INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The Tat protein encoded by the human immunodeficiency virus type 1 (HIV-1) genome strongly stimulates the synthesis of full-length HIV-1 transcripts by increasing the processivity of RNA polymerase II (Pol II). Tat recognizes the trans-acting response element (TAR) RNA stem-loop structure near the 5' end of the nascent viral transcript and is proposed to function by recruiting a cellular cofactor(s) to Pol II in a TAR loop sequence-specific manner (for reviews, see references 8, 17, and 18).

Tat transactivation requires the C-terminal domain (CTD) of the largest subunit of Pol II (3, 29, 32, 41). Since hyperphosphorylation of the CTD correlates with processive transcriptional elongation (6) and Tat activity is sensitive to kinase inhibitors (e.g., DRB) that inhibit CTD phosphorylation (24, 25), it was proposed that Tat stimulates elongation by recruitment of a CTD kinase(s) to Pol II (reference 18 and references therein). One of the CTD kinases implicated in Tat transactivation is P-TEFb (positive transcription elongation factor b). It was originally identified as a general elongation factor in Drosophila melanogaster and later in mammalian cells (26, 27, 45). Active P-TEFb is composed of a cyclin-dependent kinase, Cdk9, and its partner cyclin T (cyclin T1 [CycT1] or the minor forms T2a and T2b [34, 37]). The in vivo assembly of the Cdk9-CycT1 heterodimer of P-TEFb requires a chaperone-dependent folding pathway involving the sequential actions of Hsp70 and a kinase-specific chaperone complex, Hsp90-Cdc37 (30). Immunodepletion of P-TEFb from HeLa nuclear extracts or inclusion of DRB in an in vitro transcription assay eliminates both basal and Tat-activated HIV-1 transcription (24, 42, 45). Supplementing the depleted extract with purified wild-type P-TEFb but not a kinase-defective mutant P-TEFb restores both activities (42). By using kinase inhibitors or introduction of dominant negative Cdk9 mutants into cells, the functional significance of P-TEFb and its kinase activity in Tat activation has also been demonstrated in vivo (13, 24).

In addition to Cdk9, the CycT1 subunit of P-TEFb also plays a critical role in Tat activation. First, CycT1 mediates the interaction of P-TEFb with Tat and the binding of recombinant CycT1 to Tat enhances the affinity and specificity of the CycT1-Tat-TAR interaction and confers dependence on TAR loop sequences (10, 37). Second, the interaction of human but not rodent P-TEFb with Tat through the cyclin box region in human CycT1 mediates a Tat-specific and species-restricted activation of HIV-1 transcription (1, 2, 9, 10). Third, overexpression of human CycT1 in nonpermissive rodent cells rescues Tat activation of HIV-1 transcription (37). This activity is attributed to a critical cysteine residue at position 261 that is present in human CycT1 but absent in the rodent homologue (1, 9, 10). This cysteine residue is located in a region termed TRM (Tat-TAR recognition motif), which is important for the interaction of human CycT1 with Tat and TAR (10). While the exact N-terminal boundary of TRM has not been determined, the C-terminal boundary of TRM is located between amino acids 250 and 262 at the C-terminal edge of the cyclin box (amino acids 1 to 272).

In addition to P-TEFb, several other cellular proteins important for Tat transactivation have also been characterized. Among these are Tat-SF1 (20, 44), the human homologue of the Saccharomyces cerevisiae transcription factor SPT5 (38), TFIIH (5, 11, 32), TFIIF (19), and a Tat-associated histone acetyltransferase (reference 17 and references therein). Tat-SF1 was biochemically identified as a Tat cofactor, and it stimulates Tat activation both in vivo and in vitro (20, 44). It was also found to have a general elongation activity under certain conditions (23, 31). Tat-SF1 interacts with Tat (44) and P-TEFb (42) and was recently found to interact with human SPT5, Pol II, and the RAP30 subunit of TFIIF as well (20). The human SPT4 and SPT5 proteins form a complex called DSIF (DRB-sensitivity-inducing factor [35]). It arrests the elongation of Pol II at sites proximal to the promoter with the help of the negative elongation factor NELF (39). It has been shown that P-TEFb positively regulates Pol II elongation by, at least in part, suppressing the activity of DSIF in a phosphorylation step that is DRB sensitive (36). While that study identified DSIF as a negative factor, other experiments have also revealed a positive role of DSIF in Tat-activated HIV-1 transcription (20, 38). Recently, Parada and Roeder (31) reported the identification of a novel RNA Pol II-containing complex that supports Tat-activated HIV-1 transcription. Interestingly, this complex contains several previously identified elongation factors, P-TEFb, Tat-SF1, and DSIF, in addition to Pol II and other unidentified polypeptides. Exactly how these elongation factors and Pol II cooperate to stimulate HIV-1 transcription is still unclear. Nevertheless, an emerging picture of the involvement of multiple factors for the control of elongation is evident, as recent studies indicated that these factors may exist in the same complex and function in combination or in a sequential manner to promote Tat stimulation of Pol II processivity.

