The Functional Role of an Interleukin 6-inducible CDK9·STAT3 Complex in Human γ-Fibrinogen Gene Expression*

The signal transducer and activator of transcription 3 (STAT3) is an IL-6-inducible transcription factor that mediates the hepatic acute phase response (APR). Using γ-fibrinogen (FBG) as a model of the APR, we investigated the requirement of an IL-6-inducible complex of STAT3 with cyclin-dependent kinase 9 (CDK9) on γ-FBG expression in HepG2 hepatocarcinoma cells. IL-6 induces rapid nuclear translocation of Tyr-phosphorylated STAT3 that forms a nuclear complex with CDK9 in nondenaturing co-immunoprecipitation and confocal colocalization assays. To further understand this interaction, we found that CDK9-STAT3 binding is mediated via both STAT NH2-terminal modulatory and COOH-terminal transactivation domains. Both IL-6-inducible γ-FBG reporter gene and endogenous mRNA expression are significantly decreased after CDK9 inhibition using the potent CDK inhibitor, flavopiridol (FP), or specific CDK9 siRNA. Moreover, chromatin immunoprecipitation (ChIP) experiments revealed an IL-6-inducible STAT3 and CDK9 binding to the proximal γ-FBG promoter as well as increased loading of RNA Pol II and phospho-Ser2 CTD Pol II on the TATA box and coding regions. Finally, FP specifically and efficiently inhibits association of phospho-Ser2 CTD RNA Pol II on the γ-FBG promoter, indicating that CDK9 kinase activity mediates IL-6-inducible CTD phosphorylation. Our data indicate that IL-6 induces a STAT3·CDK9 complex mediated by bivalent STAT3 domains and CDK9 kinase activity is necessary for licensing Pol II to enter a transcriptional elongation mode. Therefore, disruption of IL-6 signaling by CDK9 inhibitors could be a potential therapeutic strategy for inflammatory disease.

The hepatic acute phase response (APR) 2 is a coordinated response to tissue injury, infection or malignancy that initiates a global switch in the transcription of secreted proteins expressed by the vertebrate liver (1). Here, cytokines produced at the site of injury activate the de novo expression of genes encoding acute phase proteins (APPs), proteins important in homeostasis, opsonization, and wound repair. Of the APPs, fibrinogen (FBG) is known to play a key role in the APR by mediating hemostasis, participating in clot formation, platelet aggregation and clot retraction (2), processes important in promoting tissue repair at the site of injury. The FBG heterocomplex is encoded by three chains; of these, the FBG ␥ chain (␥-FBG) is of interest because it contains binding sites for platelet integrin ␣ IIb ␤ 3 and leukocyte integrin ␣ M ␤ 2 , leading to platelet aggregation and leukocyte recruitment in inflammation (3,4). In addition, its high binding affinities for vascular endothelial growth factor (5), fibroblast growth factor-2 (6), and interleukin-1␤ (7) contribute to wound healing. Finally, ␥-FBG contains a fibrin polymerization site, which is involved in fibrin clot formation and platelet aggregation. Because of the crucial role of ␥-FBG in multiple processes, the transcriptional mechanisms controlling inducible ␥-FBG expression have been extensively investigated.
The cytokine interleukin-6 (IL-6) has emerged as a major mediator of de novo acute phase reactants synthesis, and for ␥-FBG in particular (8 -11). Here, IL-6 produced and secreted at the site of injury, circulates and binds to the hepatic highaffinity IL-6 receptor (IL-6R)-␣. The liganded IL-6R␣ then forms an oligomeric complex with gp130 (transducin) and subsequently activates the Janus-(JAK) and Tyk tyrosine kinases. Tyrosine-phosphorylated gp130, in turn, binds the cytoplasmic signal transducer and activator of transcription (STAT)-1 and -3 isoforms via their Src homology (SH)-2 domains. As additional substrates of the JAK/Tyk kinases, STATs are tyrosinephosphorylated, allowing them to dimerize via intermolecular SH3 interactions, whereupon they translocate into the nucleus (12). Gene deletion experiments have shown that the actions of STAT3 are necessary for inducible expression of a network of APP genes, including C-reactive protein, serum amyloid A, angiotensinogen, and ␥ -FBG (13)(14)(15).
The mechanism of how tyrosine-phosphorylated STAT3 (pTyr-STAT3) induces gene expression is partly understood. Upon entry into the nucleus STAT3 associates with the p300/ CBP coactivator, a protein that acetylates STAT3 on its NH 2terminal modulatory domain to stabilize the STAT3-p300/CBP complex (16,17). This complex, in turn modifies chromatin via its histone acetyl transferase activity, serving as an intermolecular bridge to recruit other chromatin modifying proteins like BRG1 (16,18,19). The recruitment of an enhancesome results in enzymatic modifications of core nucleosomes, affecting chromatin structure and inducing preinitiation complex formation. By this mechanism, STAT3 controls genes important in diverse cellular functions such as cell cycle control (20,21), antioxidant cellular defenses (22), and the APR (13)(14)(15).
In this study, we observed that IL-6 strongly induces a nuclear complex of STAT3 with cyclin-dependent kinase 9 (CDK9) and sought to further understand its role in APP regulation using ␥-FBG as a model gene. CDK9 is a major component of a complex known as the positive transcriptional elongation factor (P-TEFb) involved in derepression and activation of RNA Pol II (23). We found IL-6 induces STAT3 binding to CDK9 in a mechanism mediated by both the NH 2 -and COOHterminal domains of STAT3. Inhibition of CDK9 activity or its expression decrease IL-6-inducible ␥-FBG transcription. Chromatin immunoprecipitation (ChIP) experiments provide direct evidence that IL-6 induces CDK9 recruitment to the ␥-FBG promoter along with enhanced RNA Pol II and phospho-Ser 2 CTD Pol II loading on the coding region. Moreover, IL-6-inducible Pol II binding is abolished by the CDK9 inhibitor, flavopiridol (FP). Taken together, our data indicate that the IL-6inducible STAT3⅐CDK9 complex is essential for ␥-FBG induction during the APR. This phenomenon suggests STAT3 promotes transcription elongations as an additional mechanism for induction of APPs.
