Peroxisome Proliferator-Activated Receptor γ Recruits the Positive Transcription Elongation Factor b Complex to Activate Transcription and Promote Adipogenesis

Abstract

Positive transcription elongation factor b (P-TEFb) phosphorylates the C-terminal domain of RNA polymerase II, facilitating transcriptional elongation. In addition to its participation in general transcription, P-TEFb is recruited to specific promoters by some transcription factors such as c-Myc or MyoD. The P-TEFb complex is composed of a cyclin-dependent kinase (cdk9) subunit and a regulatory partner (cyclin T1, cyclin T2, or cyclin K). Because cdk9 has been shown to participate in differentiation processes, such as muscle cell differentiation, we studied a possible role of cdk9 in adipogenesis. In this study we show that the expression of the cdk9 p55 isoform is highly regulated during 3T3-L1 adipocyte differentiation at RNA and protein levels. Furthermore, cdk9, as well as cyclin T1 and cyclin T2, shows differences in nuclear localization at distinct stages of adipogenesis. Overexpression of cdk9 increases the adipogenic potential of 3T3-L1 cells, whereas inhibition of cdk9 by specific cdk inhibitors, and dominant-negative cdk9 mutant impairs adipogenesis. We show that the positive effects of cdk9 on the differentiation of 3T3-L1 cells are mediated by a direct interaction with and phosphorylation of peroxisome proliferator-activated receptor γ (PPARγ), which is the master regulator of this process, on the promoter of PPARγ target genes. PPARγ-cdk9 interaction results in increased transcriptional activity of PPARγ and therefore increased adipogenesis.

POSITIVE TRANSCRIPTION elongation factor b (P-TEFb) facilitates transcription elongation through phosphorylation of RNA polymerase II (RNA pol II) carboxyl-terminal domain (CTD). The active core of P-TEFb comprises cyclin-dependent kinase 9 (cdk9) and a C-type regulatory cyclin (cyclin T1, T2, K) (reviewed in Ref. 1). Cdk9 exists in mammalian cells in two isoforms, p42 cdk9 and p55 cdk9, with a tissue-specific expression pattern, responding to different signals (2). Both isoforms phosphorylate Ser-2 of multiple heptapeptide repeats of CTD RNA pol II releasing the repression action of the 5,6-dichloro-1-β-ribofuranosyl-benzimidazole (DRB) sensitivity-inducing factor and the negative elongation factor. P-TEFb is required not only as a basic transcription elongation factor, but it is also recruited by some transcription factors to activate transcriptional elongation from specific promoters. For example, the requirement of cyclin T1/cdk9 complex by the HIV-1 Tat protein to efficiently elongate the viral RNA has been extensively studied (3, 4). Other transcription factors targeting P-TEFb to the promoters of its target genes include MyoD (5), c-Myc (6), signal transducer and activator of transcription 3 (7), nuclear factor-κB (NFκB) (8), androgen receptor (AR) (9), and the aryl hydrocarbon receptor (10), implicating the cyclin T/cdk9 complex in the regulation of cellular processes such as heat-shock response, antigen presentation and processing, apoptosis, muscle cell differentiation, cell growth, or cell proliferation. Moreover, interaction of c-Myc with cyclin T1 and cdk9 results in efficient transcription of c-Myc target genes, and inhibition of cdk9 with DRB, at a concentration that does not inhibit the general transcription process, results in the block of c-Myc-induced proliferation and apoptosis (6). Of particular interest is the participation of P-TEFb in differentiation processes. Interaction of cdk9 with MyoD results in increased differentiation of muscle cells. Interestingly, MyoD interacted with, and was phosphorylated by, the cyclin T2/cdk9 complex, whereas c-Myc and NFκB were shown to recruit the cdk9/cyclin T1 complex (5). Hence, the specificity of the interaction might be dictated by the cyclin T component of the P-TEFb complex (1). Cdk9 is also required for the monocyte differentiation program. In these cells, cyclin T1 is highly induced upon stimulation with phorbol 12-myristate 13-acetate (11). High levels of expression of cyclin T1 and T2, as well as cdk9, in differentiated tissues suggest the participation of P-TEFb in the activation or maintenance of specific differentiation processes (12, 13). Interestingly, white adipose tissue showed a very high level of expression of cyclin T1 (12). All these studies prompted us to investigate the role of P-TEFb in the adipocyte differentiation process.

Adipogenesis is a particular system, which involves two major events: preadipocyte proliferation and adipocyte differentiation. Upon reaching confluence, proliferative preadipocytes become growth arrested by contact inhibition. Those cells reenter cell cycle after hormonal induction, arrest proliferation again, and undergo terminal adipocyte differentiation. This last stage is characterized by the coordinated expression of specific genes that will finally determine the specific adipocyte phenotype of the cells. The major regulator of terminal adipocyte differentiation is the peroxisome proliferator-activated receptor γ (PPARγ) (reviewed in Refs. 14 and 15), which is induced during the initial phases of these terminal stages of adipogenesis by CCAAT enhancer-binding protein-β and -δ, as well as by E2F1 (16). PPARγ, upon activation by either fatty acid derivatives or antidiabetic thiazolidinediones, drives the expression of several adipocyte-specific genes, such as the fatty acid binding protein (aP2), thus transforming the cell into the characteristic lipid-rich adipocyte (17). Subsequent studies have demonstrated that ectopic expression of PPARγ further induces adipocyte differentiation (18). This pivotal role of PPARγ in adipocyte differentiation is also highlighted by the phenotype observed in humans with mutations in the PPARγ gene and by PPARγ-deficient mice, which are essentially devoid of white adipose tissue (19).

In this study we show that cyclin T1, T2, and cdk9 are expressed in adipocytes. Furthermore, cdk9 activity is increased during adipocyte differentiation. We show that inhibition of cdk9 activity impairs adipogenesis, whereas ectopically expressed cdk9 is highly adipogenic in 3T3-L1 cells. The adipogenic effects of cdk9 are mediated by the induction of PPARγ transcriptional activity through direct phosphorylation. Finally, we show that PPARγ is complexed with P-TEFb on promoters of PPARγ target genes.

