Upregulation of Cyclin T1/CDK9 Complexes During T Cell Activation

Judit Garriga1,2, Junmin Peng4, Matilde ParrenÄ o1,2, David H Price4, Earl E Henderson1,3 and Xavier GranÄ a*,1,2

1Fels Institute for Cancer Research and Molecular Biology, Temple University School of Medicine, 3307 North Broad Street, Philadelphia, Pennsylvania 19140, USA; 2Department of Biochemistry, Temple University School of Medicine, 3307 North Broad Street, Philadelphia, Pennsylvania 19140, USA; 3Microbiology and Immunology, Temple University School of Medicine, 3420 North Broad Street, Philadelphia, Pennsylvania 19140, USA and 4Department of Biochemistry, University of Iowa, Iowa City, Iowa 52242, USA

Cyclin T1 has been identi®ed recently as a regulatory date in response to intracellular or extracellular signals subunit of CDK9 and as a component of the transcrip- in mammalian cells, although it has been shown that tion elongation factor P-TEFb. Cyclin T1/CDK9 com- the levels of cyclin C mRNA are stimulated by serum plexes phosphorylate the carboxy terminal domain and cytokines (Lew et al., 1991; Liu et al., 1998). All (CTD) of RNA polymerase II (RNAP II) in vitro. Here cyclin/CDK pairs involved in transcription are able to we report that the levels of cyclin T1 are dramatically phosphorylate the C-terminal domain (CTD) of RNA upregulated by two independent signaling pathways polymerase II (RNAP II) in vitro. Multiple kinases triggered respectively by PMA and PHA in primary seem to phosphorylate the CTD of RNAP II in vivo in human peripheral blood lymphocytes (PBLs). Activation a manner that appears to be essential for progression of these two pathways in tandem is sucient for PBLs to through dierent transcriptional stages (Dahmus, enter and progress through the cell cycle. However, the 1996). In this regard, transcriptional initiation and expression of cyclin T1 is not growth and/or cell cycle elongation can be distinguished by the level of CTD regulated in other cell types, indicating that regulation of phosphorylation. cyclin T1 expression is dependent on tissue-speci®c A positive transcription elongation factor (P-TEFb) signaling pathways. Upregulation of cyclin T1 in has been described that enables RNA polymerase II to stimulated PBLs results in induction of the CTD kinase enter productive elongation by phosphorylating the activity of the cyclin T1/CDK9 complex, which in turn CTD and counteracting negative factors which limit correlates directly with phosphorylation of RNAP II in polymerase processivity (Marshall et al., 1996; Mar- vivo, linking for the ®rst time activation of the cyclin T1/ shall and Price, 1995). Cloning of the subunits of P- CDK9 pair with phosphorylation of RNAP II in vivo.In TEFb has revealed that it is a cyclin/CDK pair (Peng addition, we report here that endogenous CDK9 and et al., 1998a; Wei et al., 1998; Zhu et al., 1997). The cyclin T1 complexes associate with HIV-1 generated Tat small subunit of Drosophila P-TEFb was cloned using in relevant cells and under physiological conditions sequence information obtained from the puri®ed (HIV-1 infected T cells). This, together with our results protein (Zhu et al., 1997) and was found to be the showing that HIV-1 replication in stimulated PBLs homolog of the human CDC2-related kinase PI- correlates with the levels of cyclin T1 protein and TALRE (GranÄ a et al., 1994a). We have reported associated CTD kinase activity, suggests that the previously that PITALRE is a primarily nuclear Ser/ cyclin T1/CDK9 pair is one of the HIV-1 required host Thr-Pro-directed kinase (Garriga et al., 1996b) with cellular cofactors generated during T cell activation. activity that is not cell cycle regulated in HeLa and ML-1 (GranÄ a et al., 1994a). Initial studies found most Keywords: PITALRE; CDK; RNA polymerase II; of the active PITALRE in high molecular weight HIV; Tat; peripheral blood lymphocytes complexes containing a number of unidenti®ed proteins. We had suggested that one of these proteins, p95, might be a positive regulatory subunit because the kinase activity of ectopically overexpressed Introduction PITALRE seemed to be limited by the amount of p95 (Garriga et al., 1996a). Subsequently, the large subunits Cyclins are the regulatory subunits of cyclin dependent of both Drosophila (Peng et al., 1998b) and human kinases (CDKs). Several of these complexes have been (Peng et al., 1998a; Wei et al., 1998) P-TEFb were identi®ed hereto with roles in cell cycle regulation and cloned and found to be cyclin proteins. Because the transcription (GranÄ a and Reddy, 1995; Jones, 1997; kinase subunit (PITALRE) required the cyclin subunit Morgan, 1997). The activity of cyclin/CDK pairs with for activity, PITALRE was renamed CDK9 and the roles in cell cycle control is generally regulated by large subunits were called T-type cyclins. growth factors or is dependent on the cell cycle stage. Recent evidence clearly implicates P-TEFb in Tat However, no changes in the activity of cyclin/CDK transactivation of the HIV-1 promoter. Tat pull-down pairs involved in transcription have been reported to experiments provided early evidence of a Tat-associated kinase (TAK) (Herrmann and Rice, 1993, 1995). Tat acts by binding to the nascent transcript containing the TAR sequence and increasing the processivity of RNAP II (Jones, 1997; Jones and Peterlin, 1994). Since *Correspondence: X GranÄ a Received 23 September 1998; revised 29 October 1998; accepted 2 the latter function is similar to that of P-TEFb November 1998 (Marshall et al., 1996; Marshall and Price, 1995) it Upregulation of cyclin T1/CDK9 during T cell activation J Garriga et al 3094 was not surprising that CDK9 was found to associate it is associated with components of the basal with the activation domain of Tat in vitro and in HeLa transcriptional machinery as was found for CDK7 cells overexpressing Tat (Yang et al., 1997; Zhu et al., and CDK8 (Morgan, 1997). To this end, we utilized 1997). Additionally, P-TEFb