Molecular Endocrinology, doi:10.1210/me.2005-0171
Molecular Endocrinology 20 (4): 809-830
Copyright © 2006 by The Endocrine Society
A Novel Partial Agonist of Peroxisome Proliferator-Activated Receptor-
(PPAR) Recruits PPAR-Coactivator-1
, Prevents Triglyceride Accumulation, and Potentiates Insulin Signaling in Vitro
Elke Burgermeister,
Astride Schnoebelen,
Angele Flament,
Jörg Benz,
Martine Stihle,
Bernard Gsell,
Arne Rufer,
Armin Ruf,
Bernd Kuhn,
Hans Peter Märki,
Jacques Mizrahi,
Elena Sebokova,
Eric Niesor and
Markus Meyer
Department of Vascular and Metabolic Diseases (E.B., A.S., A.F., J.M., E.S., E.N.), Department of Discovery Chemistry (J.B., M.S., B.G., A.R., A.R., B.K., H.P.M.), and Department of Exploratory Development (M.M.), Pharmaceuticals Division, Fa. Hoffmann-La Roche AG, CH-4070 Basel, Switzerland
Address all correspondence and requests for reprints to: Dr. Markus Meyer, Fa. Hoffmann-La Roche AG, Pharmaceuticals Division, Department of Exploratory Development (PDME), Building 15, Room 1.042A, Grenzacherstrasse 124, CH-4070 Basel, Switzerland. E-mail: markus.meyer{at}roche.com.
 |
ABSTRACT
|
|---|
Partial agonists of peroxisome proliferator-activated receptor-
(PPAR), also termed selective PPAR modulators, are expected to uncouple insulin sensitization from triglyceride (TG) storage in patients with type 2 diabetes mellitus. These agents shall thus avoid adverse effects, such as body weight gain, exerted by full agonists such as thiazolidinediones. In this context, we describe the identification and characterization of the isoquinoline derivative PA-082, a prototype of a novel class of non-thiazolidinedione partial PPAR ligands. In a cocrystal with PPAR it was bound within the ligand-binding pocket without direct contact to helix 12. The compound displayed partial agonism in biochemical and cell-based transactivation assays and caused preferential recruitment of PPAR-coactivator-1
(PGC1) to the receptor, a feature shared with other selective PPAR modulators. It antagonized rosiglitazone-driven transactivation and TG accumulation during de novo adipogenic differentiation of murine C3H10T1/2 mesenchymal stem cells. The latter effect was mimicked by overexpression of wild-type PGC1 but not its LXXLL-deficient mutant. Despite failing to promote TG loading, PA-082 induced mRNAs of genes encoding components of insulin signaling and adipogenic differentiation pathways. It potentiated glucose uptake and inhibited the negative cross-talk of TNF on protein kinase B (AKT) phosphorylation in mature adipocytes and HepG2 human hepatoma cells. PGC1 is a key regulator of energy expenditure and down-regulated in diabetics. We thus propose that selective recruitment of PGC1 to favorable PPAR-target genes provides a possible molecular mechanism whereby partial PPAR agonists dissociate TG accumulation from insulin signaling.
|
INTRODUCTION
|
|---|
PEROXISOME PROLIFERATOR-ACTIVATED RECEPTOR- (PPAR) (1) is a ligand-activated transcription factor of the nuclear receptor (NR) superfamily (1). PPAR forms heterodimers with the retinoid X receptor (RXR) that transactivate PPAR-responsive elements (PPREs) of target genes involved in insulin signaling, lipid/glucose metabolism, immune response, cell cycle, and differentiation of epithelial or mesenchymal cells. The receptor is selectively activated by physiological fatty acid derivatives, such as 15-deoxy-
12,14-prostaglandin J2, and by a panel of chemically diverse full agonists such as glitazars and thiazolidinediones (TZDs). The TZDs rosi- and pioglitazone are insulin-sensitizing drugs approved for therapy of type 2 diabetes mellitus in humans.
PPAR has been firmly validated as a master regulator of fat cell differentiation (adipogenesis) in vitro and in vivo (1, 2). Phenotypes of rare complete loss of function mutations in severely insulin-resistant lipodystrophic human individuals [e.g. dominant-negative point mutations in the activation function 2 (AF2)] and tissue-selective knockout mice emphasize the role of the receptor as indispensable for development and maintenance of a functional adipose tissue (3). Unfortunately, the positive relationship between PPAR activity and adipogenesis accounts for the unwanted adverse effect of weight gain during TZD application in already diabetic and obese individuals. Other serious side effects include formation of edemas, enhancing the incidence of congestive heart failure (4) and PPAR ligands as possible risk factors for carcinogenesis (5). In contrast to adipogenesis, the correlation between insulin sensitization and PPAR activity is bell shaped with a maximum at suboptimal activation of the receptor (3). For example, individuals with the rare gain of function Pro115Gln allele are obese and insulin resistant. On the other hand, both the common partial loss of function mutation Pro12Ala in humans and heterozygotic PPAR deficiency in knockout mice confer resistance to diet-induced obesity and improve insulin sensitivity. Thus, neither full agonism (obesity) nor full antagonism (lipodystrophy) results in optimal insulin sensitization.
This fact is exploited by partial agonists/antagonists, also termed selective PPAR modulators (SPPARMs). These agents are expected to preclude the development of adverse effects mentioned earlier, while maintaining antidiabetic efficacy (3, 6). Hence, the identification and preclinical development of SPPARMs with an improved safety profile were strongly encouraged over the last years (7). The SPPARM concept is essentially based on differential recruitment of certain cofactors, i.e. NR coactivators or corepressors (8), to the receptor, resulting in a tissue- and promoter-selective expression of a favorable panel of target genes (3, 6). Accordingly, coactivators were grouped into "beneficial" or "adverse" factors regarding their insulin-sensitizing [e.g. steroid receptor coactivator 1 (SRC1)] vs. proadipogenic [e.g. SRC2/transcription intermediary factor 2 (TIF2), DRIP205/TRAP220] actions in vitro and in vivo (2, 3, 4, 5, 6, 7, 8, 9). For example, knockout mice for TIF2 exhibit a lean phenotype and are resistant to diet-induced insulin resistance and obesity, whereas SRC1 null mice are susceptible (9). Mouse embryonic fibroblasts lacking the mediator complex subunit DRIP205/TRAP220 fail to undergo adipogenic differentiation (2). However, the exact molecular mechanism by which adipogenesis is uncoupled from insulin sensitization is not yet fully understood.
