Ligand-Dependent Interaction of PPARδ With T-Cell Protein Tyrosine Phosphatase 45 Enhances Insulin Signaling

Insulin initiates its cellular actions by binding to its cell surface transmembrane receptor, which is a tyrosine kinase (1). Upon binding of insulin to the extracellular α-subunit, the intrinsic protein tyrosine kinase (PTK) activity of the intracellular β-subunit is activated, leading to its autophosphorylation and subsequent phosphorylation of insulin receptor (IR) substrate (IRS) proteins and other adaptor molecules (2). This tyrosine phosphorylation provides docking sites for the recruitment of several src homology 2 domain–containing proteins, such as the p85 regulatory subunit of phosphatidylinositol 3-kinase (PI3K), which mediate the biological activities of insulin in its target organs (2,3). Under the pathological conditions associated with obesity and type 2 diabetes, impaired responses to insulin are coupled with resistance to its actions in target organs, including liver, muscle, and adipose tissue (3).

Evidence is emerging that protein tyrosine phosphatases (PTPs) play a pivotal role in IR signaling by dephosphorylating the IR within minutes of its activation to terminate IR-mediated insulin signaling (4,5). T-cell protein tyrosine phosphatase (TCPTP) and PTP1B are both implicated as negative regulators of IR (4,6), but they have distinct biological activities by virtue of their substrate specificity and different subcellular locations (4). TCPTP, which was originally identified in human T cells (7), has two alternative splice variants: a 48-kDa form (TCPTP48), which contains a hydrophobic carboxy-terminus and is localized to the endoplasmic reticulum (ER), similar to PTP1B, and a shorter 45-kDa form (TCPTP45), which lacks the hydrophobic C-terminus and is targeted to the nucleus (8).

TCPTP45 is localized to the nucleus in resting cells but is translocated to the cytoplasm and plasma membrane in response to external stimuli, such as insulin, where it can access its substrates (9). Accordingly, TCPTP45 can dephosphorylate the IR in the cytoplasm and at the plasma membrane, whereas ER-targeted TCPTP48 dephosphorylates the IR after it has been endocytosed, similar to the action of PTP1B (5,9). In addition to using receptor PTKs such as IR as substrates, TCPTP also uses nonreceptor PTKs as substrates, including signal transducer and activator of transcription 3 (STAT3) (10,11). Indeed, recent studies demonstrated that STAT3, a target of TCPTP45, is implicated in the insulin resistance induced by interleukin 6 (IL-6) through upregulation of suppressor of cytokine signaling 3 (SOCS3), which contributes to the negative regulation of insulin signaling by interacting with the IR and IRS (12).

The biological activity of peroxisome proliferator–activated receptor (PPAR) δ, a nuclear receptor, has been credited to transcriptional programs that regulate the expression of target genes (the so-called genomic action) (13,14). However, there have been recent reports that could not explain the biological functions of PPARδ by a genomic action alone (15,16). Here, we report that short-term activation of PPARδ by high-affinity ligand results in the formation of a stable complex with TCPTP45, thereby blocking the insulin-responsive translocation of nuclear TCPTP45 into the cytoplasm and preventing access of TCPTP45 to IR and its inhibitory effect on insulin signaling. In addition, this protein-protein interaction blunts IL-6–induced insulin resistance by accelerating TCPTP45-mediated deactivation of the STAT3-SOCS3 signal in a process mediated by sequestration of TCPTP45 within the nuclear compartment. These observations demonstrate the role of a novel TCPTP45-mediated pathway in the insulin-sensitizing action of PPARδ.

Cell and tissue lysates were prepared and then precleared with protein G Sepharose. Precleared lysates were mixed with the specified antibodies and then precipitated with protein G Sepharose. The immunoprecipitates and total lysates (input) were washed with PBS and analyzed by immunoblot.

Male ICR mice (4 weeks old, 15–20 g) were purchased from Orient Bio (Seongnam, Korea) and maintained in pathogen-free environmental conditions on a 12-h light-dark cycle at 22 ± 2°C. All procedures were approved by the Konkuk University Institutional Animal Care and Use Committee (approval number: KU16144). After 1 week of acclimation, the mice were divided into two groups and were fed a normal diet (ND) (Altromin Spezialfutter GmbH & Co., Lage, North Rhine-Westphalia, Germany) or a high-fat diet (HFD) (Research Diets, New Brunswick, NJ) for 10 weeks, during which water and diet were available ad libitum.

