3-Phosphoinositide-Dependent Protein Kinase-1 Activates the Peroxisome Proliferator-Activated Receptor-γ and Promotes Adipocyte Differentiation

Abstract

Adipocyte differentiation is regulated largely through the actions of the peroxisome proliferator-activated receptor (PPAR) γ nuclear receptor and the insulin signaling pathway. 3-Phosphoinositide-dependent protein kinase-1 (PDK1) serves as a critical regulatory point in insulin signaling through its ability to phosphorylate the activation loop of several protein kinase families. The present study was undertaken to determine the interrelationships between the PDK1 and PPARγ signaling pathways, and their association with adipocyte differentiation. Coexpression of PDK1 and PPARγ1 in 293T cells stimulated PPARγ response element-dependent reporter gene activity in either the presence or absence of ligand. PDK1-mediated stimulation of PPARγ1 activity was comparable in magnitude to the coactivator activated in breast cancer-1, and was blocked by either the corepressor silencing mediator of retinoid and thyroid hormone receptor or dominant-negative PAX8-PPARγ1. Heterologous Gal4-PPARγ1 assays indicated that PDK1 interacted with the ligand binding domain, and physically associated with PPARγ1; however, PDK1-mediated stimulation was not dependent on phosphorylation of PPARγ1 by PDK1. PDK1 stimulatory activity was eliminated by mutation of the α-helical hydrophobic motifs in PDK1, L²⁶⁸XII, and V³¹³XXLL, and expression of the α-helical region encompassing these motifs stimulated PPARγ response element-dependent transcription. PDK1-PPARγ interaction was confirmed by chromatin immunoprecipitation analysis of the lipoprotein lipase and adipocyte fatty acid-binding protein promoters. In cells expressing PDK1 and PPARγ, binding to PPARγ response elements occurred, which was enhanced by treatment with a PPARγ agonist. Expression of PDK1 in 3T3-L1 or COMMA-1D mammary epithelial cells promoted adipocyte differentiation in the presence of a PPARγ agonist that was comparable to the response of PPARγ1-transfected cells in the presence of agonist; expression of PDK1 and PPARγ resulted in a synergistic effect. Adipocyte differentiation in the presence of a PPARγ agonist was markedly attenuated in PDK1 null cells. These results suggest that PDK1 can function as a PPARγ1 coactivator independently of its catalytic activity and establishes an important mechanistic link between adipocyte differentiation and the insulin signaling pathway.

PDK1 (3-PHOSPHOINOSITIDE-dependent protein kinase-1) was first characterized as a protein kinase B (AKT) kinase in extracts of skeletal muscle and brain (1, 2) and serves as a central mediator of growth factor receptor-coupled activation of phosphatidylinositol 3-kinase (PI3-K) (3) through transphosphorylation of a common motif in the activation loop of AKT and other protein kinase families (4, 5). The PI3-K pathway plays a critical role in insulin signaling, where its impairment can lead to type II diabetes (6). Insulin-dependent processes, such as adipocyte differentiation, are equally dependent on activation of the PI3-K/AKT axis, as shown by the ability of constitutively active AKT to promote spontaneous adipocyte differentiation in 3T3-L1 cells (7). Indeed, an inactivating mutation in AKT2 is associated with a familial form of insulin resistance and diabetes (8).

Peroxisome proliferator-activated receptor (PPAR) γ is a member of the ligand-dependent nuclear hormone receptor family (9) and serves as a receptor for the thiazolidinedione class of antidiabetic agents that enhance insulin sensitization in peripheral tissues (10). PPARγ is particularly abundant in adipose tissue (11) and stimulates adipogenesis in fibroblasts (12) through activation of adipocyte gene expression (13–15). Overexpression of constitutively active PPARγ in 3T3-L1 preadipocytes induces adipocyte differentiation, and this effect is blocked by dominant-negative PPARγ (16). Although PPARγ-mediated transactivation is enhanced by insulin (17), the mechanistic link between nuclear receptor activation, insulin signaling, and adipocyte differentiation has not been established. In the present study, evidence is presented for the coassociation of PDK1 and PPARγ1, the ability of PDK1 to act as a PPARγ coactivator, and the ability of PDK1 to induce adipocyte differentiation in the presence of a PPARγ agonist.

