INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The human immunodeficiency virus type 1 (HIV-1) regulatory protein Tat, the transactivator of transcription, dramatically increases the production of viral RNA (for a review, see reference 26). This effect is dependent on the transactivation response element (TAR) located downstream of the promoter in the HIV long terminal repeat (LTR). TAR is an RNA element, and TAR RNA binds directly to Tat (17, 72). Although Tat may also exert a stimulatory effect on transcriptional initiation, its predominant effect in vivo appears to be at the level of transcriptional elongation (42, 46, 47). In one model, Tat serves to promote the formation of highly processive RNA polymerase complexes at the HIV LTR. In the absence of Tat, HIV-directed transcripts tend to terminate prematurely at apparently random sites (45). This model is supported by experiments conducted with cell-free transcription systems (28, 44, 50, 51). A unique feature of the stimulation of HIV-directed transcription in vitro is the observation that it is preferentially inhibited by 5,6-dichloro-1--D-ribofuranosylbenzimidazole (DRB), an adenosine analogue that targets RNA polymerase II (pol II)-mediated elongation in vitro and in vivo (5, 51).

Efforts to uncover the mechanism of Tat action have included extensive searches for cellular components that interact with Tat in vivo and in vitro. This approach disclosed structural and functional interactions with several general transcription factors as well as other proteins with as-yet-unknown roles in transcription (8, 23, 38, 41, 43). Outstanding among these Tat-interacting proteins is a novel Tat-associated protein kinase, TAK, found by Herrmann and Rice (35, 36). This kinase is able to phosphorylate the carboxy-terminal repeat domain (CTD) of pol II and is highly sensitive to DRB.

The mammalian CTD consists of a series of 52 heptad repeats, YSPTSPS, located on the large subunit of pol II. It participates in gene expression at several levels and appears to coordinate the machinery of transcription and posttranscriptional RNA processing (37, 55, 57). The CTD can be extensively phosphorylated at multiple sites, predominantly on serine but also on threonine and tyrosine residues (11). Its phosphorylation plays a dominant role in transcription regulation: the hyperphosphorylated form of the large subunit (IIo) is associated with transcription elongation complexes, while only the hypomodified form (IIa) can assemble into the preinitiation complex (60). During the transition from initiation to elongation, or soon after, IIa is converted to IIo (10). The role of the CTD in transcription is not entirely understood, however, and it is possible that more than one round of phosphorylation is required, perhaps with intervening dephosphorylation. The CTD is essential for cell viability in yeast and for transcription from TATA-less promoters in vitro but is nonessential for TATA-containing promoters in vitro (1, 48, 59, 70). It is required for Tat transactivation (7, 61, 75).

At least 10 kinases which can phosphorylate the CTD in vitro are known, and 3 of them have an established connection with the transcription machinery. These are CDK7/cyclin H, subunits of the CDK-activating kinase (CAK) complex, which is a subassembly of transcription factor TFIIH; CDK8/cyclin C, a component of the pol II holoenzyme; and CDK9/cyclin T, now identified as components of the positive transcription elongation factor P-TEFb. P-TEFb was first recognized in Drosophila Kc cell extract as an activity needed to overcome abortive elongation and allow formation of long transcripts (53). P-TEFb is distinguished from other transcription factors by its sensitivity to very low doses of DRB (52). It differs from TFIIH, which exhibits less sensitivity to DRB, in that TFIIH functions in initiation and promoter clearance while P-TEFb has the attributes of an elongation factor (18, 53, 54, 62).

The possibility that TAK corresponds to human P-TEFb was suggested by the functional similarity between the stimulation of elongation by Tat and the shift to productive elongation by P-TEFb. Circumstantial support for this inference was provided by the observations that both P-TEFb and TAK are CTD kinases and that low concentrations of DRB inhibit P-TEFb, TAK, and the Tat effect (35, 51-53). Conclusive evidence came when Zhu et al. (80) cloned the catalytic subunit of Drosophila P-TEFb and showed that its human homologue (PITALRE) is identical to the kinase subunit of TAK. PITALRE, now called CDK9 (67, 73), was first cloned by Graña et al. (29) as a CDC2-related kinase of unknown function.