For intraperitoneal glucose tolerance testing, mice were fasted for 16 h, and then glucose (2 g/kg body mass) was administered intraperitoneally. Blood was taken from the tail vein at 0, 30, 60, 90, and 120 min after the glucose injection, and the circulating glucose levels were determined using a OneTouch automatic glucose monitor (LifeScan, Milpitas, CA). For intraperitoneal insulin tolerance testing (IPITT), insulin (0.5 units/kg of body mass) in sterile saline was injected intraperitoneally, and the circulating glucose levels were determined as described above.

Data are expressed as mean ± SE. Statistical significance was determined by one-way ANOVA, followed by the Tukey-Kramer or Student t test, as appropriate.

In a yeast-hybrid assay that used PPARδ as bait, we identified a cDNA fragment encoding a full-length amino acid sequence of TCPTP45 in a universal human cDNA library. The interaction of PPARδ with TCPTP45 was confirmed by coimmunoprecipitation assays performed in primary hepatocytes, hepatoblastoma-derived HepG2 cells, differentiated C2C12 myotubes, and 3T3-L1 adipocytes. When cells were treated with GW501516, a specific PPARδ agonist, immunoprecipitation of PPARδ from whole-cell lysates resulted in the coimmunoprecipitation of TCPTP45. However, PTP1B, another tyrosine-specific phosphatase that targets the IR β-subunit (IRβ), was not coimmunoprecipitated with PPARδ (Fig. 1A–D). This GW501516-dependent interaction was almost completely abolished in HepG2 cells stably expressing PPARδ small hairpin (sh)RNA, indicating that ligand-dependent activation of PPARδ is a prerequisite for its interaction with TCPTP45 (Fig. 1E and Supplementary Fig. 1A). This interaction of PPARδ with TCPTP45 was also confirmed using reciprocal antibodies for immunoprecipitation and immunoblotting (Fig. 1F). No proteins were immunoprecipitated using control IgG. Similar results were obtained in HEK293T cells cotransfected with Flag-tagged PPARδ and hemagglutinin (HA)-tagged TCPTP45, in which the GW501516-dependent interaction of PPARδ and TCPTP45 was present at 30 min and reached a maximum at 45 min (Fig. 1G and Supplementary Fig. 1B). In addition, when HEK293T cells were transfected with individual plasmids expressing PPARα, δ, or γ, together with TCPTP45, in the presence or absence of specific ligands for each PPAR isoform, PPARδ but not PPARα and γ demonstrated a strong interaction with TCPTP45 in a coimmunoprecipitation assay, indicating that only the PPARδ isoform can interact with TCPTP45 (Supplementary Fig. 1C).

A direct interaction between PPARδ and TCPTP45 was further confirmed in glutathione S-transferase (GST) pull-down assays, in which most of the PPARδ was pulled down with GST-fused TCPTP45 in the presence of GW501516 (Fig. 2A). Thus, we generated a panel of TCPTP45 deletion mutants to identify the regions of TCPTP45 that are responsible for its interaction with PPARδ (Fig. 2B). Whereas full-length TCPTP45 (FL) displayed a strong interaction with PPARδ, the deletion mutants TCPTP45 (D1) and TCPTP45 (D2) showed no significant interaction, indicating that the COOH-terminal region of TCPTP45 is essential for this interaction (Fig. 2C). Because the COOH-terminal region of TCPTP45 has a bipartite nuclear localization sequence (NLS), we constructed three NLS mutants to identify the critical amino acid residue(s) responsible for the association of TCPTP45 with PPARδ. Site-directed mutants were constructed by introducing amino acid substitutions in the N-terminal part of NLS residues 350, 351, and 352 of TCPTP45 (TCPTP45^350-A3), and further substitutions in the COOH-terminal part of NLS residues 378, 379, and 380 (TCPTP45^378-A3), and in both the N- and COOH-terminal parts of NLS residues 350, 351, 352, 378, 379, and 380 (TCPTP45^350/378-A3), replacing arginine or lysine residues with alanine (Fig. 2D).

