Interaction between P-TEFb and the C-Terminal Domain of RNA Polymerase II Activates Transcriptional Elongation from Sites Upstream or Downstream of Target Genes

Transcriptional elongation by RNA polymerase II (RNAPII) isregulated by the positive transcription elongation factor b(P-TEFb). P-TEFb is composed of Cdk9 and C-type cyclin T1 (CycT1),CycT2a, CycT2b, or CycK. The role of the C-terminal region ofCycT1 and CycT2 remains unknown. In this report, we demonstratethat these sequences are essential for the activation of transcriptionby P-TEFb via DNA, i.e., when CycT1 is tethered upstream ordownstream of promoters and coding sequences. A histidine-richstretch, which is conserved between CycT1 and CycT2 in thisregion, bound the C-terminal domain of RNAPII. This bindingwas required for the subsequent expression of full-length transcriptsfrom target genes. Thus, P-TEFb could mediate effects of enhancerson the elongation of transcription.

INTRODUCTION

Top

Introduction

Transcriptional elongation by RNA polymerase II (RNAPII) isa highly regulated process between negative transcription elongationfactors (N-TEF), which pause RNAPII, and the positive transcriptionelongation factor b (P-TEFb), which facilitates its elongation(5, 45, 48). Transcription begins with the assembly of the preinitiationcomplex (PIC) on the promoter, which includes core elements(CP) and promoter-proximal regions (PPR) (4). CP consists ofa TATA box, a TFIIB recognition element, an initiator (INR),and downstream promoter elements. PPR contain upstream sequences,where Sp1 and other activators bind (4). They recruit the PICthrough general transcription factors (GTF; TFIIA to -J) andmediators (SRB, etc.), which interact with the nonphosphorylatedC-terminal domain (CTD) of RNAPII and mediate the initiationof transcription (23, 31, 39, 47). TFIIH helps RNAPII to clearthe promoter and partially phosphorylates the CTD (17, 56).Thus, transcription is initiated, but further elongation isblocked by N-TEF (11, 45), which includes at least two differentcomplexes, the 5,6-dichloro-1ß-d-ribofuranosylbenzimidazole(DRB) sensitivity-inducing factor (DSIF) and the negative elongationfactor (NELF) (19, 21, 57, 62). DSIF contains SPT4 (16 kDa)and SPT5 (160 kDa) (19, 21, 57). NELF is composed of four subunits,of which the smallest, NELF-E (or RD), contains an RNA recognitionmotif (62). At present, the mode of action of N-TEF is not wellunderstood. Somehow it pauses RNAPII, resulting in arrestedtranscription (28, 44, 58, 61, 62). Finally, it is P-TEFb, consistingof Cdk9 and cyclin T1, T2a, T2b, or K (CycT1, differentiallyspliced CycT2a and CycT2b, and CycK), that overcomes effectsof N-TEF and releases RNAPII from its arrest (8, 42, 45). P-TEFbphosphorylates the CTD, DSIF, and possibly other targets inthese arrested complexes, thus catalyzing the transition frominitiation to elongation of transcription (8, 11, 21, 27, 45,59, 60).

Transcriptional enhancer sequences (enhancers) are defined ascis-acting DNA elements that increase levels of gene expression(4, 6, 36). Their effects are dependent on PPR and CP but independentof location, orientation, and distance with respect to the siteof initiation of transcription (4, 6, 36). They can functionindependently of chromatin, i.e., on episomal plasmids and inin vitro transcription systems (49). Moreover, their effectsare dependent on the CTD of RNAPII, which in primates contains52 repeats of the heptapeptide YSPTSPS. Whereas sequences with52 and 31 repeats support effects of different enhancers, thosewith 5 repeats have been found to be inactive in these assays(14). Thus, the CTD functions as a surface for the integrationof different signals from activators and repressors on transcription.When unphosphorylated, the CTD is covered by mediators and recruitscoactivators (50, 51). In sharp contrast, the phosphorylatedCTD plays a central role in the cotranscriptional processingof mRNA by binding to the elongator complex (2). To date, anaccepted model of how enhancers function is not available. Thelooping model suggests that proteins bound to enhancers interactdirectly with the PIC on promoters. Alternatively, the scanningmodel speculates that enhancer-binding proteins travel alongthe DNA until they meet their cognate promoter (4).

A well-defined system that incorporates important features ofeukaryotic promoters and enhancers and reveals the central rolethat P-TEFb plays in the elongation of transcription by RNAPIIis represented by the human immunodeficiency virus (HIV). Onthe HIV long terminal repeat (LTR) are found binding sites forTFE3, LEF-1, NF-AT, NF- ${kappa}$ B, Sp1, TFIID, TFIII, and LBP-1 (24).Thus, there exists a classical arrangement of the enhancer,PPR, and CP (24). In the absence of enhancer binding proteinsand Tat, the transcription of HIV is arrested at the transactivationresponse (TAR) element stem loop RNA, which is present at the5' ends of all viral transcripts (24). With the help of P-TEFb,Tat binds to TAR and promotes the elongation rather than theinitiation of HIV transcription (3, 9, 13, 22, 24, 29, 59).NF- ${kappa}$ B and other activators that bind to the HIV enhancer cansubstitute for the function of Tat (18) and/or act synergisticallywith it to increase levels of viral replication (40, 52). NF- ${kappa}$ Balso increases the expression of the immunoglobulin light chainkappa (Ig ${kappa}$ ) gene (43). As in HIV, RNAPII is engaged but doesnot elongate past this cellular promoter. The binding of NF- ${kappa}$ Bto Ig ${kappa}$ enhancer then releases RNAPII from its pausing site (46).A similar regulation of transcription is found in the heat-shockgenes from Drosophila melanogaster and the human proto-oncogenec-myc (32, 55). Tat transactivation also resembles enhancersin its requirement for the CTD of RNAPII for its effects (38).Thus, it is not surprising that the major histocompatibilitycomplex class II (MHC II) transactivator (CIITA), which mediateseffects of B-cell-specific and gamma interferon-inducible enhancers(54) and NF- ${kappa}$ B (1), also binds P-TEFb.

