differentiation and analyzed for activity of endogenous Pparγ. B, Preconfluent 3T3-L1 cells were electroporated with control or ChREBP siRNA, and ChREBP isoform expression was determined at day 0 by qPCR. Cells were then induced to differentiate for 7 days (C) and assessed for adipocyte conversion by Oil Red-O staining, phase-contrast microscopy, and protein levels of PPARγ, C/EBPα, and aP2. D, Preconfluent 3T3-L1 cells were infected with a retrovirus expressing DN and its expression validated by immunoblotting at day 0. E, Cells were induced to differentiate, and adipocyte conversion after 8 days was assessed by Oil Red-O staining, phase-contrast microscopy (left panel) and expression of adipocyte marker genes (right panel); *, P < .05.
Interfering with ChREBP activity partially inhibits adipocyte differentiation
We next depleted ChREBP by electroporation of siRNA into 3T3-L1 preadipocytes, leading to a reduction of the mRNA expression of both isoforms at day 0 (Figure 4B). Effectiveness of the used siRNA was further validated in similarly electroporated 3T3-L1 and C3H10T1/2 adipocytes that, in contrast to preadipocytes, allow visualization of ChREBP protein levels. siRNA-mediated depletion was observed at both ChREBPα mRNA and protein level 48 hours after electroporation (Supplemental Figure 3, A and B, left and right panels) and remained effective for at least 6 days (Supplemental Figure 3A, left and right panels). Adipocyte differentiation of preadipocytes electroporated with ChREBP siRNA resulted in reduced Oil Red-O staining (Figure 4C, left panel) and lower expression of adipocyte-specific proteins such as PPARγ, C/EBPα, and aP2 at day 7 (Figure 4C, right panel; densitometry shown in Supplemental Figure 3C). We further validated this finding by expressing a dominant-negative ChREBP (DN), unable to bind DNA (21), in 3T3-L1 precursor cells (Figure 4D). Of note, ChREBP-independent effects of the DN-mutant on differentiation by inhibiting MAX-like protein X cannot be excluded. Expression of ChREBP-DN reduced Oil Red-O staining and mRNA expression of marker genes such as Pparγ2, C/ebpα, aP2, lipoprotein lipase (Lpl), and Cd36 antigen (Cd36) after 8 days of differentiation (Figure 4E, left and right panels). In summary, ChREBP depletion by siRNA or expression of a DN-mutant in precursor cells reduced lipid accumulation and partially inhibited adipocyte differentiation.
ChREBP-CA in mature adipocytes induces PPARγ activity by involving FASN and the PPARγ-LBD
We then studied ChREBP activation in mature adipocytes by adenoviral expression of GFP or a GFP-ChREBP-CA fusion protein in fully differentiated 3T3-L1 cells. Overexpression was validated by fluorescence microscopy (Supplemental Figure 4A), measuring ChREBP protein levels (Figure 5A), and by the induction of known ChREBP target genes (Figure 5B). We found that the most responsive transcript analyzed was ChREBPβ (Figure 5B). ChREBP-CA induced gene expression changes in adipocytes translated into a modest increase in DNL by approximately 15% (Figure 5C), when assessed by the incorporation of 14C-acetate into extractable lipids. Intriguingly, ectopic ChREBP-CA substantially induced PPRE-driven luciferase activity and this induction was partially blocked by the specific FASN inhibitor C75 (Figure 5D) used at a concentration that was previously shown to be nontoxic to both 3T3-L1 preadipocytes and adipocytes (39). This suggests that ChREBP-driven PPARγ activation may depend on fatty acid synthesis. The reduction of PPARγ activity in C75 treated control adipocytes indicates that fatty acid synthesis is required also for basal PPARγ activity in these cells (Figure 5D). As in preadipocytes, the increase in endogenous PPARγ activity was unlikely due to nonspecific effects of ectopic ChREBP-CA, because RARE-driven luciferase activity was unaltered (Supplemental Figure 2C). We then asked whether the PPARγ-LBD would be responsive to ChREBP activation. Indeed, adipocytes expressing ChREBP-CA exhibited a several fold increased activity of a PPARγ-LBD-driven Gal4 reporter (Figure 5E). Activation of the PPARγ-LBD was blunted in the absence of the plasmid expressing the LBD, suggesting specificity, and additive with the activation elicited by the synthetic PPARγ agonist pioglitazone (Supplemental Figure 4, B and C). These data suggest that ChREBP-CA induces activation of the PPARγ-LBD in a fatty acid synthase -dependent manner.
