Regulation of Polymerase II Transcription by 7SK snRNA: Two Distinct RNA Elements Direct P-TEFb and HEXIM1 Binding

General procedures.

Unless otherwise stated, all manipulations were performed according to standard laboratory protocols (27). Human HeLa and G3H (21) (provided by Q. Zhou, University of California, Berkeley) cells were grown in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum (Invitrogen). A standard calcium phosphate transfection protocol was used to introduce expression vectors into HeLa and G3H cells.

Expression constructs.

To generate the p7SK expression construct, the human 7SK snRNA gene carrying 244-bp upstream and 5-bp downstream flanking sequences was PCR amplified by using oligonucleotides A (TGGAGACTGCAGTATTTAGC) and B (ATAGAATTCGGGTCAAAAGAAAGGCAGAC) as 5′- and 3′-specific primers, respectively, and human genomic DNA as a template. The nucleotide sequence of the human 7SK gene was obtained from GenBank (accession number X05490). The amplified DNA was digested with PstI and EcoRI and inserted into the same sites of pBluescript II KS(+) (Stratagene). The same approach was used for construction of p7SKd6, p7SKd17 to p7SKd19, p7SKm10, p7SKm13, p7SKm14, p7SK5′hp, p7SKm11/m11c, and p7SKm12/m12c, except that appropriate mutagenic oligonucleotides were used as 3′-specific primers and p7SK was used as a template. For construction of all other 7SK expression vectors, a two-step PCR amplification approach was used. In the first reaction, either the 5′ or the 3′ half of the human 7SK gene was amplified by using an appropriate mutagenic internal primer and oligonucleotide A or B. The resulting DNA products were used as megaprimers in the second amplification step in combination with the appropriate terminal primer (oligonucleotide A or B). The amplified products were digested with PstI and EcoRI and cloned into the same sites of pBluescribe. The sequences of the utilized mutagenic oligonucleotide primers are available upon request. To generate pHA-HEXIM1, hemagglutinin (HA) epitope was added to the N-terminal region of HEXIM1 by PCR amplification using oligonucleotides H1 (ACCCAAGCTTACTCTACTAGCCATGTACCCATACGACGTCCCAGACTACGCTGCCGAGCCATTCTTGTCAGA) and H2 (TCAGCGGCCGCAAGTTTCAGTCTAGTCTCCA) as primers and a full-length HEXIM1 cDNA clone as a template (Invitrogen). The PCR product was digested with HindIII and NotI and inserted into the same sites of the pcDNA3 expression vector (Invitrogen). The identity of each construct was verified by sequence analysis.

IP and in vitro reconstitution.

About 5 × 10⁶ cells were washed in ice-cold TBS buffer (150 mM NaCl, 40 mM Tris-HCl, pH 7.4). Cells were resuspended in 500 μl of cold NET-2 buffer (150 mM NaCl, 0.05% NP-40, 50 mM Tris-HCl, pH 7.4) and sonicated for 3 × 30 s with 60-s intervals at setting 3 with a Branson sonifier. Cell debris was removed by centrifugation, and the supernatant was incubated with 20 μl of protein A-agarose beads (Sigma) swollen in NET-2 buffer and saturated with anti-HA antibody (12CA5; Roche). After immunoprecipitation (IP), the beads were washed five times with 1 ml of NET-2 buffer and RNAs were extracted by phenol-chloroform extraction.

For in vitro reconstitution experiments, about 30 μl of protein A-agarose beads coated with an anti-CycT1 antibody (ab2098; Abcam) was incubated with a G3H cell extract under the conditions described above, except that the extract had been pretreated with 50 units of micrococcal nuclease (Amersham Biosciences) for 20 min at 30°C. After immunoprecipitation of P-TEFb, the beads were saturated with 5 μg of Escherichia coli tRNA, washed five times with NET-2 buffer, and incubated for 1 h at 4°C with a mixture of about 200 pg of in vitro-transcribed, internally labeled mutant or wild-type 7SK RNAs and about 1.5 pmol of purified His-tagged recombinant HEXIM1 protein (17). The beads were washed five times with NET-2 buffer, binding of 7SK RNAs was tested by polyacrylamide gel electrophoresis, and retention of HA-tagged CycT1 and His-tagged HEXIM1 was verified by Western blot analysis performed with anti-HA (12CA5; Roche) and anti-His antibodies (ab1187; Abcam). To generate template DNA for in vitro transcription of 7SK, 7SK5′+3′hp, 7SKd6, and 7SKd12 RNAs, the 7SK-coding regions of p7SK, p7SK5′+3′hp, p7SKd6, and p7SKd12 were PCR amplified by using appropriate oligonucleotide primers carrying the T7 RNA polymerase promoter.