Results

P-TEFb Is Expressed and Active in Differentiated Adipocytes

While evaluating the participation of the P-TEFb complex in adipocyte differentiation we found that cdk9 was expressed at all stages of differentiation of 3T3-L1 preadipocytes, both at the RNA and protein level (Fig. 1, A and B). Interestingly, Western blot analysis as well as RNA quantification showed that the expression of p55 cdk9 protein isoform was strongly increased during adipocyte differentiation, whereas p42 cdk9 expression was unchanged (Fig. 1, A and B). No differences in the expression of cyclin T1 or cyclin T2 were observed (data not shown). aP2 mRNA expression was quantified as a marker of differentiation (Fig. 1C). To further study the expression of P-TEFb during adipogenesis, immunofluorescence analyses were performed. During early stages of differentiation of 3T3-L1 cells, cdk9 was evenly distributed between the cytoplasm and the nucleus (Fig. 1D; d 0 and d 1). Starting at d 3, cdk9 expression became more prominent in the nuclear compartment (Fig. 1D). Strikingly, at this stage of differentiation, coexpression of cdk9 with PPARγ in the nucleus of differentiating cells could be observed (Fig. 1D; d 3 and d 8). Similar results were observed when the expression of cyclins T1 and T2 were analyzed (supplemental Fig. 1 published on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). Furthermore, immunoprecipitated cdk9 from differentiated adipocytes was able to phosphorylate purified recombinant RNA pol II CTD, showing a gradual increase in cdk9 activity, as measured by the level of RNA pol II CTD phosphorylation, indicating that cdk9 was indeed active in adipocytes (Fig. 1E). This was consistent with the increase in nuclear localization of cdk9 at the same stages of adipocyte differentiation (Fig. 1D).

Fig. 1.

Expression of Cdk9 during 3T3-L1 Differentiation A, Western blot analysis of nuclear extracts prepared at different days of adipocyte differentiation of 3T3-L1 cells. The proteins detected with specific antibodies are indicated. Histone H1 is used as a load control marker. B and C, Quantification of mRNA expression levels by real-time PCR of the long form (cdk9p55) and total cdk9 (B), and of the adipogenic marker aP2 (C) at the indicated times of differentiation. Results were normalized by the expression level of 18s RNA. D, Comparative analysis of PPARγ and cdk9 expression by immunofluorescence in 3T3-L1 cells induced to differentiate. Days of differentiation indicate proliferating (d −1), confluent (d 0), reentry into cell cycle (d 1), early differentiation (d 3), and terminally differentiated (d 8) cells. PPARγ-expressing cells are labeled in red whereas cdk9 expressing cells are in green. Nuclei were visualized with Hoechst staining. E, In vitro kinase assay using immunoprecipitated cdk9 from 3T3-L1 cells at different stages of differentiation. Purified human RNA pol II CTD was used as substrate. ppCTD, Phosphorylated RNA pol II CTD; rel., relative.

These results suggested a role of P-TEFb in adipose tissue biology and differentiation.

DRB Inhibition of cdk9 Results in Impaired Clonal Expansion and Terminal Adipocyte Differentiation

To further assess the role of cdk9 in the differentiation of 3T3-L1 cells, cdk9 activity was inhibited using DRB. 3T3-L1 cells treated with either vehicle or DRB were compared for their ability to differentiate into adipocytes. After 8 d in differentiation media, normal lipid accumulation was observed in control cells, whereas a dose-dependent decrease in lipid accumulation was observed in cells treated with DRB, as showed by oil red O staining (Fig. 2A). Quantitative RT-PCR performed on differentiated 3T3-L1 cells confirmed a DRB dose-dependent decrease in expression of the PPARγ target gene aP2, which is a marker of adipocytes (Fig. 2B). 3T3-L1 preadipocytes reenter cell cycle after hormonal induction of differentiation (the clonal expansion phase). Because this is a required event of these cells before terminal differentiation into adipocytes, we next explored the hypothesis that cdk9 participates in adipogenesis through the control of cell cycle during the clonal expansion phase. A decrease in cell proliferation during the clonal expansion phase (d 1 and 2) was observed, as measured by bromodeoxyuridine incorporation assays, in cells treated with DRB compared with cells treated with vehicle (Fig. 2C). This indicated that inhibition of cdk9 by DRB suppressed, at least in part, the clonal expansion phase of adipogenesis. Furthermore, the decrease in cyclin B1 mRNA expression at d 1 of differentiation in cells treated with DRB suggested that cdk9 could participate in the control of the G₂/M transition (Fig. 2D). No differences in the expression of cyclin D1 or dihydrofolate reductase were observed at this stage, whereas a significant reduction in the expression of these genes was observed at d 2 of differentiation (Fig. 2D). To further elucidate whether the role of cdk9 was limited to the regulation of the clonal expansion phase of adipocyte differentiation, cdk9 activity was inhibited starting at d 3 of differentiation. At this particular stage, 3T3-L1 cells are quiescent and have already gone through the clonal expansion phase (data not shown). Five days after treatment (d 8), oil red O staining indicated an inhibitory dose-dependent effect of DRB in lipid accumulation (Fig. 2E). A dose-dependent decrease in aP2 mRNA expression in DRB-treated cells further demonstrated inhibition of adipogenesis (Fig. 2F). These results suggested that cdk9 has a dual role in adipogenesis, i.e. a first role promoting the clonal expansion phase, and a second role promoting terminal differentiation.

Fig. 2.