CHREBP-CA expression induces PPARγ activity in adipocytes. A, Mature 3T3-L1 adipocytes were infected with adenoviruses expressing GFP or a GFP-ChREBP-CA fusion protein. Overexpression was validated by immunoblotting 48 hours later. B, mRNA was isolated and expression of known ChREBP target genes (left panel) and ChREBPβ (right panel) determined by qPCR. C, Similarly infected adipocytes were analyzed for 14C-acetate incorporation into extractable lipids. Mature 3T3-L1 adipocytes were electroporated with (D) a PPRE-driven luciferase reporter or (E) a PPARγ-LBD containing Gal4-luc reporter system, 6 hours later adenovirally infected overnight, then treated for 24 hours as indicated, and luciferase activity was determined. *, P < .05; #, P < .05 vs the corresponding vehicle treatment.
ChREBP regulates PPARγ target genes involved in thermogenesis in adipocytes and WAT
Full PPARγ agonists like the thiazolidinediones (TZDs) (40) rosiglitazone and pioglitazone up-regulate a large set of target genes in adipocytes (41, 42). Pioglitazone induced aP2, Olr1, uncoupling protein (Ucp)1 and Ucp3, and cell death-inducing DNA fragmentation factor α-like effector A (Cidea) expression in adipocytes (Figure 6A, left panel), as reported previously (42–45). If ectopic ChREBP expression results in altered PPARγ activity, then these target genes should be regulated accordingly. Indeed, ChREBP-CA expression in mature 3T3-L1 adipocytes induced Olr1 and genes involved in thermogenesis, whereas expression of aP2 was down-regulated (Figure 6A, right panel). We then addressed whether a depletion of endogenous ChREBP down-regulates these PPARγ target genes. Similar to the treatment with a known PPARγ antagonist PD068235 (46), ChREBP depletion by electroporating siRNA in adipocytes reduced both Ucp1 and Cidea expression (Figure 6B). When combined, pioglitazone and ChREBP-CA exhibited an additive induction of Cidea expression, whereas reduction of Cidea expression upon ChREBP depletion was partially rescued by pioglitazone (Supplemental Figure 5, A and B, respectively). To investigate PPARγ target gene induction in WAT, GFP or GFP-ChREBP-CA was overexpressed in isolated fat explants (24). Cultured fat explants have been shown to maintain gene expression and responsiveness to hormonal stimuli for up to 2 weeks (24); 72 hours after adenoviral infection, fat explants showed approximately 50% GFP-positive cells (Supplemental Figure 5B). ChREBP-CA overexpression elicited a robust increase of ChREBPβ and other ChREBP target genes (Figure 6C, left panels), suggesting that ex vivo expression of ChREBP-CA in WAT is functional. Strikingly, also Ucp1, Ucp3, and Cidea expressions were induced (Figure 6D, right panel), showing that the induction of PPARγ target genes involved in thermogenesis by ectopic ChREBP-CA expression is not restricted to 3T3-L1 adipocytes.
ChREBP regulates PPARγ target genes in mature adipocytes. A, 3T3-L1 adipocytes (d 7) were incubated with 1μM pioglitazone for 48 hours or infected with adenoviruses expressing GFP or a GFP-ChREBP-CA fusion protein for the same duration and analyzed for expression of selected PPARγ target genes by qPCR. B, 3T3-L1 adipocytes were incubated with 50μM of the PPARγ antagonist PD068235 or electroporated with the indicated siRNAs and analyzed for the expression of selected PPARγ target genes and ChREBPα 48 hours later by qPCR. C, epi WAT-derived fat pad explants were infected with indicated adenoviruses ex vivo and investigated after 72 hours. mRNA was isolated and analyzed for expression of ChREBP and PPARγ target genes by qPCR. *, P < .05.