Most of the kinases in this family have cyclin-regulating subunits, and the cyclin partners of human and Drosophila CDK9 were recently cloned and studied (66). The human cyclins T1, T2a, and T2b were all identified as functional partners of CDK9, and almost all of the CDK9 in HeLa nuclear extracts is associated with either cyclin T1 or T2 (67). Cyclin T1 can also bind directly to the activation domain of Tat (73). Several additional proteins are also associated with CDK9 in immunocomplexes (24, 78, 80), suggesting that human P-TEFb is a multiprotein complex or that there are other CDK9-containing complexes in addition to P-TEFb. It is also likely that Tat binds other factors in addition to CDK9/cyclin T (79). The gene for human cyclin T1 maps to chromosome 12, and this chromosome is necessary for a strong Tat effect on transcription from the HIV LTR in mouse and Chinese hamster ovary (CHO) cells (73). Transfection of CHO and mouse cells with the cyclin T1 gene overcomes the poor Tat response that was previously noticed in these cells (73). The species specificity of the Tat response is also partially complemented in mouse cells harboring human chromosome 6, but the factor(s) responsible is unknown.

Our study addresses the interaction of CDK9/cyclin T1 with Tat and its CTD substrate. We demonstrate that human Tat-associated P-TEFb contains cyclin T1 and that rodent P-TEFb is deficient in binding to Tat. Human cyclin T1 as part of the P-TEFb complex in CHO cells complements this deficiency and thus contributes to HIV-1 species specificity. We also show that human TAK is part of a large complex that phosphorylates serine 5 in the CTD heptad and is itself phosphorylated on serine and threonine residues. Our data suggest that TAK and CAK differentially phosphorylate the CTD substrate.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Materials. Nucleoside triphosphates were purchased from Pharmacia-LKB. [-³²P]ATP was obtained from ICN Pharmaceuticals Inc. Thin-layer cellulose plates were purchased from EM Science, Gibbstown, N.J. Immobilon-P polyvinylidene difluoride membrane was from Millipore. CTD3 peptides were synthesized at the University of Medicine and Dentistry of New Jersey Molecular Resource Facility with an ABI 433A synthesizer and were purified by high-pressure liquid chromatography on a C₁₈ column. Affinity-purified anti-CDK9 immunoglobulin G (IgG) (anti-PITALRE-CT) was purchased from Santa Cruz Biotechnology. Drosophila RNA pol II and cyclin T1 and T2 antisera were gifts from D. Price (67, 80).

Cells, plasmids, and viruses. Rodent-human hybrid cell lines and the parental human lymphoblast cell line (GM07890) were obtained from Coriell Cell Repositories, Camden, N.J. Glutathione S-transferase (GST)-Tat constructs (68) were obtained from the AIDS Research and Reference Reagent Program, National Institutes of Health, and from the laboratory of K. Jones (GST-Tat-1K41A). Baculoviruses harboring recombinant CDK7, cyclin H, and MAT-1 were gifts from D. Morgan (20, 21).

Preparation of cell extracts. Cell extracts were prepared essentially as described by Harlow and Lane (31). Cell monolayers were washed with phosphate-buffered saline and then lysed for 30 min with occasional rocking by using prechilled EBCD buffer (50 mM Tris [pH 8.0], 120 mM NaCl, 0.5% Nonidet P-40 [NP-40], 5 mM dithiothreitol [DTT]) containing 4 mM MgCl₂, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 µg (each) of aprotinin, pepstatin A, and leupeptin per ml. The lysate was centrifuged for 10 min at 2,500 × g, and the supernatant was collected. Cytoplasmic (S-100) and nuclear extracts were prepared as described by Dignam et al. (15) except that the dialysis step was omitted for the cytoplasmic extract. Glycerol was added to 10%, and the extracts were stored at 80°C. For gel filtration and glycerol gradient experiments, the nuclear extract was clarified by centrifugation at 100,000 × g for 30 min in a Beckman SW 50.1 rotor.

