HIV-1 Tat interaction with RNA polymerase II C-terminal domain (CTD) and a dynamic association with CDK2 induce CTD phosphorylation and transcription from HIV-1 promoter.

Human immunodeficiency virus, type 1 (HIV-1), Tat protein activates viral gene expression through promoting transcriptional elongation by RNA polymerase II (RNAPII). In this process Tat enhances phosphorylation of the C-terminal domain (CTD) of RNAPII by activating cell cycle-dependent kinases (CDKs) associated with general transcription factors of the promoter complex, specifically CDK7 and CDK9. We reported a Tat-associated T-cell-derived kinase, which contained CDK2. Here, we provide further evidence that CDK2 is involved in Tat-mediated CTD phosphorylation and in HIV-1 transcription in vitro. Tat-mediated CTD phosphorylation by CDK2 required cysteine 22 in the activation domain of Tat and amino acids 42-72 of Tat. CDK2 phosphorylated Tat itself, apparently by forming dynamic contacts with amino acids 15-24 and 36-49 of Tat. Also, amino acids 24-36 and 45-72 of Tat interacted with CTD. CDK2 associated with RNAPII and was found in elongation complexes assembled on HIV-1 long-terminal repeat template. Recombinant CDK2/cyclin E stimulated Tat-dependent HIV-1 transcription in reconstituted transcription assay. Immunodepletion of CDK2/cyclin E in HeLa nuclear extract blocked Tat-dependent transcription. We suggest that CDK2 is part of a transcription complex that is required for Tat-dependent transcription and that interaction of Tat with CTD and a dynamic association of Tat with CDK2/cyclin E stimulated CTD phosphorylation by CDK2.

HIV-1 1 Tat is a viral protein that interacts with transactivation response (TAR) RNA, a hairpin-loop structure at the 5Ј-end of all nascent viral transcripts (reviewed in Refs. 1 and 2). Tat stimulates transcriptional elongation (3), which is reg-ulated by phosphorylation of the largest subunit of RNA polymerase II (RNAPII). The C-terminal domain (CTD) of RNA-PII consists of heptapeptide YSPTSPS repeats, which are phosphorylated on Ser-2 and Ser-5 residues during transcription (reviewed in Ref. 4). CTD is phosphorylated by host cell cycle-dependent protein kinases CDK7 and CDK9 (reviewed in Ref. 5). General transcription factor TFIIH-associated CDK7 phosphorylates Ser-5 during initiation of transcription, whereas positive transcription elongation factor b-associated CDK9 phosphorylates Ser-2 during elongation of transcription (5). Tat associates with the bulge of TAR RNA and also binds to cyclin T1, a cyclin partner of CDK9, which in turn interacts with the loop of TAR RNA (6). This allows CDK9/cyclin T1 to be recruited by Tat to the HIV-1 promoter (7). Tat also modifies the substrate specificity of CDK9 to achieve additional Ser-5 phosphorylation (8).
Although the evidence for the role of CDK9/cyclin T1 in Tat-mediated HIV-1 transcription is overwhelming, our recent data suggest that there may be an additional CTD kinase involved in the Tat response. We have purified a Tat-associated T-cell-derived kinase (TTK) that phosphorylates CTD (9 -11). TTK stimulates Tat transactivation in vitro (11) and in vivo (10,11). TTK containes CDK2, which phosphorylates CDK7 (11).
In the work presented here, we analyze the effect of Tat on CTD phosphorylaton by CDK2/cyclin E and the function of CDK2 in transcription assays of HIV-1 promoter in vitro. Our finding demonstrated that interaction of Tat with CTD and a dynamic interaction with CDK2/cyclin E stimulated CTD phosphorylation by CDK2. Also we showed that CDK2 was part of transcription complex and was required for Tat-dependent transcription in vitro.
Tat Phosphorylation Assays-Purified Tat (200 ng) was phosphorylated in the CTD kinase assay as described above. Phosphorylated Tat was resolved on 4 -20% SDS-PAGE and was then subjected to autoradiography and quantitation with the Phosphor Imager (Packard Instruments, Wellesley, MA).