PPAR-coactivator-1 (PGC1) is a central regulator of energy expenditure, mitochondrial biogenesis, and oxidative phosphorylation (OXPHOS) (10). The pleiotropic functions of PGC1 in liver (e.g. gluconeogenesis), muscle (e.g. fiber switch), and brown adipose tissue (e.g. thermogenesis) have been thoroughly investigated and reviewed recently (11). PGC1 is up-regulated during ex vivo differentiation of human adipocytes (12) and by TZDs in murine 3T3-L1 adipocytes (13). PGC1 is down-regulated in muscle of insulin-resistant and diabetic patients (14, 15). Reduced expression of PGC1 has been observed as well in adipose tissue of morbidly obese subjects compared with lean controls (12) and of insulin-resistant individuals (16). Single nucleotide polymorphisms (17) and haplotypes (18) of the PGC1 locus influence development of insulin resistance and diabetes.
Knockout mice generated by two independent groups corroborated the essential function of PGC1 in energy homeostasis (19, 20). However, the impact of PGC1 deficiency on white adipose tissue (WAT) and insulin signaling was less clear and not devoid of contradictions.
In our study, we therefore further investigated the function of PGC1 in insulin signaling. We examined two cellular model systems in which PPAR1 and PGC1 are coexpressed: murine C3H10T1/2 mesenchymal stem cells before and after adipogenic differentiation and human hepatoma HepG2 cells. Herein we tested whether PPAR-effector functions can be diverted from proadipogenic and/or lipogenic toward insulin-sensitizing and/or energy expenditure pathways by ligand-driven recruitment of PGC1. In this context, we identified a novel SPPARM, the isoquinoline derivative PA-082, which was characterized in terms of its biochemical properties, binding mode, and cellular effects. We demonstrated that structurally diverse SPPARMs facilitate preferential interaction of PPAR with PGC1. In sum, our data supported the idea that this cofactor may be critical in uncoupling insulin signaling from triglyceride (TG) accumulation.
|
RESULTS
|
|---|
Identification and Characterization of a Novel Partial Isoquinoline PPAR Agonist PA-082
Based on an in silico screening approach of a PPAR agonist library, 1-(3,4-dimethoxy-benzyl)-6,7-dimethoxy-4-[4-(2-methoxy-phenyl)-piperidin-1-ylmethyl]-isoquinoline (PA-082) was identified as a lead hit for a novel class of partial PPAR ligands (Fig. 1A
) (21). To further characterize this compound, scintillation proximity receptor-binding assays (SPAs) were performed (Fig. 1B). PA-082 displaced [3H]GW2570 (GI262570/farglitazar), a non-TZD full PPAR agonist (22), from purified recombinant glutathione-S-transferase (GST)-ligand-binding domain (LBD)-PPAR fusion protein with a potency [inhibition constant (Ki) = 0.8 µM) similar to that of GW0072 (Ki = 0.5 µM), a reference partial PPAR agonist (23). The full agonist and TZD rosiglitazone bound to PPAR with a Ki of 1.2 µM under the same conditions (data not shown). PA-082, like rosiglitazone, was unable to displace radioligand binding to either PPAR or PPAR
(data not shown). To examine its activity in a cell-based assay, CV1 cells were transiently cotransfected with a GAL4-LBD-PPAR fusion construct and a GAL4-responsive luciferase reporter plasmid and were then incubated for 24 h with increasing concentrations of ligands. PA-082 (EC50 = 260 nM) and GW0072 (EC50 = 470 nM) activated PPAR-driven reporter gene expression with 40% of the efficacy obtained by rosiglitazone (EC50 = 250 nM) at a final concentration of 10–5 M (Fig. 1C). Both PA-082 and GW0072 competitively antagonized transactivation of the reporter gene driven by 100 nM rosiglitazone (Fig. 1D).
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 1. Biochemical Properties of PA-082, a Novel PPAR Partial Agonist
A, Chemical structure of PA-082: 1-(3,4-dimethoxy-benzyl)-6,7-dimethoxy-4-[4-(2-methoxy-phenyl)-piperidin-1-ylmethyl]-isoquinoline. B, Radioligand binding assay. Displacement curves of [3H]GW2570 from GST-PPAR-LBD protein by the partial agonists PA-082 and GW0072. Values are in counts/min. C and D, CV1 cells were cotransfected with GAL4-PPAR-LBD and GAL4-UAS-luc plasmids and stimulated with PPAR ligands for 24 h. C, Representative concentration-response curves of reporter gene transactivation by partial agonists compared with rosiglitazone. D, Competition of partial agonists against 100 nM rosiglitazone. Values are expressed in relative light units (RLU) of luciferase activity compared with vehicle-treated control cells. a.u., Arbitrary units; Rosi, rosiglitazone.
|
|
Preferential Recruitment of PGC1 to PPAR by PA-082
To examine the cofactor recruitment profile of PA-082, a cell-free fluorescence resonance energy transfer (FRET) system was established. Herein, ligand-dependent binding of biotinylated 25-mer peptides to GST-LBD-PPAR was measured, each peptide harboring one LXXLL motif of SRC1, TIF2, SRC3, or PGC1, respectively. Concentration-response curves (Fig. 2, A–D) demonstrated that rosiglitazone (EC50 PGC1/SRC1 = 75/60 nM) bound to SRC1 and PGC1 with equal affinity. In contrast, PA-082 (EC50 PGC1/SRC1 = 195/79 nM) and GW0072 (EC50 PGC1/SRC1 = 157/58 nM) recruited PGC1 with a 70–83% efficiency compared with rosiglitazone at a final concentration of 10–6 M but were less efficient (<30%) in binding to SRC1. A similar pattern was observed for recruitment of TIF2 and SRC3. The alanine mutant LXXAA of SRC1 served as a negative control and was not recruited to the receptor (data not shown). FRET values were then expressed as effect of peptide binding in percent compared with the maximal effective concentration of rosiglitazone (at 10–6 M) and summarized (Table 1). All tested full agonists exhibited little differential selectivity between the four peptides, as marked by the low (
1) effect ratios between PGC1/SRC1, PGC1/TIF2, and PGC1/SRC3. In contrast, all partial agonists tested were characterized by high (
1.5 to 6) effect ratios, reflecting preferential recruitment of PGC1 over cofactors of the p160 family. To test for PPAR antagonism, FRET assays were repeated with a corepressor peptide harboring the CoRNR-box motif LXLX2IIX3L of nuclear receptor corepressor 1 (NCoR1). When NCoR1 was loaded on GST-PPAR-LBD apo-receptor, the initial FRET signal was reduced in a concentration-dependent manner by the ligands reflecting its displacement from the receptor (Fig. 2E). GW0072 (EC50 = 480 nM) and PA-082 (EC50 = 610 nM) displaced NCoR1 with a similar 85–99% efficacy like rosiglitazone (EC50 = 210 nM). In a setting in which NCoR1 was detached from GST-PPAR by 1 µM rosiglitazone, the selective PPAR-antagonist GW9662 (EC50 = 276 nM) recruited NCoR1 back to the receptor in a dose-dependent manner (Fig. 2F). In contrast, the partial agonists were unable to counteract rosiglitazone-mediated release of NCoR1. These data demonstrated that PA-082, similar to GW0072 (23), is not a PPAR antagonist but rather a partial agonist with a "high PGC1/low SRCs/no NCOR1" cofactor recruitment profile.
