Tat and TAR RNA enhance phosphorylation of affinity-purified core RNAPII by recombinant P-TEFb (CycT1-CDK9) in vitro. (A) In vitro kinase reactions were carried out with 200 ng of affinity-purified core RNAPII, 30 ng of baculovirus-expressed FLAG-tagged CDK9 and, where indicated above each lane, 15 ng of human CycT1 (+; aa 1 to 303), 15 ng of CycT1 mutant C261A (mt; aa 1 to 303), 30 ng of wild-type HIV-1 Tat (+; aa 1 to 86), 30 ng of HIV-1 Tat C22G (mt; aa 1 to 86), 30 ng of HIV-1 Tat activation domain (48; aa 1 to 48), 30 ng of C22G mutant HIV-1 Tat activation domain (48*; aa 1 to 48), and either 24 ng of wild-type HIV-1 TAR RNA (wt; nt +1 to +80) or 24 ng of loop mutant (lm; nt +1 to +80) HIV-1 TAR RNA. The migration positions of RNAPII complexes containing either the hypophosphorylated (IIa) or the hyperphosphorylated (IIo) CTD are indicated with arrows. (B) RNAPII CTD phosphorylation was analyzed by Western blot using monoclonal antisera specific to the CTD heptapeptide repeat phosphorylated at either position Ser-5 or Ser-2 (Babco). Reactions contained 30 ng of wild-type HIV-1 Tat and 24 ng of HIV-1 TAR RNA.

The Tat ARM and activation domain are required to stimulate GST-CTD phosphorylation by CycT1-CDK9 in vitro. (A) Phosphorylation of GST-CTD by wild-type and mutant Tat proteins was analyzed with in vitro kinase experiments. Kinase reactions contained 400 fmol of CycT1 (aa 1 to 303), 750 fmol of CDK9, 140 fmol of GST-CTD, 400 fmol of Spt5 and, where indicated, 3 pmol of HIV-1 Tat (aa 1 to 48), HIV-1 Tat (aa 1 to 72), HIV-1 Tat (aa 1 to 86); HIV-2 Tat (aa 1 to 77), and HIV-2 Tat (aa 1 to 130). The relative migration positions of hypophosphorylated (CTDa) or hyperphosphorylated (CTDo) GST-CTD are indicated with arrows. The bottom panel shows the level of CDK9 autophosphorylation in each reaction. (B) Analysis of the ability of TAR RNA or the isolated Tat transactivation domain to interfere with Tat-enhanced GST-CTD phosphorylation by P-TEFb. Conditions are as in panel A except for lanes 7 to 9, which contain 300 ng of rI-rC and 50 ng of either wild-type of loop mutant HIV-1 TAR RNA (nt +1 to +80) as competitor, as indicated. The presence (+) or absence (−) of HIV-1 Tat-1 (aa 1 to 86) or HIV-1 Tat (aa 1 to 48) is also indicated above each lane. In the reaction shown in lane 3, the activation domain of Tat (aa 1 to 48; circled) was incubated with CycT1-CDK9 prior to addition of wild-type Tat (aa 1 to 86), whereas in the reaction shown in lane 4 the wild-type Tat protein (aa 1 to 86; circled) was incubated with CycT1-CDK9 prior to the addition of the mutant Tat (aa 1 to 48). (C) Analysis of the ability of HIV-1 Tat to stimulate CycT1-CDK9 or CycH-CDK7 kinase activity. The individual GST-Cyc–CDK complexes were isolated from SF9 cells coinfected with recombinant baculovirus vectors and purified by chromatography on glutathione-Sepharose beads. (Left panel) In vitro kinase reaction conditions are as in A and contained 20 ng of HIV-1 Tat (86 aa; lanes 2, 4, 6, and 8), 50 ng of Spt5 (lanes 5 to 8), and 60 ng of a preformed complex containing either CycT1(1–303)-CDK9 or CycH-Cdk7, as indicated above each lane. Lanes 9 and 10 show standard in vitro kinase reactions (18) containing GST-CTD and the indicated protein kinase complexes.