The formation of a TAR loop-dependent P-TEFb-Tat-TAR ribonucleoprotein complex is recognized as an essential step towards the assembly of productive Pol II elongation machinery at the HIV-1 promoter. Previous analyses of the TAR loop-dependent complex used free, recombinant CycT1 (10, 37). Since the active form of P-TEFb for Tat transactivation consists of a heterodimer of Cdk9-CycT1, we wanted to understand how P-TEFb forms a stable complex with Tat and TAR. In this report, we showed that formation of the P-TEFb-Tat-TAR ternary complex requires the deactivation of two autoinhibitory mechanisms in P-TEFb. Autophosphorylation of Cdk9 overcomes the first autoinhibition by creating a favorable P-TEFb conformation and exposing the TRM region in CycT1 for efficient interaction with TAR RNA. The second autoinhibition is caused by the intramolecular interaction between the N- and C-terminal regions of CycT1, which blocks the access of TAR RNA to CycT1 TRM. Relief of this autoinhibition is provided by the interaction of the CycT1 C-terminal region with Tat-SF1. Upon release from the intramolecular interaction, the C-terminal region also interacts with Pol II and is required for efficient HIV-1 transcriptional elongation, suggesting that it may link the basal transcription elongation apparatus to the P-TEFb-Tat-TAR ternary complex. These results reveal novel control steps for the assembly at the HIV-1 promoter of a multicomponent elongation complex for Tat transactivation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

DNA constructs and antibodies. Different hemagglutinin (HA)-tagged CycT1 cDNA fragments were generated by PCR and subcloned into the BamHI and EcoRI sites of the pcDNA3 expression vector (Invitrogen). Glutathione S-transferase (GST)-CycT11-HA was generated by subcloning the fragment containing a C-terminally truncated CycT1 (with only amino acids 1 to 333) (CycT11) into the BamHI and EcoRI sites of the pGEX-4T-1-HA expression vector, which contains the HA epitope cloned into the SalI and NotI sites of pGEX-4T-1 (Amersham Pharmacia). GST-CycT1-C was generated by subcloning the CycT1 cDNA fragment encoding a portion of the CycT1 C-terminal region (amino acids 402 to 701) into the BamHI and EcoRI sites of the pGEX-2T expression vector. C-terminally Flag-tagged Tat-SF1 cDNA was cloned into the EcoRI site of the pSV7d expression vector (Sigma). Anti-Cdk9 and anti-Pol IIa antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, Calif.), and polyclonal anti-CycT1 antibodies were generated in rabbits against the GST-CycT1-C fusion protein.

Expression and purification of P-TEFb complexes and Tat-SF1. For P-TEFb-containing CycT1 deletion mutants, plasmids bearing the gene expressing HA-tagged, C-terminally truncated CycT1 were transiently transfected into 293T cells. P-TEFb complexes were affinity purified from the cell lysates 48 h later as described previously (42). Cdk9-Flag-HA-CycT1 complexes were generated by coexpression of HA-CycT1 and Cdk9-Flag in 293T cells followed by two sequential affinity purification steps (with anti-HA and anti-Flag antibodies) and peptide elution. A mutant P-TEFb with a kinase-defective Cdk9-HA subunit was generated and purified as described previously (42).

Tat-binding assay. Binding of P-TEFb complexes to immobilized GST-Tat(1-48) or activation domain mutant GST-Tat(1-48, C22G) was performed as described previously (2).

TAR RNA EMSA. ³²P-labeled TAR RNA was synthesized by T7 RNA polymerase from the HindIII-digested DNA templates pT7TAR (wild type) and pT7TAR(+31/+34) as described previously (42). Electrophoretic mobility shift assays (EMSA) were carried out essentially as described previously (10), and when noted in the figures, 100 µM ATP or [-S]ATP was also included in the incubation mixture prior to electrophoresis. To study the effect of Tat-SF1 on P-TEFb-Tat-TAR complex formation, approximately 50 ng of affinity-purified, Flag-tagged Tat-SF1 was preincubated with P-TEFb (containing ~30 ng of Cdk9) for 10 min before being added to the gel shift reaction mixture.

Partial protease digestion of P-TEFb. Purified P-TEFb complexes were first incubated with ATP in the kinase buffer under the same conditions described for the EMSA reaction. Trypsin diluted in D_{0.1 M KCl} buffer (20 mM HEPES-KOH [pH 7.9], 10% glycerol, 100 mM KCl, 0.2 mM EDTA, 0.05% NP-40, 1 mM DTT) was then added to the reaction mixture at a final concentration of 7 ng/ml, and the incubation continued for another two minutes at 30°C. HA-tagged CycT1 fragments were visualized by Western blotting with antibodies specific for HA.

Transcription assay. Immunodepletion of P-TEFb from HeLa nuclear extracts was carried out with immobilized anti-Cdk9 and anti-CycT1 antibodies as described previously (42). In vitro transcription reactions with mixtures containing HeLa nuclear extract depleted of P-TEFb, affinity purified P-TEFb complexes, HIV-1 promoter templates, and Tat were carried out exactly as described previously (42).