Effects of Cdk9 on Adipogenesis A–E, Representative micrographs of oil red O staining of 3T3-L1 cells differentiated in vitro for 8 d in the presence or absence of the indicated concentrations of the specific cdk9 inhibitor DRB added either at the induction of differentiation (A) or 2 d after induction (E). mRNA of differentiated cells was analyzed for the expression of the adipocyte marker aP2 by quantitative PCR in response to DRB added either before (B) or after (F) the clonal expansion phase. Results were normalized by the expression of the β-actin RNA. Cell cycle status of the cells was analyzed by quantification of bromodeoxyuridine incorporation either in the absence or presence of 30 μm DRB added before the clonal expansion phase (C). mRNA expression of cyclin B1, dihydrofolate reductase, and cyclin D1 was quantified at different times of adipocyte differentiation (d0, d1, d2) in the absence or in the presence of DRB added before the clonal expansion phase (D). Cyc, Cyclin; rel., relative.

Impaired Expression of cdk9 Modulates Adipogenesis

The observed effects of DRB in adipogenesis could be the result of the inhibition of kinases other than cdk9. To prove the implication of cdk9, a 3T3-L1 cell line overexpressing a dominant-negative, kinase-dead cdk9 (DNcdk9) was created. Oil red o staining indicated that after 8 d in differentiation medium, these cells lost their capacity to differentiate into adipocytes, compared with cells expressing an empty vector (Fig. 3A). Overexpression of DNcdk9 was assessed by PCR assay (Fig. 3A, bottom panel). Inhibition of adipogenesis was further proved by decreased aP2 mRNA expression in DNcdk9 cells (Fig. 3B).

Fig. 3.

Cdk9 Promotes Adipogenesis A, Micrographs of oil red O staining of 3T3L1 adipocytes expressing either an empty vector (control) or a vector encoding a DNcdk9 mutant 8 d after induction of differentiation. The expression of DNcdk9 is visualized by PCR using T7 and cdk9_42 reverse primers (see Materials and Methods) (lower panel). B, mRNA expression of aP2, analyzed by real-time PCR, of cells used in panel A. Results were normalized by the expression of the 18s RNA. C, Representative micrographs of oil red O staining of 3T3L1 adipocytes expressing either an empty vector (pcDNA3-3T3L1) or a vector encoding cdk9 (cdk9-3T3L1) 4 d after induction of differentiation. Overexpression of cdk9 in cdk9-3T3L1 cells is visualized by Western blot (lower panel). Histone H1 was used as a load control marker. D, mRNA expression of aP2, analyzed by real-time PCR, of cells used in panel C. Results were normalized by the expression of the 18s RNA. hist, Histone.

Because inhibition of cdk9 activity resulted in decreased adipocyte differentiation, we next tested whether overexpressing cdk9 could accelerate this process. To test this hypothesis, we created stable cell lines expressing the pcDNA3-cdk9 and control vector. Overexpression of cdk9 in these cells was demonstrated by Western blot analysis (Fig. 3C, bottom panel). After 4 d of differentiation only a small proportion of cells expressing a control vector were differentiated (Fig. 3, A and B), whereas a strong increase in lipid accumulation and up-regulation of adipogenic mRNA markers was observed in cells overexpressing cdk9 (Fig. 3, C and D), further demonstrating that cdk9 is an adipogenic factor.

Cdk9 Increases PPARγ Activity

Because P-TEFb has an impact on adipogenesis, we tested whether these effects could be mediated by PPARγ, which is the master regulator of adipocyte differentiation. To analyze the influence of cdk9 on PPARγ activity, we performed cotransfection experiments using a PPARγ-responsive, luciferase-based reporter construct [PPRE-thymidine kinase (TK)-Luc] and expression vectors for PPARγ and cdk9. A cdk9 dose-dependent induction of luciferase activity was observed in the presence of PPARγ both in the presence and absence of the PPARγ agonist rosiglitazone (Fig. 4A). To further prove that the effects of cdk9 on the PPARγ-responsive promoter were mediated by PPARγ, a chimeric Gal4-PPARγ construct containing the Gal4 DNA-binding domain fused to either the PPARγ AB that contains the ligand-independent PPARγ transactivating domain, or the PPARγ ligand binding domain (LBD) that contains the ligand-dependent transactivating domain were used in combination with a Gal4-responsive, luciferase-based reporter construct [upstream activating sequence (UAS)-TK-Luc]. Consistent with the results using a PPRE-TK-Luc, a robust dose-dependent increase in luciferase activity was observed when cdk9 was cotransfected with the Gal4-PPARγ AB construct (Fig. 4B). No effects of cdk9 were observed when the Gal4-PPARγ LBD construct was used either in the presence or absence of rosiglitazone (Fig. 4B). These results demonstrated that cdk9 modulates PPARγ activity. This was further proved by the observation that cdk9 inactivation by cotransfection of a DNcdk9 expression vector resulted in the attenuation of PPARγ activation by rosiglitazone (Fig. 4C). Moreover, inactivation of cdk9 kinase activity by DRB resulted in a complete inhibition of PPARγ activity (Fig. 4D). Finally, to ascertain that cdk9 activates PPARγ-mediated transcription, chromatin immunoprecipitation (ChIP) studies on the aP2 promoter, which is a known PPARγ target gene, were performed in both nondifferentiated and differentiated 3T3-L1 adipocytes. A 200-bp fragment of the mouse aP2 promoter, containing the binding site of PPARγ, was amplified by PCR when anti-cdk9, anti-PPARγ, or antiacetylated histone H4 antibodies were used to immunoprecipitate chromatin from differentiated 3T3-L1 cells (Fig. 4E, upper panel). No amplification product was observed when immunoprecipitated chromatin from confluent, nondifferentiated 3T3-L1 preadipocytes was used as template, or when chromatin was immunoprecipitated using an irrelevant antibody, or when a non-PPRE-containing region of the aP2 or glyceraldehyde-3-phosphate dehydrogenase (GAPDH) promoter were used as template (Fig. 4E), demonstrating the specificity of the binding. The results of the ChIP assays demonstrate that the complex cdk9-PPARγ is present on the promoter of a PPARγ target gene. Furthermore, the presence of acetylated histone H4 on the PPARγ-binding site of the aP2 promoter suggests that, in the presence of cdk9, this promoter is active. These results suggest that the stimulatory role of P-TEFb during adipogenesis is likely the result of its ability to modulate PPARγ activity.