CTD Mobility Shift Assays-Assays were performed as we described previously (17). GST-CTD was labeled by CDK2/cyclin E as described above, except 50 nM ARAFGVPVRTYAHEVVTLWYRA peptide (18) was added to stimulate CTD hyperphosphorylation. The kinase reaction was stopped with 7 mM EDTA prior to the addition of Tat. Then, 50 ng of 32 P-labeled GST-CTD was incubated under the condition of the kinase assay, but with purified Tat or Tat peptide for 15 min at 30°C. Before application on a gel, SDS (2%) was added as described previously (17). Complexes were resolved on 4% acrylamide gel and then subjected to autoradiography using the Phosphor Imager (Packard Instruments).
Biotinylation of Template DNAs-The HIV-1 long terminal repeat (LTR) template (nucleotides Ϫ92 to ϩ180) was amplified by PCR from a plasmid (pDH125) carrying the entire HIV-1 dual tropic HIV-1 genome with the forward primer 5Ј-biotinylated ACTTTCCGGGGAG-GCGTGATC-3Ј and the reverse primer 5Ј-GCCACTGCTAGAGATTT-TCCACACTG-3Ј (Molecular Resource Facility, University of Medicine and Dentistry of New Jersey). The PCR product was phenol/chloroformed and precipitated prior to use.
Pull-down of HIV-1 Transcription Complexes-Transcription reac-tion mixtures (50 l) contained 300 g of HeLa whole cell extract, 1.0 g of biotinylated HIV-LTR (Ϫ92/ϩ180) DNA, 1.0 g of poly(dI-dC) and Tat (1 g Western Blot Analysis-The purified HIV-1 transcriptional complexes assembled on the immobilized templates were heated for 10 min at 100°C in SDS-loading buffer. The released proteins were resolved on 4 -20% SDS-PAGE and then transferred to polyvinylidene fluoride membranes (Millipore, Allen, TX). The membrane was analyzed with antibodies against RNAPII, CDK2 and CDK9.

CDK2/Cyclin E Phosphorylates Ser-2 and Ser-5 of the CTD Heptapeptide
Repeat-We have recently demonstrated that CDK2/cyclin E phosphorylates CTD (11). CDK2 binds to a (S/T)PX(K/R) consensus phosphorylation site (19). The CTD heptapeptide (YSPTSPS) contains two SP repeats that could potentially be recognized by CDK2/cyclin E. To analyze the phosphorylation sites, we used synthetic peptides containing two CTD repeats (hepta-2). We also used mutated CTD peptides containing a Ser-5 mutation in the first hepta repeat (A5), a Ser-2 mutation in the second hepta repeat (A9), a Ser-5 mutation in the second hepta repeat (A12) or two Ser-7 mutations in the first and second hepta peptides (A7/A14) (20). Mutations in Ser-5 of the first repeat (A5) and Ser-2 in the second repeat (A9) greatly reduced (5-to 7-fold) hepta-2 phosphorylation by CDK2/cyclin E (Table I). In contrast, the Ser-5 mutation in the second hepta repeat (A12) or the two Ser-7 mutations in the first and second hepta peptides (A7/A14) only moderately affected (less than 2-fold) hepta-2 phosphorylation ( Table I). Addition of Roscovitine, a purine analog that competes with ATP, inhibited hepta-2 phosphorylation in accord- Tat Enhances CTD Phosphorylation by CDK2/Cyclin E-Tat has been reported to induce CTD phosphorylation by CDK7 (11,18,22,23) and CDK9 (8,24). We analyzed the effect of Tat on CTD phosphorylation by CDK2/cyclin E. In the absence of Tat, CDK2/cyclin E poorly phosphorylated recombinant GST-CTD (Fig. 1A, lane 1). Tat dramatically enhanced both hypoand hyperphosphorylation of GST-CTD by CDK2/cyclin E in a concentration-dependent manner (Fig. 1A, lanes 2-4). The hyperphosphorylated form of CTD (CTDo), appeared as an apparent higher molecular weight band on SDS-PAGE, whereas the hypophosphorylated CTD (CTDa), migrated