Detecting CycT1-associated proteins in HeLa nuclear extract. Approximately 0.15 mg of HeLa nuclear extracts was dialyzed against D_{0.1M KCl} buffer and then incubated with 0.5 µg of GST or the various GST-CycT1 fusion proteins immobilized on glutathione-Sepharose beads for 3 h at 4°C. After extensive washes with D_{0.2M KCl} buffer and then GE buffer (120 mM NaCl, 100 mM Tris-HCl [pH 8.0], 10% glycerol, 0.2 mM EDTA, 0.1% NP-40, 2 mM DTT), the bound materials were eluted with 15 mM glutathione. Eluted materials were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting using antibodies specific for Tat-SF1, Pol IIa, SPT5, and RAP30.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Requirement of ATP for formation of a stable TAR loop-dependent P-TEFb-Tat-TAR complex. Previously, we noticed that on a per CycT1 molecule basis, the stability of a CycT1-Tat-TAR complex as determined by EMSA was significantly higher than that of the P-TEFb-Tat-TAR complex (2, 42), implicating a negative effect of Cdk9 on formation of such a complex. To investigate the role of Cdk9 in this process, we affinity purified the Cdk9-HA-CycT11 heterodimer from 293T cells transiently expressing the HA-tagged CycT11 protein by anti-HA antibody immunoprecipitation (-HA IP) followed by HA peptide elution (2, 42). The purified protein fraction was then tested for its ability to bind to Tat and TAR in a gel mobility shift assay (Fig. 1A). The reason for using the C-terminally truncated CycT11 protein (amino acids 1 to 333) instead of full-length CycT1 (CycT1FL) (amino acids 1 to 726) in this experiment will become clear shortly. CycT11 contains the conserved cyclin box (amino acids 1 to 272), the presence of which is sufficient for binding to Cdk9 (2, 10, 34). It also contains the TRM region important for Tat and TAR binding (10).

View larger version (35K): [in this window] [in a new window]   FIG. 1.   Requirement of ATP hydrolysis for formation of a stable P-TEFb-Tat-TAR ribonucleoprotein complex. (A) The Cdk9-HA-CycT11 heterodimer was affinity purified from 293T cells transiently transfected with an HA-tagged CycT11 cDNA by -HA IP followed by HA peptide elution. The purified fraction containing the heterodimer and free HA-CycT11 was incubated with 32P-labeled wild-type TAR RNA (lanes labeled w) or the loop mutant TAR+31/+34 (lanes labeled m) in the absence () or presence (+) of Tat and/or ATP as indicated. The reactions were analyzed by EMSA. Lanes 9 and 12 also contained anti-Cdk9 antibodies (-Cdk9), which caused the retardation of the Cdk9-HA-CycT11-Tat-TAR complex and also partially destabilized the complex, probably because of the polyclonal nature of the antibodies. (B) Silver-stained SDS-polyacrylamide gel showing the presence of Cdk9-HA-CycT11 and free HA-CycT11 in the -HA IP fraction (lane 1). A nonspecific protein (*) was revealed in a control fraction prepared in parallel from untransfected 293T cells (lane 2). (C) Free HA-CycT11 present in the -HA IP fraction formed an ATP-independent complex with Tat and TAR in gel shift reactions. Recombinant CycT11 (rT11) was used as a reference (lanes 5 to 7) for determining the identity of the ATP-independent complex formed with the -HA IP fraction (lanes 8 to 10). (D) Requirement of ATP hydrolysis for P-TEFb-Tat-TAR assembly. All reaction mixtures contained the -HA IP fraction, Tat, and 32P-labeled wild-type TAR RNA. [-S]ATP was used in place of ATP in lane 3.

Most of cellular P-TEFb requires ATP hydrolysis for stable interaction with Tat and TAR. The stable association of Cdk9-HA-CycT11 with Tat and TAR requires ATP hydrolysis and not just simple ATP binding, because replacement of ATP in the binding reaction mixture with a nonhydrolyzable form of ATP, [-S]ATP, largely inhibited complex formation (Fig. 1D, compare lanes 2 and 3). Because P-TEFb is a Cdk-cyclin kinase complex, it is most likely a protein phosphorylation event that results in the assembly of the P-TEFb-Tat-TAR complex. To rule out the possibility that the requirement of ATP was due to the loss of phosphates on P-TEFb during cell lysis and the subsequent purification procedure, multiple phosphatase inhibitors were used in the experiment. P-TEFb thus purified still required ATP for high-affinity binding to Tat-TAR (data not shown), suggesting that most of the cellular P-TEFb may lack the key phosphorylation that is important for complex formation.