Partial purification of TAK. TAK purification was as described previously (80) with an additional chromatographic step. The peak fractions from the POROS 20 SP column were pooled, and the KCl concentration was adjusted to 100 mM. The pooled fractions were loaded on a Reactive Blue (Sigma) column equilibrated with column buffer (25 mM HEPES [pH 7.6], 0.1 mM EDTA, 1 mM DTT, 0.1 mM PMSF, 4 mM MgCl₂, 10% glycerol, and protease inhibitors [1 µg {each} of aprotinin, pepstatin A, and leupeptin per ml]) containing 100 mM KCl. The active fractions were eluted in the same buffer containing 0.5 to 1 M KCl.

TAK assay. The production of GST-Tat fusion proteins and the TAK activity assay were carried out as described by Zhu et al. (80). For nucleotide usage experiments, unlabeled nucleotides were added at various concentrations. The gels were dried, quantified on a Packard Instant Imager, and exposed to X-ray film.

Immunoprecipitation (IP)-kinase assays. Protein A-Sepharose beads (Pharmacia) were washed three times with EBCD buffer and incubated at 4°C for 30 min with anti-CDK9 antibody (0.5 to 1 µg of IgG/10 µl of packed beads). The beads were washed four times with EBCD buffer containing 0.2 mM PMSF, incubated for 1 h with cell extract at 4°C, washed five times with EBCD containing 0.03% sodium dodecyl sulfate, and subjected to a kinase assay as described previously (80).

Western blot analysis. Proteins from polyacrylamide gels were transferred to nitrocellulose membranes by using the semidry transfer protocol (4). The membranes were blocked for 1 h with blocking buffer (20 mM Tris [pH 7.6], 150 mM NaCl, 2.3% low-fat milk, 0.5% bovine serum albumin, 0.5% NP-40, 0.05% Tween 20) and incubated for 4 to 6 h at room temperature or overnight at 4°C with primary antibody. This was followed by three 10-min washes with TBST (20 mM Tris [pH 7.6], 150 mM NaCl, 0.05% Tween 20) and incubation for 20 min with blocking buffer. Secondary antibody (Bio-Rad goat anti-rabbit IgG-horseradish peroxidase conjugate) was added to the blocking buffer at a 1:3,000 dilution, and the membranes were incubated for an additional hour. After three 15-min washes with TBST, membranes were reacted with chemiluminescence reagent (NEN Life Science Products) and exposed to Kodak Biomax film.

EDTA-Mg experiments. EDTA was added to 293 cell S-100 fractions (15 µl [225 µg of protein]) to a final concentration of 5 or 10 mM in a total volume of 100 µl of EBCD buffer containing 0.2 mM PMSF. After incubation at room temperature for 5 to 30 min, MgCl₂ was added to a final concentration of 10 or 20 mM. The samples were incubated for an additional 5 to 30 min at room temperature, after which they were added to 50 µl of beads containing GST-Tat48 in EBCD buffer without magnesium; this was followed by the standard TAK assay. Thus, the final concentrations of MgCl₂ and EDTA were two-thirds of those specified.

Phosphoamino acid analysis. Phosphoamino acid analysis was performed essentially as described by Ausubel et al. (4). Briefly, phosphorylated peptide and CDK9 were transferred from gels to polyvinylidene difluoride membranes, hydrolyzed in 6 M HCl, dried in a vacuum, and resuspended in 2 to 5 µl of water. In some experiments, the membranes were treated with alkali before acid hydrolysis. A portion of the digest containing 1,000 to 5,000 cpm was spotted on a cellulose thin-layer chromatography plate, and 1 µl of a solution containing about 1 µg of each unlabeled phosphoamino acid was spotted as a reference. The amino acids were resolved by two-dimensional electrophoresis (HTLE 7000; CBS Scientific). The amino acid standards were visualized by ninhydrin staining, and ³²P-labeled amino acids were detected by autoradiography and quantified with a Packard Instant Imager.