Cell Culture and Stimulation-The human hepatoblastoma cell line HepG2 (ATCC, Manassas, VA) was cultured in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% (v/v) fetal bovine serum, 2 mM L-glutamine, 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, and penicillin (100 units/ml)/streptomycin (100 g/ml). Cells are maintained at 37°C in a humidified atmosphere of 5% CO 2 . Cells were serum-starved for at least 16 h before treatment. FP was added 1 h before IL-6 stimulation.
Transfection and Luciferase Activity Assay-Transient transfection was performed using Lipofectamine PLUS reagent (Invitrogen) according to the manufacturer's instruction. For reporter assay, HepG2 cells were plated into 6-well plates and cotransfected with ␥-FBG-LUC reporter gene and the transfection efficiency control plasmid pSV2PAP. Twenty-four hours later, cells were stimulated with IL-6 with or without FP pretreatment. Both luciferase and alkaline phosphatase activities were measured 48 h after transfection. All transfections are carried out in three independent experiments. For co-immunoprecipitation, indicated expression plasmids were co-transfected into 10-cm 2 dishes using the same protocol. Cells were treated with IL-6 24 h after transfection prior to protein extraction.
Preparation of Subcellular Extracts-Sucrose cushion-purified nuclear extract (NE) was prepared as described (16). In brief, cells were harvested in PBS and centrifuged to collect pellets. Pellets were resuspended in double cell volume of solution A and centrifuged to obtain the supernatants as cytoplasmic fraction. The nuclear pellets were purified on sucrose cushions (16). After high salt extraction in buffer C (16), the nuclei were centrifuged at 12,000 ϫ g at 4°C for 20 min. The supernatants were saved as NE. The protein concentrations were measured by Coomassie dye binding (Protein Reagent, Bio-Rad).
Western Blotting-Proteins were fractionated by SDS-PAGE, and transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA). Membranes were blocked in 5% milk for 0.5-1 h and incubated with indicated primary Ab at 4°C overnight. Membranes were washed in TBS-0.1% Tween 20 and incubated with secondary Ab at 20°C for 1 h. Signals were detected by the enhanced chemiluminescence assay (ECL; Amersham Biosciences) or visualized by the Odyssey Infrared Imaging system. ␤-Actin is used as a loading control.
Co-immunoprecipitation-1-2 mg of HepG2 NE were precleared with 40 l of protein A-Sepharose beads (Sigma) for 1 h at 4°C. Immunoprecipitation was performed in the presence of 5 g of the indicated primary Ab at 4°C overnight. Immune complexes were captured by adding 50 l of protein A-Sepharose beads and rotated at 4°C for 2 h. After the supernatant was discarded, protein A-Sepharose beads were washed with cold PBS for 4 -5 times, and immunoprecipitates were fractionated by SDS-PAGE.
siRNA Transfection-HepG2 cells were plated into 6-well plates at a density of 2.5 ϫ 10 5 cells/well. On the following day, the cells were transfected with either siRNA targeting human CDK9 or the negative control siRNA (Ambion, Austin, TX) using TransIT-siQUEST transfection reagent (Mirus, Madison, WI) according to the manufacturer's instructions. 48 h later, the transfected cells were exposed to IL-6 stimulation prior to total cellular RNA extraction for Real-Time (RT)-PCR. Two Step ChIP Assay-Two step ChIP was performed as described (27). In brief, 4 -6 ϫ 10 6 HepG2 cells per 100-mm dish were washed twice with PBS after stimulation. Proteinprotein cross-linking was first performed with disuccinimidyl glutarate (DSG, Pierce) followed by protein-DNA cross-linking with formaldehyde. After cells were washed and collected in 1 ml of PBS, pellets were lysed by SDS lysis buffer and sonicated 4 times, 15 s each at setting 4 with 10 s break on ice until DNA fragments lengths were between 200 and 1000 bp. Equal amounts of DNA were immunoprecipitated overnight at 4°C with 4 g of the indicated Ab in ChIP dilution buffer. Immunoprecipitates were collected with 40 l of protein A magnetic beads (Dynal Inc., Brown Deer, WI), and washed sequentially with ChIP dilution buffer, high-salt buffer, LiCl wash buffer, and finally in 1ϫ TE buffer. DNA was eluted in 250 l of elution buffer for 15 min at room temperature. Samples were de-crosslinked in de-cross-linking mixture at 65°C for 2 h. DNA was phenol/chloroform extracted, precipitated by 100% ethanol and used for RT-PCR.
Quantitative Real-time PCR (Q-RT-PCR)-Total cellular RNA was extracted by Tri Reagent (Sigma). 2 g of RNA was used for reverse transcription using SuperScript III First-Strand Synthesis System from Invitrogen. 2 l of cDNA products were amplified in a 20-l reaction system containing 10 l of iQ SYBR Green Supermix (Bio-Rad) and 400 nM primer mix. All the primers were designed by PrimerExpress v2.0. For ␥-FBG mRNA expression, the primers 5Ј-GGCAACTGTGCTGAAC-AGGAT-3Ј and 5Ј-GATGGCCAGCGTGACACTT-3Ј were used. The sequences of primer sets used in genomic assays are shown in Table 1. All the reactions were processed in MyiQ Single Color Real-Time PCR thermocycler using two step plus melting curve program, and the results were analyzed by iQ5 program (Bio-Rad). For quantitative real-time genomic PCR (Q-gPCR), a standard curve was generated using a dilution series of genomic DNA (from 40 ng to 25 g) for each primer pair. The fold change of DNA in each immunoprecipitate was determined by normalizing the absolute amount to input DNA reference and calculating the fold change relative to that amount in unstimulated cells.