Fig. 4.

Cdk9 Modulates PPARγ-Mediated Transactivation in Vitro and in Vivo A, Activity of the PPRE-TK-luc reporter transfected in NIH-3T3 cells in the presence or absence of the PPARγ agonist rosiglitazone. Cells were transfected with an expression vector encoding PPARγ and with the indicated amounts of expression vector coding for cdk9. The luciferase activity was measured and normalized to the expression of a β-gal-encoding plasmid. Activity is presented relative to the values obtained in cells transfected with PPARγ and treated with dimethylsulfoxide. B, Relative luciferase activity as determined after transfection of NIH-3T3 cells with the Gal4-responsive reporter construct UAS-TK-Luc. Cells were transfected with an expression vector for Gal4-PPARγ A/B fusion protein or Gal4-PPARγ LBD fusion protein in the absence or presence of increasing concentrations of a cdk9 and cyclin T1 expression vectors (collectively termed “P-TEFb” in the figure) and in the absence or presence of rosiglitazone as indicated. Results were normalized for the expression of a β-gal reporter. Values are the mean of three independent experiments. C, NIH-3T3 cells were transfected with PPRE-TK-luc and expression vectors coding for PPARγ and a dominant-negative form of cdk9 (DNcdk9) in the presence or absence of PPARγ agonist rosiglitazone. D, Same transfection as in panel A using PPRE-TK-luc and PPARγ vectors. Cells were treated with increasing concentrations of the cdk9 inhibitor DRB (10–50 μm). Results were normalized to the expression of a β-gal-encoding plasmid. E, ChIP assay demonstrating binding of cdk9 to the aP2 promoter. Cross-linked chromatin from either confluent preadipocytes (lower panel) or 3T3L1 adipocytes differentiated during 7 d (upper panels) was incubated with antibodies against PPARγ, cdk9, acetylated histone H4, with purified rabbit IgGs, or without any antibody (mock). Immunoprecipitates were analyzed by PCR using primers specific for the aP2 promoter region containing a PPRE (aP2 prom), a region of the aP2 promoter not containing the PPRE, or the GAPDH promoter (GAPDH prom). The input, included in the PCR, represents 20% of the total chromatin. rosi, Rosiglitazone; RLU, relative light units.

Cdk9 Physically Interacts with PPARγ

To test whether the induction of PPARγ activity in the presence of P-TEFb is the consequence of a direct interaction between PPARγ and cdk9, nuclear extracts from COS cells transfected with cdk9 and PPARγ expression vectors were immunoprecipitated with cdk9 or irrelevant antibodies. PPARγ protein could be detected in cdk9 immunoprecipitates as assessed by Western blot analysis (Fig. 5A), suggesting that a PPARγ-cdk9 complex is present in these cells. To identify the PPARγ domain responsible for the interaction, glutathione-S-transferase (GST)-PPARγ fusion proteins containing different PPARγ domains were incubated with in vitro translated ³⁵S-radiolabeled cdk9 (Fig. 5B). Interestingly, we found that cdk9 binds to both the A/B and DE domains of PPARγ (Fig. 5, C and D). In particular, cdk9 bound to a region from amino acids 1–72 and 292–399. The same results were observed when cyclin T1 or T2 was used instead of cdk9 (data not shown). These results further demonstrated a direct interaction between cdk9 and PPARγ.

Fig. 5.

PPARγ Interacts with cdk9 A, Coimmunoprecipitation of PPARγ and cdk9 from Cos cells transfected with PPARγ and cdk9 expression vectors. Extracts were immunoprecipitated with a cdk9 antibody or purified rabbit IgGs (mock) and revealed by an anti-PPARγ antibody. One twentieth of total extract is shown as a control (input). B, Schematic representation of the deletion GST-PPARγ constructs used in the subsequent experiments. C and D, GST pull-down assay showing the interaction of cdk9 with the A/B (C) or DEF (D) domains of PPARγ. In vitro translated ³⁵S-radiolabeled cdk9 protein was incubated with glutathione-sepharose-bound GST-PPARγ (deletion constructs) fusion proteins or GST alone. Bound proteins were separated by SDS-PAGE and detected by autoradiography. Input represents total in vitro translated protein. IP, Immunoprecipitation.