together with nonphosphorylated CTD (Fig. 1A, lanes 2-4). Mutation in cysteine-22 (C22G) rendered Tat inactive to stimulate phosphorylation of CTD by CDK2 (Fig. 1A, lanes 5-7). In a control experiment, Tat stimulated GST-CTD phosphorylation by CDK9/cyclin T1 in a concentration-dependent manner (not shown) in accordance with the earlier report (24). We subsequently examined the effect of Tat on phosphorylation of RNA-PII CTD. Again, CDK2/cyclin E did not phosphorylate RNAPII CTD in the absence of Tat (Fig. 1B, lane 5). Addition of Tat stimulated RNAPII CTD phosphorylation in a concentrationdependent manner (Fig. 1B, lanes 6 -8). In this experiment RNAPII CTDo and RNAPII CTDa could not be resolved due to the fewer number of CTD repeats in yeast RNAPII (4). Taken together, the results presented in this section indicate that Tat dramatically induced CDK2 to phosphorylate CTD.
Tat-mediated Stimulation of CTD Phosphorylation Requires Amino Acids 42-72 of Tat-A recent report suggested that two domains of Tat, the N-terminal domain (amino acids 1-48) and an arginine-rich motif (ARM, amino acids 49 -57), are required to hyperphosphorylate CTD by CDK9/cyclin T1 (24). We analyzed the requirement of Tat domains for CTD phosphorylation by CDK2 using chemically synthesized peptides corresponding to several domains of Tat. The N-terminal domain of Tat (amino acids 1-49) did not stimulate CTD phosphorylation (Fig. 2, lanes 5-7). Tat with the truncation of first 42 amino acids (Tat 42-72) stimulated CTD hypo-and hyperphosphorylation (Fig.  2, lanes 8 -10), although to a lesser extent as compared with the effect of the full-length Tat (Fig. 2, lanes 2-4). To analyze whether a part of the Tat 42-72 peptide could also stimulate CTD phosphorylation, we used peptides that spanned the 42-72 sequence. A peptide containing only the RNA-binding domain of Tat (Tat-(49 -60)) did not stimulate CTD hyperphosphorylation (Fig. 2, lanes 14 -16). Peptides flanking the RNAbinding domain (Tat 42-54 and Tat 56 -70) did not have a stimulatory effect either (Fig. 3, lanes 11-13 and 17-19). Therefore, Tat 42-72 is the minimal activation domain required to achieve CTD phosphorylation.
CDK2/Cyclin E Phosphorylates Tat by Dynamically Associating with Amino Acids 16 -24 and 36 -49 of Tat-Previously, HIV-2 Tat was shown to be a substrate for a CTD kinase (25). To study if CDK2/cyclin E dynamically interacted with Tat we analyzed whether Tat was phosphorylated by CDK2. We discovered that Tat was phosphorylated by CDK2/cyclin E in a concentration-dependent manner (Fig. 3A, lanes 2-4). To analyze which domain of Tat may form dynamic contacts with CDK2/cyclin E during phosphorylation reaction, a competition assay was performed with the excess of Tat peptides, which spanned different domains of Tat. The peptide containing amino acids 11-24 of Tat (Tat-(11-24)) completely blocked Tat phosphorylation (Fig. 3B, lane 3). In contrast, the peptide containing amino acids 6 -14 of Tat (Tat-(6 -14)) was not inhibitory. Therefore amino acids 15-24 are critical for the CDK2/ cyclin E-mediated phosphorylation of Tat. Neither Tat-(49 -60) nor Tat-(56 -70) interfered with Tat phosphorylation (Fig. 3B, lane 4 and 5). We observed that Tat-(36 -72) inhibited Tat phosphorylation (Fig. 3B, lane 6). Therefore, we concluded that amino acids 36 -49 of Tat also dynamically interact with CDK2/cyclin E during phosphorylation reaction. Taken together, our observations demonstrate that CDK2/cyclin E dynamically interacts with the N-terminal domain of Tat.