Discussion
Here, we report that ChREBP activation induces PPARγ activity and promotes adipocyte differentiation. We found that ectopic ChREBP-CA expression was associated with increased transcriptional activity of PPARγ by activating its LBD, and that ChREBP-CA’s effect on PPARγ activity was reduced in the presence of a FAS inhibitor. This could suggest that ChREBP controls the generation of PPARγ-activating lipid species that, in turn, promote adipocyte differentiation, which adds to the notion of DNL-derived lipids with signaling function (14, 47, 48). Whether this lipid species could be a ligand for PPARγ, or enhancing PPARγ transactivation by other mechanisms such as posttranslational modifications, are crucial questions that remain to be determined. The observed additive effects of ChREBP-CA expression with pioglitazone in regard to PPARγ-LBD activation and target gene expression argue against the generation of a ligand that competes with TZD for binding and favor the latter hypothesis. However, noncompetitive binding of 2 chemically different ligands to the PPARγ LBD simultaneously, resulting in its additive activation, has been shown (49). Other mechanisms besides generating lipids that act via PPARγ, such as ChREBP-driven posttranslational modifications of PPARγ, could also explain our observations. Importantly, PPARγ activation in vivo induces insulin sensitization (15), which could account for previously observed correlations between adipose ChREBP expression/activity and insulin sensitivity. Impairing ChREBP may reduce the generation of PPARγ activating lipids, adipocyte differentiation, and insulin sensitivity. This is in accordance with reduced fat mass and pronounced insulin resistance in ChREBP-deficient mice (6, 7), which is unlikely caused by decreased availability of lipids for triglyceride formation and storage, because total DNL (lipid incorporation of 3H from injected 3H2O) in WAT and liver of these mice were normal (7).
A variety of lipids have affinity for PPARγ but the identity of endogenous high-affinity ligands is still enigmatic (50). DNL controlled by SREBP and FASN were previously linked to PPARγ ligand generation (36, 37). Our findings also implicate the glucose sensor ChREBP in controlling the activation of PPARγ by mechanisms as discussed above. Intriguingly, SREBP and ChREBP partially overlap in their target genes and both induce Fasn expression (33). Thus, both insulin (via direct induction of PPARγ and via SREBP) and glucose (via ChREBP) could feed forward to drive adipocyte differentiation ensuring triglyceride storage during anabolic states. A limitation of this study is that we did not identify a PPARγ-activating lipid species that is generated in a ChREBP-CA-dependent manner and the precise mechanisms of PPARγ activation. Previously identified structures could be potential candidates (37, 50, 51). In addition, we studied ChREBP functions by using highly active ChREBP-CA in several assays, a mutant that is likely to induce the very maximum of possible ChREBP dynamics. Although ChREBP mutants have been proven useful tools for studying its molecular function (8), dissecting the physiological activation of ChREBP by glucose metabolites in regard to this new mechanism would be interesting. Also, yet unknown effects of ChREBP-driven DNL may contribute to enhanced adipogenesis, independently of PPARγ activation.
ChREBPβ was constitutively active and more potent than ChREBPα in activating a heterologous reporter when luciferase activity was normalized to protein levels, which is consistent with earlier results (7). Deleting another 62 amino acids of the N terminus of CHREBP resulted in an even higher activity of ChREBP-CA, potent enough to induce ChREBP target genes in 3T3-L1 preadipocytes. Both ChREBPβ and ChREBP-CA expressed ectopically showed lower protein levels than ChREBPα despite comparable mRNA levels in HEK293 and 3T3-L1 cells, pointing towards reduced stability. Despite its substantial contribution to total ChREBP mRNA expression in sc WAT, we found endogenous ChREBPβ only at very low levels in differentiating 3T3-L1 and C3H10T1/2 cells. However, it was the most highly induced transcript by ChREBP-CA in adipocytes and represents a sensitive readout for ChREBP activity. Future studies will help to further dissect the biological function of this isoform.
Ectopic ChREBP-CA expression in adipocytes or WAT induced PPARγ target genes like Ucp1 that promote browning of WAT, whereas both a PPARγ antagonist or depletion of endogenous ChREBP in mature adipocytes decreased most of these genes. Whether the induction by either TZD ligands or ChREBP-CA results in increased thermogenesis of in vitro cultivated adipocytes is not the focus of this study and would be hard to estimate, because basal UCP1 expression in these cells is very low. Furthermore, UCP1 requires additional adrenergic cues for its activation (52). Besides activating PPARγ by expressing ectopic ChREBP-CA, we cannot rule out the involvement of other pathways that target thermogenic gene expression. A potential mediator is fibroblast growth factor (FGF)21 (53), which is induced upon ChREBP activation in hepatocytes (54) and also in adipocytes (data not shown). Because PPARγ also controls FGF21 expression (55), and FGF21 itself enhances PPARγ activity by preventing its sumoylation (56), it is difficult to predict the contribution of each pathway to the observed induction of thermogenic genes. The down-regulation of the PPARγ target gene aP2 by ChREBP-CA was surprising and suggests that ectopic ChREBP-CA expression reduces PPARγ activation. However, ChREBP-CA may affect other transcriptional regulators of this gene that cause its down-regulation. Another mechanism could be that a generated lipid species activating PPARγ induces gene-specific activation or repression, a known property of a subgroup of synthetic ligands referred to as selective PPARγ modulators (41, 57). Interestingly, these selective modulators induce insulin sensitization in vivo without affecting fat mass (58) or even reducing it (41), which is in contrast to the typical weight gain and increase in fat mass associated with TZD ligands.