RESULTS

PDK1 Enhances PPARγ Transactivation

PPARγ-dependent transcription was assessed with a PPARγ response element (PPRE)-dependent reporter gene in the absence and presence of the highly selective PPARγ ligand GW7845 (Fig. 1A). Expression of PDK1 alone did not affect endogenous PPARγ activity, but coexpression of PDK1 and PPARγ1 resulted in a 2- to 3-fold stimulation of reporter activity compared with PPARγ1 alone in the presence or absence of ligand. Dominant-negative Pax8-PPARγ1 suppressed transactivation in all instances (Fig. 1A). The stimulatory activity of PDK1 on PPARγ was comparable to the activity of the coactivator activated in breast cancer-1 (AIB1) (18, 19), but PDK1 and AIB1 did not produce an additive effect, suggesting that they interacted at a common site within PPARγ (Fig. 1B). The corepressor silencing mediator of retinoid and thyroid hormone receptor (SMRT) completely blocked ligand-independent activation of PPARγ and partially inhibited ligand-dependent stimulation in the presence or absence of PDK1 (Fig. 1B). Immunoprecipitation (IP)/Western blot assays were next carried out to determine whether activation of PPARγ by PDK1 was due to direct protein-protein interaction (Fig. 1C). PDK1 coimmunoprecipitated with PPARγ, and vice versa, suggesting specific physical interaction.

Fig. 1.

PDK1 Activates PPARγ Transcriptional Activity A, Cells were transfected with a p3XPPRE-TK-Luciferase plasmid and either empty vector (Vector), PDK1 (PDK1), PPARγ (PPAR), or PAX8-PPARγ (PAX8-PPAR) alone and in combination. Cells were treated 24 h after transfection with 1 μm GW7845, and luciferase activity determined 24 h later. Activity is expressed as the fold change in activity compared with cells transfected with empty vector in the absence or presence of GW7845. Each value is the mean ± sem of three to six determinations. B, Cells were transfected as in panel A except that AIB1 (AIB1) and SMRT (SMRT) were evaluated alone and in combination with PDK1 and PPARγ. Each value is the mean ± sem of three to six determinations. C, Cells were cotransfected with PPARγ (PPAR) and PDK1 (PDK1), and cell lysates immunoprecipitated with either IgG (IgG), anti-PDK1 (PDK1) or anti-PPARγ (PPARγ). Immunoprecipitates were separated by SDS-PAGE in 10% Tris-glycine gels, transferred onto nitrocellulose and Western blotting (WB) carried out with either PPARγ or PDK1 antibodies. DMSO, Dimethylsulfoxide.

PDK1 Binds to the Ligand Binding Domain (LBD)

To map the PPARγ domain interacting with PDK1, heterologous reporter assays with Gal4-PPARγ fusion proteins were carried out (Fig. 2A). PDK1 stimulated Gal4-PPARγ activity to a comparable degree as the wild-type receptor (Fig. 1), and a more pronounced effect was noted with Gal4-PPARγ LBD (Fig. 2A). Both Gal4-PPARγ and Gal4-PPARγ/LBD coimmunoprecipitated with PDK1 (Fig. 2B) corroborating the results with the native protein (Fig. 1C). These data indicate that coactivation with the Gal4 fusion proteins by PDK1 is a direct interaction with the receptor LBD.

Fig. 2.

PDK1 Interacts with the PPARγ LBD A, Cells were transfected with pG5Luciferase and either empty vector (vector), PDK1 (PDK1), Gal4-PPARγ (Gal4-PPAR), Gal4-PPARγ/LBD (Gal4-PPAR/LBD) alone or in combination. Luciferase activity was determined as described in Fig. 1A. Activity is expressed as the fold change in activity compared with cells transfected with empty vector in the absence or presence of GW7845. Each value is the mean ± sem of three determinations. B, PDK1 binds to the PPARγ LBD. Cells were transfected with PDK1 (PDK1) and either an empty Gal4 vector (Gal4 vector), Gal4-PPARγ (PPARγ) or Gal4-PPARγ/LBD (PPARγ/LBD). Cell lysates were immunoprecipitated with an anti-Gal4 (Gal4) antibody, and the immunoprecipitate separated by SDS-PAGE, transferred onto nitrocellulose, and probed with an anti-PDK1 antibody. WB, Western blot.

PDK1 Does Not Phosphorylate PPARγ

To determine whether the stimulatory effect of PDK1 on PPARγ transactivation was due to phosphorylation, a putative PDK1 phosphorylation motif in PPARγ, S²²⁵FxxT resembling T³⁰⁸FxxT in the AKT activation loop, was mutated (Fig. 3A). The S225A mutation did not affect coactivation of PPARγ by PDK1. Furthermore, kinase-dead PDK1 (KD-PDK1) did not transactivate PPARγ (Fig. 3A). To directly determine whether PPARγ served as a PDK1 substrate in vitro kinase assays were also carried out with rPDK1 and a glutathione-S-transferase (GST)-PPARγ fusion protein or recombinant SGK (serum- and glucocorticoid-activated protein kinase) as substrates (Fig. 3B). Although PDK1 did phosphorylate SGK, it did not phosphorylate GST-PPARγ, again indicating that PPARγ coactivation by PDK1 was not phosphorylation dependent.

Fig. 3.