Tat Binds to CTD-It has been suggested that Tat binds to RNAPII CTD through the RNA-binding arginine-rich motif of Tat, based on the observation that TAR RNA inhibited Tatmediated CTD phosphorylation by CDK9/cyclin T1 (24). Therefore it was of interest to directly investigate binding of Tat to CTD. For this purpose, we utilized a modified gel shift assay that was previously developed to study the binding of TAR RNA to short peptides derived from double-stranded RNAactivated kinase (17). In the modified assay, we analyzed retardation of phosphorylated CTD as a function of Tat binding. CTD migrated on 4% polyacrylamide gel as a diffuse band (Fig.  4A, lane 1). Addition of full-length Tat resulted in a dose-dependent shift of CTD (Fig. 4A, lanes 2-4). Mutation in the N-terminal domain of Tat (C22G) reduced the efficiency of the shift (Fig. 4A, lanes 5-7). Truncation of first N-terminal 36 or 42 amino acids significantly reduced the ability of mutated Tat to bind to CTD (Fig. 4B, lanes 3 and 4). Further truncated Tat-(56 -70) or isolated RNA-binding domain of Tat-(49 -60) did not bind to CTD (Fig. 4B, lanes 5 and 6). However, the isolated activation domain of Tat, Tat-(1-44) bound to CTD (Fig. 4C, lane 5), although less efficiently than full-length Tat (Fig. 4C, lane 2). Because the N-terminal peptides, Tat- (6 -14) and Tat-(11-24), did not bind CTD (Fig. 4C, lanes 3 and 4), then amino acids 24 -44 may contain a CTD binding site in addition to the one or more sites located in the amino acids 42-72.
Therefore, Tat interacts with CTD via two binding sites located within amino acids 24 -72.
CDK2 Associates with RNAPII in HeLa Nuclear Extract-To assess whether CDK2 is part of a transcription complex we analyzed whether CDK2 is associated with complexes that contain RNAPII. Following fractionation of HeLa nuclear extract on Superose 6 size exclusion column, CDK2 was found to be present in the high molecular weight fractions that also contained RNAPII (Fig. 5A, fractions 15-19). To assess whether CDK2 and RNA PII were part of the same macromolecular complex, an affinity column that immobilized 8WG16 anti-RNAPII monoclonal antibodies was utilized. The HeLa nuclear extract was applied to the affinity column and the retained proteins were eluted with sodium acetate buffer at pH 3.4. Immunoblotting analysis showed that both RNAPII and CDK2 were partially retained by the column, whereas the ␣-subunit of the translation initiation factor eIF-2, which served as a negative control, was not retained at all (Fig. 5B). The data suggest that RNAPII and a species of CDK2 are part of the same macromolecular complex and that CDK2 could be involved in the control of RNAPII-mediated transcription.
CDK2 Associates with HIV-1 Promoter-To analyze whether CDK2 is part of the preinitiation complex assembled on HIV-1 promoter, preinitiation complex was formed on a biotinylated HIV-1 LTR template that was subsequently immobilized on streptavidin-agarose beads. The retained proteins were analyzed by immunoblotting analysis. RNAPII and CDK9, but not CDK2, associated with the HIV-1 LTR template (Fig. 6, lane 6). Addition of Tat did not stimulate association of CDK2 with the HIV-1 LTR template (Fig. 6, lanes 3 and 5). To analyze whether CDK2 associates with elongation complex, we performed a similar experiment, except that GTP, CTP, and ATP were added to the transcription reaction to promote initial assembly of the transcriptional apparatus. Also Tat was added to the reaction. In the absence of UTP, RNAPII is paused on the fifth nucleotide of TAR RNA. Analysis of proteins associated with this paused RNAPII complex showed that both CDK2 was present in the paused complex (Fig. 6, lanes 2 and 4). These observations suggest that CDK2 is present in early elongation complex.