was found to be essential antibodies to various subunits of basal transcriptional for Tat transactivation both in vitro (Zhu et al., 1997) factors and to RNAP II. Coincidentally, a polyclonal and in vivo (Mancebo et al., 1997). antibody (s235) raised to a peptide corresponding to Multiple cyclin partners for human CDK9 have the carboxyl-terminal end of the large subunit of been identi®ed allowing for dierent forms of P-TEFb TFIIF, RAP74, was found to cross-react strongly and (Peng et al., 1998a). Two genes (T1 and T2) encode speci®cally with the PITALRE-associated protein p95 cyclin T1, T2a and T2b with the latter two diering (see Materials and methods). As demonstrated in from each other only at the carboxyl-terminus due to Figure 1a, s235 and anti-CDK9 antibodies immuno- dierential splicing (Peng et al., 1998a). P-TEFb, precipitate a complex containing p95 and CDK9 in the comprised of CDK9 and any one of the cyclin T absence of competing antigenic peptides (left and lower subunits, is able to stimulate transcription in vitro and panels). A dierent antibody raised to the amino- in vivo from a CMV promoter when co-transfected terminus of RAP74 (anti-RAP74-NT) recognized with CDK9 in HeLa cells (Peng et al., 1998a). While RAP74 in a total protein extract and also in vitro this work was in progress a direct interaction between translated RAP74 (Figure 1a, right panel). However, Tat and cyclin T1 was demonstrated in vitro (Wei et RAP74, was not detected in either CDK9 or s235 al., 1998). This interaction enhances the anity and immunoprecipitates. To ascertain whether the protein speci®city of Tat binding to the HIV RNA element complexes brought down by anti-CDK9 and s235 TAR. It is not clear yet if cyclin T2a or T2b can exhibited similar kinase activities, we performed kinase associate with Tat. assays in the presence or absence of exogenous Activation of T-cells allows strong transcriptional substrates as described in Materials and methods. activation of the HIV promoter and induces productive The kinases immunoprecipitated by both antibodies replication of the virus (Kaufman et al., 1987; Nabel exhibited similar speci®city in that both could and Baltimore, 1987; Siekevitz et al., 1987; Tong- phosphorylate the CTD of RNAP II (Figure 1b, left Starksen et al., 1987). It has also been shown that a panel) and pRB (not shown) and without any CTD activity that associates with GST-Tat is exogenous substrate phosphorylation of p95 and upregulated upon activation of T-cells (Nekhai et al., autophosphorylation of CDK9 was observed with 1997; Yang et al., 1997). Since CDK9 has been found both immunoprecipitates (Figure 1b, right panel). to associate with Tat in vitro and in transient Since the origin of the s235 antibody is not essential transfection experiments in cell lines that support Tat to the results reported here, further characterization of transactivation, it is likely that induction of the GST- this antibody will be reported elsewhere (see Materials Tat-associated CTD activity upon activation of and methods). primary peripheral blood lymphocytes (PBLs) results Three cyclin partners for human CDK9 have from activation of a cyclin/CDK9 complex. However, recently been identi®ed and designated cyclins T1, the mechanism of activation of such a complex in T2a and T2b (Peng et al., 1998a). Since p95 has a stimulated PBLs remains unknown. A number of molecular weight similar to these cyclins, we consid- dierent possibilities can be envisioned including ered whether p95 could correspond to one of these upregulation of the protein levels of a rate limiting proteins. We performed immunoprecipitations with subunit (either by the cyclin, the CDK or both) and anti-CDK9, anti-cyclin T1 and anti-cyclin T2 antibo- activation of preexisting subunits and/or complexes by dies followed by Western blot with s235 (Figure 1c). post-translational modi®cations and/or other means s235 recognized recombinant cyclin T1 and the (GranÄ a and Reddy, 1995; Morgan, 1997; Xiong, 1996). cyclin T1 immunoprecipitated by anti-CDK9 and Here we report that the induction of the cyclin T1/ anti-cyclin T1 antibodies (Peng et al., 1998a). s235 CDK9 complex in activated PBLs is the result of did not cross-react with cyclins T2a and T2b. dramatic upregulation of the cyclin T1 regulatory Essentially identical results (not shown) were obtained subunit. CDK9 levels increase as well, but to a lesser by probing the Western blot with antibodies generated extent. In addition, this complex is targeted by Tat in T against cyclin T1 directly (Peng et al., 1998a). These cells in vivo. Upregulation of cyclin T1 results in results demonstrate that p95 is cyclin T1 and that the dramatic induction of CDK9 kinase activity and s235 antibody is a suitable reagent for its detection (see correlates with phosphorylation of RNAP II in vivo Materials and methods). and with productive replication of HIV-1 in PBLs. This is the ®rst report describing regulation of the kinase Endogenous cyclin T1 and CDK9 associate with Tat-1 in activity of a cyclin/CDK pair involved in transcrip- HIV-1 infected T cells tional control in response to the activation of a signaling pathway in mammalian cells. It has been found that CDK9 is present in anti-Tat-1 immunoprecipitates from HeLa cells ectopically overexpressing Tat (Yang et al., 1997). Since T cells are a primary target for HIV and Tat-1 is essential for HIV Results replication, we sought to ascertain whether endogenous cyclin T1 and CDK9 interact with Tat-1 in T-cells The PITALRE (CDK9)-associated protein p95 is infected by HIV-1. Tat-1, CDK9 and cyclin T1 cyclin T1 complexes were immunoprecipitated with the indicated Before PITALRE was identi®ed as the CDK9 subunit antibodies (Figure 2), and the immunoprecipitates were of P-TEFb, we undertook experiments to determine if resolved by SDS ± PAGE followed by immunoblot Upregulation of cyclin T1/CDK9 during T cell activation JGarrigaet al 3095