Thus, we hypothesized that P-TEFb plays a general role in mediatingeffects of enhancers. Indeed, in this report, we demonstratethat P-TEFb activates transcription from proximal and distalsites upstream and downstream of the promoter and coding sequenceof a reporter gene. PPR, CP, and the CTD of RNAPII were requiredfor these effects of P-TEFb. A direct correlation between thebinding of a histidine-rich stretch in the C-terminal regionof CycT1 and the CTD could be made with this transcriptionalactivation, which also required the kinase activity of Cdk9.Importantly, adding these sequences of CycT1 to CycK, whichlacks a long C-terminal extension, allowed this P-TEFb complexto activate transcription from the same distal sites. Theseeffects were on the elongation rather than initiation of transcription,indicating that PPR and CP were sufficient to recruit and positionRNAPII. We conclude that the interaction between C-terminalregions of RNAPII, CycT1, and distal sites enables P-TEFb toact as a transcriptional coactivator. Thus, P-TEFb can linkactivators at enhancers and PIC at promoters, which resultsin the productive elongation of transcription.

MATERIALS AND METHODS

Top

Materials and Methods

Plasmid constructions. The plasmid reporter pG6(5'Pro) was described previously (15).It contained six upstream activation sequence (UAS) repeats,which bind to residues 1 to 147 of the Gal4 DNA binding domain(DBD). UASs were located upstream of the three Sp1 sites, atposition -119 of the HIV type 1 (HIV-1) LTR, relatively to thestarting point of transcription. No TAR sequences were presentin pG6(5'Pro). The HIV LTR was fused to the chloramphenicolacetyltransferase (CAT) reporter gene. pG6(3'Pro) containedthe same sequences as pG6(5'Pro), the only difference beingthat UAS sites were placed 3' to CAT and promoter sequences.pG5 reporter plasmid contained the EIb TATA box and was describedpreviously (53).

The different plasmid effectors of Gal-CycT1 code for the chimerasof the full-length CycT1 [Gal-CycT(1–726)] and truncatedforms Gal-CycT(1–590), Gal-CycT(1–420), and Gal-CycT(1–300)(see Fig. 1A), all fused at their N termini to the DBD of Gal4(1–147)and inserted at the BamHI/XbaI sites of pEF-Bos-myc(EF). pSG424-Gal-CycK(1–357)and pSG424-Gal-CycK(1–357)T(300–726) code for thechimeras of residues 1 to 357 CycK and of the same portion ofCycK fused to CycT1 residues 300 to 726, respectively, linkedto the Gal4 DBD protein at their N termini, and inserted atthe EcoRI/XbaI sites of pEF-Bos-myc(EF). The plasmid effectorsGal-Cdk9 and Gal-Cdk9(D167N) expressed the chimeras of Cdk9and its kinase-deficient mutant Cdk9(D167N) fused at their Ntermini to the DBD of Gal4 and inserted at the BamHI/XbaI sitesof pEF-Bos-myc(EF). pSG424-Gal-VP16 codes for the activationdomain of VP16 fused to Gal4 DBD at its N terminus. For in vitrotranscription and translation, the different CycT1 truncationforms were subcloned into pEF-Bos-T7 in the BamHI/EcoRI sites.For in vitro glutathione S-transferase (GST) pull-down assays,CycT1 and its C-terminal truncation forms were cloned into pGEX-2TK(Amersham Pharmacia Biotech, Piscataway, N. J., Madison, Wis.)in the BamHI/EcoRI sites to express GST fusion proteins of CycT1as described elsewhere (10).

View larger version (41K):
[in this window]
[in a new window]
FIG. 1. The C-terminal region of CycT1 is required for P-TEFb to activate transcription via DNA. (A) Plasmid effectors and targets used in this study. The plasmid effector GT(1–726) directed the expression of the full-length CycT1 protein from the EF-Bos promoter. CycT1 was fused at its N terminus to the Gal4 DBD (Gal, positions 1 to 147) and the c-Myc epitope tag (myc, black bar). Below the full-length chimera are Gal fusion proteins with the C-terminal truncations of CycT1. Also presented are the plasmid effectors GCdk9 and GCdk9(D167N), which expressed wild-type Cdk9 (striped bar) and the kinase-deficient mutant Cdk9(D167N) (white bar) fused at their N termini to Gal (gray bar). Also depicted are the EF-Bos promoter (EF), two cyclin boxes (positions 1 to 250, black box) of CycT1, and the polyadenylation site (pA) from simian virus 40. The plasmid target pG6(5'Pro) contained the HIV LTR linked to the CAT reporter gene. Six repeats of the synthetic Gal4-binding sites (UAS) were introduced upstream of the three Sp1 sites. These UAS repeats replaced the enhancer (EN) in the HIV LTR. Also presented is the promoter, which contained the CP, i.e., the TATA box and INR sequences, and the PPR, i.e., Sp1 sites, all placed 5' to the CAT reporter gene (white bar). The arrow indicates the site of initiation of transcription. The plasmid target pG6(3'Pro) was identical to pG6(5'Pro) except that the UAS repeats were placed 3' to CAT and promoter sequences. (B) P-TEFb activates transcription when presented upstream of the promoter via DNA. HeLa cells coexpressed pG6(5'Pro) and the indicated hybrid Gal-CycT1 (GT), Gal-Cdk9 (GCdk9), and dominant-negative Gal-Cdk9(D167N) proteins. The Gal4 DBD (Gal) was also expressed alone. CAT activities were measured as described in Materials and Methods. Data are relative to basal levels of the reporter gene alone (lane 1). They represent two independent experiments, and standard errors of the means are given. (C) P-TEFb activates transcription when presented via DNA downstream of the promoter and coding sequence. HeLa cells coexpressed pG6(3'Pro) with the indicated hybrid Gal-CycT1 proteins, wild-type Gal-Cdk9, and kinase-deficient Cdk9(D167N). CAT activities are relative to basal levels of the reporter gene alone (lane 1). They represent two independent experiments, and standard errors of the means are given. (D) Western blotting analysis with anti-Myc antibody revealed expression levels of hybrid Gal-CycT1 (lanes 3 to 6), wild-type Gal-Cdk9 (lane 7), and dominant-negative Gal-Cdk9(D167N) proteins (lane 8).