Incubation with ATP enhances binding of the Tat-CycT1-CDK9 complex to TAR RNA in vitro. (A) RNA gel-shift analysis of complexes containing recombinant Tat, CycT1, and CDK9 on HIV-1 TAR RNA (nt +1 to +80) in the presence or absence of ATP. The reactions in lanes 3 and 4 contained 12 ng of GST-CycT1(1–303), 30 ng of FLAG-CDK9, and 30 ng of HIV-1 Tat (aa 1 to 86) in the absence (lane 3) or presence (lane 4) of ATP. The reactions in lanes 5 and 6 contained 30 ng of baculovirus coexpressed CycT1(1–728)-CDK9 complex and 30 ng of HIV-1 Tat (aa 1 to 86), incubated in the absence (lane 5) or presence (lane 6) of ATP. Arrows indicate the positions of the different Tat–P-TEFb–TAR complexes. (B) Analysis of autophosphorylation of recombinant P-TEFb complexes containing either full-length CycT1 (lane 1) or the CycT1 cyclin domain (aa 1 to 303; lane 2). Each P-TEFb complex was incubated with [γ-³²P]ATP, separated by SDS-PAGE, and visualized by autoradiography. (C) Analysis of the effect of ATP on the binding of HIV-1 Tat to phosphorylated or unphosphorylated CycT1(1–303)-CDK9 complexes. GST-CycT1 (aa 1 to 303) was coupled to beads and incubated with CDK9 and HIV-1 Tat (aa 1 to 86) in the absence (lane 1) or presence (lane 2) of ATP. Complexes were washed stringently as described in Materials and Methods, and the proteins were visualized by Western blot. The CycT1(1–303), FLAG-CDK9, and HIV-1 Tat (aa 1 to 86) proteins were visualized with monoclonal antisera to GST, FLAG, and the Tat ARM, respectively. The amount of CDK9 and Tat which bound to GST-CycT1 was estimated by comparing to 50% of the input GST-CycT1 (lane 3), 50% of the input FLAG-CDK9 (lane 4), and 30% of the input HIV-1 Tat (lane 5) in each reaction.

P-TEFb autophosphorylation is essential to support binding of different Tat proteins to TAR RNA and requires functional CDK9 kinase activity. (A) RNA gel shift experiments analyzing the ATP dependence of different Tat–P-TEFb complexes. Where indicated above each lane, binding reactions contained 12 ng of CycT1(1–303); 30 ng of wild-type CDK9 or the catalytic mutant D167N CDK9; 30 ng of HIV-1 Tat (aa 1 to 86; HXB2 isolate) (lanes 2 to 5 and lanes 14 to 17), HIV-1 Tat (aa 1 to 101; SF2 isolate) (lanes 6 to 9), or HIV-2 Tat (aa 1 to 130; Rod isolate) (lanes 10 to 13); and 15 ng HIV-1 TAR RNA (nt +1 to +80). Binding of proteins to TAR RNA was performed in the presence (+) or absence (−) of ATP as designated below each lane. (B) Enhanced binding of the Tat–P-TEFb complex to TAR RNA requires ATP hydrolysis. Conditions are as in panel A. AMP-PNP is the nonhydrolyzable ATP analog, adenylylimido-diphosphate. The Tat–P-TEFb–TAR complexes are indicated with arrows.

Analysis of CDK9 autophosphorylation sites by two-dimensional phosphoamino acid and phosphotryptic peptide mapping experiments. (A) Two-dimensional phosphoamino acid analysis of recombinant autophosphorylated CDK9. (B) Autophosphorylated CDK9 was digested with trypsin, and peptide fragments were separated by a two-dimensional TLC. The two major phosphorylated peptides are labeled peptide 1 (P1) and peptide 2 (P2). The bottom left corner of the autoradiogram denotes the origin prior to electrophoresis. (C) Phosphoamino acid content of P1 and P2. Peptides P1 and P2 were isolated by two-dimensional chromatography and analyzed for phosphoamino acid context. (D) These data, together with radioactive sequencing of P2 which revealed phosphorylation at residue 3, indicate that the major site of phosphorylation is located in a tryptic peptide (aa 345 to 358; boxed) located near the C terminus of human CDK9.