Figure 1 The CDK9-associated protein p95 is cyclin T1. (a) Protein extracts from HeLa cells were immunoprecipitated with anti- CDK9 and s235 antibodies, resolved by SDS ± PAGE and immunoblotted with anti-CDK9, s235 and anti-RAP74-NT (the antibody used for the Western blot is indicated below each panel). In vitro translated RAP74 (IT RAP74) and 40 mg of total protein extract (TE) were loaded as controls. The migration of CDK9, p95, RAP74, IgGs and molecular weight markers is indicated. A non- speci®c band detected in s235 immunoprecipitates in the presence of the antigenic peptide is indicated by an asterisk (*). (b) s235 and anti-CDK9 antibodies immunoprecipitate complexes with the same substrate speci®city. Kinase assays on CDK9 and s235 immunoprecipitates from HeLa protein extracts were performed as described in the Materials and methods section, in the presence of the CTD domain of RNAP II (GST-CTD) (left panel) or in the absence of exogenous substrates (right panel). (c) The s235 antibody recognizes endogenous and recombinant cyclin T1 directly and does not cross-react with cyclins T2a and T2b. HeLa nuclear extracts were immunoprecipitated with anti-GST, anti-CDK9, anti-cyclin T1 and anti-cyclin T2, resolved by SDS ± PAGE and subjected to Western blot with s235 antibodies. Recombinant cyclin T1, cyclin T2a and cyclin T2b were also analysed by Western blot. Identical results were obtained by Western blot using anti-cyclin T1 antibodies (not shown). Relevant proteins and the migration of size markers are indicated at the right using anti-Tat-1 antibodies. Figure 2 demonstrates that HIV-1 produced Tat-1 associates with endogenous cylcin T1 and CDK9 in SUP-T1 cells productively infected with HIV-1 and in the HIV-1 shedding T cell line MOLT-4 IIIB persistently infected with HIV- 1 IIIB. These complexes were not detected in uninfected SUP-T1 cells, which obviously do not express any Tat protein. Conversely, Western blot analysis using anti-cyclin T1 and s235 antibodies detected cyclin T1 in anti-Tat-1 immunoprecipitates of MOLT- 4 IIIB cells and SUP-T1 cells infected with HIV-1 IIIB. However, cyclin T1 was not detected in anti-Tat-1 immunoprecipitates from uninfected SUP-T1 cells (not shown). Thus, these data demonstrate that in vivo, HIV Tat-1 targets both cyclin T1 and CDK9 when these proteins are expressed at their endogenous levels in Figure 2 Endogenous cyclin T1 and CDK9 associate with Tat-1 relevant cells. In this regard, it is important to note in HIV-1 infected cells. MOLT-4 IIIB (a HIV-1 shedding T cell line), SUP-T1 cells and SUP-T1 cells infected with HIV-1 IIIB that the levels of cyclin T1 and CDK9 in SUP-T1 cells (24 h post-infection) were lysed and protein extracts immunopre- are very similar to the levels of these proteins in cipitated with anti-CDK9-CT, s235 and anti-Tat-1 antibodies. activated PBLs (data not shown). While this paper was Anti-CDK9 immunoprecipitates were performed in the presence in preparation, it was reported that cyclin T1 associates (+) or absence (7) of the antigenic peptide. Forty mg of total directly with Tat in vitro in such a way that cyclin T1 protein extract (TE) was loaded as control. Immunoprecipitates were resolved by 15% SDS ± PAGE followed by Western blot enhances the anity and speci®city of the Tat/TAR with anti-Tat-1 antibodies. Tat-1 and the light chain of IgGs are interaction (Wei et al., 1998) (see Discussion). indicated at the right Upregulation of cyclin T1/CDK9 during T cell activation J Garriga et al 3096 lated during the cell cycle of mammalian cells (Grana Cyclin T1 protein levels are dramatically upregulated in Ä and Reddy, 1995). Results reported here show that PMA and PHA-treated PBLs and this correlates with phosphorylation of RNAP II in vivo cyclin T1 is upregulated upon activation by two dierent mitogenic signals, PMA, which involves It has been shown that a CTD kinase activity that PKC activation, or PHA, which increases intracellular associates with GST-Tat in vitro is induced upon Ca2+ concentration through the TRC/CD3 receptor. activation of PBLs by PMA and/or PHA (Yang et al., However, neither of these two mitogens alone is 1997). Since CDK9 has been found to associate with sucient to induce quiescent PBLs to enter S phase Tat in vitro and in transient experiments in cell lines and eventually proliferate. Upon treatment of quiescent that support Tat transactivation (Yang et al., 1997; PBLs with PHA, these cells enter G1, however, they Zhu et al., 1997), we postulated that induction of the cannot progress through S-phase without the addition GST-Tat associated CTD kinase upon activation of of a second signal (Firpo et al., 1994). We treated PBLs might result from generation and/or activation of quiescent PBLs with PMA, PHA or PMA and PHA the cyclin T1/CDK9 complex. In general terms, together and then analysed the expression of cyclin T1 activation of a cyclin/CDK pair occurs by upregula- by Western blot. As indicated in Figure 4a, PMA and tion of the rate limiting subunit, most often the cyclin PHA alone induce cyclin T1 expression, although regulatory subunit, by post-translational modi®cations signi®cantly higher levels of cyclin T1 protein are aecting either subunit and/or by abrogation of the detected in PBLs treated with PMA and PHA interaction of CDK-inhibitors (CKIs) with either the together. We also measured S-phase entry by FACs CDK or the cyclin/CDK pair (GranÄ a and Reddy, 1995; analysis and by measuring the kinase activity Morgan, 1997; Xiong, 1996). Since our initial associated with cyclin A, which is essential for DNA experiments in 293 cells overexpressing CDK9 sug- replication. As expected, an increase in the number of gested that p95 (cyclin T1) could be limiting for cells in S and G2/M phases was detected when PBLs activation of CDK9 (Garriga et al., 1996a), we sought were treated with both PMA and PHA, but PBLs to determine if the levels of cyclin T1 are regulated treated with either PMA or PHA alone remained with during T cell activation. Thus, PBLs were obtained as a G1 DNA content (Figure 4b). In agreement with this described in Materials and methods section and result, we detected kinase activity in cyclin A com- incubated with PMA, PHA or TNF-a for 72 h. plexes only from PBLs treated with PMA and PHA Activation of PBLs with either PHA or PMA, which together (Figure 4c). are known to stimulate T cells through independent Since cyclin T1 seems to be upregulated at some pathways, resulted in a sharp upregulation of time during the G0/G1 transition upon PMA or PHA cyclin T1, together with a more modest increase in stimulation of PBLs and because even higher levels of the level of the CDK9 subunit (Figures 3 and 4). Since cyclin T1 protein are detected in PBLs progressing in vitro biochemical evidence indicates that CDK9 is a through S and G2/M phases, we sought to analyse CTD kinase (Peng et al., 1998b; Zhu et al., 1997) we whether cyclin T1 protein expression was regulated analysed the phosphorylation status of RNAP II in during cell cycle entry and progression of non T cell vivo using two dierent antibodies that distinguish the human cell types. T98G and HaCaT cells were serum