C-terminal truncation of CycT1 in addition to P-TEFb phosphorylation stabilizes the P-TEFb-Tat-TAR complex. The gel mobility shift assay described above used P-TEFb complexes containing the C-terminally truncated CycT1 molecule CycT11. To examine the effect of the C-terminal region of CycT1 on ATP-dependent P-TEFb-Tat-TAR complex formation, we generated P-TEFb complexes containing HA-tagged CycT1FL (HA-CycT1FL) or a series of mutant CycT1s with C-terminal deletions (HA-CycT11 to -4) (Fig. 2A). These proteins were transiently expressed in 293T cells and affinity purified by -HA IP. Western blotting indicated that deletions of the C-terminal region of CycT1 up to position 333 did not affect its ability to interact with Cdk9 (Fig. 2B), consistent with the notion that the cyclin box of CycT1 is sufficient for Cdk9 binding.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   The P-TEFb-Tat-TAR complex is stabilized by C-terminal truncation of CycT1 in addition to P-TEFb phosphorylation. (A) Diagram showing the domain structures of CycT1FL and the C-terminal truncation mutant constructs CycT11 to -4. The C-terminal boundary of TRM is located between amino acids 250 and 262 at the C-terminal edge of the conserved cyclin box domain (amino acids 1 to 272). The N-terminal boundary of TRM is unclear. (B) The C-terminal truncation of CycT1 did not affect its association with Cdk9. HA-CycT1FL or HA-CycT11 to -4 and their associated Cdk9 proteins were affinity purified from 293T cells transiently expressing the various HA-tagged CycT1 proteins by anti-HA IP. The levels of Cdk9 and HA-CycT1 in these preparations were analyzed by Western blotting with anti-Cdk9 and anti-HA antibodies. A faint band right below CycT1FL is a nonspecific cross-reactive protein. (C) Equal amounts of Cdk9-HA-CycT1FL and Cdk9-HA-CycT11 to -4 were tested for their abilities to form ternary complexes with Tat and 32P-labeled wild-type (lanes w) or the loop mutant (lanes m) TAR RNAs in a gel mobility shift assay.

Autophosphorylation of Cdk9 is required for P-TEFb-Tat-TAR complex formation. Given the fact that high-affinity P-TEFb-Tat-TAR assembly requires ATP hydrolysis, we wanted to confirm that it is the Cdk9 kinase of P-TEFb and not some other contaminating kinases or ATPases in the affinity-purified fraction which hydrolyzes ATP and stimulates P-TEFb-Tat-TAR assembly. We prepared P-TEFb heterodimers that contained either wild-type or a kinase-defective Cdk9 with a point mutation changing Asp167 to Asn (Fig. 3A). (Since the HA tag is appended to Cdk9, no free CycT1 is present in these -HA IP fractions.) Because of the inhibitory activity of the CycT1 C-terminal region, a large amount of wild-type P-TEFb containing CycT1FL was used to detect the formation of the ternary complex (Fig. 3B, lanes 1 and 2). Compared to wild-type P-TEFb, the kinase-defective Cdk9(D167N)-CycT1 heterodimer failed to form a stable complex with Tat and TAR both in the presence and absence of ATP (Fig. 3B, lanes 3 and 4). This binding defect, in addition to the failure of this kinase-defective P-TEFb to phosphorylate the Pol II CTD, probably contributed to its severe deficiency in mediating Tat activation (42).

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Autophosphorylation of Cdk9 is required for efficient P-TEFb-Tat-TAR complex formation. (A) P-TEFb complexes containing either HA-tagged wild-type Cdk9 (WT) or a kinase-defective Cdk9 (D167N) were affinity purified by -HA IP and normalized for their CycT1 levels by Western blotting with anti-CycT1 antibodies. (B) The kinase activity of Cdk9 is required for P-TEFb-Tat-TAR formation. The two P-TEFb complexes prepared for panel A were compared in a gel mobility shift assay for forming P-TEFb-Tat-TAR complexes in the presence (+) or absence () of ATP. Since the HA tag was attached to Cdk9, no free CycT1 was present in these -HA IP fractions. (C) Autophosphorylation of P-TEFb is required for P-TEFb-Tat-TAR complex formation. Purified Cdk9-HA-CycT1 immobilized on anti-HA antibody beads was incubated with ATP in a kinase reaction mixture. Upon removal of ATP, the phosphorylated complex was eluted from the beads and incubated with Tat and 32P-labeled wild-type TAR RNA in a reaction mixture without ATP. An equal amount of unphosphorylated P-TEFb was used as a control. (D) Phosphorylation of CycT1 by Cdk9 is not required for P-TEFb-Tat-TAR formation. Cdk9-HA-CycT1FL and Cdk9-HA-CycT11 to -4 complexes were prepared, and their levels were normalized as described for Fig. 2B. Equal amounts of these complexes were analyzed in in vitro kinase reactions. Note that HA-CycT1 was not phosphorylated by Cdk9. However, Cdk9-HA-CycT11 formed an ATP-dependent complex with Tat-TAR.

Autophosphorylation of Cdk9 does not affect P-TEFb-Tat binding but induces conformational changes in P-TEFb for better TAR recognition. Autophosphorylation of Cdk9 may facilitate P-TEFb-Tat-TAR formation by increasing the binding of P-TEFb to either Tat protein or TAR RNA. To distinguish between these two possibilities, we compared unphosphorylated, in vitro autophosphorylated, and alkaline phosphatase-treated Cdk9-HA-CycT1 heterodimers (see Materials and Methods) for their abilities to interact with wild-type GST-Tat(1-48) proteins. GST-Tat(1-48, C22G), which contains a point mutation in the Tat activation domain, was used as a control to demonstrate the specificity of the interaction. Western blotting with anti-CycT1 antibodies indicated that autophosphorylation of Cdk9-HA only minimally increased the binding of Cdk9-HA-CycT1 to wild-type Tat (Fig. 4A, compare lanes 2 and 4) but that phosphatase treatment slightly decreased the binding (compare lanes 2, 4, and 5). In contrast, adding the same amount of phosphatase to a binding reaction mixture prior to the gel shift electrophoresis inhibited the formation of the Cdk9-T11-Tat-TAR complex but not the complex containing free T11 (T11-Tat-TAR) (Fig. 4B). These results indicated that autophosphorylation of Cdk9 did not significantly affect the P-TEFb-Tat interaction. Rather, it probably enhanced the binding of the P-TEFb-Tat complex to TAR RNA to facilitate P-TEFb-Tat-TAR complex formation.