Immunofluorescent Staining-HepG2 cells were grown in 6-well tissue culture plates containing sterile coverslips (Fisher). After treatment, cells on coverslips were rinsed by PBS twice and fixed with 4% paraformaldehyde for 20 min at room temperature. After fixation, cells were washed with PBS three times and treated with 0.2% Triton X-100 for 15 min at room temperature prior to immunostaining. All slips were blocked with blocking buffer containing 1% bovine serum albumin and 0.1% Triton X-100 for 1 h and then washed once with PBS. The primary antibody was diluted in blocking buffer (1:50 for anti-pTyr705 STAT3, 1:100 for anti-STAT3 (c-20), and 1:200 for anti-CDK9 (c-20)) and applied to the cells on coverslips. Cells were incubated in the primary antibody at 4°C overnight. Next day, all slips were washed three times for 10 min each with PBS and then incubated with the diluted fluorescence-labeled secondary antibody (1:100, secondary antibodies for STAT3 and pTyr705STAT3 were from Jackson ImmunoResearch, secondary antibody for CDK9 staining was from Invitrogen) in blocking buffer for 1 h at room temperature in the dark. Cells were washed three times with PBS in low lighting and dried at room temperature. Coverslips were mounted on slides using Dako-Cytomation Fluorescent Mounting Medium and observed under confocal microscope (Zeiss LSM510 META system). Images were captured at a magnification of ϫ40 or ϫ60.

IL-6-inducible ␥-FBG Expression Is Mediated by STAT3-
Previous studies showed that IL-6 potently up-regulates ␥-FBG expression in the human hepatoma cell line, HepG2 (8 -11). To confirm this finding, ␥-FBG expression was measured over a time course of IL-6 stimulation by Q-RT-PCR. A 2.5-fold increase in ␥-FBG mRNA abundance was detected as early as 2 h after IL-6 treatment, and the mRNA level continued to increase until a plateau of 12-fold relative to control was observed at 24 h (Fig. 1A). To determine the transcriptional contribution, a luciferase reporter driven by 643 bp of the ␥-FBG promoter containing three functional type II IL-6 response elements (IL-6REs) was constructed. This Ϫ607/ϩ36 ␥-FBG-LUC plasmid was transiently transfected into HepG2 cells and stimulated in the absence or presence of various doses of IL-6 (10 and 50 ng/ml) for two different times (12 and 24 h).
For the cells stimulated with 10 ng/ml IL-6, we observed an 8-fold induction of normalized luciferase reporter activity relative to control after 12 h, and 13-fold over control after 24 h of stimulation ( Fig. 1B). At 50 ng/ml IL-6, the inducible activity of the ␥-FBG promoter was increased by 15-fold at 12 h and 24-fold at 24 h (Fig. 1B). To determine the contribution of IL-6inducible transcription mediated by STAT3, increasing concentrations of dominant-negative (DN-) STAT3 (Tyr 705 to Phe, Ref. 15) were co-transfected with the Ϫ607/ϩ36 ␥-FBG-LUC reporter gene. As little as 0.1 g of DN-STAT3 decreased IL-6inducible reporter activity by more than 70% (Fig. 1C). Cotransfected DN-STAT3 had no significant effects on the basal activity of the ␥-FBG-LUC reporter gene. Together these data indicated that ␥-FBG is an IL-6-inducible gene, mediated at least in part by STAT3-dependent transcriptional induction.
IL-6 Induces a Nuclear STAT3⅐P-TEFb Complex-A series of studies have shown that P-TEFb can interact with various transcription factors or nuclear receptors, such as nuclear factorkappa B (NF-B) (28), c-Myc (29), androgen receptor (30), and peroxisome proliferator-activated receptor-␥ (PPAR␥) (31). To determine whether STAT3 associates with P-TEFb, HepG2 cells were stimulated in the presence or absence of IL-6 and NE was subjected to nondenaturing co-immunoprecipitation (Co-IP) assay using anti-CDK9 as the primary Ab. The immune complexes were fractionated on an SDS-PAGE, and STAT3 was detected by Western blot ( Fig. 2A, upper panel). We observed STAT3 binding only in the IL-6-stimulated NE. By contrast, cyclin T1 was also observed in the complex, but there was no difference in cyclin T1 abundance between IL-6-stimulated and unstimulated NEs ( Fig. 2A, middle panel), indicating that the CDK9⅐cyclin T1 complex formation is independent of IL-6 stimulation.
Because IL-6 was required for the association of STAT3 and CDK9, we next asked whether activated STAT3 isoforms were interacting with CDK9. After nondenaturing CDK9 IP, the immune complexes were probed with an antibody that specifically recognized phospho-Tyr 705 STAT3. A strong signal was specifically detected in the IL-6-stimulated CDK9 complexes (Fig. 2B, upper panel). By contrast, inactive cytosolic STAT3 failed to interact with CDK9 although CDK9 is expressed constitutively in both the cytosol and nucleus (data not shown). Together, these data suggest that Tyr phosphorylation and nuclear translocation are essential for the formation of the STAT3⅐CDK9 complex.