To extend the study to full-length cofactor proteins, expression plasmids for SRC1, TIF2, and PGC1 were cotransfected with GAL4-PPAR-LBD and GAL4-responsive reporter plasmid. Luciferase activity was potentiated by rosiglitazone (
25-fold) in presence of all three cofactors (Fig. 3, A and B). SRC1 and TIF2 overexpression resulted in a strictly ligand-dependent increase of PPAR activity, whereas basal ligand-independent coactivation by PGC1 was clearly evident at low ligand concentrations. Interestingly, the partial agonism of PA-082 (Fig. 3, C and D) was restored above the level of the full agonist rosiglitazone in the presence of ectopically expressed wild-type PGC1 (50-fold) but not by SRC1 or TIF2 (10 to 15-fold). A similar quantitative effect was recorded for GW0072 (data not shown). The LXXLL-deficient PGC1 mutant L2A completely blunted the effects of all ligands investigated (Fig. 3, B and D), underlining the indispensable function of the L2 motif in the interaction of PGC1 with PPAR. Similar results were obtained using a heterodimer-driven reporter system in which expression plasmids for PPAR, RXR, and a reporter plasmid containing a trimer of the PPRE from the rat acyl-coenzyme A-oxidase (ACO) promoter were cotransfected (Fig. 3E). PGC1, but not the L2A mutant, enhanced the activity of partial agonists (2-fold) on this PPRE compared with cells transfected with empty vector. Dominant-negative (DN)-PPAR1, an AF2-truncated mutant used previously (24), completely abolished reporter activity driven by all ligands. These findings underlined the requirement of functional PPAR receptor for the genomic action of its ligands.
PA-082 Enhances Physical Interaction between PGC1 and PPAR
To test for PGC1/PPAR complex formation, human embryonic kidney (HEK)293 cells were transiently cotransfected with tagged full-length hemagglutinin (HA)-PGC1, green fluorescent protein (GFP)-PPAR, or GFP as control and were stimulated for 24 h with PPAR ligands (at 10–6 M). Coimmunoprecipitation (CoIP) was then performed in whole-cell lysates using anti-HA and anti-GFP monoclonal antibodies (Fig. 4A). In the sample from vehicle-treated control cells (lane 2) ligand-independent interaction of PPAR with PGC1 was detectable, whereas bead controls (lanes 5 and 6) remained clear. A ligand-dependent (2- to 3-fold) increase of receptor-cofactor complexes was observed in lysates from cells treated by the compounds (lanes 3 and 4). A summary from the relative OD values of bands in gels demonstrated that GW0072 and PA-082 were as effective as rosiglitazone in precipitating PPAR/PGC1 complexes (Fig. 4B). To visualize recruitment of endogenous p160 cofactors, GST-pull-down assays were undertaken. Whole-cell extracts from HEK293 cells were incubated for 2 h with recombinant GST or GST-PPAR-LBD fusion protein and 10 µM of the respective ligand in the reaction setup. Rosiglitazone increased pulldown of all three cofactors, whereas in the presence of GW0072, more PGC1 was precipitated than SRC1 and TIF2 (Fig. 4C). These results corroborated that structurally distinct SPPARMs enhance binding of PGC1 to PPAR and that their ability to transactivate promoters is restored to a certain extent by this cofactor.
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 4. PA-082 Promotes Direct Interaction of PPAR with PGC1
A, CoIP of HA-PGC1 and GFP-PPAR. HEK293 cells was cotransfected with full-length, tagged expression plasmids and stimulated with PPAR ligands (10–6 M). Representative Western blots upon reciprocal CoIP (left panel) and input cell lysates (right panel) are shown. The change factor of the OD in bands of gels compared with the vehicle-treated control is indicated. B, Quantitation of CoIP results from two independent experiments. Values are mean fold ± SD. *, P < 0.05 vs. vehicle. C, Pull down of endogenous p160 cofactors and PGC1 and from HEK293 whole-cell lysates with GST (left panel) or GST-PPAR-LBD in the absence and presence of 10 µM ligand (right panel). Changes in OD are indicated as in panel A. IB, Immunoblot; IP, immunoprecipitation; mAb, monoclonal antibody; Rosi, rosiglitazone; Usp, unspecific band; Veh, vehicle.
|
|
PA-082 Prevents TG Accumulation during de Novo Adipogenic Differentiation of Murine Mesenchymal C3H10T1/2 Stem Cells
Adipocytes are the major target cells of PPAR agonists in vitro and in vivo (1, 2, 3). We thus used murine C3H10T1/2 stem cells, a pluripotent mesenchymal cell line that can be committed by appropriate stimuli into the adipocyte, myocyte, or osteoblast lineage (25). Transient cotransfection of subconfluent fibroblast-like C3H10T1/2 cells with GAL4-PPAR-LBD and GAL4-reporter plasmids resulted in a ligand-dependent increase in luciferase activity (Fig. 5A). GW0072 (EC50 = 86 nM) and PA-082 (EC50 = 64 nM) achieved 45 and 52% efficacy of rosiglitazone (EC50 = 22 nM) at a final concentration of 10–6 M. These findings demonstrated that PA-082 acted as a partial agonist in this cellular system.
We then examined the expression levels of PGC1 during de novo adipogenic differentiation. Postconfluent C3H10T1/2 cells were incubated for 6 d with insulin (200 nM) and PPAR ligands (10–6 M) or were chemically differentiated with a cocktail of 3-isobutyl-1-methylxanthine (IBMX), dexamethasone, and insulin, as indicated in Materials and Methods. In Western blots of whole-cell lysates, expression of PPAR1, PPAR2, and PGC1 was detectable in undifferentiated cells, but was considerably increased upon adipogenic differentiation (Fig. 5B
). This pattern was similar to the one obtained with 3T3-L1 adipocytes. Western blots had been validated with a human/mouse-specific affinity-purified rabbit polyclonal antiserum against the C-terminal 777–797 amino acids (aa) of human PGC1 [which was later used for chromatin immunoprecipitation (ChIP)] and a polyclonal antiserum against the N-terminal 1–300 aa, as indicated in Materials and Methods. Quantitative RT-PCR (QPCR) with primers specific for PGC1 corroborated its expression in this cell system (data not shown).