Expression of recombinant CDK7 complexes and preparation of Sf9 cell extracts. Monolayers of Sf9 insect cells (2 × 10⁷ cells) were infected with recombinant baculoviruses harboring the gene for CDK7 or cyclin H or with a mixture of viruses encoding the CAK complex (CDK7, cyclin H, and MAT-1) at a multiplicity of infection of 5 to 10 PFU/cell. Two days after infection, the cells were lysed in hypotonic buffer (10 mM HEPES [pH 7.4], 10 mM NaCl, 1 mM EDTA, 1 mM PMSF, and 1 µg of leupeptin per ml), NaCl was added to a 150 mM final concentration, and the lysates were clarified by centrifugation.

Fractionation of cytoplasmic and nuclear extracts. Nuclear or cytoplasmic extracts were diluted into 0.5 ml of HKMEG column running buffer (20 mM HEPES [pH 7.6], 150 mM KCl, 4 mM MgCl₂, 0.1 mM EDTA, 10% glycerol, 0.01% NP-40, 0.1 mM PMSF, 1 mM DTT, 1 µg [each] of aprotinin, leupeptin, and pepstatin A per ml) and applied to a 45-ml column of Sephacryl S-300 Superfine (Pharmacia) equilibrated in the same buffer. The void volume was determined by using blue dextran, the column was calibrated by using the high-molecular-mass marker kit from Sigma (MW GF-100), and the molecular masses of complexes were determined by extrapolation. Fractions (0.5 ml) were collected at flow rate of 0.5 ml/min. Glycerol gradient ultracentrifugation was carried out in 4 ml of a 20 to 50% glycerol gradient in HKMEG buffer. Nuclear or cytoplasmic extracts (3 mg of protein) were diluted into 0.2 ml of HKMEG and loaded on the gradient. An identical gradient was loaded with high-molecular-weight marker proteins, and both gradients were centrifuged at 45,000 rpm for 18 h in a Beckman SW 50.1 rotor. Fractions (0.1 ml) were collected by pumping from the bottom of each tube.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Association of CTD kinase with HIV-1 Tat. The Tat-associated kinase TAK binds to full-length Tat (Tat86), first-exon Tat (Tat72), or its activation domain (Tat48) and hyperphosphorylates a model substrate, GST-CTD (35). We used the TAK assay, a GST-Tat pull-down kinase assay, to examine the interactions of TAK with its substrates and different GST-Tat fusion proteins. Figure 1A shows that TAK phosphorylates intact pol II (lane 7), as well as a synthetic peptide, CTD3 [ACS(YSPTSPS)₃KK], containing three copies of the CTD heptad (lanes 1 and 3). TAK binding requires all three regions of the activation domain and is abolished by mutations in the cysteine-rich region (Tat48Cys22) (lanes 2 and 8), the core region (Tat86Lys41) (lane 4), and the N terminus (80).

View larger version (40K): [in this window] [in a new window]   FIG. 1.   Specificity of CTD phosphorylation by TAK/P-TEFb. (A) Detection of TAK activity with CTD3 and RNA pol II substrates. The indicated GST-Tat or GST-Tat mutant fusion proteins (lanes 1 to 4, 7, and 8) were used to isolate TAK from cytoplasmic fractions of 293 cells by the GST pull-down procedure. For comparison, IP-kinase assays were conducted with immunoprecipitates prepared with anti-CDK9 or control antibodies (lanes 5 and 6, respectively). Kinase reaction mixtures contained 13.5 µg of CTD3 (lanes 1 to 6) or 20 ng of purified Drosophila RNA pol II (lanes 7 and 8) as the substrate. (B) CTD3 phosphorylation by TAK eluted from immunoprecipitated CDK9 complexes. The immunoprecipitates were washed extensively, and complexes were eluted for 18 h with the indicated amounts of CDK9 C-terminal peptide. The eluate was subjected to TAK assay, and the supernatant was examined by gel electrophoresis and autoradiography. The control was the standard TAK assay with 293 cell S100 and GST-Tat48. (C) Detection of CDK9 in the eluates after binding to GST-Tat48 beads. The proteins bound to the beads in panel B were examined by Western blotting with anti-CDK9 antibody.