Phosphorylation Is Not Required for PDK1-Dependent Activation of PPARγ A, Cells were transfected with a p3XPPRE-TK-Luciferase plasmid and either empty vector (Vector), PPARγ (PPAR), PPARγ/S225A (PPAR/S225A) in the presence and absence of PDK1 (PDK1) or kinase-dead PDK1 (KD-PDK1). Luciferase activity was determined as described in Fig. 1A. Activity is expressed as the fold change in activity compared with cells transfected with empty vector in the absence and presence of GW7845. Each value is the mean ± sem of three determinations. B, rPDK1 was incubated in kinase buffer with [γ³²P]-ATP and either GST-PPARγ or rSGK as substrate for 30 min at 30 C. The reaction was stopped with SDS-sample buffer, separated by SDS-PAGE in a 10% Tris-glycine gel, transferred onto nitrocellulose and phosphorylation determined by autoradiography. Autophosphorylated PDK1 and phosphorylated SGK were present, but PPARγ1 was not phosphorylated. DMSO, Dimethylsulfoxide; WB, Western blot.

PDK1 Contains Coactivator Motifs

The PDK1 catalytic domain (20) contains two hydrophobic domains similar to the LXXLL motif found in PPARγ coactivators (21), ²⁷⁵LQYII²⁷¹ and ³¹³VXXLL³¹⁷ present in α-helices F and H, respectively. Introduction of double point mutations I271R/I272R and L316R/L317W in PDK1 eliminated its ability to coactivate PPARγ (Fig. 4A). To determine whether the PDK1 sequence encompassing α-helices F through H could directly coactivate PPARγ, residues 256–322 in the PDK1 catalytic domain were amplified and coexpressed with PPARγ, and reporter activity was determined (Fig. 4B). Remarkably, PDK1 [256–322] exhibited almost full coactivator activity, providing direct evidence that this domain can mediate nuclear receptor coactivation through a unique phosphorylation-independent mechanism.

Fig. 4.

PDK1 Coactivation Resides in α-Helices F and H PDK1 α-helices F and H are depicted at the top of the figure. A, Cells were transfected with p3XPPRE-TK-Luciferase and either empty vector (vector), PPARγ (PPAR), PDK1/I271R-I272R (PDK1/I271R-I272R) or PDK1/L316R-L317W (PDK1/L316R-L317W) alone and in combination. Luciferase activity was determined as described in Fig. 1A. Activity is expressed as the fold change in activity compared with cells transfected with empty vector in the absence or presence of GW7845. Each value is the mean ± sem of three determinations. B, Cells were transfected with p3XPPRE-TK-Luciferase and either empty vector (Vector), PPARγ (PPAR) or PDK1 [256–322] (PDK1 [256–322]) alone and in combination. Luciferase activity was determined as described in Fig. 1A. Activity is expressed as the fold change in activity compared with cells transfected with empty vector in the absence and presence of GW7845. Each value is the mean ± sem of three determinations. DMSO, Dimethylsulfoxide.

PDK1 and PPARγ Associate at PPAR Response Elements

To assess the coactivation potential of PDK1 on PPARγ, chromatin immunoprecipitation (ChIP) analysis of the PPRE within the lipoprotein lipase (LPL) and adipocyte fatty acid-binding protein (aP2) promoters (22, 23) was assessed in cells expressing PDK1 and PPARγ (Fig. 5). IP with either PDK1 or PPARγ antibodies revealed transcription complexes on the LPL and aP2 PPRE promoter regions, which were enhanced by treatment with the PPARγ agonist GW7845. These results are consistent with PPRE-dependent reporter analysis in cells expressing PDK1 and PPARγ (Fig. 1), and further indicate the nuclear localization of PDK1.

Fig. 5.

ChIP Analysis of PPRE-Containing Promoter Regions Cell lysates from COMMA-1D mammary epithelial cells expressing either empty virus or PPARγ and PDK1 were treated with 0.1 μm GW7845 or 0.01% dimethylsulfoxide and analyzed by ChIP assay as described in Materials and Methods. Lysates were immunoprecipitated with either a PPARγ or PDK1 antibody and the PPRE-containing promoter regions of aP2 and LPL were amplified by PCR.