Binding assays. Assays for the direct binding between the hybrid GST-CTD anddifferent truncation proteins of CycT1 were performed as describedpreviously (10). In brief, 5 µg of the hybrid GST-CTDwas incubated at 4°C for 3 h in the presence of ³⁵S-labeledin vitro transcribed-translated CycT1 or its C-terminal truncationforms (Promega) in 300 µl of binding buffer (buffer A)consisting of 20 mM HEPES-KOH (pH-7.9), 0.1% Triton X-100, 0.05%NP-40, 0.3% bovine serum albumin, 5 mM dithiothreitol, 5 mMEDTA, and 200 mM KCl. Following incubation, binding reactionproducts were washed extensively with buffer A containing 500mM KCl. Washed beads were subjected to sodium dodecyl sulfate-polyacrylamideelectrophoresis (SDS-PAGE) and analyzed by autoradiography.

For interaction between CycT1 and the endogenous RNAPII, HeLanuclear extracts were incubated with the different GST-CycT1fusion proteins at 4°C for 3 h in ELB binding buffer (50mM HEPES-KOH [pH 7.9], 0.1% Triton X-100, 5 mM dithiothreitol,5 mM EDTA, 150 mM NaCl). Following incubation, reaction productswere washed extensively with ELB binding buffer containing 500mM NaCl. Beads were subjected to 5% SDS-PAGE, followed by Westernblotting with anti-RNAPII antibody (8WG16; BAbCO, Richmond,Calif.).

For interaction between CycT1 and the CTD of RNAPII, Raji cells(35) (kind gift of D. Eick) expressing hemagglutnin (HA) epitope-taggedRNAPII, where the CTD contains 52 or 5 repeats of the heptapeptidesequence, were transfected with 30 µg of effector plasmids.Lysates were prepared from these cells with ELB buffer and wereimmunoprecipitated with anti-Myc antibody (Santa Cruz Biotechnology,Santa Cruz, Calif.). Proteins were subjected to SDS-PAGE, transferredto a nitrocellulose membrane, and probed with anti-HA antibody(Santa Cruz Biotechnology).

Transient-transfection assays. HeLa cells were transfected with target plasmids (0.2 µg)with or without the different effector plasmids (1 µg)with Lipofectamine as recommended by the manufacturer (GIBCO/BRL,Gaithersburg, Md.). All transfections were balanced to totalof 1.2 µg of DNA with pEF-Bos-myc plasmid DNA. At 48 hafter transfection, cells were lysed in lysis buffer (250 mMTris-HCl [pH 7.5] and 0.1% Triton X-100), and CAT activitieswere measured by a liquid scintillation assay as described previously(9). Western blot analysis was performed to evaluate expressionlevels of the proteins (see Fig. 1D). For the inhibition ofkinase activity of Cdk9, cells were treated with DRB (Sigma,St. Louis, Mo.) (diluted in dimethyl sulfoxide) at various concentrations,24 h before and 48 h after transfection. CAT activities weremeasured from cell lysates as described previously (9).

Raji cells expressing the different RNAPII CTD were grown inRPMI 1640 medium supplemented with 10% fetal calf serum (FCS),100 µg of streptomycin per ml, 2 mM L-glutamine, 1 mgof G418 per ml, and 0.1 µg of tetracycline per ml. Beforetransfection, cells were washed three times with medium containingonly 1% FCS and were grown for an additional 1 day with mediumcontaining 10% FCS but no tetracycline. This enabled the expressionof the different RNAPII-CTD forms. The next day, the differentCycT1 plasmid effectors were transfected by electroporation(250V; 950 µF), and cells were grown for an additional2 days. Cells were lysed with ELB buffer, and the lysates weresubjected to immunoprecipitation with the anti-Myc antibodyfollowed by Western blotting with anti-HA antibody (35).

Immunoprecipitation and Western blotting. For determining the expression levels of proteins, 293T cellswere transfected with various plasmids (2 µg). At 24 hafter transfection, cells were lysed with ELB buffer. Proteinswere subjected to SDS-PAGE, transferred onto Immobilon-NC membranes(Millipore, Bedford, Mass.), and reacted with the appropriateantibodies. Proteins were visualized by enhanced chemiluminescence(Amersham Inc., Evanston, Ill.).