CDK2/Cyclin E Is Required for Tat-mediated Transcription from HIV-1 Promoter-The requirement of CDK2 in HIV-1 transcription was tested in a reconstituted transcription reaction. The reaction was assembled with biotinylated HIV-1 LTR template immobilized on streptavidin-Sepharose beads and with a mixture of basal factors, including RNAPII, e-TFIID (TBP and TAF proteins), rTFIIA, rTFIIB, rTFIIF, hTFIIH, hRHA, and rp300. Under these conditions, we observed very little activated transcription with basal factors (Fig. 7, lane 1). Strikingly, addition of recombinant CDK2/cyclin E stimulated activated transcription to about 20-fold (Fig. 7, lane 2). Addition of CDK9/cyclin T1 stimulated activated transcription to about 9-fold (Fig. 7, lane 3). Next we analyzed whether CDK2/ cyclin E present in HeLa nuclear extract was essential for Tat-dependent transcription on immobilized template. Addition of Tat to HeLa nuclear extract stimulated transcription 65-fold (Fig. 7, compare lane 5 with lane 4). Immunodepletion of CDK2/cyclin E from the HeLa nuclear extract completely abolished the Tat-mediated activation of transcription (Fig. 7,  lane 6). Addition of recombinant CDK2/cyclin E to the immunodepleted nuclear extract restored the Tat-activated transcription in a concentration-dependent manner (Fig. 7, lanes  7-9). Therefore, collectively, these results suggest that the CDK2/cyclin E complex is necessary for Tat-activated transcription in vitro. DISCUSSION This study describes a potentially novel mechanism by which HIV-1 Tat stimulates CTD phosphorylation by CDK2/cyclin E. More specifically, we have found that full-length Tat binds to FIG. 5. CDK2 associates with RNAPII. A, CDK2 comigrates with RNAPII in gel-filtration assay. HeLa nuclear extract was fractionated on Superose 6. Nuclear extract input, and the column fractions were resolved on 4 -20% SDS-PAGE and analyzed by immunoblotting with antibodies against RNAPII (8WG16) and CDK2. CDK2 and RNAPII coeluted in fractions 15-19. I, input; rCDK2, recombinant CDK2. Numbers at the top correspond to the fraction numbers. B, CDK2 associates with RNAPII. Elution profile of HeLa nuclear extract (750 g) separated on anti-RNAPII (8WG16) affinity column. Flow-through and eluted fractions were concentrated in microcon-10 spin columns, resolved on 10% SDS-PAGE, and analyzed by immunoblotting with antibodies against RNAPII CTD (8WG16), CDK2, and eIF-2␣. RNAPII and CDK2 were retained on the column and eluted in acidic pH as described under "Experimental Procedures." In contrast, eIF-2␣ was not retained by the column. CTD and is also dynamically associated with CDK2/cyclin E, which greatly stimulated recombinant CDK2/cyclin E to phosphorylate CTD. We found that CDK2 associates with RNAPII and is part of transcription complex assembled on the HIV-1 LTR template. Finally, we have shown that CDK2 was required for Tat-dependent transcription.
Recent studies provided substantial evidence that a general transcription elongation factor-b-associated CDK9/cyclin T1 plays a key role in the Tat-mediated activation of HIV-1 transcription (reviewed in Refs. 1 and 2). Tat associates with the bulge of TAR RNA and also binds to cyclin T1, a cyclin partner of CDK9, which in turn interacts with the loop of TAR RNA (6,22). This allows CDK9/cyclin T1 to be recruited by Tat to the HIV-1 promoter and stimulate transcription elongation by RNAPII (7). Although the evidence for the role of CDK9/cyclin T1 in Tat-mediated HIV-1 transcription is overwhelming, the collective data from our laboratories suggest that there may be an additional CTD kinase involved in the Tat response.