View larger version (28K): [in this window] [in a new window]   FIG. 4.   P-TEFb phosphorylation does not affect P-TEFb-Tat binding but induces conformational changes in P-TEFb for better TAR recognition. (A) Equal amounts of in vitro-phosphorylated P-TEFb (Cdk9-HA-CycT1; lane 4), unphosphorylated P-TEFb (lanes 1 to 3), and alkaline phosphatase (AP; 10 U)-treated P-TEFb (lane 5) complexes were incubated with wild-type (WT) GST-Tat(1-48) proteins bound to glutathione-Sepharose beads. After washes, the amount of P-TEFb bound to Tat was examined by Western blotting with anti-CycT1 antibodies (-CycT1). The GST-Tat(1-48, C22G) mutant protein, which contains a point mutation in the Tat activation domain, was used as a control (lane 3). (B) Alkaline phosphatase treatment inhibited the formation of the Cdk9-T11-Tat-TAR complex but not the complex containing free T11. The binding and gel shift conditions are the same as described for lane 11 of Fig. 1A, except that 10 U of AP was included in the binding reaction mixture in lane 3 prior to electrophoresis. (C) P-TEFb phosphorylation resulted in changes in trypsin sensitivity in the CycT1 C-terminal region. Purified Cdk9-Flag-HA-CycT1 complexes were first subjected to in vitro kinase reactions in the presence or absence of ATP, and then trypsin was added to the reaction mixtures. The cleaved products were detected by Western blotting with an anti-HA monoclonal antibody (-HA) which recognizes the HA tag at the N terminus of CycT1. The position of a cross-reactive protein doublet is indicated by an *.

The N- and C-terminal regions of CycT1 interact in an intramolecular manner to inhibit P-TEFb-Tat-TAR complex formation. The data presented above suggested that P-TEFb needs to overcome two separate barriers in order to bind TAR RNA in conjunction with Tat. The unphosphorylated state of Cdk9 constitutes the first barrier, which can be overcome through Cdk9 autophosphorylation. The induced conformational change in P-TEFb probably facilitates P-TEFb-TAR binding. The second barrier resides in the C-terminal region of CycT1, which interfered with the binding of the P-TEFb-Tat complex to TAR RNA (Fig. 2C). To explore the mechanism for this inhibition, we asked whether the C-terminal region of CycT1 can fold back to interact with the CycT1 N-terminal region, thereby masking the TAR-binding surface. First, a GST pull-down assay was performed to test the interaction of CycT11, which contains the N-terminal cyclin box and its immediate flanking region (Fig. 2A), with GST-CycT1-C, which contains a CycT1 C-terminal fragment (amino acids 402 to 701). As shown in Fig. 5A, CycT11 was able to interact with GST-CycT1-C but not with GST alone, indicating a possible interaction between the N- and C-terminal regions of CycT1.

View larger version (35K): [in this window] [in a new window]   FIG. 5.   The CycT1 C-terminal region interacts with the N-terminal region to inhibit P-TEFb-Tat-TAR complex formation. (A) HA-tagged recombinant CycT11 proteins were incubated with equal amounts of GST or GST-CycT1-C (CycT1 C-terminal fragment, amino acids 402 to 701) bound to glutathione-Sepharose beads. After washes, bound CycT11 was detected by Western blotting with the anti-HA monoclonal antibody 12CA5. Twenty percent of the CycT11 used in the binding reaction mixture was shown as input. (B) Binding of CycT1-C to recombinant CycT11 (rT11) and the Cdk9-CycT11 complex (Cdk9/T11) in trans did not impede their interactions with Tat and TAR. Recombinant CycT11 (~200 ng) or Cdk9-CycT11 (~300 ng) was incubated in the presence or absence of CycT1-C (~400 ng) for 10 min, followed by the addition of Tat (100 ng) and 32P-labeled TAR RNA. The reaction products were analyzed by gel mobility shift assay. (C) Binding of CycT1-C to Cdk9-CycT1FL disrupted the intramolecular interaction in CycT1 and stabilized the P-TEFb-Tat-TAR complex. Reaction mixtures containing CycT1-C (~400 ng), Cdk9-CycT1FL (~300 ng), Tat (100 ng), TAR, and ATP were analyzed as described for panel B. Lane 1' is an eight-times-longer exposure of lane 1 and shows the position of Cdk9-CycT1FL-Tat-TAR in the gel.