RNA analysis.

For Northern blots, RNAs were separated on a 6% sequencing gel and electrotransferred onto a Hybond-N membrane (Amersham). For detection of wild-type and mutant 7SK snRNAs, oligonucleotides P1 (AGGCAGACTGCCACATGCAG), P2 (GTGTCTGGAGTCTTGGAAGC), and P3 (CATGGAGCGGTGAGGGAGGA), complementary to different regions of 7SK, were used as probes after being terminally labeled with T4 polynucleotide kinase and [γ-³²P]ATP (3,000 Ci/mmol; Amersham). The RNase A/T1 mapping procedure was described earlier (10). To synthesize sequence-specific antisense RNA probes, the appropriate 7SK expression constructs were linearized by PstI and used as templates for transcription by T7 RNA polymerase. To generate an RNA probe for mapping of luciferase mRNA, a 150-bp fragment of the luciferase gene was PCR amplified by using oligonucleotides L1 (TGGAGACTGCAGGCGCCATTCTATCCTCTAGA) and L2 (ATAGAATTCAACGGACATTTCGAAGTATTCC). The amplified fragment was inserted into the PstI and EcoRI sites of pBST, and the resulting pLUC plasmid, after linearization with PstI, was transcribed by T7 polymerase. Production of an antisense RNA probe for U6-5.8 RNA has been reported (8).

Luciferase reporter gene assay.

Human HeLa cells were cotransfected with 1 μg of luciferase reporter plasmid, pSV4LUC (4), carrying the luciferase coding region under the control of the simian virus 40 (SV40) promoter with 5 μg of pU6-5.8 and 10 μg of pBST, p7SK, p7SKd12, or p7SK5′+3′hp; 48 h later, transfected cells were washed with phosphate-buffered saline and lysed in reporter lysis buffer, and luciferase activity was determined by standard chemiluminescence detection procedures using a Sirius luminometer (2). Values were normalized to the total cell protein content.

Fluorescence in situ hybridization.

Synthesis of amino-modified oligonucleotides, chemical conjugation with FluoroLink Cy3 monofunctional reactive dye (Amersham Biosciences), fluorescence in situ hybridization of human HeLa cells, and image acquisition and processing were performed according to the protocols of the laboratory of R. Singer (http://singerlab.aecom.yu.edu). The following oligonucleotides were used to detect transiently expressed 7SK5′+3′hp (TT*ATGCAGCGCCTCATTT*CGCACATGGAGCGGTT*T) and wild-type 7SK (TT*GGATGTGTCTGGAGTCTT*GGAAGCTTGACTACT*T) snRNAs. Amino-allyl-modified T residues are indicated by asterisks. Nuclear DNA was stained with 0.1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI).

Nucleotide sequence accession numbers.

The following GenBank accession numbers were obtained for the sequences of chicken (AJ890101), zebrafish (AJ890102), Tetraodon nigrovidis (AJ890103), and Fugu rubripes (AJ890104) 7SK RNAs identified in our laboratory (unpublished results).

Two distinct regions of 7SK snRNA are required for in vivo binding of P-TEFb.

To define the structural elements of human 7SK snRNA responsible for either direct (6, 37) or indirect (16) binding of P-TEFb in living cells, a series of mutant 7SK RNAs carrying nested internal deletions were transiently expressed in human G3H cells expressing an HA epitope-tagged version of CycT1 (21). The P-TEFb binding capacity of each ectopically expressed RNA was determined by co-IP with the HA-CycT1/CDK9 complex, hereafter called HA-P-TEFb. To generate the p7SK expression plasmid, the human 7SK gene was PCR amplified and inserted into a plasmid vector (Fig. 1A). The internal deletions d1 to d14 introduced into the coding region of p7SK are shown on a previously proposed secondary structure of human 7SK snRNA (31). Since the Pol III-mediated synthesis of 7SK snRNA is controlled by upstream promoter elements, deletions in the coding region were not expected to alter the transcriptional activity of the truncated 7SK genes (15, 19).