We have reported that Tat associates with a CTD kinase isolated from human primary T-lymphocytes (TTK) (9). Microinjections of TTK into human primary fibroblasts stimulated Tat-dependent expression of a reporter LacZ gene placed under the control of HIV-1 LTR (10). It was tempting to suggest that TTK contained CDK9, but we did not find CDK9 by immunoblotting assays (10). Moreover, biochemically TTK was distinct from CDK9, because TTK phosphorylated CDK7 and promoted association of CDK7 and cyclin H (10).
Analysis of TTK by biochemical fractionation showed that TTK uniquely copurified with CDK2 and not with CDK9 or CDK7 (11). Tat induced the TTK and CDK2 to phosphorylate CTD, specifically at serine 2 residues (11), which was in contrast to the reported phosphorylation of Ser-5 by CDK9 in the presence of Tat (8). TTK restored Tat-mediated transcription activation of HIV-1 LTR in an HeLa nuclear extract immunodepleted of CDK9 but not in the HeLa nuclear extract doubledepleted of CDK9 and CDK7 (11). Therefore CDK2 was likely to be involved in the regulation of Tat-mediated transcription.
Another line of evidence, which points to CDK2, comes from the analysis of the cell cycle regulation of HIV-1 transcription. We have observed that Tat-mediated transcription is regulated during the cell cycle (10,26). We have demonstrated that transcription from HIV-1 LTR is Tat-dependent at the G 1 and less so at G 2 /M phase of the cell cycle (26). We observed highest Tat response of HIV-1 LTR transcription in the cells progressing from G 0 to G 1 phase (10). Overexpression of specific CDK inhibitors, p16 and p27, blocked Tat transactivation (10). Overexpression of dominant negative mutants of CDK2 and CDK4 but not CDK1 blocked Tat transactivation (10). These data indicate that Tat-dependent transcription may in part be regulated by CDK2, which activity is highest at the G 1 phase. Finally, we have observed that the activity of CDK2/cyclin E is increased in HIV-1-infected quiescent peripheral CD4 lymphocytes (27). Collectively, our published data (9 -11, 26, 27) points to a possibility that Tat-dependent transcription may be regulated by CDK2.
We report here that Tat stimulated CDK2 to phosphorylate CTD in a concentration-dependent manner. Little phosphorylation was observed in the absence of Tat. Analysis of the Tat regions responsible for stimulation of the kinase showed that two regions of Tat were required for efficient stimulation of CTD hypo-and hyperphosphorylation: cysteine 22 and the peptide, containing amino acids 42-72 of Tat. Mutation in the cysteine 22 decreased the efficiency of Tat stimulation, implying that the N terminus of Tat is important for stimulation. On the other hand, analysis of Tat truncation mutants, lacking N terminus amino acids showed that Tat-(42-72) partially stim-ulated CTD phosphorylation by CDK2/cyclin E. The full-length Tat showed 2-fold higher level of CTDa and CTDo phosphorylation at 2 M concentration, compared to CTD phosphorylation at 6 M Tat-(42-72) (Fig. 2B). Therefore Tat-(42-72) is ϳ10times weaker activator than the full-length Tat. Therefore, it is likely that there is a binding site located in the N-terminal 42 amino acids of Tat, in addition to the one or more binding sites located within amino acids 42-72. These binding sites may interact with CTD and/or CDK2/cyclin E to stimulate CTD phosphorylation. Our results are in agreement with recent demonstration that full-length Tat, but not the isolated Nterminal domain of Tat, stimulated CTD hyperphosphorylation by CDK9 (24).