After showing that PGC1 and PPAR are coexpressed during adipogenic conversion, we tested the effect of PPAR agonists on TG accumulation. Lipid loading was strongly induced over 6 d by rosiglitazone and GW2570, as evident by a 15- to 35-fold increase in TG content, whereas PA-082 and GW0072 failed to do so (<5-fold) (Fig. 6A). Partial agonists competitively antagonized TG accumulation (70% at 10–5 M) driven by 100 nM rosiglitazone (Fig. 6B). These results correlated with reduced Oil red O staining of lipid droplets in cells treated with partial agonists (Fig. 6C). Expression of mRNA encoding adipose fatty acid-binding protein (A-FABP/aP2), a major adipocyte marker gene, was up-regulated to a lesser extent by partial agonists (30- to 40-fold) than by rosiglitazone (100-fold), as measured by real-time QPCR (Fig. 6D). Similar data were obtained from 3T3-L1 cells (data not shown). To detect ligand-dependent binding of PPAR and PGC1 to the aP2 locus, ChIP experiments (Fig. 6E) were performed. We used a rabbit polyclonal antiserum against the N terminus of human/mouse PPAR1 followed by subsequent PCR amplification of the two adjacent PPREs in the enhancer of the murine aP2 gene (ARE6/7). Binding of PPAR to ARE6/7 was specifically pronounced in rosiglitazone-differentiated adipocytes, but not in cells treated with PA-082 (Fig. 6E, upper panel) or GW0072 (data not shown). ChIP repeated with a human/mouse-specific affinity-purified rabbit polyclonal antiserum against the C-terminal 777 to 797 aa of human PGC1 resulted in a similar pattern (Fig. 6E, middle panel). Accessibility of chromatin in the presence of rosiglitazone was visualized by use of an acetylhistone H4 antibody (Fig. 6E, lower panel). The PPRE (URE1) from the enhancer of the murine uncoupling protein-1 (UCP1) was not precipitated by either compound (data not shown). These findings underlined the reduced capacity of partial agonists for TG loading during adipogenic differentiation of C3H10T1/2 stem cells, but suggested an unexpected role of PGC1 at the aP2 promoter in this cell system.
View larger version (48K):
[in this window]
[in a new window]
|
Fig. 6. PA-082 Prevents TG Accumulation during de Novo Adipogenesis of C3H10T1/2 Cells
A–E, Differentiation was performed with PPAR ligands as in Fig. 5B. A, Concentration-response curves of TG accumulation. Values are mean fold increase of TG content ± SD compared with vehicle-treated control cells (n = 3). *, P < 0.05 vs. vehicle. B, Competitive antagonism of TG accumulation induced by rosiglitazone at a final concentration of 100 nM. Values are expressed as mean % of TG content ± SD compared with rosiglitazone-treated cells (n = 3). *, P < 0.05 vs. rosiglitazone alone. C, Oil red O staining of lipid droplets (scale bar = 50 µm). D, Expression of mouse aP2 mRNA determined by QPCR. Values were normalized to S12 and are mean fold increase of aP2 mRNA ± SD compared with vehicle-treated undifferentiated control cells (n = 3). *, P < 0.05 vs. vehicle. E, ChIP of mouse aP2 (ARE6/7) enhancer elements by antisera specific for PPAR, PGC1, or acetylhistone H4. ARE, Adipocyte regulatory element; Pre-, preadipocyte; Rosi, rosiglitazone; Veh, vehicle;
|
|
PA-082 Induces mRNA of Genes Involved in Insulin Signaling
Expression profiling of mRNA by QPCR revealed clusters of genes differentially regulated by full vs. partial agonists (Table 2). In C3H10T1/2 cells, de novo differentiated for 6 d into the adipocyte lineage as described earlier (Fig. 5), enzymes involved in TG reesterification and storage, phosphoenolpyruvate carboxykinase (PEPCK) and glycerol kinase (GK), were more strongly increased by rosiglitazone than by PA-082. Neither rosiglitazone nor PA-082 significantly up-regulated mRNAs of genes involved in fatty acid oxidation, such as carnitine palmitoyltransferase 1 (CPT1), energy expenditure (UCP1), or mitochondrial oxidative phosphorylation, as exemplified by estrogen-related receptor (ERR). Instead, the mRNAs encoding the lineage commitment transcription factors PPAR2 and CAAT/enhancer binding protein- (C/EBP) were induced both by rosiglitazone and PA-082. Both ligands also increased mRNAs of adapter proteins in the insulin signaling cascade, c-cbl associated protein (CAP), and insulin receptor substrate 2 (IRS2). This altered gene balance points toward a net bias toward lipogenesis in cells differentiated with the full agonist. In contrast, cells differentiated with the partial agonist exhibited lower TG load, while maintaining expression of mRNAs relevant for lineage commitment and insulin signaling.
Overexpression of PGC1 Reduces TG Load but Activates Expression of Genes Involved in Insulin Signaling
The ChIP data (Fig. 6E) indicated an unexpected function of PGC1 in the regulation of the aP2 gene. We therefore examined the effect of overexpression of wild-type PGC1 and its LXXLL-deficient mutant L2A on TG accumulation and gene expression during de novo differentiation of C3H10T1/2 cells. Subconfluent stem cells were transiently transfected with plasmids for 24 h, and then reseeded in full medium and grown to postconfluence for 48 h before start of differentiation. QPCR (Fig. 7A) and TG (Fig. 7B) analyses were performed 5 d later. One day after transfection a 20- to 40-fold overexpression of RNAs for HA-PGC1-WT and -L2A mutant was measurable in postconfluent cells (data not shown). The HA-PGC1 messages were still elevated by 2- to 4-fold 5 d later (Fig. 7A, top panel). WT-PGC1, but not L2A, enhanced both basal and ligand-dependent levels of aP2 and CAP mRNA, whereas UCP1 mRNA remained unchanged (Fig. 7A, lower panels). The absolute expression levels of aP2 and CAP mRNA remained lower in cells differentiated with PA-082 than with rosiglitazone. However, the relative potentiating effect of WT-PGC1 compared with cells transfected with empty vector was stronger in cells differentiated with PA-082 (2- to 5-fold) than with rosiglitazone (1.5-fold). These results indicate that PGC1 itself, and in synergy with PPAR ligand, positively regulates aP2 and CAP mRNA expression. In the same setting WT-PGC1 decreased rosiglitazone-induced TG accumulation by 50% compared with mock transfected cells or cells transfected with WT-PPAR, whereas the L2A mutant was ineffective (Fig. 7B). DN-PPAR prevented TG accumulation to an extent similar to that of PGC1. These results demonstrated that PGC1 overexpression or its recruitment by ligand prevent TG loading in the early commitment phase of adipogenic differentiation from C3H10T1/2 mesenchymal precursor cells without compromising gene expression relevant for insulin signaling (CAP) and fatty acid binding (A-FABP/aP2).