Species specificity of the interaction between Tat and CDK9/cyclin T-containing complexes. TAK activity is found in human HeLa and 293 cell extracts (36, 80) and Jurkat lymphocyte extracts (our unpublished data) as well as in monkey (COS) cell extracts (36). Since neither hamster nor mouse cells support efficient Tat transactivation (58), we examined CHO and 3T3 cell extracts for TAK activity. Figure 2A shows that extracts prepared from 3T3 and CHO cells (lanes 4 and 5) gave rise to greatly reduced levels of CTD3 phosphorylation compared to 293, HeLa, and COS cell extracts (lanes 1 to 3). Kinase assays demonstrated that the diminished TAK activity of the rodent cell extracts correlated with very poor phosphorylation of CDK9 bound to the GST-Tat48 beads (Fig. 2A, lanes 1 to 5) despite the presence of comparable levels of CDK9 in the extracts, as shown by Western blotting (Fig. 2C, lanes 1 to 5). Mixing experiments, in which increasing amounts of CHO cell extracts were added to 293 cell extracts, ruled out the possibility that the CHO extracts contain an inhibitor of TAK activity (data not shown). Furthermore, CDK9 immunoprecipitated from rodent cell extract phosphorylated CTD3 with an efficiency similar to that of human CDK9 immunoprecipitates (Fig. 2D), demonstrating that rodent cells have active CDK9 complexes. Thus, the species specificity must be conferred by another component of these complexes.

View larger version (62K): [in this window] [in a new window]   FIG. 2.   TAK in CHO-human hybrid cell extracts. (A) CTD3 phosphorylation by TAK pulled down from different cell extracts: human (293, HeLa, and GM07890), monkey (COS), rodent (3T3 and CHO), and the rodent-human hybrids CHO(6) and CHO(12). (B) Western blotting of proteins bound to GST-Tat48 beads from the indicated cell extracts by using anti-CDK9 and anti-cyclin T1 antibodies. (C) Western blot analysis of proteins in unfractionated cell extracts with the same antibodies as in panel B. (D) IP-kinase assays conducted with the same cell extracts with CTD3 as the substrate.

CDK9 exclusively phosphorylates serine residues in CTD3 but is phosphorylated on both serine and threonine. The CTD of RNA pol II is phosphorylated by various CTD kinases on serine, threonine, or tyrosine residues (12, 13). We used CTD3 as a model substrate to identify the CTD residues that are phosphorylated by TAK complexes. CTD3 gives rise to two phosphorylated bands when separated in sodium dodecyl sulfate gels (see, for example, Fig. 1B). Two-dimensional phosphoamino acid analysis of the upper and lower phosphorylated bands revealed phosphorylation of serine residues only (Fig. 3A and B, respectively). Identification of the specific serine residue that is targeted is described below. Contrary to an earlier report (14), CDK9 itself is phosphorylated on both serine and threonine residues when bound to Tat (Fig. 3D) or anti-CDK9 antibody (Fig. 3C). The same result was obtained with chromatographically purified CDK9 (Reactive Blue column fraction; data not shown). Phosphotyrosine was not detected. These findings agree with the conclusion that CDK9 (PITALRE) is a Ser/Thr proline-directed kinase (25), since in the heptad serines but not threonines are followed by prolines (Fig. 4A). Moreover, the detection of phosphorylated threonine in CDK9 is consistent with the prediction, based on its sequence (29), that the kinase contains a typical T loop in which Thr 186 is a potential phosphorylation site.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Phosphoamino acid analysis of CTD3 peptide and CDK9. CTD3, phosphorylated in TAK assays, resolved into two bands upon gel electrophoresis. The upper band (A) and lower band (B) were analyzed separately. Positions of phosphoamino acid markers are circled. CDK9 was phosphorylated in IP-kinase assays (C) or TAK assays (D) and examined in the same way.