Truncation of the CDK9 C terminus of CDK9 destroys autophosphorylation and ATP-enhanced binding of Tat–P-TEFb to TAR RNA in vitro. (A) A five-alanine substitution (CDK9-5A) or a truncation (CDK9Δ323) in the C terminus of CDK9 does not alter the ability to phosphorylate GST-CTD in in vitro kinase experiments. Kinase reactions contained 750 fmol each of CycT1 (aa 1 to 303) and CDK9, 140 fmol of GST-CTD, and 3 pmol of HIV-1 Tat (aa 1 to 86). The CDK9 5A protein contains Ser-Thr to alanine substitutions at residues 347, 350, 353, 354, and 357. The positions of hypophosphorylated (CTDa) or hyperphosphorylated (CTDo) GST-CTD are indicated with arrows. The bottom panel shows the level of CDK9 autophosphorylation in each reaction. (B) CDK9Δ323 is unable to modulate the affinity of the Tat–P-TEFb complex for TAR RNA. Binding of recombinant proteins to HIV-1 TAR RNA was analyzed by gel shift experiments as described for Fig. 3A. Complex formation was analyzed in the absence or presence of either HIV-1 Tat (two left panels) or HIV-2 Tat (right panel), as indicated above each lane. The Tat–P-TEFb complex is indicated with an arrow.

PK-A phosphorylation or substitution of S/T residues with negatively-charged amino acids restores TAR RNA-binding activity of complexes containing the kinase deficient CDK9 mutant (D167N). (A) For lanes 4 to 6, the CycT1(1–303)-D167N complex was phosphorylated with PKA and analyzed for its ability to bind TAR RNA in gel mobility shift experiments. Binding reactions contained 12 ng of CycT1 (aa 1 to 303), 30 ng of wild-type CDK9 or D167N CDK9, 30 ng of HIV-1 Tat (aa 1 to 86), and 15 ng of HIV-1 TAR RNA (nt +1 to +80) in the presence or absence of ATP. For lanes 7 to 13, a series of 5- or 9-aa substitution mutations were introduced into the C-terminal tail of D167N or CDK9 that replaced Ser or Thr residues with glutamate (E) residues, as indicated schematically at the bottom of the figure. Complexes of CycT1(1–303)-CDK9 were purified following coexpression in baculovirus and analyzed for TAR RNA-binding activity in gel shift experiments as described in the legend to Fig. 4. The mutant R343,R344A replaced two arginine residues in the region preceeding P1 with Ala residues. (B) In vitro transcription experiment assessing the ability of CycT1(1–303)-D167N and CycT1(1–303)–D167N-5E complexes to inhibit Tat transactivation in vitro. HeLa nuclear extracts were incubated in the presence or absence of GST-Tat, as indicated above each lane, under conditions shown previously to support Tat-dependent transactivation in vitro (52). Where indicated, reactions also contained 60 ng of recombinant CycT1(1–303)-CDK9, CycT1(1–303)-D167N, or CycT1(1–303)–D167N-5E complex. Arrows indicate TAR-dependent wild-type (wt) and TAR-independent (ΔTAR) HIV-2 RNA runoff transcripts formed in vitro.

Model depicting binding of the Tat-CycT1-CDK9 complex to HIV-1 TAR RNA. The general orientation of the CycT1-CDK9 complex was approximated based on the X-ray crystal structure of cyclin A-CDK2 (28, 44; PDB ID code 1JST), prepared with the program RasMol. The TRM of CycT1 (green line) extends beyond the second cyclin fold and is implicated in the zinc-dependent interaction with Tat (18). The TRM contains two arginine residues and several hydrophobic amino acids that may recognize the loop of TAR RNA (18). CDK2 does not possess a structure comparable to the C terminus of CDK9, which is indicated with a red line at the end of CDK2. The Tat activation domain (blue line) interacts with CycT1 TRM in part through the proposed zinc bridge and as well as through interactions with cyclin domain residues, and the ARM is required for binding to the bulge of TAR RNA. At left, the Tat–P-TEFb complex containing unphosphorylated CDK9 is shown to bind poorly to TAR RNA, as is observed for complexes containing the native full-length CycT1 protein, and all P-TEFb complexes containing HIV-2 Tat. The right panel depicts high-affinity binding observed upon CDK9 autophosphorylation (see the text for details).