PDK1 Is Required for Adipocyte Differentiation

To assess a functional role for PDK1 in adipocyte differentiation, 3T3-L1 preadipocytes were transfected with the various constructs and primed with insulin and a suboptimal concentration of the PPARγ ligand rosiglitazone (Fig. 6). Cells transfected with either PDK1 or PPARγ exhibited adipocyte differentiation in the presence of rosiglitazone, and expression of both proteins produced a synergistic effect. Differentiation was low in cells transfected with either the empty vector or KD-PDK1. Adipocyte differentiation was not observed in the absence of rosiglitazone (results not shown). These results suggest that PDK1 was capable of inducing adipocyte differentiation only in cells primed with a PPARγ ligand, and that differentiation was ligand dependent, as shown previously (24). To determine whether PDK1 expression affected the level of PPARγ, 3T3-L1 cells were analyzed by Western blotting (Fig. 6C). PDK1 did not significantly alter PPARγ levels, suggesting that the stimulatory effect of PDK1 on differentiation was not due to altered PPARγ expression. Adipocyte differentiation was also examined in mouse mammary epithelial cells in the presence of the PPARγ ligand GW7845 (Fig. 6D). Cells transduced with either an empty virus or PDK1 did not undergo differentiation in the presence of GW7845; however, PPARγ-expressing cells and notably cells expressing both PDK1 and PPARγ exhibited significant differentiation that was blocked by treatment with the PI3K inhibitor LY294002.

Fig. 6.

PPARγ and PDK1-Dependent Adipocyte Differentiation in Preadipocytes and Mammary Epithelial Cells A, 3T3-L1 cells were transfected with either empty vector (Vector), PDK1 (PDK1), PPARγ (PPARγ), PPARγ (PPARγ), PPARγ+PDK1+ (PPARγ+PDK1), KD-PDK1 (KD-PDK1) or PPARγ+KD-PDK1+ (PPARγ+KD-PDK1). After transfection, cells were treated with 5 μg/ml insulin for 24 h, and subsequently with 0.5 μm rosiglitazone for 7 d. Cells were then washed with PBS, fixed, and stained with Oil Red O for lipids as described in Materials and Methods. Magnification, ×100. B, Adipocyte differentiation in panel A was quantified by counting the number of Oil Red O stained cells in each of three microscopic fields. Each value is the mean ± sd of three fields. C, 3T3-L1 cells were transfected with PDK1 as in panel A and PPARγ levels determined by Western blot. D, Mouse mammary epithelial cells were retrovirally transduced with either empty virus (Vector), PDK1 (PDK1), KD-PDK1 (KD-PDK1), PPARγ (PPARγ), PPARγ+PDK1+ (PPARγ+PDK1) or PPARγ+KD-PDK1 (PPARγ+KD-PDK1), and grown in 12-well plates for 24 h. Differentiation was induced as described in panel A except that 1 μm GW7845 was used as the PPARγ agonist in all instances. Cells expressing PPARγ (PPARγ) and PPARγ+PDK1 (PPARγ+PDK1) were also treated with 10 μm LY294002 in the presence of GW7845. Cells were stained with Oil Red O as in panel A. Magnification, ×100. E, Quantitation of adipocyte differentiation in mouse mammary epithelial cells. Cells were treated as described in panel D and lipids quantitated using AdipoRed and measuring fluorescence. Each value is the mean ± sd of three determinations.

The dependence of adipocyte differentiation was also assessed in PDK1 null embryonic stem cells in the presence and absence of rosiglitazone (Fig. 7). PPARγ ligand rosiglitazone induced intense adipogenesis in wild-type ES cells that was markedly attenuated in PDK1 (-/-) cells, suggesting that PDK1 has a dominant role in promoting adipocyte differentiation.

Fig. 7.

Adipocyte Differentiation in PDK1 Null Cells Rosiglitazone (Rosi)-induced adipocyte differentiation in embryonic stem cells with homozygous disruption of PDK1. Wild-type or PDK1 (−/−) ES cells were grown in 12-well plates for 24 h and treated with 10 nmtrans-retinoic acid for 9 d, followed by treatment with 1 μm rosiglitazone for 10 d. Cells were stained with Oil Red O for the presence of fat droplets in adipocytes as described in Materials and Methods. Upper four panels, ×100; lower panels, ×40.

PDK1 and PPARγ Modeling

PDK1 contains two LXXLL-like domains, ²⁷⁵LXXII²⁷¹ and V³¹³XXLL³¹⁷ in α-helices F and H, respectively, within the catalytic domain (20) (Fig. 8). Based on the known crystal structure of PDK1 and PPARγ, it is proposed that in order for helices F and H to interact with PPARγ, they have to be exposed in a folded conformation where either ²⁷⁵LXXII²⁷¹ or V³¹³XXLL³¹⁷ interacts with the coactivator domain, and the other α-helix provides the necessary hydrophobic core to orient the helical domain interaction (Fig. 8).

Fig. 8.

Predictive Model of PDK1 Binding to PPARγ A, In the PDK1 catalytic domain, the F-helix is shown as a magenta ribbon, the H-helix is shown as a green ribbon and the G-helix is shown as a cyan ribbon (20, 50 ). The loops between helices F, G, and H are shown as orange ribbons. The ATP molecule is shown as a ball-and-stick model. The four mutated residues are labeled. B, Using the reported crystal structure of the complex between a coactivator and PPARγ (48, 50 ), an energy minimized model of the complex between PPARγ and PDK1 α-helices F, G, and H was generated. A ribbon representation of the model of a coactivator sequence (shown in magenta) interacting with the coactivator binding site of PPARγ is shown. The binding of the coactivator domain to PPARγ keeps helix-12 in the agonist conformation. PPARγ is represented by the gray colored ribbon, except for helix-12 (yellow) and the coactivator binding region (blue). C, A perpendicular view (B).