To exclude the possibility that the Co-IP findings were artifactual because of biochemical fractionation, we confirmed this interaction using confocal colocalization assays. For this purpose, HepG2 cells were transfected with a plasmid encoding CDK9 fused to a monomeric strawberry fluorescence protein pcDNA-FStraw-CDK9, and stimulated with IL-6 prior to fixa- increases endogenous ␥-FBG mRNA expression. Serum-starved HepG2 cells were treated with IL-6 (10 ng/ml) for indicated times. Shown is the result of Q-RT-PCR assays plotting the fold change of ␥-FBG in IL-6-treated cell normalization to GAPDH. B, IL-6 induces the reporter activity of ␥-FBG in a time-and dose-dependent manner. HepG2 cells were transiently transfected with ␥-FBG-LUC reporter and control plasmid pSV2PAP. 24 h after transfection, cells were treated with two different doses of IL-6 (10 ng/ml or 50 ng/ml) for 12 or 24 h followed by assay for reporter gene expression. C, DN-STAT3 inhibits IL-6-induced ␥-FBG-LUC reporter activity. Cells were transfected as in B except that different amounts of DN-STAT3 were co-transfected with the reporter gene. Amount of transfected DNA was kept equivalent using an empty expression plasmid. Data shown are means Ϯ S.D. from three independent transfections. The data were analyzed by Student's t test. *, p value Ͻ0.05; **, p value Ͻ0.01. tion. Cells were then stained with anti-STAT3 Ab and secondary FITC-labeled Ab. In the transfected cells, straw-CDK9 is diffusely and constitutively localized in nucleus but excluded from the nucleoli (Fig. 2D, bottom middle). This distribution pattern of Straw-CDK9 is identical to that of the endogenous CDK9 by immunofluorescence labeling (Fig. 2D, top middle). In the unstimulated cells, the majority of STAT3 was detected in cytoplasm (Fig. 2E, top). By contrast, after IL-6 treatment, there is an obvious accumulation of nuclear STAT3 (Fig. 2E, bottom). STAT3-CDK9 co-localization is indicated by the merged overlay (Fig. 2E, bottom right). A similar assay was performed staining for phospho-Tyr 705 STAT3. By contrast with anti-STAT3 labeling, no detectable phospho-Tyr 705 STAT3 was observed in unstimulated cells (Fig. 2F, top). Upon IL-6 stimulation, phospho-Tyr 705 STAT3 was strongly localized to the nucleus, where it colocalized with nuclear CDK9 (Fig. 2F, bottom right). These data confirmed the Co-IP studies and indicated that activated STAT3 co-localizes with CDK9 in the nucleus.
The STAT3 NH 2 Terminus Is Sufficient for CDK9 Complex Formation-To map the domain(s) of STAT3 responsible for CDK9 interaction, we first sought to confirm that ectopically expressed V5 epitope-tagged STAT3 (V5-STAT3) associated with endogenous CDK9 in an IL-6dependent manner. For this purpose, an expression vector encoding fulllength V5-STAT3 was transiently transfected into HepG2 cells. NEs were prepared in the absence or presence of IL-6 stimulation, and subjected to nondenaturing Co-IP assays. As seen in Fig. 3A, V5-STAT3 is captured by nondenaturing IP of CDK9 in an IL-6-dependent manner. Note that no complex is seen using IgG as the immunoprecipitating Ab, demonstrating assay specificity. Fig. 3B shows that cyclin T1 is also captured by nondenaturing IP of V5-tagged STAT3. These data indicated that the ectopic V5-STAT3 also inducibly associated with endogenous P-TEFb, containing CDK9 and cyclin T1.
To identify the regions of STAT3 interacting with CDK9, a series of expression vectors encoding COOH domain-deleted V5-STAT3 proteins were constructed (the relevant domains are schematically shown in Fig. 3C). The V5-STAT3 deletion mutants were then expressed in HepG2 cells and the CDK9-bound mutated STAT3 proteins were detected using nondenaturing Co-IP. We observed that all of the STAT3 COOH-terminal deletion mutants (containing amino acids 1-320, 1-465, 1-585, and 1-688) complexed with endogenous CDK9 (Fig.  3D). To further dissect the domain in the NH 2 terminus, an expression vector encoding STAT3-(1-130) was constructed and tested for binding by Co-IP (Fig. 3E). STAT3-(1-130) bound endogenous CDK9 in a manner similar to STAT3-(1-320). This finding suggests that NH 2 -terminal domain of STAT3 is sufficient for the association of STAT3 and CDK9. To determine if the NH 2 terminus was necessary for CDK9 interaction, we tested whether the NH 2 -terminal-deleted STAT3 containing amino acids 131-770 (termed ⌬130) could still FIGURE 2. Activated STAT3 complexes with CDK9 in HepG2 nuclear in the presence of IL-6. A and B, STAT3 and CDK9 form a complex in nuclear in an IL-6-dependent manner. Serum-starved HepG2 cells were treated with IL-6 (10 ng/ml) for 30 min, and NEs were isolated using sucrose gradient fractionation. 2 mg of NE were immunoprecipitated by anti-CDK9 Ab or normal rabbit IgG as control. The immune complexes were fractioned by 10% SDS-PAGE and immunoblotted with anti-STAT3 (A), anti-phospho-STAT3 (Y705) (B), or anti-Ac-K 87 STAT3 Abs (C). The blots were then reprobed with cyclin T1 and CDK9 Abs. Because the polyvinylidene difluoride membrane was reprobed with anti-cyclin T1 antibody without stripping, the STAT3 band (indicated by * in A, middle panel, second lane) remained in IL-6-stimulated cells. D, the distribution of straw-CDK9 is similar to that of endogenous CDK9. Endogenous CDK9 (upper panel) was stained by a polyclonal rabbit IgG directed against CDK9 and Alexa 568 goat anti-rabbit secondary Ab. For straw-CDK9 (lower panel), HepG2 cells were transfected with pcDNA-FStraw-CDK9 and 24-h later split into 6-well plates containing sterile coverslips. Nuclei were stained by DAPI. E and F, activated STAT3 colocalizes with CDK9 under IL-6 stimulation. Cells were treated with IL-6 (10 ng/ml) for 30 min and then stained for STAT3 or pTyr 705 -STAT3. STAT3 was detected by a rabbit Ab directed against STAT3 (c-20) and a FITC-conjugated goat anti-rabbit secondary Ab. pTyr 705 -STAT3 was recognized by a mouse anti-phospho-Tyr 705 STAT3 Ab (B7) and a FITC-conjugated goat anti-mouse secondary Ab. Straw-CDK9 was transfected and expressed as described in D. Shown is confocal immunofluorescent imaging of representative cells. Empty arrows indicated the co-localization of STAT3 or pTyr 705 -STAT3 with CDK9. DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51 JOURNAL OF BIOLOGICAL CHEMISTRY 37095 interact with CDK9. As seen in Fig. 3F, STAT3 (⌬130) still bound CDK9. Previous work has indicated that the STAT3 COOH transactivation domain, spanning residues 716 -770 could bind to in vitro-translated CDK9 (20). We therefore tested whether the STAT3-CDK9 interaction depended on both the NH 2 -and COOH-terminal domains. For this purpose, we constructed an expression vector encoding STAT3-(130 -688), missing both NH 2 and COOH-terminal activation domains. As we expected, STAT3-(130 -688) did not bind CDK9 (Fig. 3G). Together these data indicated that although the STAT3 NH 2 terminus was sufficient for CDK9 complex formation, both NH 2 and COOH termini participate in complex formation.