The C-terminal region of CycT1 interacts with Tat-SF1 and Pol II in HeLa nuclear extracts. The above results indicated that the key to relieving the second autoinhibition of P-TEFb-Tat-TAR complex formation is to disrupt the intramolecular interaction in CycT1. Other than by artificially disrupting this interaction by causing competition with recombinant CycT1-C or removing the C-terminal region, we investigated the possibility that the CycT1-associated proteins may relieve the autoinhibition by interacting with and stabilizing the C-terminal inhibitory region of CycT1. First, we examined whether the C-terminal region of CycT1 may interact with components of a recently characterized Pol II elongation complex (31). Equal amounts of immobilized GST, GST-CycT1FL, GST-CycT11, and GST-CycT1-C proteins were incubated with HeLa nuclear extracts, and their abilities to bind to several known elongation factors and Pol II were analyzed by immunoblotting. Small fractions of RNA Pol IIa and Tat-SF1 in the nuclear extract were found to interact with CycT1FL and its C-terminal region (Fig. 6A, lanes 5 and 9) but not with the N-terminal region (CycT11; lane 7) or GST alone (lane 3). Unlike Pol IIa and Tat-SF1, little SPT5 (a subunit of DSIF) or RAP30 (a subunit of TFIIF) in the nuclear extract was found to associate with CycT1 under this condition (Fig. 6A), although a weak interaction between P-TEFb and DSIF was detected in a purified system (data not shown).

View larger version (31K): [in this window] [in a new window]   FIG. 6.   Tat-SF1 interacts with the C-terminal region of CycT1 and represses the autoinhibitory activity of this region. (A) The C-terminal region of CycT1 interacted with Tat-SF1 and RNA Pol IIa in HeLa nuclear extracts (NE). HeLa nuclear extracts were incubated with equal amounts of GST, GST-CycT1FL, GST-CycT11, and GST-CycT1-C proteins bound to glutathione-Sepharose beads. After washes, the bound proteins were analyzed by Western blotting with antibodies specific for Tat-SF1, Pol IIa, RAP30, and SPT5. Five percent of the nuclear extracts used in the binding reaction mixture were shown as input. The nonspecific bands shown in lanes 4 and 5 in anti-Tat-SF1 and anti-RAP30 antibody panels are cross-reactive bacterial proteins. (B) Silver-stained SDS-polyacrylamide gel showing Flag-tagged Tat-SF1 affinity purified from transfected 293T cells. (C) Binding of Tat-SF1 to the C-terminal region of CycT1 enhanced the P-TEFb-TAR interaction. Equal amounts of Cdk9-HA-CycT1FL (Cdk9/T1FL; lanes 1', 1, and 2) and Cdk9-HA-CycT11 (Cdk9/T11; lanes 4 and 5) were incubated with Tat, 32P-labeled TAR, and ATP in the presence or absence of purified Tat-SF1. Reaction products were analyzed by gel mobility shift assay. Lane 1' is an eight-times-longer exposure of lane 1 and shows the position of Cdk9-T1FL-Tat-TAR.

Binding of Tat-SF1 to the CycT1 C-terminal region stabilizes the P-TEFb-Tat-TAR complex. Since the CycT1 C-terminal region interacts with Pol IIa and Tat-SF1, we investigated whether the association with Tat-SF1 would stabilize the binding of the P-TEFb-Tat complex to TAR. We affinity purified Flag-tagged Tat-SF1 protein from transiently transfected 293T cells, and the purified protein was analyzed by SDS-PAGE and silver staining (Fig. 6B). When tested in a gel mobility shift assay, Tat-SF1 did not recognize TAR RNA by itself (data not shown) or in the presence of Tat (Fig. 6C, lane 3), in spite of having two putative RNA recognition motifs (44). Because of the strong autoinhibition caused by the intramolecular interaction in CycT1, wild-type P-TEFb (Cdk9-CycT1FL) produced a very weak, albeit ATP-dependent (data not shown), P-TEFb-Tat-TAR complex visible only after a prolonged exposure of the autoradiogram (Fig. 6C, lane 1' is an eight-times-longer exposure of lane 1). Importantly, preincubation of Flag-tagged Tat-SF1 with P-TEFb supershifted the P-TEFb-Tat-TAR complex and significantly enhanced the assembly of a multiprotein complex that most likely contained Tat-SF1, Cdk9-CycT1FL, Tat, and TAR (Fig. 6C, lane 2). The involvement of Tat-SF1 in forming this supershifted complex was also revealed by the observation that inclusion of anti-Flag or anti-Tat-SF1 antibody in the binding reaction mixture inhibited the complex formation (data not shown). This stimulatory effect of Tat-SF1 depended on the presence of ATP in the reaction mixture, suggesting that Tat-SF1 cannot overcome the first autoinhibitory step in P-TEFb, which is relieved only through Cdk9 autophosphorylation. In the same reaction as that shown in lane 2 of Fig. 6C, Tat-SF1 also interacted with and increased the binding to TAR RNA by free CycT1 (CycT1FL) present in the -HA-CycT1 IP fraction. As predicted, the cooperative TAR binding promoted by Tat-SF1 requires the C-terminal region of CycT1, as Cdk9-CycT11 interacted strongly with Tat and TAR independently of Tat-SF1 (Fig. 6C, lanes 4 and 5). These results demonstrated an important role of Tat-SF1 in overcoming the intramolecular inhibition in CycT1 by binding to the CycT1 C-terminal region. In contrast to Tat-SF1, inclusion of purified calf thymus RNA Pol II in a binding reaction mixture did not lead to an enhanced binding of P-TEFb to Tat-TAR (data not shown).