CTDA (hypophosphorylated CTD) and CTD0 (hyper- starved and re-stimulated as described previously phosphorylated CTD) forms of RNAP II. The monoclonal B3 antibody recognizes only hyperpho-

sphorylated RNAP II (CTD0) (Mortillaro et al., 1996), while the polyclonal anti-RNAP II antibody recognizes both forms. Western blot analysis shows that upregulation of cyclin T1 correlates with phosphorylation of the RNAP II in vivo (Figure 3), which is in agreement with the model that RNAP II is a bona ®de in vivo substrate of CDK9. Incubation of PBLs with TNF-a, which cannot activate resting T-cells because of the lack of TNF-a receptors, did not induce cyclin T1 and this correlated with lack of induction of RNAP II phosphorylation (Figure 3), suggesting that only those signals that induce cyclin T1 expression are able to induce RNAP II phosphorylation. Interest- ingly, this is the ®rst report showing regulation of the protein levels of a cyclin regulatory subunit involved in transcriptional control in response to any intracellular or extracellular stimulus.

Cyclin T1 is induced by two independent mitogenic Figure 3 Upregulation of cyclin T1 correlates with phosphoryla- pathways that cooperate to fully activate T-cells, but its tion of RNAP II in PMA and PHA-treated PBLs, which are expression is not modulated during the cell cycle in other signals that induce viral replication. Human PBLs were untreated or treated with PHA, TNF-a and PMA for 72 h. Forty mgof cell types protein extracts were subjected to Western blot analysis by using The protein levels of a number of cyclins including the following antibodies: B3, which recognizes RNAP II0 (Mortillaro et al., 1996); anti-RNAP II, which recognizes cyclin E, cyclin A and cyclin B, as well as their RNAP IIA and II0; s235; anti-cyclin A; anti-CDK9 and anti- associated kinase activities, are dramatically modu- CDK2. Relevant proteins are indicated at the right Upregulation of cyclin T1/CDK9 during T cell activation JGarrigaet al 3097 (Mayol et al., 1995, 1996). We found that the levels of Cyclin T1/CDK9 associated-CTD-kinase activity is cyclin T1 are not modulated during cell cycle entry and induced by single mitogenic signals (either PMA or progression of serum starved and re-stimulated T98G PHA) in PBLs. However, combination of these two cells (Figure 5a). We also found that the levels of signals exhibits a strong synergistic eect on the kinase CDK9 protein and associated kinase activity do not activity of the cyclin T1/CDK9 complex and correlates change in serum starved and re-stimulated T98G cells with HIV-1 replication in these cells (data not shown). A slight induction in the levels of cyclin T1 protein is shown in re-stimulated HaCaT We have shown here that either PMA or PHA alone is cells when compared to serum starved cells (Figure 5b). sucient to induce cyclin T1 protein expression, a However, we have not detected this slight induction mechanism that could explain a possible induction of consistently (data not shown). No signi®cant changes the activity of the cyclin T1/CDK9 complex and thus, in cyclin T1 protein levels are detected through the cell the induction of kinase activity trapped by GST-Tat in cycle (Figure 5b). In addition, we also analysed the activated PBLs. To test this possibility directly, PBLs levels of cyclin T1 in cycling 293 cells synchronized by were non-stimulated or stimulated with PMA, PHA or treatment with hydroxyurea (Ashihara and Baserga, PMA and PHA together for 48 h and protein lysates 1979) and nocodazole (Mayol et al., 1996). Hydro- were obtained. The CTD kinase activity associated xyurea arrests cells reversibly at the G1/S transition, with CDK9 and cyclin T1 was determined in CDK9 and cells progress synchronously through S phase and and cyclin T1 immunoprecipitates using GST-CTD as mitosis after placing them in fresh medium without the exogenous substrate. We also determined the cyclin E drug. Nocodazole arrests cells in a state of mitotic and cyclin A associated kinase activities as controls pseudometaphase. Mitotic cells placed in nocodazole- (see Materials and methods). Figure 6a shows that free medium progress synchronously through the next activation of the CTD-kinase activity present in CDK9 cell cycle. The levels of cyclin T1 protein did not and cyclin T1 complexes occurred in response to oscillate, at any phase during the cell cycle of 293 cells stimulation by a single mitogenic signal (either PMA (Figure 5c and d). Altogether, these results indicate or PHA). In contrast, the cyclin E and cyclin A that cyclin T1 is not a typical cell cycle regulatory associated activities are not induced by the action of cyclin. Thus, modulation of cyclin T1 levels are either mitogen alone. Nevertheless, when the two probably restricted to a number of speci®c cellular mitogens are used together and PBLs enter and signals and cell types. progress through the cell cycle, the levels of cyclin T1