DISCUSSION

PPARγ regulates a variety of metabolic functions that modulate growth, differentiation, and apoptosis in a variety of cell types (9, 25, 26). One function in particular is its ability to regulate adipogenesis (27, 28) through its interaction with several coactivators (29, 30) and possibly by its negative regulation by ERK2 and JNK (31). PDK1 occupies a central point between receptor-mediated activation of PI3K and several downstream effectors of insulin signaling (32), such as AKT (6, 33), through phosphorylation of the activation loop. Due to the seemingly common roles of PDK1 and PPARγ in insulin signaling and adipocyte differentiation, and the ability of PDK1 to shuttle between the nucleus and cytosol (34), we investigated whether PDK1 could interact with PPARγ and modulate its transcriptional activity.

Remarkably, PDK1 activated and physically associated with PPARγ independently of phosphorylation. The inability of KD-PDK1 to act as a coactivator was likely due to the adoption of a different conformation for the G and H α-helices compared with the wild-type enzyme as shown for PDK1(S241A) (20, 35). Recently, a phosphorylation-independent mechanism was reported for the interaction between the N-terminal noncatalytic domain of PDK1 and Ral-GEF (36). In addition, the activation of PPARγ by MAPK kinase 5 independently of phosphorylation was associated with its interaction with the hinge region of PPARγ (37).

Although PDK1 can translocate to the nucleus, it is not known whether nuclear sequestration is associated with a specific transcriptional function (34). We show for the first time that PDK1 and PPARγ interact at the PPRE of two PPARγ-dependent promoter regions, indicating nuclear localization. PDK1 appeared to function as a coactivator in a manner similar to AIB1 because both proteins activated transcription in a mutually exclusive manner (Fig. 1B). Coactivation is believed to be dependent on ligand binding, where a conformational shift in the AF2 domain (19) allows association of the coactivator with the AF2-modified hydrophobic pocket in the LBD region to allow formation of the coactivator complex (38). Although coactivation occurred in the absence of ligand, PDK1-PPARγ interaction was markedly enhanced by GW7845 in both reporter and ChIP assays (Figs. 1 and 5). This suggests that PDK1 prevented the recruitment of corepressors such as SMRT (39), an effect that was further enhanced by ligand. Indeed, SMRT completely suppressed PDK1-mediated coactivation (Fig. 1B).

Although PDK1 did not phosphorylate PPARγ, it did physically associate with the receptor that associated with PPRE in two PPARγ-activated genes (Fig. 5). All coactivators contain an LXXLL motif in an α-helical domain (38), and PDK1 contains the LXXLL-like domains ²⁷⁵LXXII²⁷¹ and V³¹³XXLL³¹⁷ in α-helices F and H, respectively (20) (Fig. 4). Expression of α-helices F through H (PDK1[256–322]) acted similarly to the holoenzyme, and mutation of either LXXLL-like motifs eliminated PPARγ activation, suggesting they act in a cooperative manner. It is not clear which domain is primary because mutation of either motif eliminated activity (Fig. 4), suggesting a destabilization of the hydrophobic core formed by helices F, G, and H.

Activation of PPARγ in either preadipocytes or mammary epithelial cells resulted in adipogenesis. These results are consistent with previous studies showing the adipocyte-enhancing effect of PPARγ2 expression in fibroblasts (12, 40), and the dependence of adipocyte differentiation during development on PPARγ (41) through its regulation of programmed adipocyte-specific gene expression (15). Overexpression of PDK1 in preadipocytes treated with a PPARγ ligand led to enhanced differentiation (Fig. 6), an effect that was blocked by kinase-dead PDK1. PDK1 similarly enhanced PPARγ-mediated adipocyte differentiation in mammary epithelial cells, a process that was blocked by the PI3K inhibitor LY294002, and is consistent with a recent report indicating the ability of secretory mammary epithelial cells to undergo adipocyte differentiation (42). Interestingly, PDK1 null cells exhibited an attenuated response to rosiglitazone, indicating the importance of the PDK1 signaling pathway in adipocyte differentiation. PPARγ agonists such as rosiglitazone regulate PPARγ-mediated transcription by promoting the dissociation of corepressors and recruitment of coactivators (43). Our data further suggest that PDK1 may regulate adipogenesis by serving as a coactivator of PPARγ.