The C-terminal region of CycT1 is required for efficient HIV-1 transcriptional elongation. The binding of the CycT1 C-terminal region to Pol II and Tat-SF1 and the requirement of this region for the assembly of a multiprotein complex at TAR RNA suggested that this region may be important for basal and Tat-activated HIV-1 transcription. We analyzed the abilities of Cdk9-CycT1FL and Cdk9-CycT11 to mediate Tat activation in HeLa nuclear extracts immunodepleted of the endogenous P-TEFb. Immunoblotting analysis indicated only a trace amount of P-TEFb left in the depleted extract (data not shown), which was probably responsible for the very low level of HIV-1 transcription observed in this extract (Fig. 7B, lanes 1 and 2). Supplementing the depleted extract with two concentrations (1× and 3×) of wild-type P-TEFb (Cdk9-CycT1FL) allowed Tat to specifically increase the level of transcripts elongating beyond 1,000 nucleotides from an HIV-1 promoter containing the wild-type TAR element (pHIV+TAR-G400 [43]) but not from an internal control promoter with a mutant TAR (pHIVTAR-G100) (Fig. 7B, lanes 5, 6, 9, and 10). The reduced fold of Tat activation as a result of more P-TEFb (3×) being added to the reaction mixture (2.8-fold in lanes 5 and 6 versus 6.1-fold in lanes 9 and 10 after normalization to internal controls) was probably due to an efficient elongation mediated by high levels of P-TEFb, which partially bypassed the requirement for Tat.

View larger version (28K): [in this window] [in a new window]   FIG. 7.   The C-terminal region of CycT1 is required for efficient HIV-1 transcriptional elongation. (A) P-TEFb complexes containing either HA-tagged CycT11 or CycT1FL were normalized for their Cdk9 and CycT1 levels by Western blotting with anti-Cdk9 and anti-HA antibodies. (B) Equal amounts of the two P-TEFb complexes were added to transcription reaction mixtures containing P-TEFb-depleted HeLa nuclear extracts (NE) as well as the DNA templates pHIV+TAR-G400 and pHIVTAR-G100 in the presence (+) or absence () of Tat protein. The amount of P-TEFb analyzed in lanes 3 to 6 was three times (3×) higher than that in lanes 7 to 10. +TAR-G400 and TAR-G100 are RNase T1-resistant RNA fragments transcribed from the two G-less DNA cassettes (400 and 100 bp) inserted, respectively, into pHIV+TAR-G400 and pHIVTAR-G100 at a position 1 kb downstream of the HIV-1 promoter region.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The formation of a TAR loop-dependent P-TEFb-Tat-TAR complex is essential for Tat-specific and TAR-dependent stimulation of HIV-1 transcription. The data presented here provide a mechanistic view of how a high-affinity P-TEFb-Tat-TAR complex is assembled. Interestingly, assembly of this complex is a regulated process involving the relief of two autoinhibitory mechanisms in P-TEFb. Most of the P-TEFb heterodimers isolated from human cells appear to be intrinsically inactive in forming stable P-TEFb-Tat-TAR complexes. P-TEFb undergoes conformational changes in at least two controlled steps and requires the help from another elongation factor(s) in order to form a high-affinity complex on TAR RNA.

Previous alanine-scanning mutagenesis of the CycT1 TRM identified residues that are critical for the interaction with Tat and others that are required specifically for binding of the Tat-CycT1 complex to TAR RNA (10). Thus, the CycT1 TRM makes independent contacts with Tat and TAR. As depicted in the diagram shown in Fig. 8, the data presented here are consistent with the model that the TRM subdomain required for P-TEFb-TAR interaction is blocked, either directly or indirectly, by both unphosphorylated Cdk9 and the C-terminal region of CycT1, which folds back to interact with the N-terminal region. In contrast, the part of TRM specific for Tat binding appears to be accessible irrespective of the phosphorylation state of Cdk9 (Fig. 4A) and the intramolecular interaction in CycT1 (2). Relief of the first autoinhibition requires Cdk9 autophosphorylation, which alters the conformation in CycT1 and/or Cdk9 and unmasks the critical TRM subdomain for possible interaction with TAR RNA. It is interesting that a mutant P-TEFb with a kinase-defective Cdk9 subunit was incapable of forming a stable complex with Tat and TAR (Fig. 3C), suggesting that the failure of this mutant elongation factor to mediate Tat activation is not simply due to its defective CTD kinase activity (42) but rather to its inability to bind TAR RNA and to be recruited to the HIV promoter at an earlier stage.