CDK9⅐STAT3 Complex in ␥-FBG Transcription
To further explore the importance of the NH 2 -terminal domain for the STAT3 transcriptional activation function, STAT3-(⌬130) was cotransfected with ␥-FBG-LUC reporter gene and luciferase reporter activity measured. As shown in Fig. 3H, 0.5 g of STAT3-(⌬130) decreased IL-6-inducible reporter activity by nearly 30% and 1.0 g dramatically reduced reporter activity by 97% compared with empty vector controls. In addition, the basal level of luciferase activity was also inhibited by STAT3-(⌬130) in a dose-dependent manner. This result indicates that the NH 2 terminus is essential for STAT3 to activate the transcription of downstream genes. Our interpretation of this finding is that STAT3-(⌬130) competes with endogenous STAT3 for promoter binding, and is unable to effectively activate transcription because of its reduced binding affinity for transcriptional elongation factors.

CDK9 Activity Is Required for IL-6-induced Expression of ␥-FBG-
To investigate the functional role of CDK9 in IL-6-induced expression of ␥-FBG, we inhibited CDK9 kinase activity by a chemical inhibitor, flavopiridol (FP). FP is a highly selective P-TEFb inhibitor with a K i of 3 nM (32). We first tested the effect of FP on IL-6-inducible ␥-FBG transcription. HepG2 cells transfected with the Ϫ607/ϩ36 ␥-FBG-LUC reporter plasmid were pretreated with either vehicle (Me 2 SO) or FP (500 nM) for 1 h prior to IL-6 stimulation. Both the basal and IL-6-induced activities of ␥-FBG-LUC reporter were dramatically decreased when FP was added (Fig. 4A). In addition, FP also inhibited IL-6 induced endogenous ␥-FBG mRNA (Fig. 4B). Consistent with these results, expression of a kinase-deficient DN-CDK9 also inhibited the induction of ␥-FBG-LUC reporter in a dose-dependent manner, with greater than 50% inhibition seen with 0.25 g of expression vector (Fig. 4C). These results indicated that IL-6-inducible expression of ␥-FBG was highly dependent on CDK9 kinase activity.
To more specifically confirm the essential role of CDK9 in the ␥-FBG induction, short interfering RNA (siRNA) transfection was used to specifically silence endogenous CDK9 expression. As seen in Fig. 4D, transfection of CDK9 siRNA (siCDK9) significantly reduced CDK9 protein levels to less than 20% compared with control siRNA. To examine the effect of CDK9 knockdown on IL-6-inducible ␥-FBG expression, abundance of ␥-FBG mRNA was measured in a time course of stimulation by Q-RT-PCR. In contrast to control transfectants, the induction of ␥-FBG mRNA in siCDK9transfected cells was significantly decreased at every time point (Fig.  4E). From these results, we concluded that IL-6-induced ␥-FBG expression requires CDK9 expression and activity. Although CDK9 is considered to be a general regulator of Pol II-dependent gene transcription, siRNA against CDK9 did not affect GAPDH mRNA expression (data not shown). This indicates that ␥-FBG and the housekeeping genes may have different requirements for CDK9. IL-6 Induces P-TEFb Recruitment to the ␥-FBG Gene-Three functional type II IL-6 response elements (REs) have been identified on the about 600 bp of the human ␥-FBG promoter upstream of the transcription initiation site, and all contribute to full IL-6 inducibility (8). To identify the interaction of STAT3 with these sites, three sets of primers spanning distinct STAT3responsive region in ␥-FBG promoter (RE1, RE2, and RE3) were designed (sequences in Table 1) and optimized by quantitative real time FIGURE 4. CDK9 activity is required for IL-6-stimulated expression of ␥-FBG. A, A CDK9 inhibitor, FP, inhibits ␥-FBG-LUC reporter gene activity. HepG2 cells were transfected with ␥-FBG-LUC reporter gene and pSV2PAP as an internal control. 24 h after transfection, cells were stimulated with IL-6 alone (Neg) or pretreated with FP (500 nM) for 1 h followed by IL-6 stimulation. The control cells were pretreated with the vehicle dimethyl sulfoxide. 24 h after IL-6 stimulation, cells were collected to measure reporter gene activity. Shown is normalized reporter activity. B, FP blocks IL-6-induced ␥-FBG mRNA expression. HepG2 cells were pretreated with FP or dimethyl sulfoxide as described above before 24 h of IL-6 stimulation. ␥-FBG and GAPDH mRNA expressions were assayed by Q-RT-PCR. The fold change of ␥-FBG in IL-6-treated cells over IL-6-unstimulated control was obtained after correction for the amount of GAPDH. C, DN-CDK9 inhibits the IL-6 induction of the ␥-FBG reporter gene. Different amounts of DN-CDK9 were cotransfected with ␥-FBG-LUC reporter gene. Cells were then treated with IL-6 (10 ng/ml) for 24 h or left unstimulated prior to reporter gene assay. Data shown are means Ϯ S.D. from three independent transfections. D, CDK9 siRNA transfection efficiently inhibits CDK9 expression. HepG2 cells grown in 6-well plates were transiently transfected with 100 nM CDK9 siRNA (siCDK9), control siRNA (Con) or transfection reagent alone (Neg). 