When these mutant proteins were individually incubated with lysates prepared from HEK293T cells that had been transfected with PPARδ, TCPTP45^350-A3 displayed a weaker ability than TCPTP45^378-A3 to interact with PPARδ. Furthermore, the interaction between PPARδ and TCPTP45 was nearly eliminated when PPARδ was incubated with TCPTP45^350/378-A3, indicating that the bipartite NLS of TCPTP45 is indispensable for the interaction with PPARδ (Fig. 2E). Consistent with this, the localization of TCPTP45 was limited to the cytoplasm of HepG2 cells transfected with HA-tagged TCPTP45^350/378-A3 (Fig. 2F).

Because TCPTP48 also has a bipartite NLS region, we examined the association between PPARδ and TCPTP48 in HepG2 by coimmunoprecipitation. As expected, PPARδ interacted with TCPTP48, suggesting that such protein-protein interactions via the bipartite NLS region may occur nonspecifically under any conditions in which these molecules come close together by cellular destruction (Supplementary Fig. 2A). However, when the nuclear or cytoplasmic fractions of HepG2 cells stimulated with GW501516 were analyzed to confirm the specificity of TCPTP, an interaction between PPARδ and TCPTP was observed in both fractions (Supplementary Fig. 2B and C). These results indicate that although TCPTP45 and TCPTP48 can both interact with PPARδ via their bipartite NLS, their subcellular location is a unique determinant for a specific interaction with PPARδ.

To further evaluate the specificity of the bipartite NLS of TCPTP45 in the interaction with PPARδ, the bipartite NLS of TCPTP45 was replaced with SV40 large T antigen NLS (17). When the cells were transfected with TCPTP45^350/378-SV40, although most of TCPTP45^350/378-SV40 was observed in the nuclear fraction, the interaction between PPARδ and TCPTP45^350/378-SV40 was not detected in the presence of GW501516 (Supplementary Fig. 2D–F).

Because PPARδ interacted with both TCPTP45 and TCPTP48 in lysates prepared from HepG2 cells, we examined the cellular colocalization of PPARδ and TCPTPs to confirm the exact nature of the interaction with PPARδ. By confocal fluorescence microscopy using fluorescent fusion protein versions of PPARδ and TCPTP45^D182A or TCPTP48^D182A, which are substrate-trapping mutants that can form complexes with tyrosine-phosphorylated substrates (18), we observed that green fluorescent protein (GFP)–tagged PPARδ was colocalized with Discosoma spp. red fluorescent protein (DsRed)–tagged TCPTP45 in the nucleus of CHO cells stably expressing the IR (CHO-IR). Upon stimulation with insulin, most of the DsRed-TCPTP45 was rapidly translocated into the cytoplasm, but this insulin-stimulated translocation was almost completely abolished in cells treated with GW501516 (Supplementary Fig. 3A). By contrast, neither colocalization of GFP-PPARδ and DsRed-TCPTP48 nor translocation between intracellular compartments was observed, stressing the importance of subcellular localization in the interaction between PPARδ and TCPTP45 (Supplementary Fig. 3B).

To investigate whether GW501516-activated PPARδ affects the insulin-stimulated interaction of TCPTP45 with IRβ, coimmunoprecipitation assays were performed using Flag-PPARδ and HA-TCPTP45^D182A in CHO-IR and HepG2 cells. Upon stimulation with insulin, IRβ was coimmunoprecipitated with TCPTP45^D182A in both cells, whereas an interaction between PPARδ and TCPTP45^D182A was not observed in the absence of GW501516. By contrast, insulin-stimulated complex formation between IRβ and TCPTP45^D182A was reduced in cells treated with GW501516 because of increased interaction between PPARδ and TCPTP45^D182A. Furthermore, GSK0660-mediated antagonism of GW501516 caused the dissociation of PPARδ and TCPTP45^D182A, with a concomitant increase in the formation of IRβ and TCPTP45^D182A complexes (Fig. 3A and B). Similar results were obtained from HepG2 cells stably expressing PPARδ or control shRNA (Fig. 3C). As expected, insulin-stimulated translocation of TCPTP45^D182A from the nucleus to the cytoplasm was almost completely abolished in the presence of GW501516 with a concomitant reduction in the interaction of TCPTP45^D182A and phosphorylated IRβ, and this GW501516-mediated inhibition of TCPTP45^D182A translocation and interaction with IRβ was prevented by GSK0660 (Fig. 3D).