For immunoprecipitation experiments, cells were lysed with ELBbuffer and immunoprecipitated with the appropriate antibody.Following binding to the antibody, reaction mixtures were incubatedwith protein A-Sepharose beads at 4°C. Immunoprecipitatereaction products were washed extensively with ELB buffer andsubjected to SDS-PAGE followed by Western blotting.

RPA. COS cells were transfected with pG6TAR(5'Pro) (10 µg)and with Gal-CycT1 fusion proteins and C-terminal-truncationforms (10 µg). At 24 h after transfection, cells wereharvested and RNA was purified. A 30-µg portion of RNAwas used for RNase protection assays (RPA). Antisense probesMTX-147 and MTX-89 (kind gift of J. Karn) were linearized withXbaI and BamHI, respectively. Probes were transcribed in thepresence of [ ${alpha}$ -³²P]UTP by using MAXIscript (Ambion, Austin, Tex.)with T3 RNA polymerase to produce labeled RNA probe. Followingsynthesis, probes were subjected to polyacrylamide-urea gelelectrophoresis and purified before use. RPA were performedusing the RPAIII kit as recommended by the manufacturer (Ambion).Protected RNA fragments were separated on 11% polyacrylamide–ureasequencing gels. Gels were dried and subjected to autoradiography(38, 60).

RESULTS

Top

Results

P-TEFb can activate transcription from sites upstream and downstream of the promoter and coding sequence of a target gene. To evaluate the hypothesis that P-TEFb can mediate effects ofenhancers on the elongation of transcription, we fused CycT1and Cdk9 to the DBD of Gal4 (Fig. 1A). These chimeras allowedus to present P-TEFb via UAS that bind Gal. This artificialDNA tethering reproduced the recruitment of P-TEFb by CIITAand RelA (1, 25) and should mimic the scenario with many as-yet-uncharacterizedactivators. Full-length CycT1 and its C-terminally truncatedderivatives, fused to Gal, were then examined for their abilityto activate transcription. We used a reporter gene which containedsix UAS repeats located upstream of the HIV promoter, whereTAR was deleted, and the CAT reporter gene [Fig. 1A, pG6(5'Pro)].Importantly, these UAS repeats replaced the NF- ${kappa}$ B binding sites,which represent the enhancer in the HIV LTR. Gal-CycT1 chimerascontaining the wild-type 726 (Fig. 1B, lane 3) or N-terminal590 (lane 4) residues activated transcription from pG6(5'Pro)35- to 40-fold above basal levels. Further C-terminal truncationsof CycT1 reduced the ability of the hybrid Gal-CycT1 proteinsto activate transcription from pG6(5'Pro). Thus, Gal-CycT1 fusionproteins containing the N-terminal 420 (Fig. 1B, lane 5) or300 (lane 6) residues displayed little or no activity on pG6(5'pro).Importantly, levels of expression of our mutant chimeras wereeven higher than that of the wild-type fusion protein (Fig.1D, compare lanes 4, 5, and 6 to lane 3). We conclude that theC-terminal region of CycT1 is essential for the ability of CycT1to activate transcription via DNA. Furthermore, the hybrid Gal-Cdk9protein (Fig. 1B, lane 7) but not a kinase-deficient hybridGal-Cdk9 protein (lane 8) activated transcription 50-fold fromthe same reporter gene. Levels of activation by the Gal-Cdk9chimera were similar to levels of activation by the Gal-VP16fusion protein (data not presented). These results demonstratethat the direct recruitment of an active Cdk9 to a site upstreamof a promoter activates transcription. We conclude that longCycT1 as well as Cdk9 can function when presented via DNA, indicatingthat P-TEFb mediates transcriptional effects via the C-terminalregion of CycT1.

Given these results and since there is no directionality toUAS repeats, we hypothesized that P-TEFb should also functionwhen placed at some distance downstream of a target gene. Toexamine this notion, six UAS repeats were positioned 3' of theCAT gene, distant from the promoter and coding sequence [Fig.1A, pG6(3'Pro)]. In this system, only the Gal-CycT(1–726)(Fig. 1C, lane 3) and Gal-CycT(1–590) (lane 4) chimerasactivated transcription via pG6(3'Pro). Levels of activationwere 25- to 30-fold above basal levels. C-terminal deletionsof CycT1, i.e., the hybrids Gal-CycT(1–420) (Fig. 1C,lane 5) and Gal-CycT(1–300) (lane 6), had lower or noactivity on this plasmid target. Moreover, the Gal-Cdk9 chimera(Fig. 1C, lane 7) but not the kinase-deficient Gal-Cdk9 protein(lane 8) also supported transcriptional activation from thisdistal site. Similar to the results with pG6(5'Pro), the C-terminalregion of CycT1 was required for effects of our chimeras onpG6(3'Pro). We conclude that P-TEFb can function via sites whichare located upstream or downstream of and at a distance fromthe promoter and coding sequence.

Sp1 is required for the activation of transcription by P-TEFb when presented via DNA. As P-TEFb could replace NF- ${kappa}$ B when UAS were placed just upstreamof the three Sp1 sites on the HIV LTR, we examined whether PPRis required for P-TEFb to mediate its effects. It should benoted that Sp1 sites are required for NF- ${kappa}$ B to activate HIV transcription(24) and for the function of the 72-bp repeats of simian virus40 (20). To this end, five UAS repeats were placed next to theCP (E1b TATA box) and upstream of the CAT reporter gene (Fig.2A, pG5). Importantly, no Sp1 sites were present in pG5. Noneof our Gal-CycT1 fusion proteins activated transcription frompG5 (Fig. 2B, lanes 1 to 6). However, the Gal-VP16 chimera activatedtranscription via the same plasmid target (Fig. 2B, lane 7).Moreover, deletions of Sp1 sites in pG6(5'Pro) gave the sameresult, with three sites functioning optimally and their progressiveremoval resulting in no activity (data not presented). Thissuggests that Sp1 and CP recruit the PIC to initiate transcription.We conclude that PPR and CP are essential for the ability ofP-TEFb to activate transcription by RNAPII via DNA.