72 h after transfection, equivalent amounts of protein from the whole cell lysates were used for immunoblot. Top panel, CDK9 staining; bottom, ␤-actin staining as a loading control. E, siRNA transfection was performed as described in D. 48 h after siRNA transfection, cells were treated with IL-6 (10 ng/ml) for indicated times. The ␥-FBG mRNA expression and the fold of induction were measured and calculated as described in Fig. 1A. The results from Q-RT-PCR are expressed as means Ϯ S.D. from duplicates. The data were analyzed by Student's t test. *, p value Ͻ0.05; **, p value Ͻ0.01.  (Fig. 5B). We next examined the effect of IL-6 on inducible CDK9 binding to the ␥-FBG promoter. In a pattern similar to that observed for STAT3, IL-6 induces 2-fold increase of CDK9 binding to the upstream ␥-FBG IL6 RE1 (Fig. 5C), but no significant recruitment was observed on the RE2 and RE3 (data not shown). The same DNA was examined for changes in CDK9 binding to the TATA box region (spanning nt Ϫ66 to ϩ6), exon 5 (nt ϩ2524 to ϩ2606), and exon 7 (nt ϩ4148 to ϩ4239). Here, IL-6 induced 2.3-fold binding to the TATA box, a 4-fold increase on exon 5 and a 2.5-fold increase on exon 7 within 30 min after stimulation, suggesting that CDK9 may accompany the elongating polymerase during transcription (Fig. 5C).
IL-6 Induces Pol II Recruitment to the ␥-FBG Gene-To further understand the role of CDK9 recruitment in IL-6 stimulation, the binding of RNA Pol II and phospho-Ser 2 CTD Pol II was examined by two step ChIP assay. IL-6 induced a 3-fold increase in total Pol II binding on the ␥-FBG TATA box, and strongly induced Pol II loading on the coding sequences (Fig.  5D). Because CDK9 is a kinase for serine 2 of the Pol II CTD, two step ChIP was performed using anti-phospho-Ser 2 CTD Pol II Ab. We observed a 3-fold increase of phospho-Ser 2 CTD Pol II binding to the TATA box, and greater than 8-fold increase on exon 5 (Fig. 5E). We noted that the distribution pattern of CDK9 was similar to that of phospho-Ser 2 CTD Pol II, supporting the notion that CDK9 is the IL-6-inducible Ser 2 -CTD kinase.
To further establish this relationship, we investigated the effects of FP on IL-6-inducible total and phospho-Ser 2 CTD Pol II recruitment. In this experiment, HepG2 cells were pretreated with FP (500 nM) before IL-6 stimulation. The chromatin was processed for two step ChIP assay using anti-Pol II (Fig. 5F) and phospho-Ser 2 CTD Pol II (Fig. 5G) Abs. We found that IL-6induced occupancy of the TATA box, exon 5, and exon 7 by RNA Pol II as well as phospho-Ser 2 CTD Pol II was significantly inhibited by FP. These results suggest that CDK9 is required for RNA pol II recruitment and licensing it to enter transcription elongation mode, thereby promoting IL-6-inducible ␥-FBG expression. To exclude the possibility that FP interferes with STAT3 and CDK9 recruitment, their binding to the ␥-FBG gene in the presence of FP was assayed by two step ChIP. IL-6-induced STAT3 and CDK9 occupancy of the ␥-FBG promoter and coding region were not significantly affected by FP pretreatment (Fig. 5, H and I).
FP Inhibits Basal and IL-6-inducible Phospho-Ser 2 CTD Pol II Formation-Because RNA Pol II is not the only substrate of CDK9 (33-37), we tested whether CDK9 could also phosphorylate STAT3. HepG2 cells treated in the absence or presence of FP were then IL-6-stimulated. Western blots were performed using anti-phospho-Tyr 705 and anti-phospho-Ser 727 STAT3 Abs (the latter modification is known to be essential for maximal transcriptional activity of STAT3 (38)). We observed the strong IL-6-inducible STAT3 Tyr and Ser phosphorylation were unaffected by FP (Fig. 6, A and B). As an additional determination, we observed that the IL-6-inducible STAT3⅐CDK9 association was unaffected by FP treatment (Fig. 6C), suggesting that CDK9 kinase activity is not essential for complex formation. Because CDK9 is thought to be a major Pol II Ser 2 -CTD kinase we examined its effect on total Pol II-and phospho-Ser 2 CTD Pol II abundance. Although FP did not affect total Pol II abundance (Fig. 6D), the general levels of phospho-Ser 2 CTD  Table 1 for sequence information). B to I, serum-starved HepG2 cells were treated with IL-6 (10 ng/ml) alone for 30 min (from B to E) or FP (500 nM) pretreated for 1 h (from F-I), and two step ChIP assay was performed as described under "Experimental Procedures." The sequences in the promoter or coding region of the ␥-FBG gene in the immunoprecipitates were amplified by Q-gPCR using specific primer sets as shown in Table 1. B, STAT3 recruits to IL-6 REs on ␥-FBG promoter after IL-6 stimulation. C, IL-6 induces CDK9 recruitment to the IL-6 RE1, TATA box as wells as exons 5 and 7. D and E, IL-6 increases the Pol II and phospho-Ser 2 -CTD Pol II loading to the endogenous ␥-FBG gene. F and G, FP inhibits the recruitment of RNA pol II and phospho-Ser 2 CTD Pol II to the TATA box and coding region. H and I, FP pretreatment does not affect the inducible STAT3 and CDK9 binding to the ␥-FBG gene. The results are expressed as means Ϯ S.D. from duplicates. The data were analyzed by Student's t test. *, p value Ͻ0.05; **, p value Ͻ0.01. Pol II was significantly decreased (Fig. 6E). We therefore conclude that FP specifically inhibits phospho-Ser 2 CTD Pol II formation without affecting STAT3 activation or complex formation with CDK9.