GW501516 induced an interaction between PPARδ and TCPTP45; thus, particular attention was paid to the effect on insulin signaling because TCPTP45 can antagonize insulin signaling by dephosphorylating IRβ (5). When HepG2 cells were treated with insulin, IRβ was rapidly phosphorylated, and this effect was further potentiated in the presence of GW501516 (Fig. 4A). This GW501516-mediated potentiation of IRβ phosphorylation was almost completely abolished in HepG2 cells transfected with TCPTP45^350/378-A3, a site-directed mutant lacking the molecular regions that interact with PPARδ (Fig. 4B). Consistent with these results, GW501516-mediated potentiation of IRβ phosphorylation was also manifested in cells that transiently expressed TCPTP45 but not TCPTP48 (Fig. 4C). Furthermore, small interfering (si)RNA-mediated silencing of TCPTP, but not PTP1B, prevented the effect of GW501516 to potentiate IRβ phosphorylation (Fig. 4D and Supplementary Fig. 4A). Similar results are shown in primary hepatocytes (Fig. 4E).

When HepG2 cells were exposed to insulin, IRβ phosphorylation peaked after 2–5 min and then rapidly returned to basal levels. However, pretreatment with GW501516 for 30 min sustained IRβ phosphorylation until 60 min after the initial insulin exposure (Fig. 4F and H). This effect of GW501516 on the duration of IRβ phosphorylation was further potentiated in cells ectopically expressing PPARδ (Fig. 4G and H, and Supplementary Fig. 4B and C). This effect of GW501516 remained significant for as long as 6 h in cells expressing PPARδ (Supplementary Fig. 4D). This effect of GW501516 on IRβ phosphorylation was prevented by shRNA-mediated silencing of PPARδ (Fig. 4I–K) but was rescued by ectopic overexpression of PPARδ in transfectants stably expressing PPARδ shRNA (Fig. 4L–N). By contrast, treatment with WY14643 or rosiglitazone, specific ligands for PPARα and PPARγ, respectively, did not affect insulin-stimulated phosphorylation of IRβ (Supplementary Fig. 4E and F).

Ectopic expression of TCPTP45 inhibited the effect of GW501516 on IRβ phosphorylation in cells stably expressing PPARδ shRNA but not control shRNA, whereas this effect of GW501516 was rescued by ectopic expression of PPARδ (Fig. 5A and Supplementary Fig. 5A). Furthermore, the GW501516-stimulated interaction between PPARδ and TCPTP45 demonstrated in PPARδ precipitates was clearly correlated with increased phosphorylation of IRβ (Fig. 5B). However, this effect of GW501516 was almost completely abolished in the coimmunoprecipitate of TCPTP45^350/378-A3, indicating that the GW501516-stimulated interaction of PPARδ with TCPTP45 prevents the access of TCPTP45 to cytoplasmic tyrosine-phosphorylated IRβ (Fig. 5C).

IR signaling has its biological effects in target organs through phosphorylation of intracellular effector molecules such as IRS-1 and Akt (2). Consistent with the effect of GW501516 on IRβ phosphorylation, GW501516 treatment prolonged the duration of insulin-stimulated IRS-1 and Akt phosphorylation for 6 h in HepG2 cells stably expressing control shRNA (Supplementary Fig. 5B), whereas these effects of GW501516 were dramatically inhibited in the presence of PPARδ shRNA (Fig. 5D). In addition, the GW501516-stimulated interaction of PPARδ with TCPTP45 observed in PPARδ precipitates was correlated with the increased phosphorylation of IRS-1 and Akt; however, these effects of GW501516 were almost completely abolished when TCPTP45^350/378-A3 was coimmunoprecipitated, indicating that the interaction between PPARδ and TCPTP45 directly correlates with IRS-1 and Akt phosphorylation (Fig. 5E).

We next assessed glucose levels in the HepG2 cells to evaluate whether GW501516-mediated activation of PPARδ is directly coupled to the biological effects of insulin. Upon exposure to insulin for 30 min, glucose levels in the glucose production medium dramatically decreased, and this reduction in glucose output was further enhanced by treatment with GW501516 in HepG2 cells stably expressing control shRNA. However, this effect of GW501516 on glucose levels in the glucose production medium was significantly inhibited in cells stably expressing PPARδ shRNA (Fig. 5F).