The d1-to-d6 deletions tested in the first experiment covered the entire coding region of the 7SK gene except for the 5′-terminal nucleotides G1 to G5, which may influence transcription initiation, and the 3′-terminal T-rich track that serves as a terminator for Pol III. Expression of the truncated 7SK RNAs and their association with HA-P-TEFb were monitored by IP with an anti-HA monoclonal antibody, followed by Northern blotting (Fig. 1B). Probing of RNAs recovered from the supernatant (Fig. 1B, lanes S) of each IP reaction revealed that all mutant 7SK RNAs accumulated efficiently in G3H cells. Blotting of the immunoprecipitated RNAs (lanes P) showed that the 7SKd1, 7SKd2, 7SKd3, and 7SKd4 RNAs, like the endogenous wild-type 7SK snRNA, coprecipitated with HA-P-TEFb, indicating that the middle part of 7SK from C102 to C295 carries no crucial elements for P-TEFb binding. It is important to note, however, that each mutant RNA showed a slightly reduced P-TEFb binding affinity compared to the endogenous 7SK snRNA. Most likely, the central region of 7SK contributes to P-TEFb binding by facilitating the correct folding of the full-length RNA. As expected, the mutant 7SKd1, 7SKd2, 7SKd3, and 7SKd4 RNAs were also recovered in IP experiments performed with an antibody directed against Cdk9 (data not shown). More importantly, the 7SKd5 and 7SKd6 RNAs lacking the 5′- or 3′-teminal hairpin of the wild-type RNA, respectively, were fully inactive in P-TEFb binding, as demonstrated by both anti-HA (Fig. 1B) and anti-CycT1 (data not shown) IPs. These results demonstrate that the elements crucial for P-TEFb binding are contained in the two terminal hairpins of human 7SK.

To delimit more precisely the RNA sequences that constitute the P-TEFb binding motif in the long 5′ hairpin of 7SK, additional internally truncated RNAs, 7SKd7 to 7SKd14 (see Fig. 1A), were expressed in G3H cells and their HA-P-TEFb binding capacities were tested (Fig. 1C). We found that removal of the terminal loop and two adjacent basepairs from the top of the 5′ hairpin abolished the HA-P-TEFb binding capacity of the resulting 7SKd11 RNA. As expected, 7SKd7, 7SKd8, 7SKd9, and 7SKd10 RNAs featuring more extensive 5′-hairpin truncations also failed to associate with HA-P-TEFb. The same results were obtained upon removal of the d12 and d13 internal segments of the 5′ hairpin; the resulting 7SKd12 and 7SKd13 RNAs did not interact with HA-P-TEFb. Removal of the d14 basal segment of the 5′ hairpin of 7SK reduced but did not fully abolish the HA-P-TEFb binding capacity of 7SKd14, indicating that the proximal part of the 5′ hairpin contains no crucial elements for P-TEFb binding. In summary, we conclude that the 3′-terminal hairpin and the distal part of the 5′ hairpin of 7SK contain all the elements that are fundamental for in vivo recruitment of P-TEFb.

The distal part of the 5′-terminal hairpin of 7SK directs HEXIM1 binding.

The 7SK-dependent inhibition of the CTD kinase activity of P-TEFb also requires HEXIM1, which, besides interacting with CycT1, directly binds to 7SK snRNA (16, 38). To define the HEXIM1 binding elements of 7SK snRNA, an HA-tagged version of human HEXIM1 (HA-HEXIM1) was transiently expressed in HeLa cells (Fig. 2A). Interaction of HA-HEXIM1 with coexpressed mutant 7SK RNAs carrying the d1, d2, d3, d4, d5, or d6 internal deletion was investigated by IP with anti-HA antibody (Fig. 2B). Northern blot analysis showed that all mutant 7SK RNAs, except 7SKd5, coprecipitated with HA-HEXIM1, indicating that the 5′-terminal hairpin carries the HEXIM1 binding elements of 7SK.

Next, we investigated the HA-HEXIM1 binding abilities of 7SK RNAs lacking the d11, d12, d13, or d14 5′-hairpin fragment (Fig. 2C). While 7SKd14 RNA formed a strong interaction with HA-HEXIM1, neither 7SKd12 nor 7SKd13 RNA was detectable in the pellets from the IP reactions, demonstrating that the d12 and d13 internal regions are crucial for HEXIM1 binding. Likewise, a very weak HA-HEXIM1 interaction was detected for 7SKd11, indicating that the d11 apical stem-loop largely contributes to the HEXIM1 binding capacity of 7SK snRNA. These results demonstrate that the HEXIM1 binding elements of 7SK snRNA are located in the distal portion of the 5′-terminal hairpin. Since this region is also crucial for P-TEFb binding (Fig. 1C), we conclude that elements directing P-TEFb and HEXIM1 binding largely, if not fully, overlap in the 5′ hairpin of 7SK snRNA.