View larger version (35K): [in this window] [in a new window]   FIG. 4.   Selectivity of CTD3 phosphorylation by CDK9 and CDK7. (A) Sequence of wild-type (WT) CTD3 and mutant peptides. Substitutions are denoted by the underlined boldface letters. (B) Phosphorylation of the indicated CTD3 peptides in TAK assays. (C) Comparison of CTD3 phosphorylation by CDK9 and CDK7 complexes. CDK9 was monitored in TAK assays conducted with various amounts of 293 cell S100 (15 µg/ml). CDK7 was monitored by using Sf9 cell extracts (~60 µg) containing either recombinant CAK or recombinant CDK7 and cyclin H. The control was the kinase assay with mock-infected Sf9 cell extract.

Serine 5 in the CTD heptad is essential for phosphorylation by TAK. The CTD heptad contains three serine residues (Fig. 4A), and human CTD kinases specific for serine 5, serines 2 and 7, or serines 2 and 5 have been characterized (27, 65, 77). To identify the serine residue(s) that is phosphorylated by CDK9, we generated a series of CTD3 mutant peptides in which serines were replaced by alanines (Fig. 4A). In peptides S2A, S5A, and S7A, all three serines at positions 2, 5, and 7, respectively, were changed to alanines. Comparison with wild-type CTD3 shows that serine 5 is essential for heptad phosphorylation by TAK and that the serines at positions 2 and 7 are dispensable (Fig. 4B). Next, we replaced all six serines at positions 2 and 7 in the three heptad repeats; this peptide, S2,7A, was also efficiently phosphorylated by TAK (Fig. 4A and B), implying that serine 5 is the only serine in the heptad that is recognized by TAK.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Kinetics of TAK activity and nucleotide usage by TAK. (A) Time course of CTD3 phosphorylation by TAK from 293 cell S100. Reactions were stopped at the indicated time points. (B) Kinase reaction mixtures containing [-32P]ATP were supplemented with the indicated unlabeled nucleotides at the concentrations shown. Control, no unlabeled nucleotides added.

Nucleotide usage by Tat-associated CDK9/cyclin T complex. Most protein kinases utilize only ATP as a phosphate donor (40). This was shown to be true for purified Drosophila P-TEFb, which contains the CDK9/cyclin T complex (52). For comparison, we tested the nucleotide utilization by human CDK9/cyclin T complexes in competition experiments by adding an excess of unlabeled nucleotides over [-³²P]ATP and monitoring CTD3 phosphorylation. As shown in Fig. 5B, the efficiency of ATP utilization in the TAK assay was at least fivefold higher than that of dATP and at least 50-fold higher than that of GTP and dGTP. The pyrimidine nucleotides CTP, dCTP, UTP, and dTTP were not utilized at 500 µM (Fig. 5B) or even at concentrations of as high as 1 mM (data not shown). These data are very similar to results obtained with purified Drosophila P-TEFb (52), indicating that this elongation factor and Tat-bound human CDK9 complexes have the same phosphate donor specificity. GTP alone (without ATP) can be used as the phosphate donor for both Thr and Ser phosphorylation in CDK9 and for Ser phosphorylation in CTD3 (data not shown). The phosphorylation efficiency with labeled GTP is about 50-fold less than that with labeled ATP, confirming the results obtained in the competition experiments.