View larger version (25K):
[in this window]
[in a new window]
FIG. 2. Proximal promoter elements are required for the activation of transcription by P-TEFb via DNA. (A) Depicted are the hybrid GT(1–726) protein, C-terminal truncations of CycT1, and the plasmid target, pG5, which contained the E1b TATA box placed upstream of the CAT reporter gene. (B) HeLa cells coexpressed pG5 and our Gal-CycT1 (GT) chimeras (Fig. 1A). The Gal-VP16 fusion protein (GVP16) was used as a positive control. Results are relative to the basal level (lane 1) and represent two independent experiments, where standard errors of the means are given.

RNAPII interacts with the C-terminal region of CycT1 in vivo. These results indicated that C-terminal residues of CycT1 promotetranscriptional activation via distal sites. These interactionscould occur by the looping of DNA. Indeed, some activators thatbind to enhancers and coactivators that interact with activatorsand PIC also facilitate the bending of DNA (4, 36). The CTDof RNAPII is also required for these effects (14, 50). Accordingly,by interacting with the CTD, P-TEFb could function as a coactivatorthat links enhancers to the PIC. This interaction could leadto the phosphorylation of the CTD by Cdk9 and the activationof transcription.

To evaluate this hypothesis, we examined whether P-TEFb interactswith RNAPII. Indeed, both CycT1 and Cdk9 coimmunoprecipitatedthis polymerase in HeLa cells (Fig. 3A, lanes 2 and 3). Moreover,we mapped the region of CycT1 that binds RNAPII (Fig. 3B). Invitro binding reactions using nuclear extracts and GST-CycT1fusion proteins demonstrated that neither GST alone (Fig. 3B,lane 1) nor the fusion protein containing the N-terminal 300residues (lane 8) interacted with RNAPII. The same was truefor GST-CycT1 chimeras containing the first 480 residues (Fig.3B, lane 7) and residues from positions 611 to 726 (lane 9).In contrast, full-length CycT1 (Fig. 3B, lane 2) and GST-CycT1fusion proteins containing the 551 N-terminal residues (lane3) or only sequences from positions 231 to 611 (lane 4) and350 to 611 (lane 5) bound RNAPII. Moreover, a region of thehybrid GST-CycT1 protein from positions 481 to 551 (Fig. 3B,lane 6) still interacted with RNAPII. This region contains ahistidine-rich stretch that could mediate the binding betweenCycT1 and RNAPII. Other GST-CycT1 chimeras that contained these70 residues also binds RNAPII (Fig. 3B, lanes 4 and 5). We concludethat the C-terminal region of CycT1 between positions 481 and551 interacts with RNAPII in vivo and with nuclear extractsin vitro.

View larger version (36K):
[in this window]
[in a new window]
FIG. 3. P-TEFb interacts with RNAPII via the C-terminal region of CycT1 in vivo and in vitro. (A) P-TEFb interacts with RNAPII in vivo. Total nuclear extracts from HeLa cells were immunoprecipitated with anti-Cdk9 (lane 2) or anti-CycT1 (lane 3) antibodies. One tenth of the nuclear extract was used as an input control (lane 1). Preimmune serum was used as a control for immunoprecipitations (lane 4). Samples were separated by SDS-PAGE, transferred to a membrane, and probed with the antibody 8WG16, which recognizes RNAPII. (B) CycT1 interacts with RNAPII in vitro. GST-CycT1 chimeras that were examined for their interactions with the endogenous RNAPII are depicted. Also presented are important elements in CycT1: the cyclin box, the Tat-TAR recognition motif (TRM), the coil domain, the histidine-rich stretch (His), and PEST sequences. Equal amounts of the indicated GST-CycT1 fusion proteins, which were bound to glutathione-Sepharose beads (lanes 2 to 9), and the control GST alone were incubated with HeLa nuclear extracts as described in Materials and Methods. After a GST pull-down assay, Western blotting was performed with the 8WG16 antibody against RNAPII.

The C-terminal region of CycT1 binds the CTD of RNAPII in vitro and in vivo. Next, we examined the binding between a GST-CTD fusion proteinand ³⁵S-labeled CycT1 protein, which was transcribed and translatedwith rabbit reticulocyte lysate in vitro. Only full-length CycT1(Fig. 4A, lane 2) and CycT1 containing the N-terminal 590 residues(lane 3) bound the GST-CTD chimera. Since the wild-type CycT(1–726)did not bind GST alone, this interaction was specific (Fig.4A, lane 1). Moreover, mutant CycT1 proteins containing theN-terminal 420 (Fig. 4A, lane 4) and 300 (lane 5) residues alsodid not bind the GST-CTD fusion protein. We conclude that theCTD of RNAPII binds the C-terminal residues of CycT1 in vitro.