Figure 4 Mitogen-dependent expression of cyclin T1. (a) Human PBLs were untreated or treated with PHA, PMA and PHA+PMA for 24 h or 48 h. Protein extracts were subjected to Western blot analysis by using anti-cyclin T1 antibodies. (b) The cell cycle distribution of PBLs treated as indicated in (a) was analysed by ¯ow cytometry (see Materials and methods). (c) Cyclin A kinase activity was determined from protein extracts of PBLs treated for 48 h as in (a). Cyclin A activity was determined using histone H1 as exogenous substrate on anti-cyclin A immunoprecipitates (see Materials and methods) Upregulation of cyclin T1/CDK9 during T cell activation J Garriga et al 3098

Figure 5 Expression of cyclin T1 is not modulated during the cell cycle in other cell types. T98G (a) and HaCaT (b) cells were serum starved and re-stimulated as described in the Materials and methods section and cells were harvested at the indicated time points (in hours). (c) 293 cells were arrested at the G1/S transition by adding hydroxyurea (HU) to the medium for 16 h. Progression through S phase was resumed by washing the cells and placing them in fresh medium. Cells were harvested at the indicated points post-hydroxyurea release (see Materials and methods). (d) 293 cells were incubated in the presence of nocodazole (NOC) for 24 h. Mitotic cells resumed the cell cycle after placing them in fresh medium without the drug. Cells were collected at the indicated time points (see Materials and methods). Cyclin T1 protein expression was analysed by Western blot using anti-cyclin T1 antibodies. The percentage of cells in each phase of the cell cycle was determined by ¯ow cytometric analysis and is indicated below its corresponding gel. As indicates asynchronous cells

and CDK9-associated kinase activities are increased poorly in non-stimulated PBLs producing less than one several fold, which is consistent with the higher levels syncytia per 104 cells at 72 h, which is consistent with of cyclin T1 present in these cells. In combination with the virus being in a non-integrated form (see the very low levels of cyclin T1 present in non- Discussion). Stimulation with PHA enhanced HIV-1 stimulated PBLs (Figures 3, 4 and 6), these results replication sevenfold and stimulation with PMA suggest that cyclin T1 is rate-limiting for activation of enhanced HIV-1 replication 33-fold over control non- CDK9, although the increase in CDK9 protein levels is stimulated PBLs (Figure 6b). Stimulation with PHA also likely to contribute to the activation of the com- and PMA together dramatically enhanced HIV-1 plex. In addition, these data indicate that although replication 530-fold over control non-stimulated PBLs both mitogenic pathways are sucient for activation of and 80-fold over PHA alone and 16-fold over PMA the cyclin T1/CDK9 complex, the two pathways act alone (Figure 6b). Thus, HIV replicated in both PMA synergistically to generate more cyclin T1/CDK9 and PHA treated PBLs signi®cantly better than in non- activity in proliferating T cells. activated PBLs, which correlates directly with the T cells are a primary target for HIV infection, and it higher levels of cyclin T1/CDK9 kinase activity has been shown that T cell activation induces HIV-1 detected in these cells and indicates that cellular LTR transcription strongly (Kaufman et al., 1987; DNA replication and cell division in T cells is not Nabel and Baltimore, 1987; Siekevitz et al., 1987; required for HIV replication. However, the much Tong-Starksen et al., 1987). Since the cyclin T1/CDK9 higher levels of cyclin T1/CDK9 kinase activity complex is a target of Tat in HIV infected T cells detected in PBLs treated with PMA and PHA together (Figure 2), and Tat activity is essential for HIV correlated with a several fold increase in HIV replication (Jones, 1997), we sought to ascertain replication (Figure 6b), suggesting that fully activated whether activation of cyclin T1 correlates with HIV T cells that have entered the cell cycle are more replication in PBLs. As described above for Figure 6a, competent for HIV replication. This is most relevant PBLs were unstimulated or stimulated with PMA, for resting memory T cells harboring integrated HIV PHA or PMA and PHA together for 48 h and then genomes, which are present in a latent form infected with HIV-1. HIV-1 replication was assessed by presumably due to lack of viral accessory proteins syncytia formation (Guan et al., 1996). and host cofactors (Finzi and Silliciano, 1998), among Quanti®cation of HIV replication in these cells is them, as we show here, the cyclin T1/CDK9 complex shown in Table 1 and Figure 6b. HIV-1 IIIB replicated (see Discussion). Upregulation of cyclin T1/CDK9 during T cell activation JGarrigaet al 3099

Figure 6 Activation of the cyclin T1/CDK9 complex requires a single mitogenic signal, although the combination of PMA and PHA exhibits a strong synergistic eect. The activation of cyclin T1/CDK9 activity correlates with HIV-1 replication in PBLs. Human PBLs were untreated or treated with PHA, PMA and PHA+PMA for 48 h. (a) Protein extracts were immunoprecipitated with s235, anti-CDK9, anti-cyclin A and anti-cyclin E antibodies. Kinase assays were performed on the immunoprecipitates using GST-CTD or histone H1 as exogenous substrates as indicated (see Materials and methods). (b) HIV replication correlates with upregulation of cyclin T1 in PBLs and with upregulation of the CDK9 activity associated with cyclin T1. PBLs treated for 48 h as indicated above were infected with HIV-1 IIIB, and HIV-1 replication was assessed by syncytia formation as described in the Materials and methods section. From the data in Table 1, HIV replication fold over non-stimulated PBLs at 72 h