View larger version (29K):
[in this window]
[in a new window]
|
Fig. 7. Effects of Ectopic PGC1 on TG Accumulation and Gene Expression in C3H10T1/2 cells
A and B, Subconfluent cells were transiently transfected with PGC1-WT, PGC1-L2A, PPAR1-WT, or PPAR1-DN plasmids, respectively. Two days after transfection cells were differentiated with PPAR ligands and analyzed after an additional 5 d as follows. A, Gene expression. Values from QPCR were normalized to S12 and are expressed as mean fold increase of mRNA ± SEM compared with vehicle-treated, undifferentiated control cells transfected with empty vector (n = 4). *, P < 0.05 vs. empty vector control. B, TG content. Values are means in % ± SD of mock-transfected, differentiated control cells (n = 3). *, P < 0.05 vs. mock transfected cells. Rosi, rosiglitazone; Veh, vehicle; WT, wild type.
|
|
PA-082 Facilitates Glucose Uptake and Insulin Signaling in Mature Adipocytes
As a readout for functional insulin signaling, glucose uptake experiments were performed. C3H10T1/2 stem cells were subjected to a 6-d de novo differentiation by insulin and PPAR ligands, as described earlier (Fig. 5). After a 3-h starvation, cells were stimulated for 30 min with insulin in the presence of [14C]glucose. In this setting, partial agonists facilitated insulin-stimulated glucose uptake to a lesser extent (<2-fold) than full agonists (>5-fold) (data not shown). When cells underwent de novo differentiation in the presence of increasing amounts of partial agonists on a background of 100 nM rosiglitazone, GW0072 and PA-082 competitively antagonized (80% at 10–5 M) rosiglitazone-mediated glucose uptake (Fig. 8A). These data indicated that partial agonists somehow fail to fully promote glucose uptake pathways compared with rosiglitazone during adipogenic differentiation. To examine glucose uptake into mature adipocytes after they had undergone differentiation independently of PPAR ligand, C3H10T1/2 cells were first chemically differentiated with a cocktail of IBMX, insulin, and dexamethasone and were then incubated with PPAR ligands for an additional 3 d. Here, partial and full agonists evoked a comparable concentration-dependent increase in basal (120–140% at 10–6 M) and insulin-stimulated (200% at 10–6 M) glucose uptake compared with vehicle-treated differentiated control cells (Fig. 8, B and C). These results indicated that in mature adipocytes, but not in undifferentiated stem cells, partial agonists are equivalent to full agonists in terms of facilitating glucose uptake.
View larger version (53K):
[in this window]
[in a new window]
|
Fig. 8. Facilitation of Glucose Uptake by PA-082 into Mature C3H10T1/2 Adipocytes
A, Competitive antagonism of rosiglitazone-induced glucose uptake during de novo adipogenesis. Postconfluent cells were differentiated for 6 d with 100 nM rosiglitazone plus the indicated concentrations of partial agonists, starved and stimulated with insulin and [14C]glucose for 30 min. Values are expressed as mean fold increase of cpm ± SD compared with cells differentiated with 100 nM rosiglitazone alone (n = 3). *, P < 0.05 vs. rosiglitazone alone. B–D, Glucose uptake into mature adipocytes. Cells were chemically differentiated with IBMX/Dex/insulin as in Fig. 5B and were incubated for an additional 3 d with PPAR-agonists in the absence (panels B and C) or presence (panel D) of TNF (50 ng/ml). B, Concentration-response curves. Values are expressed as mean fold increase in counts per min ± SD compared with vehicle-treated, insulin-stimulated mature adipocytes (n = 3). **, P < 0.01 vs. vehicle. C and D, Glucose uptake at a final ligand concentration of 10–6 M. Values are expressed as mean uptake in % ± SD compared with vehicle-treated, unstimulated mature adipocytes (n = 3). *, P < 0.05 vs. vehicle. Veh, Vehicle; Rosi, rosiglitazone.
|
|
PPAR ligands have been shown to counteract the negative cross-talk of TNF on insulin signaling in adipocytes (26). Thus, mature adipocytes were coincubated for 3 d with ligands (at 10–6 M) and TNF (50 ng/ml). PA-082 was more efficacious than rosiglitazone in opposing TNF-mediated inhibition of insulin-stimulated glucose uptake (Fig. 8D). Similar data were obtained from GW0072 (data not shown). To further study this negative cross-talk, we measured phosphorylation levels of AKT as a marker of functional insulin receptor signaling. Mature chemically differentiated C3H10T1/2 adipocytes were cultured for 3 d in the presence of ligands (at 10–6 M) and TNF (50 ng/ml), starved, and then stimulated with insulin for 10 min. Western blotting was performed using cytosolic lysates. None of the compounds altered phospho-AKT levels in the absence of insulin (data not shown). Other nongenomic effects, such as phosphorylation of ERKs, were undetectable at a maximal ligand concentration of 10–6 M. GW0072 (Fig. 9A, lane 4) potentiated insulin-dependent phosphorylation of AKT to a greater extent (2-fold) than pioglitazone (Fig. 9A, lane 3) and rosiglitazone (Fig. 9A, lane 5). TNF considerably decreased insulin-dependent AKT phosphorylation in vehicle-treated cells (Fig. 9B, lane 3). PA-082 potentiated (Fig. 9B, lane 5) and, in the presence of TNF, restored (Fig. 9B, lane 6; 5-fold of lane 3) AKT phosphorylation back to the level of insulin-stimulated control cells (Fig. 9B, lane 2). In contrast, rosiglitazone was less effective in potentiating the insulin effect (Fig. 9C, lane 2) and in opposing TNF (Fig. 9C, lane 3) than GW0072 (Fig. 9C, lanes 5 and 6).