Inhibition of SV40-driven luciferase expression by a minimal 7SK RNA.

The experiments presented thus far indicate that the RNA elements crucial for in vivo recruitment of HEXIM1 and P-TEFb are located exclusively in the 5′ and 3′ hairpins of 7SK snRNA. To confirm this notion, a minimal 7SK RNA, 7SK5′+3′hp, carrying the 5′- and 3′-terminal hairpins but lacking the U116-C295 internal fragment of 7SK, was transiently expressed in HeLa and G3H cells (Fig. 3A). Another truncated RNA, 7SK5′hp, composed of the 5′-terminal hairpin (G1 to G111) and the U-rich transcription terminator (U325 to U331) of 7SK, was also expressed as a control. Northern analysis showed that both 7SK5′+3′hp and 7SK5′hp accumulated efficiently in transfected cells (Fig. 3A, lanes 1, 3, and 5). IP performed with anti-HA antibody revealed that the 7SK5′hp control RNA associated with HA-HEXIM1 (lane 2) but, as expected, failed to bind HA-P-TEFb (lane 4). This demonstrates that in living cells, binding of HEXIM1 to 7SK snRNA is independent of P-TEFb recruitment. In contrast to 7SK5′hp, 7SK5′+3′hp was able to bind both HA-P-TEFb (lane 6) and HA-HEXIM1 (data not shown), further corroborating the idea that the internal part of 7SK possesses no vital elements for HEXIM1 and P-TEFb binding.

In vitro inhibition of the kinase activity of P-TEFb necessitates only two trans-acting factors, the 7SK snRNA and HEXIM1 (16, 17, 37, 38). Therefore, we tested whether 7SK5′+3′hp, which binds both P-TEFb and HEXIM1, is capable of modulating transcription elongation in living cells. Fluorescence in situ hybridization microscopy revealed identical subcellular localizations for the ectopically expressed 7SK5′+3′hp RNA and the wild-type 7SK snRNA (Fig. 3B). Both RNAs showed a slightly speckled nucleoplasmic localization pattern with predominant exclusion from the nucleolar territories, suggesting that 7SK5′+3′hp is faithfully localized in transfected HeLa cells. Next, the p7SK5′+3′hp construct and the pSV4LUC transcription reporter plasmid expressing luciferase under the control of the P-TEFb-dependent SV40 promoter were cotransfected into HeLa cells (4, 36). As controls, an empty plasmid (pBST) or an expression plasmid encoding either the wild-type RNA (p7SK) or a mutant 7SK RNA lacking both HEXIM1 and P-TEFb binding capacities (p7SKd12) was transfected along with pSV4LUC. The transfection efficiency was controlled by inclusion of the pU6-5.8S expression plasmid carrying a tagged version of the RNA Pol III-transcribed human U6 snRNA gene (8). The transfected cells were lysed, and the luciferase activities of the lysates were determined (Fig. 3C). Cells transfected with p7SK or p7SK5′+3′hp showed strongly reduced luciferase activities compared to control cells carrying either pBST or p7SKd12. Moreover, RNase protection demonstrated that the compromised luciferase activities of the p7SK5′+3′hp- and p7SK-containing cells were accompanied by a reduced accumulation of luciferase mRNA (Fig. 3D, lanes 2 and 3), since the control cells accumulated luciferase mRNA about two times more efficiently (lanes 1 and 4). The relative accumulation levels of the ectopically expressed 7SK5′+3′hp, 7SKd12, and 7SK RNAs were determined by phosphorimager quantification and normalized to the U6-5.8 transfection control RNA (Fig. 3E). The 7SK5′+3′hp RNA accumulated about twofold more efficiently than the 7SK and 7SKd12 RNAs. Consistent with this, luciferase expression was inhibited almost two times more efficiently in p7SK5′+3′hp-transfected cells. We concluded that 7SK5′-3′hp, like wild-type 7SK, is capable of inhibiting transcription from the luciferase-linked SV40 promoter and that the 5′- and 3′-terminal hairpins of 7SK snRNA contain all the elements required for modulation of Pol II-specific transcription in living cells. Finally, the observation that overexpression of 7SK snRNA can inhibit Pol II transcription demonstrates that 7SK is limiting for the assembly of the 7SK/HEXIM1/P-TEFb regulatory complex (see Discussion).