Because skeletal muscle and adipocytes are also major target organs of insulin action (1), we investigated the effect of GW501516 on insulin signaling in differentiated mouse C2C12 myotubes and 3T3-L1 adipocytes. Consistent with the results observed in HepG2 cells, GW501516 also increased the duration of IRβ and Akt phosphorylation by promoting interaction between PPARδ and TCPTP45 (Fig. 6A and B and Supplementary Fig. 6A and B). Similar results were also observed in C2C12 myotubes transfected with epitope-tagged PPARδ and TCPTP45, whereas the effects of GW501516 on the interaction of PPARδ with TCPTP45 and insulin signaling were almost completely abolished in C2C12 myotubes transfected with epitope-tagged PPARδ and TCPTP45^350/378-A3 (Supplementary Fig. 6C). Furthermore, insulin-induced translocation of Glut4 to the plasma membrane was sustained for 6 h in cells treated with both insulin and GW501516, whereas the Glut4 signal was predominantly cytoplasmic in cells treated with insulin alone at this time point (Fig. 6C and D).

Because obesity-associated chronic inflammation of adipose tissue leads to insulin resistance in peripheral tissues, including the liver, and TCPTP45 is implicated in glucose homeostasis through STAT3 signaling in the liver (10,19), we examined the effects of GW501516 on IL-6–induced insulin resistance. Insulin enhanced the phosphorylation of IRS-1 and Akt in HepG2 cells, but the insulin-stimulated phosphorylation of IRS-1 and Akt was markedly attenuated in the presence of IL-6. However, this IL-6–mediated suppression of IRS-1 and Akt phosphorylation was reversed by GW501516, implying a possible role for PPARδ in the antagonism of inflammation-induced insulin resistance (Fig. 6E).

Because STAT3 and SOCS3 are implicated in IL-6–induced insulin resistance (12,20), we examined the effects of GW501516 on the activation and expression of STAT3 and SOCS3. Insulin-stimulated phosphorylation of STAT3 and expression of SOCS3 were inhibited in HepG2 cells stably expressing control shRNA in the presence of GW501516, but these effects of GW501516 were substantially prevented by the expression of PPARδ shRNA (Fig. 6F). Consistent with these results, siRNA-mediated knockdown of SOCS3 prevented the IL-6–triggered suppression in the phosphorylation of IRS-1 and Akt, indicating that SOCS3 plays a role in the GW501516-dependent modulation of insulin resistance induced by IL-6 (Supplementary Fig. 6D). In addition, GW501516-stimulated interaction between PPARδ and TCPTP45 observed in PPARδ precipitates was directly correlated with the increased phosphorylation of Akt, with concomitant reductions in the levels of STAT3 phosphorylation and SOCS3 expression. However, these effects of GW501516 were significantly blunted in cells transfected with TCPTP45^350/378-A3 instead of TCPTP45, indicating that the GW501516-stimulated interaction of PPARδ with TCPTP45 improves IL-6–induced insulin resistance by targeting STAT3 and SOCS3 (Fig. 6G). Similar results were also observed in primary hepatocytes (Fig. 6H).

We next wished to evaluate whether GW501516-activated PPARδ can affect glucose metabolism in ND- or HFD-fed mice. In follow-up experiments, we fed mice an ND or HFD for 10 weeks and checked their body masses weekly to confirm the expected effect of HFD feeding (Supplementary Fig. 7A). When mice were administered with GW501516 or DMSO and challenged 30 min later with a bolus of glucose by intraperitoneal injection, blood glucose levels of ND- or HFD-fed mice rapidly increased to peak 30 min later. However, blood glucose subsequently decreased more rapidly in HFD-fed mice that had been administered with GW501516 than in those that had not (Fig. 7A). Analogous results were obtained in HFD-fed mice subjected to IPITT, in which the insulin-stimulated decrease in blood glucose was greater after GW501516 administration (Fig. 7B).