Partial PPAR Agonists Oppose Inhibition of Insulin Signaling by TNF in HepG2 Human Hepatoma Cells
To examine the effects of PA-082 in a different cellular context, where PPARs are relevant players, we chose the human hepatoma cell line HepG2 (26). Expression of PGC1 and PPAR1 was detectable in whole-cell lysates, but was not further induced during a 3-d incubation with PPAR ligands (at 10–6 M) (Fig. 10A). Reporter gene expression driven by the GAL4-PPAR-LBD was then determined in transiently transfected HepG2 cells. As in C3H10T1/2 cells, GW0072 (EC50 = 297 nM) and PA-082 (EC50 = 73 nM) were partial agonists in this cell system with 42% and 18% efficacy compared with rosiglitazone (EC50 = 190 nM) at a final concentration of 10–6 M (Fig. 10B). HepG2 cells showed a robust concentration-dependent increase in insulin-stimulated AKT-phosphorylation, whereas titration of TNF against a constant amount of insulin (30 nM) resulted in a progressive reduction of this response (Fig. 10, C and D). HepG2 cells were then incubated for 3 d with PPAR ligands (at 10–6 M) and increasing amounts of TNF, starved, and then stimulated for 10 min with insulin (Fig. 10E). Here, inhibition of insulin-stimulated AKT-phosphorylation by TNF was prevented by rosiglitazone and GW0072 to a similar extent. Preincubation with PPAR ligands for less than 6 h was inefficient to establish and counteract the negative cross-talk of TNF on insulin-stimulated AKT-phosphorylation (data not shown). These data, together with those obtained from DN-PPAR (Fig. 3E), underscored that nongenomic effects were not relevant at ligand concentrations used in our study (<10–5 M).
View larger version (34K):
[in this window]
[in a new window]
|
Fig. 10. Partial PPAR Agonists Oppose TNF-Mediated Inhibition of Insulin Signaling in HepG2 cells
A, Expression of PGC1. Subconfluent HepG2 cells were treated for 3 d with PPAR ligands (10–6 M). Whole-cell lysates were subjected to Western blotting. B, Partial agonism of PA-082. Subconfluent HepG2 cells were transiently transfected with GAL4-PPAR-LBD and GAL4-UAS-luc plasmids. Ligand-dependent increase in luciferase activity was measured 48 h after transfection. Values are expressed as in Fig. 1. C–E, Insulin signaling. Subconfluent cells were treated for 3 d with PPAR ligands (10–6 M) in absence or presence of TNF, starved, and then stimulated with insulin for 10 min. Western blotting was done with cytosolic lysates. C, Concentration-response curve of insulin. Changes in OD of phospho-AKT were normalized to general-AKT or -IRS1 and are plotted as mean fold ± SD. compared with vehicle-treated controls (n = 3). ***, P < 0.001 vs. vehicle. D, Representative Western blot. E, Concentration response curves of TNF alone and TNF plus PPAR ligands on insulin-stimulated AKT phosphorylation. Values are mean % ± SD compared with vehicle-treated controls (n = 3). **, P < 0.01 ligand plus TNF (10–7 g/ml) vs. vehicle plus TNF (10–7 g/ml). F, Gene expression profile. Cells were treated as in panel A, and total RNA was subjected to QPCR. Values were normalized to S12 and are expressed as mean fold increase of mRNA ± SEM compared with vehicle-treated control cells (n = 3). *, P < 0.05 vs. vehicle. Rosi, Rosiglitazone; Usp, unspecific band; Veh, vehicle.
|
|
Instead, QPCR was performed on RNA from ligand-treated HepG2 cells. Similar to C3H10T1/2 cells, CAP mRNA was up-regulated 3-fold by PA-082 and to a lesser 1.5-fold by rosiglitazone (Fig. 10F). Similar results (2-fold for CAP) were recorded with GW0072 (data not shown). As for the protein level, mRNAs for PPAR1 and PGC1 were not increased. However, the liver-specific form L-FABP was up-regulated 2- to 3-fold by rosiglitazone and PA-082, respectively. As in C3H10T1/2 cells, ectopic expression of PGC1-WT in HepG2 cells potentiated basal and ligand-dependent expression of CAP mRNA by 2-fold compared with cells transfected with empty vector (data not shown). Taken together, these findings indicated that CAP and tissue-specific FABPs (L-,A-/aP2) may represent candidate genes that account for the potency of SPPARMs being equally efficacious compared with full agonists in opposing TNF-mediated inhibition of insulin signaling and glucose uptake in both adipocytes and hepatoma cells.
Crystallographic Structure of PPAR with PA-082 Reveals Partial Agonist Binding Mode
To understand the activation of PPAR by PA-082 on a structural level, we determined its crystal structure at 2.0 Å resolution as a ternary complex with a coactivator peptide from SRC1 (Fig. 11A). The structure was solved by use of an existing in-house structure of the PPAR-LBD (Table 3). Clear difference Fo-Fc electron density showing all details of the ligand was observed after rigid body refinement. The overall fold is virtually identical to previously determined cocrystal structures or apo structures (27, 28, 29, 30, 31). Superpositions of all atoms result in an root mean SD for the rosiglitazone structure (2PRG) of 0.74 Å, the apo structure (1PRG) of 1.07 Å, and for the 2-BABA structure (1WM0) of 1.04 Å.
Interestingly, the largest differences can be observed in the position of the helix 10/11, helix 12, and the loop between those two helices, but the key residues within this area identified as possible hydrogen bonding partners for an agonist such as Tyr473 show only minor deviations. In the structure, the binding site of the coactivator peptide of SRC1 on the surface of the PPAR-LBD is identical to that of other PPAR-LBD structures (27, 28, 29, 30, 31). The peptide is also involved in the packing of the crystal, and two hydrogen bonds are formed to a symmetry-related molecule between Thr638 and Ser208, and Thr640 to the backbone carbonyl oxygen of Lys422. As described above, full agonists bind either directly to residues of helix 12 (27) or can bind more distantly as seen in the case of the 2-BABA compounds (28). As expected, PA-082 does not show any direct interaction with helix 12. The ligand is bound in an extended S-shaped conformation in the part of the binding pocket that is formed by residues of helices 3, 5, and 7. PA-082 occupies a similar area of the binding pocket as GW0072 (23) (Fig. 11, B andC). All direct interactions of the ligand with the protein are basically hydrophobic. With respect to the position of the residues Tyr473, His449, and His323, helix 12 is in an agonist conformation. Fo-Fc difference in electron density was observed in close proximity to these residues after rigid body refinement, which could be adequately described by a glycerol molecule that was present as an additive in the crystallization buffer. In the structure it forms weak hydrogen bonds to His323, His449, Tyr327, Tyr473, and to Gln286. Distant from helix 12, another hydrogen bond is formed to the nitrogen of the isoquinoline ring of the PA-082. Further interactions of the isoquinoline ring with the protein include residues Cys285, Leu330, Val339, and Met364. The methoxy-phenyl ring points into a pocket that is lined by residues Phe226, Met329, Leu330, and Leu333. The side chain of Arg286, which appears to be very flexible in all structures, moves into the solvent region and opens space for the piperidine ring of the ligand. The large dimethoxy-phenyl ring reaches into the pocket below helix 5 and is sandwiched between Phe363 and Cys285. To accommodate this large moiety, Phe363 has to move compared with other known PPAR structures and orients itself parallel to the phenyl ring of the ligand.
|
DISCUSSION
|
|---|
In this study, we introduced the novel selective PPAR modulator PA-082. The biochemical, cell-based, and crystallographic data confirmed that this isoquinoline derivative acts as a bona fide partial PPAR agonist, which preferentially recruits PGC1, a key regulator of energy homeostasis (10, 11). Using structurally unrelated SPPARMs, we further collected evidence for a common theme of PGC1 recruitment by these compounds and describe a function for this cofactor in uncoupling insulin signaling from TG accumulation in vitro.