MATERIALS AND METHODS

Cell Culture and Reagents

293T, 3T3-L1, and mouse mammary epithelial COMMA-1D cells were obtained from the Tissue Culture Shared Resource, Lombardi Comprehensive Cancer Center. 293T and 3T3-L1 cells were maintained in DMEM supplemented with 10% fetal calf serum (FCS) at 37 C under 5% CO₂. COMMA-1D cells were maintained in DMEM/F12 medium (Sigma-Aldrich Chemical Co., St. Louis, MO) supplemented with 5% FCS at 37 C under 5% CO₂. Mouse embryonic stem cells with homozygous deletion of PDK1 were kindly provided by Dr. Dario Alessi, University of Dundee (44). Antibodies were obtained from the following sources: anti-PPARγ (Santa Cruz Biotechnology, Santa Cruz, CA), anti-PDK1, anti-Gal-4 and anti-Myc-Tag (Upstate Biotechnology, Lake Placid, NY), horseradish peroxidase-conjugated secondary antibodies (Pierce Chemical, Rockford, IL). GW7845 was kindly provided by GlaxoSmithKline (Research Triangle Park, NC), and rosiglitazone was obtained through the Chemoprevention Branch, National Cancer Institute (Bethesda, MD). Insulin was purchased from Sigma-Aldrich Chemical Co.

Plasmids and Plasmid Construction

Plasmid pcDNA3.1-PPARγ1 was constructed by PCR amplification of full-length human PPARγ1 from pPax8-PPARγ1 (45) (kindly provided by Dr. Todd G. Kroll, University of Chicago, Chicago, IL) and subcloned into pcDNA3.1 (Invitrogen, Carlsbad, CA). Gal4-PPARγ1 and Gal4-LBD were constructed by PCR amplification from human PPARγ1 and cloned into pBind (Promega, Madison, WI) in-frame with the Gal4 DNA-binding sequence. pCMV5-MycPDK1 containing hPDK1 or KD-PDK1 with an N-terminal myc-tag (1) was kindly provided by Dr. Dario Alessi. All constructs were confirmed by sequencing and Western blotting of protein extracts from transfected 293T cells. pCDNA3.1-PPARγ/S225A, pCMV5-mycPDK1/I271R-I272R and pCMV5-mycPDK1/L316R-L317W were prepared by site directed mutagenesis with QuikChange (Stratagene, La Jolla, CA), and the appropriate pair of primers expressing the mutation, and confirmed by sequencing. The PDK1 fragment containing α-helices F-H and an N-terminal myc-tag (pCMV5-mycPDK1 [256–322]) was prepared by PCR amplification and the sequence confirmed by sequencing. Reporter plasmid, p3XPPRE-thymidine kinase (TK)-Luciferase (14) was generously provided by Dr. Mitchell Lazar (University of Pennsylvania, Philadelphia, PA). pcDNA3-AIB1 was kindly provided by Dr. Anna Riegel (Georgetown University, Washington, DC).

Reporter Assays

293T cells were grown in 24-well plates in DMEM containing 10% fetal calf serum; after 24 h, medium was replaced with DMEM containing 10% delipidated fetal calf serum (Sigma-Aldrich Chemical Co.). Cells were transfected using calcium phosphate precipitation (Promega) with the appropriate combination of luciferase reporter plasmid (p3XPPRE-TK-Luc for PPARγ or pG5Luc for Gal4 fusion proteins), vector expressing the gene of interest and empty control vector. After 24 h, cells were treated with 1.0 μm GW7845, and luciferase activity was measured 24 h after GW7845 treatment with the luciferase assay system (Promega).

Immunoprecipitation and Western Blotting

293T cells were cotransfected with PPARγ1 and PDK1 expression plasmids, and protein lysates prepared after 24 h by extraction in modified RIPA buffer [50 mm Tris (pH 7.4), 0.5% Nonidet P-40, 0.25% Na-deoxycholate, 125 mm NaCl, 1 mm EDTA, 1 mm EGTA, 1 mm Na₃VO₄ and protease inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN)] on ice for 1 h. Cells were sonicated for 10 sec, lysates centrifuged at 12,000 × g for 15 min at 4 C, and the supernatant removed for IP. Samples containing 500 μg protein were precleared for 30 min with 30 μl of Protein G Plus/Protein A agarose (Santa Cruz Biotechnology) and incubated for 1 h at 4 C with the appropriate antibodies or normal IgG preadsorbed to Protein G Plus/Protein A agarose. Immune complexes were collected after incubation overnight at 4 C, washed once with lysis buffer and three times with PBS. Protein was eluted by boiling in 50 μl 2× SDS sample buffer [125 mm Tris (pH 6.8), 10% 2-mercaptoethanol, 4% SDS, 20% glycerol], and eluted proteins were analyzed by SDS-PAGE and Western blotting using the appropriate antibodies.