The HEXIM1 and P-TEFb binding elements of 7SK show strong evolutionary conservation.

Consistent with its important function in transcription regulation, the 7SK snRNA shows striking sequence conservation in vertebrates. A sequence alignment of the previously characterized human, rat, and lamprey 7SK snRNAs (11, 15, 18, 24), along with four novel 7SK RNAs identified in our laboratory from chicken, Tetraodon nigrovidis, fugu, and zebra fish (unpublished data), revealed 59% sequence identity for the seven vertebrate RNAs (Fig. 4A). In general, the 5′- and 3′-terminal regions show the highest conservation, while sequences in the internal part of 7SK are more variable. Importantly, the HEXIM1 and P-TEFb binding elements defined by our analysis are located within the most conserved regions of 7SK snRNA.

The proposed secondary structure of the 5′- and 3′-end regions of human 7SK snRNA implicated in HEXIM1 and P-TEFb binding, together with the nucleotide alterations found in vertebrate 7SK RNAs, are shown in Fig. 4B. The correctness of these structures was supported by in vitro and in vivo structure probing (31). The distal region of the 5′ hairpin required for both HEXIM1 and P-TEFb binding is predicted to contain four helices (H1 to H4), a terminal and an internal loop (IL), and two bulges (U63 and 40-UU-41). This structure is preserved in all 7SK snRNAs, except that a shorter H1 helix can be folded for lamprey 7SK. The 3′-terminal hairpin is composed of a small terminal loop and two conserved helices (Hd and Hp) separated by the bulged 320-CU-321 residues. Formation of a previously proposed short helix at the base of the 3′ hairpin is not supported by phylogenetic conservation. Otherwise, the structure of the upper part of the 3′ hairpin is invariant, supporting the notion that 7SK elements directing recruitment of P-TEFb and HEXIM1 show strong structural conservation during evolution.

The RNA elements directing HEXIM1 or P-TEFb binding to the 5′ hairpin are indistinguishable.

To further characterize the HEXIM1 and P-TEFb binding elements in the 5′ hairpin of 7SK, the predicted H1, H2, H3, and H4 stems were destroyed by substitution of their descending (3′-side) strands (Fig. 5A). The resulting 7SKm1, 7SKm2, 7SKm3, and 7SKm4 RNAs were coexpressed with HA-P-TEFb and HA-HEXIM1, and their P-TEFb and HEXIM1 binding capacities were assayed (Fig. 5B). We found that the m1, m2, m3, and m4 mutations completely abolished both the HEXIM1 and P-TEFb binding abilities of 7SK snRNA. However, when the H1 and H4 stems were restored by replacement of the 5′ sides of these stems with complementary sequences (m1c and m4c), the HEXIM1 and P-TEFb binding capacities of the 7SKm1/m1c and 7SKm4/m4c double-mutant RNAs were largely restored. Two major conclusions can be drawn from these results. First, the H1 and H4 helices indeed exist in 7SK snRNA, and more importantly, these helices support HEXIM1 and P-TEFb binding in a sequence-independent manner. Our attempts to restore the H2 and H3 stems in 7SKm2/m2c and 7SKm3/m3c RNAs, however, failed to rescue the HEXIM1 and P-TEFb binding abilities of these RNAs. This may indicate that the sequence composition of the H2 and H3 stems, if they are really formed, is crucial for 7SK function. Likewise, alteration of the internal-loop (IL) sequences or changing the bulged 40-UU-41 residues abolished association of 7SKm5, 7SKm8, and 7SKm7 RNAs with both HEXIM1 and P-TEFb. In contrast, replacement of the bulged U63 with an A residue in 7SKm6 or alteration of the terminal-loop sequence of 7SKm9 reduced but did not fully eliminate the HEXIM1 and P-TEFb binding capacities of these RNAs, indicating that these sequences are not fundamental to 7SK function. Thus, we propose that a segment of the 5′ hairpin delimited by the H1 and H4 stems contains all the elements essential for HEXIM1 and P-TEFb binding. While the nucleotide composition of the H1 and H4 helices is not critical, the internal sequences are fundamental for binding of both HEXIM1 and P-TEFb. It is important to note that mutational analysis of the 5′ hairpin of 7SK failed to identify distinct binding elements for HEXIM1 and P-TEFb. With no exception, all structural alterations of 7SK which abolished or compromised HEXIM1 binding also prevented or reduced P-TEFb association. These results, taken together with the facts that CycT1 can interact with HEXIM1 (16, 17) and that HEXIM1 alone can bind to 7SK snRNA (Fig. 3A), indicate that P-TEFb is recruited to the 5′ hairpin of 7SK through an interaction formed with HEXIM1.