To determine whether the administration of GW501516 directly affected the interaction between PPARδ and TCPTP45, leading to an improvement of insulin signaling in vivo, we analyzed the interaction by coimmunoprecipitation in lysates prepared from liver, skeletal muscle, and adipose tissue. The interaction between PPARδ and TCPTP45 was greatest in the tissues of ND-fed mice administered with GW501516, with a lesser effect of GW501516 detected in HFD-fed mouse tissues. This GW501516-mediated enhancement in the PPARδ/TCPTP45 interaction was closely correlated with the phosphorylation levels of IRβ and Akt in these tissues (Fig. 7C–E). Interestingly, the circulating levels of insulin, glucose, and IL-6 were significantly higher in HFD-fed mice than in ND-fed mice, but the levels of insulin and glucose were significantly lower in mice treated with GW501516 for 30 min (Fig. 7F and G and Supplementary Fig. 7B), implying an important role for PPARδ in insulin action and whole-body glucose homeostasis in vivo.

Short-term activation of PPARδ modulates IR signaling by preventing translocation of the IR-specific phosphatase TCPTP45 into the cytoplasm because of their direct protein-protein interaction. Unlike the genomic action of the transcription factor PPARδ, little is known about nongenomic actions of PPARδ, such as its protein-protein interactions. Based on the concept that PPARδ exerts its biological actions by regulating the transcription of target genes (13), recent studies demonstrated that ligand-activated PPARδ improves insulin signaling through transcriptional regulation of target genes such as insulin-induced gene-1, pyruvate dehydrogenase kinase 4, and angiopoietin-like protein 4 in hepatocytes and myotubes (21,22).

In the most recent reports, cells were treated with specific PPARδ ligands for as long as 18 to 24 h so that the effect of PPARδ activation on transcriptional programs and downstream biological activities could be investigated (23,24). However, we show here that the activation of PPARδ by GW501516 for only 30 min was sufficient to phosphorylate IRβ and, consequently, IRS-1 and Akt. To achieve this effect, GW501516 promoted the interaction of PPARδ with TCPTP45 but could not affect the transcription of TCPTP45. These findings suggest that nongenomic actions of PPARδ, rather than its genomic actions, are important for the regulation of IR signaling. We thus propose a novel mechanism whereby PPARδ influences insulin signaling by involving the sequestration of PPARδ in the nucleus secondary to a protein-protein interaction with the IRβ-specific phosphatase TCPTP45 and, therefore, prevents binding to its cytoplasmic substrate IRβ.

Although GW501516-activated PPARδ interacted with both TCPTP45 and TCPTP48, it colocalized in the nucleus with TCPTP45 but not TCPTP48. These two alternative splice variants exhibit distinct subcellular localizations because of a difference in the seven amino acid residues at their carboxy-terminals. As a result, TCPTP48 is mainly localized to the ER, whereas TCPTP45 is restricted to the nucleus in resting cells (9). Therefore, TCPTP48 dephosphorylates IRβ after it is endocytosed into the ER in response to insulin, whereas TCPTP45 dephosphorylates IRβ in the cytoplasm and at the plasma membrane after its translocation from the nucleus (6,9). In this context, the PPARδ-mediated prevention of TCPTP45 translocation, but not that of TCPTP48, into the cytoplasm may contribute to the enhancement in insulin signaling. GW501516-activated PPARδ also did not interact with PTP1B, another tyrosine-specific phosphatase that targets IRβ in the ER (4,6). Thus, the subcellular localization, as well as the other biochemical properties of the substrate, when interacting with PPARδ, may be critical determinants of the effects of PPARδ on insulin signaling.

PPARδ, but not PPARα or γ, specifically interacted with TCPTP45 in a ligand-dependent manner to enhance insulin signaling. Previous studies demonstrated that specific ligands for PPARγ enhance insulin signaling through genomic actions regulating gene transcription, including that of adipocyte P2 (aP2), lipoprotein lipase (LPL), fatty acid-transport protein (FATP), Glut4, glucokinase, uncoupling protein (UCP)-2 and UCP-3, to regulate glucose homeostasis, adipocyte differentiation, and lipid storage in cell or animal models (25,26). Ligand-activated PPARα also ameliorated insulin resistance by upregulating genes encoding FATP, acyl-CoA oxidase, and LPL, leading to increased hepatic fatty acid uptake and lipid metabolism (27,28). These findings are consistent with the data reported here because both enhance insulin signaling, but the underlying mechanisms mediating this effect are specific to the PPAR isoform. Indeed, only a 30-min treatment with GW501516 was sufficient to activate PPARδ and enhance insulin signaling by inhibiting translocation of the IRβ-specific phosphatase TCPTP45 into the cytoplasm. This was the result of a direct interaction with TCPTP45, which we term a nongenomic action here. Consequently, the effect of each PPAR nuclear receptor on insulin signaling can be characterized according to its genomic and/or nongenomic actions.