GST-PPARγ Fusion Proteins

GST-PPARγ fusion proteins were expressed in BL21 Escherichia coli (Stratagene) and purified using glutathione-Sepharose 4B (Pharmacia Biotech). Beads were washed four times with PBS, and GST fusion proteins eluted with 10 mm reduced glutathione (Sigma-Aldrich Chemical). Protein purity was determined by SDS-PAGE in 4–20% Tris-glycine gels (Invitrogen) and stained with SimplyBlue Safestain (Invitrogen).

In Vitro Phosphorylation

PDK1 (25 ng) (Upstate Biotechnology) was incubated for 30 min with either GST-PPARγ (250 ng) or 25 ng SGK as substrate, and 10 μCi [γ³²P]ATP (Amersham-Pharmacia, Pittsburgh, PA), 100 μm ATP, 10 mm magnesium acetate, 50 mm Tris-HCl (pH 7.5), 0.1 mm EGTA, 0.1 mm EDTA, and 1 μm microcystin at 30 C. Reactions were stopped by addition of SDS sample buffer, boiled for 5 min and loaded onto a 10% Tris-glycine gel (Invitrogen). Proteins were transfer to 0.2 μm nitrocellulose and analyzed by autoradiography.

Viral Transduction of COMMA-1D Cells

Retroviral vector pCMV/hyg was generated by replacement of the Tet on/off control element on vector pRevTRE (CLONTECH, Inc., Palo Alto, CA) with the CMV promoter from pRc/CMV (Invitrogen) (46). Either pSRαMSVtkneo or pSRαSVtkneo encoding PDK1, KD-PDK1, Pax8-PPARγ, or PPARγ were cotransfected with the pSV-ψ⁻-E-MLV ecotropic vector into 293T cells. After 48 h, the supernatants were collected, mixed with an equal volume of fresh DMEM/F12 medium plus 2X supplement in the presence of 4 μg/ml polybrene, and added to COMMA-1D cells. After four rounds of infection, G418-resistant or hygromycin-resistant COMMA-1D cells were selected for 2 wk. To generate cells coexpressing PPARγ+PDK1 or PPARγ+KD-PDK1, a second round of transduction was carried out using amphotropic viruses produced in 293T cells cotransfected with pCMV/hyg-PDK1 or pCMV/hyg-KD-PDK1 and the pSV-ψ⁻-A-MLV amphotropic vector. After 2 wk of selection in 50 μg/ml hygromycin, the expression of both genes was confirmed by Western blotting.

ChIP Assay

ChIP analysis was performed following a protocol provided by Upstate Biotechnology under modified conditions. COMMA-1D cells (2 × 10⁷ cells) transduced with either empty virus or a retrovirus expressing PPARγ and PDK1 were grown in DMEM with 10% FCS. At 70–80% confluence, cells were cross-linked by adding 1.1% formaldehyde buffer containing 100 mm sodium chloride, 1 mm EDTA-Na (pH 8.0), 0.5 mm EGTA-Na, Tris-HCl (pH 8.0) directly to the culture medium for 10 min at 37 C. The medium was aspirated, the cells washed three times with ice-cold PBS containing 10 mm dithiothreitol and protease inhibitors, and lysed with warm 1% SDS buffer and incubated for 10 min on ice. The cell lysates were sonicated to shear DNA to 200-1000 bp, and samples were diluted 10-fold in ChIP dilution buffer (Upstate Biotechnology). To reduce nonspecific background, the cell pellet suspension was precleared with 60 μl of salmon sperm DNA/Protein-A agarose as a 50% slurry (Upstate Biotechnology) for 2 h at 4 C with mixing. Chromatin solutions were precipitated overnight at 4 C using 4 μg of anti-PPARγ (Santa Cruz sc-7196) and anti-PDK1 (Upstate Biotechnology) with rotation. Rabbit IgG was used as a negative control. Then 60 μl of salmon sperm DNA/Protein-A agarose slurry was added and incubated for 2 h at 4 C with rotation. The antibody/histone complex was collected by centrifugation and washed extensively using the manufacturer’s protocol. Input and immunoprecipitated chromatin were incubated at 65 C overnight to reverse cross-linking. After proteinase K digestion for 1 h, DNA was extracted using QIAGEN (Valencia, CA) spin columns, and precipitated DNA was analyzed by PCR of 30 cycles. The following primers were used for PCR to identify PPARγ-responsive elements: mouse LPL promoter, 5′-AAACCCCTCCTCTCTGCCTC-3′ and 5′-CCTCGGAGGAGGAGTAGGAG-3′; mouse aP2 promoter, 5′-CAAGCCATGCGACAAAGGCA-3′ and 5′-TAGAAGTCGCTCAGGCCACA-3′ (22, 23).