Requirement for Mg ions for Tat-TAK interaction. Initially, TAK activity was found in the nuclear fraction, but not the cytoplasmic fraction, of HeLa cells (36). However, we detected high levels of cytoplasmic TAK activity when 293, HeLa, or Jurkat cells were fractionated by the same method (16) except for omission of the last dialysis step. This dialysis was against a buffer lacking Mg²⁺. We observed that TAK activity was lost when extracts were dialyzed against this magnesium-free buffer (data not shown). To determine whether the presence of magnesium ions is responsible for the differences observed, EDTA was added to the cytoplasmic extract at two concentrations to chelate Mg²⁺ ions. As shown in Fig. 6A (lanes 2 and 3), this resulted in the loss of TAK activity, and the inhibitory effect of EDTA was reversed by adding back excess MgCl₂ (lanes 4 to 7). When the proteins bound to the beads were analyzed by Western blotting and probing with antibody against CDK9, CTD3 phosphorylation correlated perfectly with the binding of CDK9 to the beads (Fig. 6B). Similarly, TAK binding and activity could be restored by adding low concentrations of Zn²⁺ (up to 2 mM) (data not shown). To exclude the possibility that prolonged incubation in the presence of EDTA caused CDK9 degradation, the cytoplasmic extract was tested for CDK9 after the binding reaction. Western blotting showed that CDK9 was not degraded in the presence of EDTA (Fig. 6C), indicating that CDK9-containing complexes are unable to bind to GST-Tat48 in the absence of Mg²⁺. Two possibilities were considered: divalent cations are necessary for the stability of CDK9/cyclin T complexes or for the binding of such complexes to GST-Tat48. To distinguish between these possibilities, we used anti-CDK9 antibody to immunoprecipitate CDK9/cyclin T-containing complexes from the same extracts. The phosphorylation of CTD3 by the immunoprecipitates was unaffected by the removal of Mg ions (Fig. 6D), indicating that the CDK9/cyclin T complexes are not disrupted by EDTA and suggesting that the binding of TAK to the Tat activation domain is dependent on the presence of divalent cations.

View larger version (30K): [in this window] [in a new window]   FIG. 6.   Effects of Mg and EDTA on TAK activity. (A) 293 cell S100 fractions were treated with EDTA and MgCl2 at the final concentrations indicated and then subjected to TAK assay. After the kinase assay, the supernatant was examined for CTD3 phosphorylation. (B) Western blot analysis with anti-CDK9 antibody of the proteins bound to GST-Tat48 beads in the same experiment as for panel A. (C) Western blot analysis with anti-CDK9 antibody of proteins remaining in the supernatant after binding to GST-Tat48 beads in the same experiment as for panels A and B. Load, sample of 293 cell S100. (D) IP-kinase assays conducted with 293 cell S100 (15 mg/ml) subjected to treatment with EDTA and Mg2+ as indicated.

CDK9/cyclin T is present in large complexes. By gel filtration chromatography, Gariga et al. (24) found that CDK9 is present in 293 cell extracts as two multisubunit complexes of 670 and 158 kDa. In keeping with this finding, multiple polypeptides were detected when CDK9 complexes were immunoprecipitated from cell extracts (24, 29, 78, 80). On the other hand, the molecular mass of partially purified TAK has been reported to be ~110 kDa (75). Since TAK is present in both nuclear and cytoplasmic fractions, we considered the possibility that different TAK-containing complexes exist in these subcellular fractions. We employed both gel filtration chromatography and glycerol gradient sedimentation to examine TAK-containing complexes and estimate their molecular weights. To minimize possible disruption of the complexes, 293 cells were lysed by the method of Dignam et al. (15), which results in a transcriptionally active nuclear extract.

View larger version (42K): [in this window] [in a new window]   FIG. 7.   Separation of TAK complexes by gel filtration and sedimentation. (A) CTD3 phosphorylation by TAK pulled down from fractions after resolution of 293 cell S100 in a gel filtration column. Molecular mass markers (in kilodaltons) are indicated by arrows. Triangles show the estimated molecular masses of the large complex (950 kDa), CDK9/cyclin T complex (130 kDa), cyclin T (90 kDa), and CDK9 (40 kDa). (B and C) The same fractions were examined by Western blotting with antibodies to CDK9 (B) and cyclin T1 (C). Load, sample of the 293 cell S100 loaded on the column. (D) Cytoplasmic extracts from 293 cells in 150 mM KCl or adjusted to 1 M KCl were fractionated in glycerol gradients. Fractions were subjected to TAK assays, and the phosphorylated CTD3 bands were quantified, normalized, and plotted.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