Cdk9 Phosphorylates PPARγ

Cdk9 is a cyclin-dependent kinase that modulates the activity of its substrates through phosphorylation. We wanted therefore to test whether modulation of PPARγ activity implicated its phosphorylation by cdk9. Overexpression of cdk9 and cyclin T in Saos cells resulted in the accumulation of the phosphorylated form of PPARγ, as assessed by Western blot analysis (Fig. 6A), suggesting that PPARγ could be a target for cdk9. Moreover, immunoprecipitated cdk9 from 293 cells was able to in vitro phosphorylate purified full-length GST-PPARγ, whereas neither GST alone nor GST-PPARγ LBD, which contains the PPARγ LBD were phosphorylated by cdk9 (Fig. 6B). In contrast, GST-PPARγ A/B, which contains the PPARγ A/B domain was indeed phosphorylated by cdk9 (Fig. 6B). The PPARγ A/B domain contains S112, which has been shown to be phosphorylated by other serine-threonine kinases, such as cdk7. To investigate whether this residue was the target for cdk9, S112 was replaced by an alanine (S112A). GST-PPARγ A/B-S112 was not phosphorylated by cdk9 (Fig. 6B, right panel). These results demonstrate that cdk9 phosphorylates PPARγ at S112. To further elucidate how PPARγ phosphorylation could influence its transcriptional activity, we correlated the expression of the PPARγ target gene aP2 with the phosphorylation status of PPARγ. As expected, 3T3-L1 adipocytes treated with rosiglitazone increased the expression of aP2 (Fig. 6C). This effect was blunted by the addition of the PPARγ antagonist GW9662 (Fig. 6C). Strikingly, maximal aP2 induction correlated with the accumulation of the PPARγ phosphorylated forms (Fig. 6D). Similarly, the observed decrease in aP2 expression upon incubation with GW9662 correlated with a switch to nonphosphorylated PPARγ (Fig. 6D). Furthermore, incubation of 3T3-L1 adipocytes with the cdk9 inhibitor DRB resulted in a decrease in aP2 mRNA expression to the levels observed in cells treated with the PPARγ antagonist GW9662 (Fig. 6E). Interestingly, PPARγ phosphorylation was inhibited when cells were incubated with DRB (Fig. 6F). Quantification of the relative intensity of the bands demonstrated that the observed reduction in PPARγ phosphorylation levels was similar to that observed using the PPARγ antagonist GW9662 (Fig. 6, D and F). Overall, these results suggested that phosphorylation by cdk9 increases PPARγ activity. Further supporting this hypothesis was the observation that, in differentiated adipocytes, phosphorylated PPARγ is complexed with the aP2 promoter as demonstrated by ChIP assays (Fig. 6G). Coincident with phospho-PPARγ is the presence of cdk9 and acetylated histone H4 in the aP2 PPRE, indicating that, under these conditions, the promoter is active (Fig. 6G).

Fig. 6.

Cdk9 Phosphorylates and Activates PPARγ A, Western blot analysis of PPARγ phosphorylation status in Saos cells transfected with expression vectors for PPARγ, cdk9/cyc T1, or both in the absence or presence of rosiglitazone. Cells were treated 2 h before harvesting with either rosiglitazone (1 μm) or the dimethylsulfoxide vehicle. The phospho-PPARγ slow-migrating form is indicated by p. B, In vitro kinase assay using purified GST, GST-PPARγ full-length, GST-PPARγ LBD, GST-PPARγ A/B, or GST-PPARγ A/B S112A fusion protein as a substrate and immunoprecipitated cdk9 from human embryonic kidney 293 cells transfected with cdk9. Migration of the different constructs in the SDS gels is indicated by arrows. Specificity of the kinase reaction is verified by the use of the cdk9 kinase inhibitor DRB (250 μm). C, RT-PCR measuring aP2 mRNA levels in 3T3-L1 adipocytes treated for 6 h with rosiglitazone and increasing concentrations of the PPARγ antagonist GW9662. Results were normalized for β-actin expression. D, Western blot analysis of 3T3-L1 adipocytes, treated for 2 h with the indicated concentrations of rosiglitazone and GW9662, showing PPARγ phosphorylation status (left panel). A p indicates migration of phospho-PPARγ. The relative intensity of the bands of phosphorylated/nonphosphorylated PPARγ is quantified in the right panel. E, Quantitative PCR analysis of aP2 mRNA expression in 3T3-L1 adipocytes treated with either rosiglitazone at 1 μm or rosiglitazone and DRB (30 μm). Results were normalized for β-actin expression. F, Western blot analysis of PPARγ phosphorylation (slow migrating forms) upon incubation of 3T3-L1 cells with rosiglitazone or rosiglitazone and the cdk9 inhibitor DRB for 2 h at the indicated concentrations. The relative intensity of the bands of phosphorylated/nonphosphorylated PPARγ is quantified in the right panel. G, ChIP assay demonstrating binding of phosphorylated PPARγ to the aP2 promoter. Cross-linked chromatin from 3T3-L1 adipocytes differentiated during 5 d was incubated with antibodies against PPARγ, phospho-PPARγ, cdk9, acetylated histone H4, or with purified rabbit IgGs. Immunoprecipitates were analyzed by PCR using primers specific for the aP2 promoter region containing a PPRE (aP2 prom). The input, included in the PCR, represents 20% of the total chromatin. Rosi, Rosiglitazone; FL, full length.

Cdk9 Has a Role in Adipocyte Biology

In addition to its participation in adipocyte differentiation, PPARγ also plays a major role in adipocyte biology through regulation of genes implicated mainly in lipogenesis. Because cdk9 modulates PPARγ activity, we asked whether cdk9 could also participate in these processes. First, we analyzed the mRNA expression of lipogenic genes such as aP2, lipoprotein lipase (LPL), stearoyl coenzyme A desaturase (SCD1), and fatty acid synthase (FAS). DRB treatment of 3T3-L1 adipocytes resulted in the inhibition of the expression of aP2, LPL, SCD1, and FAS (Fig. 7A), whereas no changes in the expression of genes implicated in lipolysis were observed (data not shown). These results suggested that similar to PPARγ, cdk9 could also be implicated in lipogenesis. Interestingly, expression of cdk9 was strongly increased in adipose tissue of mice fed a high-fat diet (Fig. 7B). Furthermore, genetically obese db/db mice expressed high levels of cdk9 mRNA in adipose tissue compared with control nonobese mice (Fig. 7B). Similar results were observed for cyclin T1 and cyclin T2 expression (supplemental Fig. 2).

Fig. 7.

Cdk9 Participation in Adipose Tissue Biology A, mRNA expression of lipogenic genes in 3T3-L1 adipocytes treated or not with DRB measured by quantitative PCR. DRB was added for 24 h in a concentration of 50 μm. Results are normalized for the expression of β-actin mRNA. B, Expression of cdk9 mRNA measured by real-time PCR in the adipose tissue of mice fed a chow diet (C56Bl/6), a high-fat diet (HFD), or in the adipose tissue of obese db/db mice. Results are normalized for the expression of β-actin mRNA. rel., Relative.