Table 1 HIV-1 IIIB replication in stimulated and non-stimulated kinase activity in activated T cells reported previously PBLs (Yang et al., 1997). We also show for the ®rst time in Hours post- PMA+ vivo association of endogenous cyclin T1 and CDK9 infection Control a PMAa PHAa PHAa SUP-T1a with HIV-produced Tat in T cells, demonstrating that 24 0.05 0.05 0.5 11 38 Tat recruits cyclin T1/CDK9 complexes in relevant 48 0.25 4.5 1.0 25 125 cells and under physiological cellular conditions. 72 0.30 10.0 2.0 160 650 Moreover, we show that induction of cyclin T1 Syncytia formation in non-stimulated and stimulated PBLs as expression and activation of the cyclin T1/CDK9 measured by the infected centers assay (see Materials and complex does not require host DNA replication. methods). aSyncytia/104 cells However, the combination of PMA and PHA, which is mitogenic in PBLs, has a strong synergistic eect on the kinase activity of the cyclin T1/CDK9 complex and HIV replication (Figure 6a and b). Our data indicate that HIV-1 takes advantage of the physiological Discussion regulation of another host cofactor that is induced during T cell activation. We report here that cyclin T1 expression is dramati- One of the primary cellular targets for HIV infection cally upregulated by signals known to activate T cells is T cells, which are central to the pathogenesis of this and induce HIV transcription strongly. Upregulation virus. However, although quiescent T-cells are infected of cyclin T1 and to a lesser extent, CDK9, results in by HIV, these cells are refractory to HIV replication. high levels of cyclin T1/CDK9 activity and phosphor- This is mainly due to the impaired ability of the viral ylation of RNAP II in vivo, providing a mechanism to genome to integrate into the host genome because the explain induction of the GST-Tat-associated CTD viral RNA is not completely reverse transcribed Upregulation of cyclin T1/CDK9 during T cell activation J Garriga et al 3100 (Stevenson et al., 1990; Zack et al., 1990). In activated important to highlight the novelty of this aspect of the T cells, the viral DNA is integrated and HIV regulation of the cyclin T1/CDK9 complex in respect replication depends on viral accessory proteins and to other cyclin/CDK pairs. The dramatic induction of host factors (Cullen, 1998; Finzi and Silliciano, 1998). this complex during T cell activation suggests an A small pool of activated T cells harboring integrated important function for this complex in the control of HIV genomes appears to become resting memory T transcriptional elongation in a tissue-speci®c program cells, which act as reservoirs of the virus. These cells of gene expression. Whether the cyclin T1/CDK9 harbor an integrated latent viral genome and when they complex regulates the expression of particular genes are reactivated by the same antigen, HIV replication directly or is targeted to particular gene promoters by depends on the generation of those viral and host other cellular factors remains to be elucidated. factors that were limiting in resting memory T cells CDKs involved in cell cycle control are regulated by (Finzi and Silliciano, 1998). In this regard, it has been a number of independent mechanisms including shown that activation of T-cells dramatically enhances binding to a cyclin regulatory subunit, phosphoryla- HIV transcription (Kaufman et al., 1987; Nabel and tion/dephosphorylation of the CDK catalytic subunit Baltimore, 1987; Siekevitz et al., 1987; Tong-Starksen and binding to CKIs (GranÄ a and Reddy, 1995; et al., 1987). The HIV protein Tat is essential for HIV Morgan, 1997). Here we have shown that a single replication (Jones and Peterlin, 1994) and appears to mitogen (PMA or PHA) can induce cyclin T1 act by recruiting P-TEFb to the HIV promoter to expression. The same signals induce cyclins A and E increase the processivity of RNAP II and thus generate expression (Firpo et al., 1994). However, expression of full length HIV transcripts (Zhu et al., 1997). Our cyclin E or A is not sucient for activation of the results suggest that enhanced transcription of HIV in cyclin/CDK pair in either case (Firpo et al., 1994). In T activated T-cells is due, in addition to upregulation of cells, cyclin E/CDK2 activation is dependent upon the transcription factors such as NF-kB, to elevated levels combination of two dierent mitogenic signals, one of the cyclin T1/CDK9 complex, which in turn result in that upregulates the cyclin and the CDK subunits and hyperphosphorylation of RNAP II. In agreement with another that depletes p27 (Firpo et al., 1994). In the results reported here, it has been shown recently contrast, a single mitogen is sucient for activation of that the ability of P-TEFb to transactivate the HIV the cyclin T1/CDK9 complex in these cells, suggesting promoter depends on the kinase activity of CDK9, that upregulation of cyclin T1 is the main mechanism cyclin T1 and Tat (Fujinaga et al., 1998). for activation of the cyclin T1/CDK9 complex in these While this work was in progress a report was cells and that cyclin T1 acts as a truly rate-limiting published demonstrating that cyclin T1 associates positive regulatory subunit. directly with Tat in vitro in a manner that enhances the anity and speci®city of the Tat interaction to the TAR element of the nascent viral RNA (Wei et al., 1998). In addition, cyclin T1 was shown to rescue Tat Materials and methods activity when cotransfected with Tat in non-permissive rodent cells (Wei et al., 1998). Our work is consistent Cell culture with this study and extends these observations further HeLa and 293 cells were grown in DMEM supplemented by demonstrating that Tat targets the cyclin T1/CDK9 with 10% FCS. T98G and HaCaT cells were grown in complex in vivo in cells relevant to HIV replication and DMEM plus 10% FBS. The T-cell lines SUP-T1 and in physiological conditions. It remains to be deter- MOLT-4 IIIB were grown in RPMI 1640 with 10% FBS. mined however, whether other proteins are present in To obtain PBLs, blood was drawn from healthy HIV- seronegative donors, layered onto Histopaque-1077 this complex, and whether other Tat-associated kinase (Sigma) and centrifuged at 400 g for 45 min at room complexes exist. For example, it is not yet known temperature. The mononuclear layer was transferred to a whether cyclins T2a and T2b are in vivo targets of Tat. freshtube,washedtwicewithHank'sbalancedsalt We have not detected these complexes using currently solution, and ®nally resuspended in RPMI 1640 supple- available antibodies (data not shown). mented with 10% FBS. PBLs were obtained after Cyclin T1 is dramatically upregulated in PBLs upon monocyte depletion using plastic adherence and adjusted activation by either PMA or PHA. Induction of to 16106 cells/ml to immediately start the activation. For cyclin T1/CDK9 kinase activity occurs synergistically activation of PBLs, phytohemagglutinin (PHA) (Sigma) when the two signals are applied together (Figure 6a). was used at a ®nal concentration of 1 mg/ml and phorbol Interestingly, neither PMA, nor PHA (in contrast to 12-myristate 13-acetate (PMA) (Sigma) was used at 1 ng/ ml (Yang et al., 1997). TNF-a (Sigma) was used at 10 ng/ results reported earlier (Yang et al., 1997)) is capable ml. T98G and HaCaT cells were synchronized by serum by itself of inducing replication of cellular DNA in T deprivation and re-entered the cell cycle after re-stimula- cells, although they induce cell division when used tion with serum (Mayol et al., 1995, 1996). 293 cells were together. Upregulation of cyclin T1 is not linked to cell synchronized by hydroxyurea (Ashihara and Baserga, cycle entry and progression directly, because it does 1979) and nocodazole treatments (Mayol et al., 1996). not occur in response to serum stimulation even Flow cytometric analysis was performed essentially as though this treatment is sucient to stimulate exit described earlier (Mayol et al., 1995) by using an Epics from G0 and progression throughout the cell cycle in Elite system (Coulter Electronics Inc., Hiateah, FL, USA). many cell types. Cyclin T1 is constitutively expressed during entry and progression through the cell cycle of Protein assays at least two serum-responsive non T cell human cell Protein extracts were prepared by lysing cells in 50 mM lines and during the cell cycle of 293 cells. This is the Tris-Cl(pH7.4),5mM EDTA, 250 mM NaCl, 50 mM