Integrity of the 3′ hairpin is essential for P-TEFb binding.

To define the structural determinants of P-TEFb binding to the 3′ hairpin, we performed a detailed mutational analysis of this region of 7SK (Fig. 6A). Upon removal of the distal (d17) or proximal (d18) region of the 3′ hairpin or elimination of its terminal loop (d19), the mutant 7SKd17, 7SKd18, and 7SKd19 RNAs failed to interact with P-TEFb (Fig. 6B). Likewise, destruction of the distal (Hd) or proximal (Hp) stems by replacement of their 5′-side sequences with noncomplementary sequences (m12 or m11) fully obliterated the P-TEFb binding capacities of 7SKm12 and 7SKm11 RNAs. Next, we attempted to reestablish the P-TEFb binding abilities of 7SKm12 and 7SKm11 by restoring the Hd and Hp helices through replacement of their descending strands with the m12c or m11c complementary sequences. Indeed, the 7SKm12/m12c RNA regained its P-TEFb binding ability, leading to the conclusion that the Hd stem, although its nucleotide composition is not critical, plays an essential role in P-TEFb binding. In contrast, the 7SKm11/m11c RNA remained inactive in P-TEFb binding, suggesting that the nucleotide composition of the Hp stem is crucial for P-TEFb recruitment. The same results were obtained when the terminal-loop nucleotides or the C320 and U321 bulged residues were replaced (m14 and m13); neither 7SKm13 nor 7SKm14 could interact with P-TEFb. Finally, as suggested by our phylogenetic comparison (Fig. 4), alteration of the evolutionarily variable 296-AAAUGA-301 sequences at the base of the 3′-hairpin did not inhibit the P-TEFb binding capacity of 7SKm10. These results demonstrate that in the 3′-terminal hairpin of 7SK, the G302-to-C324 region, including the Hd and Hp helices, the terminal loop, and the bulged 320-CU-321 residues, constitutes the minimal structure required for establishment of an in vivo interaction with P-TEFb.

In vitro assembly of 7SK/HEXIM1/P-TEFb.

We next tested whether 7SK elements directing in vivo recruitment of HEXIM1 and P-TEFb are also important for assembly of the 7SK/HEXIM1/P-TEFb complex in the test tube. Previous reconstitution experiments showed that phosphorylation of Cdk9 is crucial for 7SK/P-TEFb interaction (6). Therefore, P-TEFb was immunoprecipitated by an anti-CycT1 antibody from G3H cell extract that had been treated with micrococcal nuclease to eliminate endogenous 7SK and to destroy 7SK/HEXIM1/P-TEFb complexes. Beads retaining P-TEFb were incubated with purified His-tagged recombinant HEXIM1 protein (17) in the presence of in vitro-synthesized internally labeled wild-type or mutant 7SK RNAs (Fig. 7A). After extensive washing, binding of 7SK RNAs was tested by polyacrylamide gel electrophoresis, and retention of HA-CycT1 and His-HEXIM1 was controlled by Western blot analysis. While the wild-type 7SK and the 7SK5′+3′hp RNAs were able to bind to P-TEFb in the presence of His-HEXIM1 (lanes 2 and 5), the 7SKd12 and 7SKd6 RNAs lacking putative HEXIM1 and P-TEFb binding elements, respectively, failed to stably associate with immobilized P-TEFb. This demonstrates that the 5′- and 3′-terminal hairpin elements of 7SK snRNA are also crucial for in vitro assembly of 7SK/HEXIM1/P-TEFb.

Finally, to confirm the functional importance of the newly determined HEXIM1 and P-TEFb binding sites of 7SK snRNA, we assayed the transcriptional-regulatory capacities of some representative mutant 7SK RNAs (Fig. 7B). As was predictable, 7SKm1 and 7SKm12 RNAs lacking HEXIM1 and/or P-TEFb binding abilities were unable to inhibit SV40 promoter-driven luciferase expression in transiently transfected HeLa cells. However, restoration of the HEXIM1 and P-TEFb binding capacities of these RNAs reestablished the transcriptional-inhibitory functions of 7SKm1/m1c and 7SKm12/m12c RNAs.