Discussion

General transcription is regulated by the activity of RNA pol II. The P-TEFb complex plays a crucial role in the control of transcriptional elongation through phosphorylation of CTD of RNA pol II. Increasing evidence suggests that P-TEFb also participates in the transcriptional regulation of specific genes through recruitment to specific promoters by particular transcription factors. Such is the case for c-Myc (20), NFκB (8), AR (9), and MyoD (5). As a result of the regulation of specific genes, P-TEFb coordinates cellular processes ranging from apoptosis, monocyte differentiation, or myogenesis (reviewed in Ref. 21). In this study, we analyzed the participation of cdk9 in adipocyte differentiation, and we found that cdk9 was required for adipogenesis. This is supported by several observations. First, expression of cdk9p55 is highly up-regulated during adipocyte differentiation (Fig. 1). Second, cdk9 activity is increased during adipogenesis (Fig. 1). Third, overexpression of cdk9 results in increased adipogenesis, whereas inhibition of cdk9 impairs the adipogenic potential of the cells (Figs. 2 and 3). Finally, cdk9 increases PPARγ activity through direct interaction with PPARγ on the promoters of PPARγ target genes (Fig. 4). The central role of PPARγ in adipose tissue differentiation and biology has been largely demonstrated. Consequently, any factor modulating PPARγ activity, and such is the case for cdk9, will have a major impact on adipocyte differentiation.

A similar scenario has been observed for the transcription factor MyoD. In this case, cdk9 phosphorylates MyoD and increases its transcriptional activity (5). Similarly, PPARγ is also phosphorylated by the P-TEFb complex, which may account for the observed effects of cdk9 on PPARγ activity. Interestingly, PPARγ is phosphorylated at S112, exactly the same residue previously reported to be phosphorylated by MAPK and cdk7 (22–24). The effects of PPARγ phosphorylation on PPARγ activity, so far, are inconsistent. MAPK phosphorylation of PPARγ has been reported to inhibit its activity (22, 23), whereas phosphorylation by cdk7 has been shown to be required for PPARγ-mediated transactivation (24). Our results are in agreement with this last study and show that PPARγ phosphorylation by cdk9 results in PPARγ activation. Three different observations support this hypothesis. First, cdk9 mediates PPARγ activation in transient transfection assays (Fig. 4). Second, PPARγ phosphorylation correlates with PPARγ maximal activity in transactivation of target genes (Fig. 6). Finally, phosphorylated PPARγ is complexed to the active promoters of target genes together with cdk9 (Fig. 6). How can phosphorylation of PPARγ at the same residue by different kinases result in either activation or repression of this transcription factor? One possible explanation is cofactor recruitment. The interaction of PPARγ with either kinase could modulate the interaction of PPARγ with different cofactors. In the case of MAPK, the interaction could result in the recruitment of corepressors, whereas for cdk7 and cdk9, the interaction could result in the recruitment of coactivators and the general transcriptional machinery. Recruitment of coactivators is probably not the case, because we did not observe any difference in cofactor recruitment to PPARγ in the absence or presence of cdk9 (data not shown). Alternatively, phosphorylation of PPARγ might be required for optimal activation. The effects observed upon MAPK inactivation on PPARγ could be indirect, such as replacement of PPARγ by a more potent transcription factor. Unfortunately, no data demonstrating the presence of PPARγ on the promoters of its target genes when MAPK inhibitors are used are available to date. Because phosphorylation of PPARγ coincides with RNA pol II CTD phosphorylation, we believe that PPARγ could tether the P-TEFb complex to the RNA pol II complex, facilitating phosphorylation and therefore elongation of the transcription of its target gene. This is the mechanism that has been suggested to explain the positive effects on the transcription of AR (9), AhR (10), or NFκB (8), mediated by the interaction of cdk9 with these transcription factors. In contrast to this, another mechanism of regulation of P-TEFb activity was described recently for the glucocorticoid receptor. In this scenario glucocorticoid receptor inhibits P-TEFb activity by competing for binding to RelA and p50 complex on the promoter of the IL-8 gene, resulting in the repression of this gene (25).

The analysis of the expression of cdk9 during adipocyte differentiation showed that the recently identified cdk9 isoform, p55, which is transcribed from an alternative promoter (2), increases during differentiation. This pattern of expression of p55cdk9 suggests that it may be a PPARγ target gene. Interestingly, we have identified a highly conserved PPAR-responsive element in the promoter of p55cdk9 (data not shown). Regulation of cdk9 expression by PPARγ would result in a positive feed-back loop increasing the availability of cdk9 to activate PPARγ-mediated transcription.

During adipocyte differentiation, which is a highly coordinated process, only a subset of genes are activated resulting in the specific adipocyte phenotype (reviewed in 15). This selective transcriptional activation supports the idea that the effects of cdk9 are rather gene specific than an effect on general transcription in the context of adipogenesis, and likely in other differentiation processes. Most interesting is the fact that cdk9 could participate in adipose tissue biology in addition to adipocyte differentiation. In differentiated adipocytes cdk9 inactivation results in the inhibition of the expression of genes implicated in lipogenesis, which is consistent with the increased expression of cdk9 that we found in mouse models of obesity. This may have important therapeutic implications, for instance in the context of obesity. Studies on the effects of specific cdk9 inhibitors in adiposity in mice are ongoing.

In summary, in this study we provide evidence that the nuclear receptor PPARγ uses the P-TEFb complex to stimulate the transcription of its target genes to drive adipogenesis.

Materials and Methods

Materials

All chemicals, except if stated otherwise, were purchased from Sigma Chemical Co. (St. Louis, MO). Pioglitazone was a kind gift of Takeda Pharmaceuticals (Osaka, Japan), and rosiglitazone was kindly provided by Novo Nordisk A/S (Bagsvaerd, Denmark). Anti-cdk9 C-20 antibody, anti-cyclin T2 S-14 antibody, anti-PPARγ E-8 antibody, anti-PPARγ H-100 antibody, and antihistone H1 FL-219 antibody were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-cdk9 ab6544 antibody, anti-cyclin T1 ab2098 antibody, and the acetylated histone H4 (Lys 9) antibody were from Abcam Ltd. (Cambridge, UK). Anti-phospho-PPARγ (Ser112) antibody was purchased from Euromedex (Mundolsheim, France), The RNA polymerase II CTD purified protein was purchased from ProteinOne (Bethesda, MD).