®rst report showing upregulation of a cyclin regulatory NaF, 0.1% Triton X-100, 0.1 mM Na3VO4,2mM PMSF, subunit by a cell-type speci®c mitogenic pathway. It is 10 mg/ml leupeptin, 4 mg/ml aprotinin and 4 mg/ml pep- Upregulation of cyclin T1/CDK9 during T cell activation JGarrigaet al 3101 statin (lysis buer). Immunoprecipitations and Western detected in s235 immunoprecipitates). Since identical blots were performed as previously described (Mayol et al., results are obtained with the cyclin T1 antibody and 1995). Brie¯y, whole protein extracts or immunoprecipi- s235, and the origin of the s235 antibody is not essential tated complexes were resolved by SDS ± PAGE (Protogel. to the results reported here, further characterization of this National Diagnostics), transferred to PVDF membranes antibody will be reported elsewhere (Garriga and GranÄ a, (Immobilon-P, Millipore) in 10 mM CAPS (pH 11) con- unpublished). taining 10% methanol, and detected with horseradish peroxidase-conjugated second antibody (Amersham) and HIV-1 IIIB infections and replication assay Enhanced Chemiluminescence reagent (Dupont). Kinase assays were performed as described earlier (Garriga et al., Infections were performed essentially as described earlier 1996a,b; GranÄ a et al., 1994a). Brie¯y, kinase assays were (Guan et al., 1996). Brie¯y, HIV-1 IIIB was prepared from performed on anti-CDK9, s235, anti-cyclin A and anti- MOLT4 IIIB cell supernatants by ultracentrifugation and cyclin E immunoprecipitates at 308Cfor20minin20mM ®ltration and then titered on SUP-T1 cells by limiting HEPES-Na (pH 7.4), 10 mM magnesium acetate, 1 mM dilution. The infected centers assay was used to quantify DTT and 20 mM ATP (2 ± 4 104 c.p.m./pmol), in the HIV-1 replication in control and mitogen stimulated PBLs absence or presence of exogenous substrates (GST- as described (Guan et al., 1996). Brie¯y, PBLs were RNAP II-CTD or histone H1) in a total volume of 35 ml. infected with HIV-1 IIIB at a multiplicity of infection of The reaction was terminated by the addition of 35 mlof approximately 0.10. Following infection for 2 h the cells 26 Laemmli sample buer. GST-RNAP II-CTD was were washed to remove cell free virus and cultured in 24 obtained as previously described (GranÄ a et al., 1994b). In well plates. At various times after infection aliquots of cells vitro translated RAP74 was obtained by using the TNT were removed, washed, and serially diluted into 96 well reticulocyte system (Promega) and pET23d-RAP74 as plates where 26105 SUP-T1 cells were immediately added template (a gift from L Lei and Z Burton). to each well containing PBLs. Syncytia were screened at 48 h using an inverted microscope. The number of syncytia represents duplicate experiments and represents one cell Antibodies productively infected with HIV-1 (Guan et al., 1996). Anti-CDK9 (anti-PITALRE-CT) antibodies were raised to the C-terminal end of CDK9 and anity puri®ed. Anti- RNAP II (SC-900) and anti-RAP74-NT (SC-234) were from Santa Cruz Biotechnology. Anti-cyclin T1 and cyclin T2 were raised to GST fusion proteins (Peng et al.,

1998a). The B3 anti-RNAP II0 was a gift from B Blencowe Acknowledgements and M Mortillaro (Mortillaro et al., 1996). The Tat-1 We thank Elizabeth Moran for critical reading of the antibody was obtained through the AIDS Research and manuscript and many helpful discussions and Ana LimoÂ n Reference Reagent Program, Division of AIDS, NIAID, and Alex Tsygankow for excellent suggestions. We thank NIH (B Cullen). Anti-cyclin A was a kind gift from J Pines John Gibas for performing the ¯ow cytometric analysis and anti-cyclin E and anti-CDK2 were from A Ko. and Jonathan Pines, Andrew Ko, Ben Blencowe, Mitie s235 was raised to the C-terminus of RAP-74 (Santa Mortillaro, Bryan Cullen, Lei Lei, Zachary Burton, Matija Cruz Biotechnology, SC-235). This polyclonal antibody Peterlin and Tom Cujec for antibodies and plasmids. JG was found to recognize strongly and speci®cally the and MP were supported by fellowships from DireccioÂ n PITALRE-associated protein p95, which we demonstrate General de InvestigacioÂ nCientõÂ ®cayTeÂ nica (Ministerio de here to be cyclin T1 (see Results). The s235 antibody EducacioÂ n y Cultura, Spain). This work was supported by recognizes an epitope in cyclin T1 that shares limited a grant from the National Institute of General Medical identity with the C-terminus of RAP-74. Interestingly, the Sciences, NIH (GM54894) and partially by a WW Smith antibodies raised to this peptide exhibited much more grant (A9802) to XG and NIH grants CA/DO75909-01 anity for cyclin T1 that for RAP-74 (RAP-74 is not (EEH) and GM35500 (DP).