View larger version (39K):
[in this window]
[in a new window]
FIG. 4. The C-terminal region of CycT1 binds the CTD of RNAPII in vitro and in vivo. (A) The C-terminal region of CycT1 binds the CTD in vitro. The indicated ³⁵S-labeled CycT1 proteins, which were transcribed and translated with rabbit reticulocyte lysate in vitro, were incubated with the GST-CTD fusion proteins bound to glutathione-Sepharose beads (lanes 2 to 5). As a control, GST alone was incubated with CycT(1–726). Binding reactions and steps leading to autoradiography were performed as described in Materials and Methods. CycT1 inputs represent 10% of amounts that were used for the binding reactions. (B) The C-terminal region of CycT1 interacts with the CTD in vivo. Raji cells stably expressing an HA epitope-tagged and ${alpha}$ -amanitin-resistant RNAPII, which contained 52 or 5 repeats of the heptapeptide YSPTSPS of the CTD, coexpressed Gal-CycT(1–590) or Gal-CycT(1–300). Lysates prepared from these cells were immunoprecipitated with the anti-Myc antibody, subjected to SDS-PAGE, transferred to membranes, and probed with the anti-HA antibody. As a control, the Gal-CycT(1–590) chimera was expressed in parental Raji cells that do not express the HA epitope-tagged RNAPII molecules (Raji-LS). Lanes 1 to 3 represent the inputs of the different HA epitope-tagged RNAPII molecules, expressing 52 (wild type [WT]) or 5 (truncated [ ${Delta}$ 5]) repeats of the CTD. Arrows indicate the wild-type and truncated large subunit of RNAPII (RPB1). Below the Western blots are the wild-type HA epitope-tagged RPB1 and ${Delta}$ 5 RPB1. The synthesis of the HA epitope-tagged and ${alpha}$ -amanitin resistant ( ${alpha}$ AR) RNAPII was directed by the cytomegalovirus (CMV) promoter.

To correlate our binding studies in vitro with functional resultsin cells, we used Raji B cells that stably express HA epitope-taggedRNAPII with different lengths of the CTD, i.e., 52 or 5 repeatsof the heptapeptide YSPTSPS (35). Previous studies demonstratedthat only RNAPII containing 52 or 31 repeats of the CTD supportedTat transactivation and effects of enhancers (14, 38). Therefore,we performed immunoprecipitations using the hybrid Gal-CycT(1–590)and Gal-CycT(1–300) proteins, respectively. Only the N-terminal590 residues of CycT1 were included because this fusion proteindid not contain the C-terminal PEST region and was expressedat higher levels than the wild-type protein (Fig. 1D, comparelanes 3 and 4). Gal-CycT(1–590) but not Gal-CycT(1–300)interacted with the full-length CTD (Fig. 4B, compare lanes5 and 7). Moreover, neither protein bound the RNAPII with only5 repeats (Fig. 4B, compare lanes 6 and 8). Coexpression studiesalso revealed that only the full-length CTD supported the functionof our Gal-CycT1 and Gal-Cdk9 chimeras on pG6(5'Pro) and pG6(3'Pro)in these Raji B cells (data not presented). We conclude thatthe C-terminal region of CycT1 binds the CTD of RNAPII in vitroand in vivo.

C-terminal region of CycT1 rescues the ability of CycK to mediate transcriptional activation. Based on these results, the CTD appeared to be the key targetfor the C-terminal region of CycT1. The binding of these twodomains was documented in several different ways and mappedto the histidine-rich stretch in CycT1. To prove the functionalsignificance of this interaction, we examined the ability ofa Gal-CycK chimera to activate transcription from distal sitesvia DNA. The same plasmid targets were used (Fig. 5A). By bindingCdk9, CycK forms an alternative P-TEFb complex (8). However,it contains only 357 residues, which include the N-terminalcyclin boxes but not the long C-terminal region of CycT1 andCycT2. Neither Gal-CycT(1–300) (Fig. 5B and C, lanes 4)nor Gal-CycK (lanes 5) activated transcription. In contrast,a Gal-CycK chimera which was fused to the C-terminal regionof CycT1 (Fig. 5B and C, lanes 6) activated transcription whenpresented via DNA to levels of the Gal-CycT(1–726) fusionprotein (lanes 3). Thus, the C-terminal residues of CycT1 rescuedthe ability of Gal-CycK fusion protein to support transcriptionalactivation from distal sites upstream and downstream of thepromoter and coding sequence.

View larger version (42K):
[in this window]
[in a new window]
FIG. 5. Addition of the C-terminal region of CycT1 to CycK rescues the activity of the Gal-CycK chimera via DNA. (A) Plasmid effector GT(1–726), which directed the expression of the Gal-CycT(1–726) fusion protein. Also illustrated are the hybrid proteins in which the N termini of CycK or of the hybrid CycK-CycT(300–726) protein were fused to the Gal4 DBD. Plasmid targets pG6(5'Pro) and pG6(3'Pro) (Fig. 1A) are also shown. (B) Addition of the C-terminal region of CycT1 to Cyc K rescues the activity of the Gal-CycK chimera from sites upstream of the promoter. HeLa cells coexpressed pG6(5'Pro) and Gal-CycT(1–726), truncated Gal-CycT(1–300), Gal-CycK(1–357), and Gal-CycK(1–357)-CycT(300–726) chimeras. Data are relative to basal levels of the reporter gene alone and represent two independent experiments; standard errors of the means are given. (C) Addition of the C-terminal region of CycT1 to CycK rescues the activity of the Gal-CycK chimera from sites downstream of the promoter and coding sequence. HeLa cells coexpressed pG6(3'Pro) and Gal-CycT(1–726), truncated Gal-CycT(1–300), Gal-CycK(1–357), and mixed Gal-CycK(1–357)-CycT(300–726) chimeras. Data are relative to basal levels of the reporter gene alone and represent two independent experiments, where standard errors of the means are given.