DISCUSSION
FBG is an APP that plays key roles in fibrin clot formation, platelet aggregation and wound repair by binding to cell surface receptors or growth factors through its ␥ chain. Previous studies have shown the acute phase induction of ␥-FBG in liver cells is mainly regulated by the cytokine-inducible STAT3 transcription factor. In this study, we further investigated the molecular mechanism by which IL-6-inducible ␥-FBG transcription is regulated by the STAT3⅐P-TEFb complex. We found that IL-6 induces a formation of STAT3⅐CDK9 complex mediated by both the STAT3 NH 2 and COOH termini. Moreover, activated pTyr 705 -STAT3 and Ac-Lys 87 STAT3 were preferentially complexed with CDK9. Quantitative two step ChIP assays indicate that IL-6 induces STAT3, CDK9, Pol II, and phospho-Ser 2 CTD Pol II recruitment to the ␥-FBG gene. Finally, our studies indicate that CDK9 is required for ␥-FBG expression because siRNA transfection and inhibition of CDK9 kinase activity both inhibit IL-6-inducible transcription. These studies indicate that P-TEFb is a critical regulator of STAT3-dependent gene activation in the APR.
STAT3 is a central transcription factor in IL-6-induced hepatic APRs. The transcription of APPs controlled by STAT3 is regulated at multiple levels. First, IL-6 induces tyrosine phosphorylation of STAT3 in its COOH terminus, leading to its dimerization and nuclear translocation. The activated STAT3 then recognizes specific motifs in the promoters of target genes and initiates assembly of the basal transcriptional apparatus (39). At this point, STAT3 recruits p300/CBP coactivators containing histone acetyltransferase activity (HAT) and the BRG1 chromatin-remodeling complex. HAT regulates transcription by acetylating the amino-terminal histone tail, increasing accessibility of chromatin-condensed templates to the transcriptional machinery (19), whereas chromatin-remodeling complexes function by altering nucleosomal structure and increasing the accessibility of Pol II to the proximal promoter (40). Both these activities promote target gene activation by relieving repression and facilitating the loading of the pre-initiation complex. The findings of our study add a new dimension to how STAT3 mediates gene expression. STAT3 not only induces transcription initiation, but also regulates transcription elongation through its association and recruiting P-TEFb to its target genes.
The P-TEFb complex has been shown to play an important role in Pol II-dependent transcription by its ability to release RNA pol II from transcriptional arrest, allowing production of full-length mRNA transcripts (41). Experiments using CDK9 inhibitors strongly indicate that CDK9 activity is required for both HIV transcription and the expression of many cellular genes. Tat, a viral transactivator encoded by HIV and other retroviral genomes (42,43), is able to recruit the CDK9⅐cyclinT1 complex to the TAR element of the HIV promoter (44), and position CDK9 to phosphorylate the negative elongation factors as well as the RNA Pol II CTD, thereby ena-bling transcriptional elongation (43,(45)(46)(47). In addition to Tat, recent studies have identified other cellular transcription factors that associate with P-TEFb, including CIITA, NF-B, c-Myc, and p53 (28,29,37,48,49). The involvement of CDK9 in regulating STAT3-dependent cell cycle regulatory genes was first reported for the p21waf1 gene (20). The authors found that DRB, another P-TEFb inhibitor, inhibits p21waf1 expression as well as RNA Pol II recruitment to p21waf1. Although these results indicated that CDK9 was required for STAT3 ability to control expression of cell cycle regulatory genes, there may be significant heterogeneity in the mechanisms of transcriptional induction between different classes of STAT3-responsive genes and to what extent the CDK9 was required for activation of the APPs was unknown. Using ␥-FBG as a model gene of the APR, our study here extends the requirement of CDK9 in STAT3dependent APP activation. Transient transfection assays revealed that CDK9 activity is required for the activation of ␥-FBG promoter transcription (Fig. 4, A and C), and Q-RT-PCR showed that CDK9 knockdown significantly suppresses the endogenous ␥-FBG expression during the 24-h time course (Fig. 4E).
Although the STAT3-CDK9 interaction was reported before (20), our study extends this previous work by: 1) demonstrating the interaction using an independent technique of confocal colocalization; 2) demonstrating that tyrosine-phosphorylated and Lys-acetylated STAT3 is found in the complex with CDK9; and, 3) discovering that the STAT3 NH 2 terminus participates in CDK9 complex formation and transcriptional activation.