To study if CDK2/cyclin E dynamically interacted with Tat we analyzed phosphorylation of Tat by CDK2/cyclin E. Tat was phosphorylated by CDK2/cyclin E in a concentration-dependent manner. Competition analysis showed that amino acids 15-24 and 36 -49 of Tat interacted with CDK2/cyclin E. The amino acids 15-24 of Tat contain a 16 SQPK 19 sequence, which resembles a putative CDK phosphorylation site (S/T)PX(K/R) (19) and therefore may serve as a CDK2 phosphorylation site. The amino acids 36 -49 of Tat contain a 41 KAL 43 sequence, which resembles a cyclin binding motif (Cy or RxL). This motif binds to a hydrophobic groove on the surface of the cyclin E and allows potential CDK2 phosphorylation site to be in proximity to CDK2 active center (28). The distance between the CDK2 phosphorylation site and the cyclin E binding site has been determined to be at least a 12 amino acids long to allow CDK2/ cyclin E binding (28). In the case of Tat, the distance between SQPK sequence and KAL sequence is 21 amino acids, which is sufficient to allow simultaneous binding of CDK2 and cyclin E to Tat. Whether these motifs interact directly with CDK2/ cyclin E still remains to be established. Also of interest would be to identify what position of Tat is phosphorylated by CDK2/ cyclin E. Finally, it would be of interest to find out a role of Tat phosphorylation in HIV-1 transcription.
The formation of Tat⅐TAR RNA⅐CDK9/cyclin T1 complex has been demonstrated in a gel-shift assay (6,24). In contrast, we did not detect binding of CDK2/cyclin E to Tat⅐TAR RNA complex in a gel-shift assay (data not shown). Therefore, the find- FIG. 8. Two-step model depicting Tat-mediated CTD phosphorylation by CDK2/cyclin E. Tat binds to CTD through amino acids 24 -72. Only binding of arginine-rich motif (ARM) to the phosphorylated CTD residues is shown. The Tat activation domain interacts with CDK2 through the proposed binding of SPQK motif and with cyclin E through the proposed KAL sequence. Upon Tat phosphorylation, the SPQK motif dissociates from CDK2 and CDK2 is redirected to the SP sequence on CTD, whereas Tat still interacts with the cyclin E and CTD. This may allow a single round as well as a multiple CTD phosphorylation by CDK2.
ing, that CDK2 phosphorylates Tat, but does not form a stable complex with Tat⅐TAR RNA suggests that CDK2-Tat interaction is dynamic. Our data point to the possibility that this dynamic interaction between Tat and CDK2/cyclin E may target CDK2 to CTD. Accordingly, we demonstrated that Tat interacted with CTD in gel-shift assay through the interactions within the amino acids 24 -72.
We observed increase of both hypo-and hyperphosphorylation of CTD (Figs. 1 and 2). For mouse CTD phosphorylated by cdc2, the CTD mobility shift takes place when at least 15 phosphates are incorporated into 52 heptapeptide repeats (29). However, in contrast to CTD phosphorylation by cdc2, which resulted in a gradual shift of CTD (29), we observed both hypoand hyperphosphorylated forms of CTD at the same time. This indicates that phosphorylation of CTD occurs in two different ways; in stochastic fashion, which generates CTDa, and in processive fashion, which results in multiple phosphorylations of the same CTD molecule. Accumulation of CTDa indicates that a single round phosphorylation is a more frequent event than multiply phosphorylation. Because we did not observe intermediately phosphorylated CTD, CTDo is likely to be a result of a cooperative phosphorylation reaction, which means that consecutive phosphorylation of already phosphorylated CTD is more efficient than the first phosphorylation event. We present a hypothetical model of the Tat-mediated CTD phosphorylation by CDK2/cyclin E (Fig. 8). Tat interacts with CTD through the arginine-rich motif (ARM, amino acids 49 -60) that may interact with the phosphorylated CTD residues and through the amino acids 24 -36 (not shown in the figure). We speculate that the activation domain of Tat dynamically interacts with CDK2 through the proposed binding of SPQK sequence and with cyclin E through the cyclin-binding KAL sequence. Upon Tat phosphorylation by CDK2, the SPQK motif dissociates from CDK2 and the kinase is redirected to the SP sequence on CTD due to the continuing interaction of Tat with cyclin E and CTD. A single round of phosphorylation reaction will generate hypophosphorylated CTD, and the amount of the CTDa will be increased with the increase of concentration of Tat, which will bring CDK2 in proximity of CTD. We speculate that, after a single round of CTD phosphorylation by CDK2, the CDK2-cyclin E-Tat complex may slide along the CTD. Ionic interactions between ARM and CTD may permit a quick movement along the CTD, similar to the described movement of lac repressor toward lac operator on DNA when the repressor is intermediately associated with DNA through ionic contacts (30). As a result, CDK2 may phosphorylate CTD at multiple sites, generating hyperphosphorylated form of CTD.