PA-082 functionally antagonized rosiglitazone-driven transactivation in CV1 cells and de novo adipogenesis and glucose uptake in C3H10T1/2 cells. However, FRET assays with a CoRNR-box peptide from NCoR1 clearly pointed out that PA-082, as reported for GW0072 (23), is not acting as a partial PPAR antagonist. Thus, its properties are unlikely to be related to corepressor recruitment, such as described for the irreversible PPAR antagonist GW9662 (32). This conclusion was supported by x-ray diffraction of the cocrystal of PA-082 with the LBD of PPAR. PA-082 binds, as predicted for a partial agonist by other published structures (23), in a stretched conformation into the hydrophobic cleft of the PPAR-LBD without direct contact to helix 12 (AF2). The compound induced a holo-receptor conformation with a proper formation of the charge clamp between helix 3, helix 12, and the SRC1 coactivator LXXLL-peptide, as described for full agonists (27). In contrast to SRC1, PGC1 forms multiple contacts with the DNA-binding domain (DBD) and LBD of many NRs and with other transcription factors (myocyte enhancer factor 2C, NRF1) and coactivators [cAMP response element binding protein (CREB)-binding protein, SRC1] (11). This interface comprises three LXXLL motifs (L1, L2, and L3) and areas in the N-terminal and central part of PGC1. L2 specifies binding to PPAR (33), whereas L2 and L3 bind to ERR (34), an important effector downstream of PGC1-mediated coactivation toward the energy expenditure/OXPHOS pathway in muscle and brown adipose tissue (BAT) (11). These extensive surface contacts account for the ability of PGC1 to coactivate NRs independent of ligand, an effect that was also evident in our FRET, CoIP, and transactivation assays. However, a recent cocrystal structure of ERR with a peptide harboring the L3 motif of PGC1 underscores that the latter is sufficient to induce the conformational change which establishes the canonical charge clamp between the LBD and the -helical cofactor peptide (35). Our FRET, QPCR, and transactivation studies also demonstrated that the L2 motif (aa 143–148, LLKKLL) of PGC1 was fully sufficient for PPAR coactivation and that its mutation to alanine (L2A) abrogated its function. Thus, to gain more detailed insight into the structural basis for preferential recruitment of this cofactor by SPPARMs, cocrystallization of PA-082 with the PPAR-LBD and peptides or subdomains of PGC1 must be envisioned in the future.
Interestingly, preference for PGC1 over p160 SRC cofactors (SRC1, TIF2, SRC3) was a characteristic of several structurally unrelated partial agonists such as GW0072 (23), FMOC-L-Leu (3, 9), and PA-082. Their partial activities were restored to the level of a full agonist such as rosiglitazone in the presence of the LXXLL-peptide (L2) of PGC1 in FRET assays and in cells upon ectopic expression of the full-length cofactor protein. This phenomenon was described previously for the PAT5A compound (31) and recently by us (36) for the sartan class of hypotensive drugs. The latter are bifunctional agents that act as dual partial PPAR agonists and angiotensin II receptor antagonists (37). In contrast, all tested full TZD and non-TZD agonists, such as -selective rosiglitazone (7) or /-dual selective GW2570 (farglitazar) (22), AZ242 (tesaglitazar) (38), and BMS-298585 (muraglitazar) (39), did not exhibit this preference. Instead, they bound to all four cofactor peptides with similar efficacy, GW2570 with a slight bias toward TIF2, a property possibly related to its strong proadipogenic activity in C3H10T1/2 cells. Interestingly, the dual / full agonist KRP-297 (40) and -selective pioglitazone (36) displayed a pattern of cofactor recruitment similar to that of SPPARMs.
PGC1 is a unique cofactor with respect to its inducibility by stress, cold, and fasting and its RNA-processing domains (10, 11). Thus, the intriguing question following our observations is whether there is any causal relationship between preferential recruitment of PGC1 by a compound and an altered downstream effector spectrum of PPAR activation in vitro and in vivo. If yes, would PGC1 be able to redirect anabolic, i.e. proadipogenic/lipogenic pathways toward a gene expression profile of enhanced insulin sensitivity and/or energy expenditure, e.g. the ERR/OXPHOS program. This may eventually help to explain differential clinical benefits of pioglitazone compared with rosiglitazone (7). To test these hypotheses in vitro, we applied two independent cell systems in which PGC1 and PPAR are coexpressed, murine C3H10T1/2 mesenchymal stem cells and their commitment to the adipocyte lineage and the human hepatoma cell line HepG2.
As a bona fide SPPARM, PA-082 effectively antagonized rosiglitazone-driven TG accumulation during de novo adipogenic differentiation of C3H10T1/2 stem cells. Supporting the concept of a switch in gene programs, we found that certain clusters of genes were differentially up-regulated by full vs. partial agonists. This pattern may explain on a molecular basis how SPPARMs reduce TG load. Both PPAR2 (41) and the C/EBP transcription factors (, ß, and ) (42) are indispensable for adipogenesis in vitro and in vivo, as shown from gene ablation experiments. Induction of expression of these transcription factors by PA-082 underscored that partial agonists are still capable of allowing commitment of C3H10T1/2 cells to the adipogenic lineage despite lower accumulation of TGs during terminal differentiation. UCP1 is a marker of uncoupling, thermogenesis, and transdifferentiation of WAT to BAT. Unexpectedly, neither UCP1 mRNA nor binding of PPAR/PGC1 complexes to the URE1 enhancer element (by ChIP) was up-regulated by PA-082. CPT1, the rate-limiting enzyme of mitochondrial fatty acid oxidation, remained unchanged as well. These findings did not favor the hypothesis of a change in PPAR-subtype specificity by PGC1, e.g. toward a PPAR-like catabolic program of energy expenditure (43). Expression of ERR itself was not altered, further excluding a role of the PGC1/ERR/ OXPHOS-pathway in this cellular context. PA-082 did not bind to PPAR or PPAR in radioligand and transactivation assays. Furthermore, no deloading of TGs was recorded when mature adipocytes were incubated for several days with PA-082 (our unpublished observation). Genes responsible for reesterification (44) of free fatty acids to TGs for storage in lipid droplets, GK, and PEPCK were more responsive to rosiglitazone than to PA-082. A similar picture was also recorded for the lipogenic transcription factor sterol regulatory element-binding protein-1 and adipose differentiation-related protein (our unpublished observation). Thus, PA-082 is likely to prevent TG storage in lipid droplets during de novo adipogenic differentiation without concomitant increase in energy expenditure via fatty acid oxidation or mitochondrial uncoupling.