Although the present findings are consistent with previous reports that GW501516-activated PPARδ ameliorates IL-6–induced insulin resistance in hepatocytes by inhibiting the STAT3-SOCS3 pathway (29), the molecular mechanisms may differ between studies because the 30-min treatment with GW501516 was sufficient to inactivate the STAT3-SOCS3 pathway in this study. Several recent studies demonstrated that activation of PPARδ by specific ligands alleviated IL-6–induced insulin resistance by suppressing SOCS3 expression, secondary to inhibition of STAT3 activation. These effects of GW501516 were observed in hepatocytes and adipocytes treated for 24 h, suggesting that the effects of PPARδ in this context may be principally dependent on its genomic activity (30,31). Indeed, TCPTP45 was reported to inhibit IL-6–induced STAT3 phosphorylation, leading to the attenuation of insulin signaling in the liver (10,11). Consistent with these reports, PPARδ-mediated retention of TCPTP45 in the nucleus facilitated the access of TCPTP45 to its substrate STAT3, resulting in reduced IL-6–mediated phosphorylation and suppression of STAT3-mediated SOCS3 expression, which is linked to insulin resistance (19,29). Thus, the insulin-sensitizing effects of GW501516 may be collectively mediated by the interaction between PPARδ and TCPTP45.

Contrary to our present findings showing the TCPTP45-dependent effect of PPARδ in insulin signaling, a previous study demonstrated that TCPTP deficiency in muscle has no effect on insulin signaling and glucose homeostasis (32). Although this discrepancy for TCPTP action in muscle insulin signaling is not fully elucidated at present, our present data evenly showed that the GW501516-dependent interaction of PPARδ and TCPTP45 potentiated the insulin-triggered phosphorylation of IRβ and Akt and translocation of Glut4 in C2C12 myotubes. In fact, previous studies reported that ligand-activated PPARδ ameliorates cellular energy homeostasis through its genomic action regulating transcription of genes, including UCP-2, UCP-3, pyruvate dehydrogenase kinase 4, carnitine palmitoyltransferase 1, and malonyl-CoA decarboxylase in skeletal muscle (13,22–25). Thus, the possibility remains that the effects of PPARδ activation on the insulin signaling may be attributed to their transcriptional regulation of multiple genes rather than to direct interaction of PPARδ and TCPTP45 in C2C12 myotubes. Further studies are necessary to clarify the exact molecule(s) associated with PPARδ-mediated insulin signaling in muscle.

Consistent with the in vitro data, mice exposed to GW501516 for as little as 30 min showed enhanced insulin-stimulated phosphorylation of IRβ and Akt in the liver, skeletal muscle, and adipose tissue. This GW501516-mediated enhancement in insulin signaling and glucose homeostasis, which occurred even in HFD-fed obese mice, was accompanied by greater interaction between PPARδ and TCPTP45, indicating that these in vivo effects of GW501516 are dependent on the association between PPARδ with TCPTP45, as in the cell models. A number of studies to date have shown that activation of PPARδ regulates insulin signaling by regulating the expression of genes involved in lipid and glucose homeostasis. However, the current study implies that PPARδ has multiple modes of action and can enhance insulin signaling through an interaction with TCPTP45, which we characterized as a nongenomic action. These findings may suggest novel alternative future therapeutic strategies for disorders related to defective insulin signaling.

Funding. This work was supported in part by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Science, ICT and Future Planning (NRF-2017R1A2A2A05001249 and NRF-2015R1A5A1009701), and by the Next-Generation BioGreen 21 Program (no. PJ01122201), Rural Development Administration, Republic of Korea.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. T.Y., S.A.H., W.J.L., S.I.H., J.-A.P., J.S.H., J.H., H.-C.S., S.G.H., C.-H.L., and D.W.H. performed the experiments. T.Y., K.S.P., and H.G.S. analyzed the results and wrote the manuscript. T.Y. and H.G.S. conceived and designed the experiments. T.Y. and H.G.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.