3T3-L1 cells undergo differentiation and form lipid droplets even in the absence of fetal bovine serum (59), suggesting that all triglyceride-incorporated fatty acids can be derived from DNL. Indeed, the expression of most enzymes involved in DNL is up-regulated during differentiation (30, 60). Many of these were found to promote in vitro adipocyte differentiation, because for instance the inhibition of cytosolic ACC activity by a biotin analog (61), reducing the activity of ATP citrate lyase by depleting microrchidia family CW-type zinc finger 2 (62), or the pharmacological inhibition or siRNA-mediated depletion of FASN (39, 63, 64) not only affected lipid accumulation, but also reduced the expression of differentiation markers. Notably, all of the above mentioned enzymes are transcriptionally induced by ChREBP (33), and the generation of PPARγ-activating lipid species may, at least in part, contribute to their proadipogenic function.
In summary, we show that ChREBP links DNL to PPARγ activity and adipocyte differentiation. In an unpublished observation, Nuotio-Antar et al (Nuotio-Antar, A.M., P. Naravat, M. Li, M. Schupp, M. Mohammad, S. Gerard, F. Zou, L. Chan, unpublished observations) independently addressed the metabolic consequences of adipose tissue-specific expression of a very similar ChREBP-CA mutant (lacking amino acids 1–196, all of Mondo conserved region I–IV) (65) in vivo. They provide evidence that adipose-specific ChREBP activation indeed confers insulin sensitization. PPARγ activation is likely to contribute to improved insulin sensitivity in these mice.
For related article see page 4020
Acknowledgments
We thank Susanne Mandrup (University of Southern Denmark, Odense), Catherine Postic (Institute Cochin, Paris), and Howard C. Towle (University of Minnesota, Minneapolis) for plasmids and helpful discussions. We also thank Lawrence Chan (Baylor College of Medicine, Houston) for valuable comments on the manuscript.
Author contributions: M.S. planned experiments, analyzed data, and wrote the manuscript; N.W., M.M., J.R., M.K., S.H., F.A.G., R.F., A.T., and I.G. planned and performed experiments and reviewed/edited the manuscript; and A.M.N.-A. provided intellectual input and reviewed/edited the manuscript.
This work was supported by the German Research Foundation (Emmy Noether Grant SCHU 2546/1–1), the Career Integration Grant from the European Union Grant CIG 291867, and project funding from the Einstein Foundation Berlin (A-2011–83), all to M.S.
Disclosure Summary: The authors have nothing to disclose.
Abbreviations
- ACC
acetyl-CoA carboxylase
- aP2
adipocyte fatty acid-binding protein 2
- C/EBPα
CCAAT/enhancer-binding protein alpha
- CA
constitutive-active ChREBP
- ChoRE
carbohydrate-response element
- ChREBP
carbohydrate-response element-binding protein
- Cidea
cell death-inducing DNA fragmentation factor α-like effector A
- DN
dominant-negative ChREBP
- epi
epididymal
- Fasn
fatty acid synthase
- FGF
fibroblast growth factor
- Gal4
galactose-inducible gene 4
- GFP
green fluorescent protein
- LBD
ligand-binding domain
- Mlx
Max-like protein X
- PPRE
PPARγ-response element
- qPCR
quantitative PCR
- RARE
retinoid acid-response element
- siRNA
small interfering RNA
- Srebp
sterol-regulatory element-binding protein
- TZD
thiazolidinedione
- Ucp
uncoupling protein
- WAT
white adipose tissue.
References
Author notes
N.W. and M.M. contributed equally to this work.