Co-IP and confocal colocalization showed that the inducible STAT3⅐CDK9 complex is rapidly formed in the nucleus within 30 min after IL-6 stimulation (Fig. 2). Moreover, only the activated, nuclear translocated STAT3 complexes with CDK9, even though CDK9 is also found in the cytoplasm. We interpret this finding to mean that the tyrosine phosphorylation produces a conformational change in STAT3, exposing the CDK9interacting domains at NH 2 and COOH termini. Currently our data do not prove that the NH 2 terminus of STAT3 binds to CDK9 through a direct protein-protein interaction. Therefore, another possibility could be that the STAT3-CDK9 interaction is indirectly mediated through other protein-protein interactions that are mapped to the STAT3 NH 2 terminus. In this regard, a recent finding from our laboratory shows that the STAT3 NH 2 terminus is sufficient for the interaction with p300/CBP, an enzyme that acetylates two lysine residues (Lys 49 , Lys 87 ) in this domain (16). These acetylations increase the stability of the STAT3⅐p300/CBP complex, and are indispensable for STAT3-dependent target gene expression (16). This finding indicates the possibility that STAT3 NH 2 terminus-CDK9 interaction is indirectly mediated by p300/CBP. The NH 2 -terminal domain is highly conserved in STAT members. According to previous studies, NH 2 terminus is required for cooperative binding of STAT4 dimers to adjacent recognition sites on DNA (50). It also regulates multiple protein-protein interactions important for the functions of STAT1 and STAT2 (51)(52)(53)(54)(55). However, little is known about the function of NH 2 terminus in STAT3. Our findings reveal that the NH 2 terminus is involved in the interactions between STAT3, the p300/CBP coactivator and the P-TEFb transcriptional elongation com-plex. The important role of the NH 2 terminus for STAT3 function can also be seen in the finding that NH 2 -terminal-deleted mutant (⌬130) repressed both the basal and IL-6-inducible activities of ␥-FBG-LUC reporter gene (Fig. 3H). Although STAT3-(⌬130) has promoter binding activity, it could not effectively induce transcription because it is unable to successfully recruit coactivators or transcriptional elongation factors.
Although it is known that P-TEFb is generally required for transcription elongation, an unanswered question is whether P-TEFb is recruited to all promoters and regulates downstream gene transcription by similar mechanisms. The existence of eight potential P-TEFb complexes resulting from different combinations of two CDK9 isoforms (56) and four types of cyclins (57) suggest the possibility that unique P-TEFb complexes might be differentially recruited by inducible transcription factors for different genes. Consistent with this notion, the requirement for CDK9 varies widely among genes. For example, HIV replication can be inhibited by FP at concentrations that have no detectable effect on cellular genes transcription (32,58). Also, a recent study found that some p53 target genes, including p21 and PUMA, are activated when CDK9 activity is inhibited, suggesting a specific subset of p53 target genes can bypass the requirement of CDK9 activity for expression (59). The further study of other STAT3 target genes may answer the question whether CDK9 activity is generally required for all STAT3-dependent genes. It will be of interest to apply ChIP assays to monitor the kinetics of CDK9 association on different promoters of STAT3 target genes, and determine whether CDK9 utilizes a general mechanism to regulate some or all of STAT3-dependent genes.
Upon P-TEFb recruitment, an important substrate is Ser 2 in the heptad repeat of the RNA Pol II CTD. This notion is supported by our finding that CDK9 binds to TATA box, exon 5 and exon 7 of ␥-FBG gene in a similar pattern as phospho-Ser 2 CTD Pol II itself (Fig. 5, C and E). Moreover our data show that FP treatment specifically and efficiently suppresses phospho-Ser 2 CTD Pol II formation without affecting STAT3⅐CDK9 interaction (Fig. 6, C and E). Although our data show that there is constitutive Ser 2 CTD phosphorylation, which is globally independent of IL-6 stimulation (Fig. 6E), this fraction of phospho-Ser 2 CTD Pol II is not strongly engaged with the ␥-FBG promoter. We suggest based on our study that the STAT3-CDK9 interaction results into increased P-TEFb targeting to the ␥-FBG gene, producing local Pol II recruitment, CTD phosphorylation, and transcriptional elongation. In fact, our experiments indicate that both the recruitment of total Pol II and phospho-Ser 2 -CTD Pol II are decreased by CDK inhibition, despite a consistent level of Pol II expression (Figs. 5F and 6D). This result suggests that Ser 2 CTD phosphorylation might affect the stability of Pol II binding to chromatin.
In addition to Pol II, CDK9 is a kinase that autophosphorylates as well as phosphorylates other targets that have been partially characterized, including Myo D, hSPT5, and p53 (33)(34)(35)(36)(37). In an effort to determine whether STAT3 is also a substrate for CDK9 phosphorylation, our findings that STAT3 phosphorylation at Tyr 705 and Ser 727 are not changed by CDK inhibition (Fig. 6, A and B) indicate that these two sites are not CDK9 targets. However, these findings do not rule out the possibility that STAT3 is a potential substrate of CDK9 at some other site(s) yet to be discovered. Further studies will be necessary to understand whether CDK9 is able to cause phosphorylation of STAT3, especially at its NH 2 and/or COOH terminus.
Interestingly, previous studies show that CDK9 and cyclin T1 expression themselves can be up-regulated in T lymphocytes stimulated with cytokines or mitogens (60 -64). For example, a combination of IL-2, IL-6, and TNF-␣ increased both CDK9 and cyclin T1 protein levels in peripheral blood lymphocytes (63). In our study, we did not observe CDK9 induction in response to IL-6, but we have found that cyclin T1 was slightly up-regulated by IL-6 (data not shown), which could be another mechanism regulating P-TEFb activity. In addition, a recent finding shows that CDK9 can be acetylated by p300/CBP in vitro and this modification affects its ability to phosphorylate Pol II (65). All these findings suggest an additional level of complexity to the regulation of transcriptional elongation.
In summary, we provide evidence that activated STAT3 associates with P-TEFb to stimulate its recruitment and transcription elongation of the ␥-FBG gene. CDK9 regulates IL-6induced ␥-FBG transcription via a mechanism involving increased binding of total and phosphorylated RNA Pol II to ␥-FBG. Considering the important roles of FBG in inflammation and cancer, this finding has functional significance, making CDK9 an appealing target for therapeutic intervention.