Interestingly, the potency of PA-082 to reduce TG loading was mimicked by overexpression of wild-type PGC1 in C3H10T1/2 cells. Ectopic PGC1 lowered TG accumulation during de novo differentiation but unexpectedly induced aP2/FABP mRNA in a ligand-independent and ligand-dependent fashion. In addition, ectopic PGC1 rescued the compromised ability of PA-082 to up-regulate aP2 mRNA. This effect was completely abrogated by the L2A mutant of PGC1, as observed before in our transactivation assays. Because UCP1 was not altered by ectopic PGC1, these observations provided additional arguments that PPAR, but not ERR, is likely to be the major player in this phenomenon. Data from ChIP experiments underscored those from QPCR that the aP2/A-FABP gene is positively regulated by PGC1 in our cell system. In contrast, two previous reports demonstrated that aP2 is not regulated by PGC1 (10, 13). In the first study, 3T3-F442A preadipose cells were stably infected with a PGC1 retroviral vector, differentiated with insulin, and mature adipocytes were incubated with 8-bromo-cAMP, 9-cis-retinoic acid, and 3,3,5-triiodo-L-thyronine, a ligand for thyroid hormone receptor (10). In this setup, PGC1 induced genes for thermogenesis but inhibited aP2 mRNA, presumably by inducing a BAT phenotype. A second study examined mature 3T3-L1 adipocytes in which aP2 was constitutively activated compared with preadipocytes and not further enhanced by rosiglitazone or ectopic PGC1 (13). Instead, we transiently transfected C3H10T1/2 mesenchymal stem cells, differentiated with PPAR ligands to adipocytes with WAT properties. The experimental designs of the latter studies thus differed from the one applied here and may explain the cell type-and context-specific regulation of the aP2 gene. Another possible explanation for the apparent controversies may reside in the versatile functions of FAPBs. aP2 is the prominent FABP in mature adipocytes and currently recognized as a marker for terminal adipocyte differentiation and TG storage in WAT (45). However, A-FABP also interacts with hormone-sensitive lipase and facilitates lipolysis, whereas A-FABP null mice exhibit increased fat mass and decreased lipolysis (46). Thus, the balance between TG storage and lipolysis will determine the final TG load of the differentiating cell. We therefore concluded that the reduced TG accumulation in C3H10T1/2 cells, which we observed in the presence of ectopic PGC1 or upon its recruitment by a PPAR ligand, is rather based on lowered TG storage (via PEPCK and GK) than on failure to undergo adipogenic commitment, as evident from functional aP2, PPAR2, and C/EBP expression. However, decreased terminal differentiation, as evident from lipid droplet staining, may be another underlying reason for reduced TG load. FABPs are also implicated in transport and delivery of fatty acid ligands from the cytosol to the nucleus where PPARs reside (47). Thus, coinduction of PGC1 with aP2 in the early commitment phase of adipogenic conversion may favor insulin signaling by enhanced delivery of PPAR ligands to the receptor.
The proinflammatory cytokine TNF promotes cellular insulin resistance by activating inhibitor-
B kinase and stress kinase (c-Jun N-terminal kinase) pathways that impair insulin receptor signaling by inhibitory serine phosphorylation of insulin receptor substrate (IRS) proteins (26, 48). Serine-phosphorylated IRS are less capable of activating phosphatidylinositol-3 kinase, AKT, and exocytosis of glucose transporter (e.g. glucose transporter 4) vesicles. TZDs counteract TNF by restoring tyrosine phosphorylation levels of IRS (48, 49). PA-082 acted as an equivalent agonist compared with other SPPARMs and TZDs in terms of insulin-stimulated glucose uptake into mature C3H10T1/2 adipocytes. It was also fully capable of opposing the negative cross-talk exerted by TNF on insulin-mediated phosphorylation of AKT/PKB kinase in mature adipocytes. The mRNAs encoding central adapter proteins involved in the insulin signaling cascade CAP and IRS2 were up-regulated both by rosiglitazone and PA-082 upon de novo adipogenesis in C3H10T1/2 cells. Although no PPRE has been identified in the gene loci of IRS1 or IRS2 so far, mRNAs have been described to be up-regulated by PPAR ligands in vitro (50) and in vivo (51). CAP is a direct PPAR-target gene with a functional PPRE in the promoter (52) and a multifunctional adapter protein that facilitates insulin receptor signaling and membrane translocation of glucose transporter 4 in vitro and insulin sensitivity in vivo. As shown for aP2 mRNA, overexpression of wild-type PGC1, but not the L2A mutant, increased basal and ligand-dependent mRNA levels of CAP. Thus, both PGC1 recruitment by a PPAR ligand and overexpression of PGC1 per se converged into a similar pattern of gene expression. These findings suggest a function of this cofactor in uncoupling TG storage from fatty acid/ligand delivery and insulin signaling in adipocytes. It may as well provide a possible explanation for the insulin-sensitizing potency of SPPARMs in the absence of fat accretion in vivo.
Fasting induces PGC1 expression in hepatocytes via the cAMP/protein kinase A/cAMP response element binding protein pathway and activates gluconeogenesis and fatty acid oxidation in conjunction with PPAR, hepatocyte nuclear factor-4 (HNF4), and FOXO1 (10, 11). Insulin/ phosphatidylinositol-3 kinase/AKT counteract this pathway by phosphorylation and nuclear exclusion of FOXO1 (53). PPAR ligands exert beneficial effects on the liver by increasing the ratio of hepatic glucose uptake/output in diabetic patients (7) and restore expression and phosphorylation of insulin signaling proteins in animals (51). As in adipocytes, SPPARMs increased expression of L-FABP and CAP mRNAs in HepG2 hepatoma cells, genes that may translate the ability of these compounds to oppose the negative cross-talk of TNF on AKT signaling. Ligands for pregnane X receptor competitively interfere with PGC1/HNF4-driven activation of gluconeogenic target genes in HepG2 cells and mouse liver (54). One could thus speculate on a similar mechanism here, where PPAR
TOP
ABSTRACT
INTRODUCTION