Adipocyte Differentiation

3T3-L1 cells were grown in DMEM supplemented with 10% fetal calf serum in 24-well plates coated with 0.2% gelatin (Sigma-Aldrich Chemical). Cells were transfected with either PDK1 or PPARγ1 expression plasmids using LipofectaminePlus (Invitrogen). When cells reached confluence after transfection, cells were treated with 5 μg/ml insulin, and 24 h later treated with 0.5 μm rosiglitazone. Medium containing rosiglitazone was replaced every other day for 7 d. Under these conditions (lack of IBMX and dexamethasone), adipocyte differentiation was virtually absent unless cells were transfected with either PPARγ or PDK1.

COMMA-1D cells were transduced with either empty virus, PPARγ, PDK1, KD-PDK1, PPARγ+PDK1 or PPARγ+KD-PDK1, and grown in 24-well plates in DMEM/F12 medium supplemented with 5% FCS. After reaching confluency, cells were treated as described for 3T3-L1 cells except that 1 μm GW7845 was used in place of rosiglitazone. After 7 d treatment, medium was removed, cells washed twice with PBS and fixed with 10% neutral buffered formalin in PBS for 15 min. Cells were then stained with 0.2% Oil Red O for 15 min or until fat droplets stained intense red, and photographed.

Wild-type mouse ES cells or ES cells containing a homozygous deletion of PDK1 were grown in 12-well plates containing DMEM supplemented with 15% FCS, 0.1 mm nonessential amino acids (Invitrogen), 10³ U/ml leukemia inhibitory factor, and β-mercaptoethanol, 7 μl/liter. After 24 h, cells were treated with 10 nm retinoic acid for 9 d (47), followed by treatment with 1 μm rosiglitazone for 10 d. Medium was then removed, cells washed twice with PBS, and fixed with 10% neutral buffered formalin in PBS for 15 min. Cells were then stained with 0.2% oil red O for 15 min or until fat droplets stained intense red, and cells photographed.

Adipocyte differentiation was also quantitated by use of the AdipoRed Assay Reagent (Cambrex, East Rutherford, NJ). After treatment of COMMA-1D cells with GW7845 as described above, cells were stained with AdipoRed reagent according to the manufacturer’s instructions. Fluorescence was measured in an Ultra384 fluorimeter (Tecan, Durham, NC) with excitation at 485 nm and emission at 535 nm. Three independent experiments were performed and the mean RFU value and sd were calculated.

Molecular Modeling

In the reported crystal structure of the catalytic domain of PDK1 (PDB ID: 1H1W) (20), two sequences were identified within the catalytic domain of PDK1 with potential LXXLL coactivator motifs, ²⁶⁸LGCII²⁷² and ³¹³VEKLL³¹⁷. The first motif lies within α-helix F, whereas the second motif resides near the C terminus of α-helix H. However, comparison of the reported crystal structure of the coactivator LXLL motif/PPARγ complex (48) with that of ²⁶⁸LGCII²⁷² within α-helix H revealed interference through van der Waals interactions due to positional effects of ²⁶⁸LGCII²⁷² within the middle of a long α-helix. However, an alternate LXLL motif denoted by ²⁷¹IIYQL²⁷⁵ occurred within α-helix F reading from the C terminus, and because these residues lie within a helix, the direction of the LXXLL moiety for interaction with PPAR becomes irrelevant. By superimposing the appropriate residues of IIYQL with that of the coactivator sequence, a model of the complex between PPARγ and the PDK1 domain encompassing α-helix F through H was generated (see Fig. 8). The coordinates for the whole complex were then subjected to energy minimization using AMBER force-field (49).

Acknowledgments

This work was supported by grants from the National Institutes of Health (NIH) (R01CA81565 and N01CN15017-44) to R.I.G. (R01CA70896, R01CA75503, R01CA86072, and R01CA86071) (to R.G.P.), and by a Comprehensive Cancer Center Support Grant from the NIH to the Lombardi Comprehensive Cancer Center, Georgetown University Medical Center.

Abbreviations

 
• AIB1

  Activated in breast cancer-1;

 
• AKT

  protein kinase B;

 
• aP2

  adipocyte fatty acid-binding protein;

 
• ChIP

  chromatin immunoprecipitation;

 
• FCS

  fetal calf serum;

 
• IP

  immunoprecipitation;

 
• KD

  kinase-dead;

 
• LBD

  ligand binding domain;

 
• LPL

  lipoprotein lipase;

 
• PDK1

  3-phosphoinositide-dependent protein kinase-1;

 
• PI3-K

  phosphatidylinositol 3-kinase;

 
• PPAR

  peroxisome proliferator-activated receptor;

 
• PPRE

  PPAR response element;

 
• SGK

  serum- and glucocorticoid-activated protein kinase;

 
• SMRT

  silencing mediator of retinoid and thyroid hormone receptors;

 
• TK

  thymidine kinase.