The kinase activity of Cdk9 is required for effects mediated by CycT1. The interaction between the C-terminal region of CycT1 and theCTD of RNAPII was required for the activation of transcription.When presented via DNA, Cdk9 had the same activity (Fig. 1B and C).To demonstrate that the kinase activity of Cdk9 is alsorequired for effects of Gal-CycT(1–726), we used DRB,a known inhibitor of Cdk9, or a kinase-deficient Cdk9 mutant[Cdk9(D167N)]. As shown in Fig. 6B, DRB inhibited the abilityof the hybrid Gal-CycT(1–726) protein to activate transcriptionvia pG6(5'Pro). These levels of inhibition were similar to theinhibition of Tat transactivation by DRB (data not presented).It is also known that at these levels, DRB does not inhibitother cyclin-dependent kinases, such as Cdk7 (43). Furthermore,the kinase-deficient Cdk9 protein [Cdk9(D167N)] also inhibitedthe activation of transcription by the Gal-CycT(1–726)fusion protein (Fig. 6C). Together, these results demonstratethat the kinase activity of Cdk9 is essential for the activationof transcription of target genes by P-TEFb from distal sites.

View larger version (20K):
[in this window]
[in a new window]
FIG. 6. The kinase activity of Cdk9 is required for the activation of transcription by P-TEFb via DNA. (A) DRB inhibits the activation of transcription by the Gal-CycT(1–726) fusion protein. The Gal-CycT(1–726) hybrid was coexpressed with pG6(5'Pro) in HeLa cells. Cells were incubated for 24 h before and 48 h after the transfection with increasing concentrations of DRB. (B) A kinase-deficient Cdk9 protein inhibits the activation of transcription by the Gal-CycT(1–726) chimera. Increasing concentrations of a plasmid expressing the dominant-negative Cdk9(D167N) protein were cotransfected with plasmids expressing Gal-CycT(1–726) and pG6(5'Pro) into HeLa cells. CAT activities are relative to basal levels of the reporter gene alone. They represent two independent experiments, and standard errors of the means are given.

CycT1 and Cdk9 increase rates of elongation of transcription via DNA. Our results demonstrated that the C-terminal region of CycT1is required for transcriptional effects when P-TEFb is presentedvia DNA. This activation of transcription could have representedincreased rates of initiation or elongation of transcriptionor both. It is established that on the HIV LTR, Tat and P-TEFbregulate the step of elongation of transcription. Nuclear run-onand steady-state RNA analyses give identical results in thissystem (1, 26). Because of the stem-loop structure, short transcripts,which correspond to TAR, have stabilities identical to thatof the full-length mRNA species. Thus, to verify that effectsof the Gal-CycT(1–726) chimera via DNA were also on thisstep of transcription, we performed RPA. RNA was purified fromcells expressing pG6TAR(5'Pro) along with different Gal-CycT1fusion proteins. The TAR sequence had no effect on the abilityof the Gal-CycT1 hybrid to activate transcription via DNA (datanot presented) (15). We used two different antisense RNA probes(60). A proximal probe hybridized to the TAR sequence and measuredlevels of initiated transcripts. A distal probe hybridized tothe CAT sequence and measured levels of elongated transcripts(Fig. 7). Each probe was annealed to the purified RNA, digestedwith RNase T1, and separated by PAGE, and the protected specieswere detected by autoradiography.

View larger version (39K):
[in this window]
[in a new window]
FIG. 7. Effects of the C-terminal region of CycT1 are on the elongation rather than initiation of transcription. RNA was purified from COS cells expressing the indicated Gal-CycT1 chimeras and the plasmid target pG6TAR(5'Pro) (lanes 1 to 5). Cells expressing Gal4-VP16 (GVP16) and Gal4 DBD (Gal) were used as the positive and negative controls, respectively. RPA were performed as described in Materials and Methods with two different RNA probes. A proximal probe corresponded to the HIV LTR from positions -10 to +59 and measured rates of initiation of transcription. A distal probe corresponded to the CAT sequence from positions 342 to 482 and measured rates of elongation.

As shown in Fig. 7, all Gal-CycT1 fusion proteins supportedsimilar levels of promoter-proximal transcription. Levels ofinitiated transcripts were similar to those obtained with theGal-VP16 fusion protein, which was used as a positive control(Fig. 7, lane 6, lower panel). However, differences betweenGal-CycT1 fusion proteins were found when elongated transcriptswere compared. Similar to the Gal-VP16 chimera, only Gal-CycT(1–726)(Fig. 7, lane 2) and Gal-CycT(1–590) (lane 3) increasedlevels of elongated transcripts (lanes 2 and 3, upper panel).In contrast, Gal-CycT(1–420) (Fig. 7, lane 4) and Gal-CycT(1–300)(lane 5) did not support these effects. These results confirmthat the effects of P-TEFb via distal sites are also on theelongation of transcription.

DISCUSSION

Top

Discussion

In this report, we present evidence that the elongation of transcriptioncan be regulated from sites upstream and downstream of targetgenes. P-TEFb, in this case composed of CycT1 and Cdk9, activatedtranscription when presented via DNA. CycT1 bound the CTD ofRNAPII in vitro and in vivo, enabling P-TEFb to mediate itseffects. This binding required more than five heptapeptide repeatsin the CTD and the histidine-rich stretch, which is conservedin the C-terminal region of CycT1 and CycT2. The kinase activityof Cdk9 was required for P-TEFb to mediate its effects. Importantly,placing just this C-terminal region of CycT1 onto another cyclinthat binds Cdk9 conferred upon CycK the ability to activatetranscription via DNA. We conclude that the C-terminal regionof CycT1 is required for the transcriptional effects of P-TEFbvia DNA. In keeping with the enhancer hypothesis, this activationdepended on promoter elements, such as CP and PPR, in this caseSp1, and the CTD of RNAPII. Thus, P-TEFb could mediate effectsof enhancers on the elongation of transcription.