References

Ashihara T and Baserga R. (1979). Methods Enzymol., 58, Herrmann CH and Rice AP. (1995). J. Virol., 69, 1612 ± 248 ± 262. 1620. Cullen BR. (1998). Cell, 93, 685 ± 692. Jones KA. (1997). Genes Dev., 11, 2593 ± 2599. Dahmus ME. (1996). J. Biol. Chem., 271, 19009 ± 19012. Jones KA and Peterlin BM. (1994). Annu. Rev. Biochem., 63, Finzi and Silliciano RF. (1998). Cell, 93, 665 ± 671. 717 ± 743. Firpo EJ, Ko A, Solomon MJ and Roberts JM. (1994). Kaufmann JD, Valandra G, Roderiquez G, Bushar G, Giri C Mol. Cell. Biol., 14, 4889 ± 4901. and Norcross MA. (1987). Mol. Cell. Biol., 7, 3759 ± 3766. Fujinaga K, Cujec TP, Peng J, Garriga J, Price DH, GranÄ aX Lew DJ, Dulic V and Reed SI. (1991). Cell, 66, 1197 ± 1206. and Peterlin BM. (1998). J. Virol., 72, 7154 ± 7159. Liu ZJ, Ueda T, Miyazaki T, Tanaka N, Mine S, Tanaka Y, Garriga J, Mayol X and GranÄ a X. (1996a). Biochem. J., 319, Taniguchi T, Yamamura H and Minami Y. (1998). Mol. 293 ± 298. Cell. Biol., 18, 3445 ± 3454. Garriga J, Segura E, Mayol X, Grubmeyer C and GranÄ aX. Mancebo HS, Lee G, Flygare J, Tomassini J, Luu P, Zhu Y, (1996b). Biochem. J., 320, 983 ± 989. Peng J, Blau C, Hazuda D, Price D and Flores O. (1997). GranÄ a X, Claudio PP, De Luca A, Sang N and Giordano A. Genes Dev., 11, 2633 ± 2644. (1994b). Oncogene, 9, 2097 ± 2103. Marshall NF, Peng J, Xie Z and Price DH. (1996). J. Biol. GranÄ aX,DeLucaA,SangN,FuY,ClaudioPP,Rosenblatt Chem., 271, 27176 ± 27183. J, Morgan DO and Giordano A. (1994a). Proc. Natl. Marshall NF and Price DH. (1995). J. Biol. Chem., 270, Acad. Sci. USA, 91, 3834 ± 3838. 12335 ± 12338. GranÄ a X and Reddy EP. (1995). Oncogene, 11, 211 ± 219. Mayol X, Garriga J and GranÄ a X. (1995). Oncogene, 11, Guan M, Zhang RD, Wu B and Henderson EE. (1996). J. 801 ± 808. Virol., 70, 7341 ± 7346. Mayol X, Garriga J and GranÄ a X. (1996). Oncogene, 13, Herrmann CH and Rice AP. (1993). Virol., 197, 601 ± 608. 237 ± 246. Upregulation of cyclin T1/CDK9 during T cell activation J Garriga et al 3102 Morgan DO. (1997). Ann. Rev. Cell. Dev. Biol., 13, 261 ± 291. Tong-Starksen SE, Luciw PA and Peterlin BM. (1987). Proc. Mortillaro MJ, Blencowe BJ, Wei X, Nakayasu H, Du L, Natl. Acad. Sci. USA, 84, 6845 ± 6849. Warren SL, Sharp PA and Berezney R. (1996). Proc. Natl. WeiP,GarberME,FangSM,FischerWHandJonesKA. Acad. Sci. USA, 93, 8253 ± 8257. (1998). Cell, 92, 451 ± 462. Nabel G and Baltimore D. (1987). Nature, 326, 711 ± 713. Xiong Y. (1996). Biochim. Biophys. Acta, 1288, 1±5. Nekhai S, Shukla RR and Kumar A. (1997). J. Virol., 71, Yang X, Gold MO, Tang DN, Lewis DE, Aguilar-Cordova 7436 ± 7441. E, Rice AP and Herrmann CH. (1997). Proc. Natl. Acad. Peng J, Marshall NF and Price DH. (1998b). J. Biol. Chem., Sci. USA, 94, 12331 ± 12336. 273, 13855 ± 13860. Zack JA, Arrigo SJ, Weitsman SR, Go AS, Haislip A and Peng J, Zhu Y, Milton JT and Price DH. (1998a). Genes Dev., Chen IS. (1990). Cell, 61, 213 ± 222. 12, 755 ± 762. ZhuY,Pe'eryT,PengJ,RamanathanY,MarshallN, Siekevitz M, Josephs SF, Dukovich M, Peer N, Wong-Staal MarshallT,AmendtB,MathewsMBandPriceDH. F and Greene WC. (1987). Science, 238, 1575 ± 1578. (1997). Genes Dev., 11, 2622 ± 2632. Stevenson M, Stanwick TL, Dempsey MP and Lamonica CA. (1990). EMBO J., 9, 1551 ± 1560.