TAK and P-TEFb. TAK and P-TEFb have a common CTD kinase activity conferred by CDK9 (74, 80). We show here that TAK, like P-TEFb (67), contains CDK9's cyclin partner, cyclin T1. However it is not clear whether P-TEFb and TAK consist solely of the CDK9/cyclin T heterodimer and whether they are identical in composition and properties. Price and colleagues showed that CDK9-depleted HeLa nuclear extracts have diminished in vitro transcription activity that can be complemented by the addition of purified Drosophila P-TEFb or recombinant human CDK9/cyclin T (67, 80). This complementation suggests that the core complex, CDK9/cyclin T, is sufficient to restore in vitro transcription elongation, although a requirement for additional components that are not efficiently depleted cannot be excluded. In this connection, it is noteworthy that, to date, Tat transactivation has been obtained only with preparations containing several additional polypeptides (49, 78).

TAK-Tat interactions. Both TAK and P-TEFb were initially identified in nuclear extracts (36, 54), while our fractionation experiments detected TAK in cytoplasmic extracts as well as nuclear extracts. The localization of at least a portion of TAK/P-TEFb in the cytoplasm is supported by the properties of transdominant Tat mutations. Such mutant Tat proteins interact with TAK (19) and inhibit wild-type Tat function in a dominant negative fashion. This phenotype is conferred by mutations in the Tat basic region, which also contains the nuclear localization signal (56). A Tat protein carrying a deletion in the basic region localized exclusively to the cytoplasm and was a dominant negative inhibitor. Inclusion of additional mutations in the activation domain eliminated this transdominant activity (63). Taken together with the observation that TAK binding requires an intact Tat activation domain, these findings imply that TAK is squelched in the cytoplasm by the transdominant Tat proteins. Interestingly, Tat interacts with another nuclear and/or cytoplasmic factor(s) which renders Tat capable of transactivating the HIV LTR in vitro without a lag period (30). Thus, as suggested previously (6, 63), it is likely that Tat is assembled with CDK9/cyclin T and perhaps other transcription regulators in the cytoplasm before entering the nucleus.

TAK, CAK, and CTD phosphorylation. The CTD kinases are a functionally heterogeneous group, including enzymes that phosphorylate serines 5, 2 and 5, or 2 and 7 in the CTD heptad, as well as factors participating in varied cellular functions and subjected to a spectrum of controls. Together with some other serine 5-specific CTD kinases, such as TFIIH and the mitogen-activated protein kinases, CDK9 belongs to the proline-directed protein kinase family (25, 71). Even though TFIIH and mitogen-activated protein kinases show the same detailed specificity for the CTD motif [PX(S/T)P], their recognition sites are not identical (71). Such differences in site recognition may be the characteristic that confers functional distinctions among these CTD kinases.

Species specificity and the mechanism of Tat transactivation. A widely cited model of Tat transactivation views the process in terms of the recruitment of P-TEFb to the elongating transcription complex by TAR-associated Tat. This model is consistent with many studies, but the findings presented here are more compatible with an alternative mode of action. We suggest that Tat first binds to TAK/P-TEFb via cyclin T1; the resultant Tat-TAK/P-TEFb complex then associates with the pol II complex, and this elongation-competent pol II assembly is stabilized, and possibly activated, by binding to both the loop and bulge structures of TAR.

View larger version (39K): [in this window] [in a new window]   FIG. 8.   Schematic representation of Tat delivery to its nuclear transactivation site. The model depicts TAK assembly in the cytoplasm. P-TEFb, composed of CDK9, cyclin T1, and other components, interacts with the activation domain (AD) of Tat. In the nucleus, the TAK complex (P-TEFb-Tat) interacts with the transcription complex via cyclin T1 and the basic domain (BD) of Tat, both of which bind to TAR. Through phosphorylation of the pol II CTD, the elongating complex becomes more processive and Tat transactivation is achieved. (See text for details.)