View larger version (37K): [in this window] [in a new window]   FIG. 8.   Model for the assembly of the P-TEFb-Tat-TAR complex through relief of two built-in autoinhibitory mechanisms in P-TEFb. See the text for details. It is important to point out that in this diagram the order of events depicted between the first and last steps is purely hypothetical.

Autophosphorylation of Cdk9 only partially exposes the subdomain of TRM for TAR recognition (Fig. 8). The efficient interaction between P-TEFb and TAR RNA also requires suppression of the autoinhibitory activity of the CycT1 C-terminal region. The intramolecular interaction of the CycT1 C-terminal region with the N-terminal half probably creates steric hindrance that blocks the access to TRM by TAR RNA (Fig. 8). To disrupt this intramolecular interaction and relieve the autoinhibition, transcription elongation factor Tat-SF1 interacts with the C-terminal region and markedly enhances the binding of the Tat-SF1-P-TEFb-Tat complex to TAR RNA (Fig. 6C). However, the effect of Tat-SF1 did not approach the level attained by the deletion of the CycT1 C-terminal domain. One possibility is that another cellular factor(s) (Fig. 8) works together with Tat-SF1 to fully stabilize the complex. Indeed, in addition to Tat-SF1, the CycT1 C-terminal region was also shown to interact with Pol IIa and perhaps other components of the Pol II elongation apparatus. Upon its release from the intramolecular interaction, this region may function as a bridge linking the P-TEFb-Tat-TAR ternary complex with the basal elongation apparatus. In support of this model, when a physiological amount of P-TEFb was analyzed in transcription reactions, the C-terminal region of CycT1 was found to be required for both basal and Tat-activated HIV-1 transcription (Fig. 7), although other minor interactions between the P-TEFb-Tat-TAR complex and the basal elongation apparatus may also contribute to the assembly of a highly processive, multicomponent Pol II elongation machinery (Fig. 8). Tat-SF1 was biochemically identified as a Tat-specific cellular cofactor (44). Our results provide a plausible explanation for the Tat-specific elongation activity of this factor. Since Tat-SF1 binds to Tat and contains two RNA recognition motifs (44), future experiments are necessary to determine whether these properties of Tat-SF1 contribute to cooperative TAR recognition by Tat-SF1, P-TEFb, and Tat.

Autoinhibition, involving intramolecular interactions that negatively regulate the function of otherwise autonomous modules, is observed in many transcription factors (for a review, see reference 14). For instance, SNAPc, a core promoter-binding factor required for transcription of RNA Pol II and Pol III snRNA promoters, was shown to have an autoinhibitory C-terminal region that represses the binding of SNAPc to DNA (28). Interestingly, like the derepression effect mediated by the interaction of Tat-SF1 with P-TEFb, repression of SNAPc DNA binding can be relieved by the interaction of SNAPc with the Oct-1 POU domain that promotes cooperative binding. Other examples of transcription factors with built-in negative control of DNA binding include Escherichia coli ⁷⁰ (7) and the largest subunit of TFIID from both Drosophila and yeast (21, 22). The similarities between P-TEFb and these transcription factors suggest the existence of a common autoinhibition and derepression mechanism to ensure specific and stable interactions of these transcription factors with either RNA recognition sequences or promoter DNA.

We noticed that most of the Cdk9-CycT1 isolated from stably transfected human 293 cells required an in vitro autophosphorylation step to form stable P-TEFb-Tat-TAR complexes, implying the lack of a key Cdk9 phosphorylation that is important for complex formation. It is not clear what effect this may have on the general and Tat-specific elongation activity of P-TEFb in vivo. Nevertheless, our observations raise an interesting possibility that the regulation of Cdk9 phosphorylation may provide a novel control step for Tat activity and HIV-1 transcription in infected cells. Future experiments to map and mutate the phosphorylated residue(s) in Cdk9 and to test the activity of the mutant construct(s) in HIV-1 transcription may allow us to test this possibility.

Although our purified in vitro system permits autophosphorylation of only Cdk9, Cdk9 can theoretically be phosphorylated and activated by other kinases in vivo. During the activation of peripheral blood lymphocytes and differentiation of promonocytic cell lines, both of which are relevant for HIV infection, a dramatic increase in HIV-1 gene expression (12) as well as P-TEFb kinase activity has been observed (12, 15, 40). The induction of P-TEFb activity has been attributed to an increase in Cdk9 and CycT1 levels in activated peripheral blood lymphocytes and an increase of CycT1 in differentiated monocytes (12, 40). However, the pharmacological reagents (phorbol esters, ionomycin, and phytohemagglutinin) used to treat the cells are known to activate a spectrum of protein kinases, one of which may in turn phosphorylate Cdk9, stabilize the P-TEFb-Tat-TAR complex, and increase the Tat-specific activity of P-TEFb. Besides cellular kinases that may modulate P-TEFb activity, specific phosphatases may also be regulated to maintain appropriate levels of phosphorylated P-TEFb in vivo in response to environmental stimuli and perhaps also during the cell cycle. Since induction of P-TEFb activity in activated T cells and differentiated macrophages may contribute directly to high levels of HIV-1 transcription and the escape from latency of transcriptionally silent proviruses (15), it is important to investigate whether induction of P-TEFb phosphorylation may contribute directly to these processes.