Our experiments confirm and extend previous studies on P-TEFb,where Gal-P-TEFb chimeras could activate transcription via UAS(34, 37). Distinct from promoter effects, our study revealedthat P-TEFb can also function as a coactivator from sites distalto the target gene. Additionally, we documented that the CTDis the target of P-TEFb. Most importantly, the ability of P-TEFbto bind the CTD could be correlated directly with its effectson the elongation of transcription. Our study also extends datafrom D. melanogaster, where the CTD is essential for effectsof P-TEFb (41). In this system, following heat shock, P-TEFbis recruited to regions upstream of promoters of over 200 transcriptionalunits, implying its requirement for the activated transcriptionof genes, where RNAPII is paused (33).

Although CycT1 and Cdk9 were tethered artificially to DNA, thissituation only mimicked several strong activators that can functionon enhancers and which bind and recruit P-TEFb to these sequences.Previously, we made this connection with CIITA, which is thecoactivator of MHC II genes (25). It functions from conservedupstream sequences in these promoters or from distal enhancersof these genes (54). Additionally, we found that NF- ${kappa}$ B also recruitsP-TEFb to the interleukin 8 gene (1). A recent report made thesame connection with the androgen receptor (30). Thus, the recruitmentof P-TEFb represents an essential step for transcriptional effectsof these activators. Although the direct tethering to DNA madeour analyses simpler, additional insights into complex DNA-proteinassembly, looping, and transcriptional activation will comefrom more detailed studies with these individual cellular andviral activators.

This study also addresses two important issues regarding HIVreplication. First, it supports the notion that DNA-bound activators,such as NF- ${kappa}$ B, recruit P-TEFb (1). This interaction explainsthe ability of HIV to start transcribing its genome in the absenceof Tat. Thus, cellular activation releases NF- ${kappa}$ B to the nucleus,where it binds to two repeated sequences upstream of the threeSp1 sites. The recruitment of P-TEFb then leads to the synthesisof Tat. Second, P-TEFb on NF- ${kappa}$ B could promote the autophosphorylationof Cdk9, which is necessary for productive interactions betweenTat, TAR, and P-TEFb. This posttranslational modification isrequired for the creation of the optimal RNA-binding surfaceof the complex between Tat and P-TEFb (7, 12). Finally, thistripartite complex leads to sustained high levels of viral replication(59).

In fulfilling the criteria of enhancers, this study advancesthe concept that P-TEFb is an important coactivator for distalactivators, i.e., that its interaction with the CTD of RNAPIIcan convey signals over long distances. Thus, our study suggestsa model of how some enhancers could function (Fig. 8). First,the PIC is assembled and positioned on promoters. Proteins boundto TATA box, INR, and Sp1 sequences play important roles inthis process (step 1). Other activators that bind to promoterand enhancer elements also remodel chromatin and recruit P-TEFb(step 2). Some proteins also bend DNA, enabling the C-terminalregion of CycT1 to interact with the CTD of RNAPII (step 3).Cdk9 then phosphorylates the CTD (step 4), and possibly otherproteins that interact with the paused RNAPII, resulting inthe elongation of transcription (step 5). In summary, the interactionbetween the C-terminal region of CycT1 and the CTD of RNAPIIallows P-TEFb to act as a coactivator that connects enhancer-bindingproteins with the paused RNAPII at promoters, which leads tothe elongation of transcription. This model brings closer theworlds of prokaryotic and eukaryotic transcription, where transcriptionalantitermination, i.e., elongation, regulates among others theexpression of bacteriophage ${lambda}$ genes (16).

View larger version (16K):
[in this window]
[in a new window]
FIG. 8. Model of P-TEFb mediation of transcriptional effects from distal sites. The rectangle at the top shows cis elements of proximal (promoters) and distal (enhancers) transcriptional regulatory elements. The promoter consists of the CP (TATA box [T] and INR), downstream promoter element, and PPR, where activators like Sp1 (blue circles) bind. PIC is assembled and positioned at the promoter (step 1, green circles). Activators, such as NF- ${kappa}$ B (purple circles), bind to the enhancer, positioned upstream (or downstream, not pictured) and in any orientation distant from the promoter and coding sequence. They affect chromatin remodeling and also recruit P-TEFb (step 2). Their ability to bend DNA enables the C-terminal region of CycT1 to bind the CTD of RNAPII (step 3). Cdk9 (orange circle) then phosphorylates the CTD of RNAPII (red circle) (step 4), resulting in the elongation of transcription (step 5).

ACKNOWLEDGMENTS

We thank Paula Zapanc-Ecimovic for secretarial assistance, DirkEick and Jonathan Karn for reagents, Robert Tjian and KeithYamamoto for advice and comments, and members of our laboratoryfor useful discussions.

Koh Fujinaga was supported by a fellowship from the UniversitywideAIDS Research Program (F99-SF-118). This work was supportedby the Howard Hughes Medical Institute.

FOOTNOTES

* Corresponding author. Mailing address: Box 0703, N215 Mt. Zion Cancer Center, 2340 Sutter St., San Francisco, CA 94143-0703. Phone: (415) 502-1905. Fax: (415) 502-1901. E-mail: matija{at}itsa.ucsf.edu.

REFERENCES

Top