Plasmids and Oligonucleotides

PBSK and pcDNA3 vectors were purchased from Stratagene (La Jolla, CA). GST-PPARγ AB, GST-PPARγ DE, PPRE-TK-Luc, and the PPARγ expression vector were described previously (26, 27). GST-PPARγ deletion mutants were generated by PCR and inserted into pGEX-4T-1 vector. A pCMV β-galactosidase vector was used as an internal control for transfection efficiency in mammalian cells. pCMV Sp6-cdk9, pCMV Sp6.1-cyclinT2, and pCMV Sp6-cyclin T1 expression vectors were purchased from Medical Research Council Geneservice (Cambridge, UK), a distributor for the Mammalian Gene Collection. The pCMV-DNcdk9 hemagglutinin-tagged vector was a kind gift from Monsef Benkirane. DNcdk9 contains a point mutation at nucleotide 563, which converts Asp to Asn. The following plasmids were kindly provided from Karsten Kristiansen: upstream activating sequence UAS-TK-luc reporter, pSG5-cdk9, pRc/CMV-cyclin T1, pcDNA3.1-PPARγ2, Gal4-PPARγ A/B, Gal4-PPARγ LBD, pGEX-5X-2 GST-PPARγ2 full-length, GST-PPARγ2 LBD, GST-PPARγ2 A/B, and GST-PPARγ2 A/B S112A. Oligonucleotides used in ChIP assays to amplify the mouse aP2 promoter in PPRE region are 5′-GAGCCATGCGGATTCTTG-3′ and 5′-CCAGGAGCGGCTTGATT GTTA-3′; for non-PPRE region of the aP2 promoter: 5′-CAGCCCCACATCCCCACAGC and 3′-GGATGCCCAACAACAGCCACAC; and for mouse GAPDH promoter amplification: 5′-AAGGCTGGTGCTGTGGAGAAACTG and 3′-GTCCCCTTGCAACATACATAACTG. Oligonucleotides used for real-time PCR experiments are the following in 5′- to 3′-orientation: cdk9p42 forward, GCCAAGATCGGCCAAGGCAC; cdk9p42 reverse, CAGCCCAGCAAGGTCATGCTC; cdk9p55 forward, CCTCTGCAGCTCCG GCTCCC; cdk9p55 reverse, CACTCCAGGCCCCTCCGCGG; T7 promoter GTAATACGACTCACTCACTATAGGG; aP2 forward, AACACCGAGATTTCCTTCAA; aP2 reverse, AGTCACGCCTTTCATAACACA; 18s forward, GTTCCGACCATAAACGATGCC; 18s reverse, TGGTGGTGCCCTTCCGTC AAT; LPL forward, GAGACCAAGAGAAGCAGCAAGATGT; LPL reverse, AGTCGGGCCAGCTGAAGTAGGAGT; SCD1 forward, TGGGTTGGCTGCTTGTG; SCD1 reverse, GCGCTGGGCAGGATGAAG; FAS forward, TGCTCCCAGCTGCAGGC; FAS reverse, GCCCGGTAGCTCTGGGTGTA.

RNA Isolation, Reverse Transcription, and Real-Time PCR

RNA was purified using an RNeasy Mini Kit of QIAGEN (Hilden, Germany). Reverse transcription of total RNA was performed at 37 C using the M-MLV reverse transcriptase (Invitrogen SARL, Cergy Pontoise, France) and random hexanucleotides primers (Promega Corp., Madison, WI), followed by a 15-min inactivation at 70 C. Quantitative PCR was carried out by real-time PCR using a LightCycler and the DNA double strand-specific SYBR Green I dye for detection (Roche, Basel, Switzerland). Results were then normalized to 18-sec levels.

Cell Culture, Cell Differentiation, and Transfections

NIH-3T3, COS, 293 and 3T3-L1 cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified atmosphere of 5% CO₂ at 37 C. 3T3-L1 cells were differentiated with DMEM, 10% serum, 0.5 mm 3-isobutyl-1methylxanthine, 10 μg/ml insulin, 1 μm dexamethasone, and 100 nm pioglitazone for 2 d. From d 3 on, cells were incubated with DMEM, 10% serum, 10 μg/ml insulin, and 100 nm pioglitazone. Oil red O staining was performed as described elsewhere (28). Transfections were performed using the Jet PEI reagent (Qbiogene, Irvine, CA). Rosiglitazone was used at the concentration of 1 μm. The luciferase and β-galactosidase activities were measured as described previously (29). For stable transfections, Jet PEI was mixed with plasmid DNA, according to the manufacturer’s instructions, and the mixtures were allowed to remain on the cells in the incubator for 4 h. G418 or puromycin (Invitrogen SARL) was added, 24 h after transfections, at a final concentration of 1 mg/ml or 2.5 μg/ml, respectively. The medium plus G418 or puromycin was replaced three times/week until no surviving cells were observed.

Protein Expression Assays

For nuclear extracts, cells were homogenized in a lysis buffer containing 10 mm HEPES (pH 7.9), 10 mm KCl, 0.1 mm EDTA, 0.1 mm EGTA, 0.1% Nonidet P-40 (MP Biomedicals, Aurora, OH), and 1 mm dithiothreitol. A protease inhibitor cocktail was added (Sigma). Lysates were centrifuged and resuspended in a buffer containing 20 mm HEPES (pH 7.9), 1 mm EDTA, 1 mm EGTA, 0.4 m NaCl, 1 mm dithiothreitol, and protease inhibitor cocktail. Protein concentrations were determined by the Bradford method (Bio-Rad, Hercules, CA). SDS-PAGE and electrotransfer were performed as described elsewhere (30). The membranes were blocked 1 h in blocking buffer (PBS, 0.5% Tween 20, 5% skimmed milk). Filters were first incubated overnight at 4 C with the indicated primary antibodies, and then for 1 h at room temperature with a peroxidase conjugate secondary antibody. The complex was visualized with enhanced chemiluminescence (Interchim, Montluçon, France).