It has been shown that the CTD heptapeptide is primarily phosphorylated on Ser-2 and Ser-5 during transcription (4). In the present paper we showed that CDK2 phosphorylates both Ser-2 and Ser-5 of CTD heptapeptide repeat when only two repeats were used as a substrate. On full-length CTD, Tat induced primarily hyperphosphorylation of Ser-2 by CDK2/ cyclin E (11). Interestingly, Tat stimulated CDK9 to phosphorylate Ser-5 in in vitro transcription assays (8,24). It indicates that there is a principal distinction in the mechanisms of CTD recognition by CDK2 and CDK9 in the presence of Tat. Also, it may indicate that the kinases are activated at different stages of transcription. In yeast, phosphorylation of Ser-5 during preinitiation of transcription allows association of capping factors with CTD (31)(32)(33). Then Ser-5 is dephosphorylated by an as yet unknown phosphatase that allows dissociation of capping factors and then Ser-2 is phosphorylated by CDK9 (34). The pattern of Ser-2 and Ser-5 phosphorylation in higher eukaryotes during HIV-1 transcription remains to be established as well as the role of Tat in this process.
We observed that CDK2 was associated with RNAPII. Fractionation of HeLa nuclear extract on the size-exclusion column showed that a portion of CDK2 coeluted with RNAPII. In addition, CDK2 associated with RNAPII on an affinity column with immobilized anti-RNAPII CTD antibodies. Therefore a portion of CDK2 is likely to be associated with the transcription complex. Surprisingly, we did not observe CDK2 in the preinitiation complex assembled on HIV-1 LTR template. Instead we found CDK2 present in the early elongation complex. This finding indicates that CDK2 may be recruited to the HIV-1 promoter after the preinitiation. Because CDK2 phosphorylates Ser-2 in the presence of Tat (11), then finding of CDK2 in the elongation complex is consistent with the observation that Ser-2 phosphorylation occurs at elongation and not at the initiation of transcription (34).
We investigated the requirement of CDK2/cyclin E for HIV-1 transcription by analyzing the effect of CDK2 in reconstituted transcription assay, which was devoid of CDK9/cyclin T1. In this system, addition of either CDK2/cyclin E or CDK9/cyclin T1 stimulated Tat-dependent transcription. In HeLa nuclear extract, immunodepleted of CDK2/cyclin E dramatically decreased Tat-dependent transcription. Addition of CDK2/cyclin E to the depleted extract fully restored Tat-dependent transcription. The results indicate that CDK2/cyclin E is required for Tat-dependent transcription in vitro. Because CDK2/cyclin E did not bind to the Tat⅐TAR RNA complex (data not shown), therefore, CDK2 may only partially reconstitute transcription activation on the HIV-1 promoter.
The role of CDK2 in stimulation of Tat-mediated transcription is intriguing due to the discussed regulation of HIV-1 transcription during the cell cycle (10,26). It is also possible that HIV-1 Tat may utilize CDK2 activity, elevated during HIV-1 infection (27), to induce transcription.
This study describes a potentially novel mechanism by which HIV-1 Tat stimulates CTD phosphorylation by CDK2/cyclin E. Uncovering an alternative pathway for Tat-mediated hyperphosphorylation of RNAPII CTD may provide additional valuable targets for anti-HIV-1 therapeutics.