Immunofluorescence in 3T3-L1 Cells

For all immunofluorescence experiments, cells were grown on coverslips. After fixation and permeabilization with 100% methanol, cells were incubated with antibodies directed against cdk9 (Abcam Ltd.), cyclin T1 (Abcam Ltd.), and PPARγ (Santa Cruz Biotechnology, Inc.). For cyclin T2 detection, cells were fixed with 4% formaldehyde and then permeabilized with 0.1% Triton X-100 (Sigma) and incubated with cyclin T2 antibody (Santa Cruz Biotechnology). Preparations were then incubated with a combination of Texas Red-conjugated antimouse IgG and fluorescein isothiocyanate-conjugated antirabbit IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA), or fluorescein isothiocyanate-conjugated antigoat IgG (Santa Cruz Biotechnology).

Coimmunoprecipitation and ChIP Assays

For coimmunoprecipitation assays, whole-cell extracts were precleared with protein A-agarose beads (Roche) during 30 min at room temperature, and an aliquot of the precleared lysates was saved as input. Extracts were then centrifuged (5 min at 3000 rpm), and supernatants were immunoprecipitated with the indicated specific antibodies overnight at 4 C, and rabbit IgGs (Sigma) were used as negative control (mock). Immunoprecipitates were then washed twice with IP buffer [150 mm NaCl, 1% Nonidet P-40, 50 mm Tris/HCl (pH 8), and protease inhibitor cocktail] and three times with washing buffer (0.25 m KCl in PBS) and subjected to SDS-PAGE electrophoresis. ChIP assays were performed as described previously (31). Briefly, proteins were formaldehyde cross-linked to DNA in confluent 3T3-L1 preadipocytes before induction of differentiation or in cells induced with differentiation medium for 7 d. Proteins were then immunoprecipitated using the indicated antibodies, rabbit IgGs or beads were used as mock, DNA was extracted from the immunoprecipitates, and PCR amplification was performed using promoter-specific oligonucleotide primers.

Pull-Down Assays

In vitro translation of pCMV-cdk9 was performed with [³⁵S]methionine (PerkinElmer, Boston, MA) in a TNT-coupled transcription-translation system, as described by the manufacturer (Promega Corp.). GST fusion or GST alone were expressed in Escherichia coli, and purified on glutathione-sepharose-4B beads (Amersham Biosciences, Uppsala, Sweden). For in vitro binding GST, GST-PPARγ AB, GST-PPARγ DE, and GST-PPARγ deletion mutants were incubated with the labeled protein in 1 ml binding buffer containing 300 mm NaCl, 0.5% Triton-X-100, 50 mm Tris (pH 8), and 2 mm EDTA at room temperature for 1 h. Beads were washed five times with the same buffer. The proteins were visualized by autoradiography after SDS-PAGE.

Kinase Assays

Immunoprecipitated cdk9 from 3T3-L1 cells at different stages of differentiation was used as kinase. Immunoprecipitates were washed once with kinase buffer [25 mm Tris/HCl (pH 7.5), 150 mm NaCl, 10 mm MgCl₂, 1 mm DTT, and protease inhibitor cocktail] in the presence of a protease inhibitor cocktail (Sigma) and phosphatase inhibitors (5 mm Na₄P₂O₇, 50 mm NaF, 1 mm vanadate). Kinase assay was performed during 30 min at 37 C in the presence of 40 μm ATP and 8 μCi γ³³P-ATP; purified RNA polymerase CTD or GST, GST-PPARγ2 A/B, GST-PPARγ A/B S112A, GST-PPARγ LBD, or GST-PPARγ2 full length was used as substrate. Reaction was stopped by boiling the samples at 95 C for 5 min in the presence of denaturing sample buffer. Samples were then subjected to SDS-PAGE; gels were then dried in a gel dryer for 1 h at 80 C and exposed to an x-ray film.

Acknowledgments

We thank members of the Fajas and the Kristiansen laboratories for helpful discussions.

This work was supported by grants from Institut National de la Santé et de la Recherche Médicale, Centre Hospitalier Universitaire de Montpellier, Association pour la Recherche contre le Cancer, Association de Langue Française pour l’Etude du Diabète et des Maladies Métaboliques, and Fondation pour la Recherche Médicale (to L.F.) and the Carlsberg Foundation, the Novo Nordisk foundation, and the Danish Natural Science Research Council (to K.K.). I.I. is supported by a Ph.D. fellowship from the Ligue Nationale Contre le Cancer.

Abbreviations

 
• AR

  Androgen receptor;

 
• cdk

  cyclin-dependent kinase;

 
• ChIP

  chromatin immunoprecipitation;

 
• CTD

  carboxyl-terminal domain;

 
• DN

  dominant negative;

 
• DRB

  dichloro-1-β-ribofuranosyl-benzimidazole;

 
• FAS

  fatty acid synthase;

 
• GAPDH

  glyceraldehyde-3-phosphate dehydrogenase;

 
• GST

  glutathione-S-transferase;

 
• LBD

  ligand binding domain;

 
• LPL

  lipoprotein lipase;

 
• PPARγ

  peroxisome proliferator-activated receptor-γ;

 
• PPRE

  PPAR response element;

 
• P-TEFb

  positive transcription elongation factor b;

 
• RNA pol II

  RNA polymerase II;

 
• SCD1

  stearoyl coenzyme A desaturase;

 
• TK

  thymidine kinase;

 
• UAS

  upstream activating sequence.