Negative Elongation Factor NELF Represses Human Immunodeficiency Virus Transcription by Pausing the RNA Polymerase II Complex*

Human immunodeficiency virus (HIV) transcription requires virally encoded Tat and the P-TEFb protein complex, which together associate with the Tat-activating region, a structured region in the nascent transcript. P-TEFb phosphorylates Proteins in the transcription elongation complex, including RNA polymerase II (pol II), to stimulate elongation and to overcome premature termination. However, the status of the elongation complex on the HIV long terminal repeat (LTR) in a repressed state is not known. Chromatin immunoprecipitation demonstrated that NELF, a negative transcription elongation factor, was associated with the LTR. Depleting NELF increased processive HIV transcription and replication. Mapping pol II on the LTR showed that pol II was paused and that NELF depletion released pol II. Decreasing NELF also correlated with displacement of a positioned nucleosome and increased acetylation of histone H4, suggesting coupling of transcription elongation and chromatin remodeling. Previous work has indicated that the Tat-activating region plays a critical role in regulating transcription from the LTR. Our results reveal an earlier stage, mediated by NELF, when repression occurs at the HIV LTR.

Human immunodeficiency virus (HIV) 3 proviral expression is regulated at the transcriptional level. Virus transcription is controlled by the upstream long terminal repeat (LTR), which includes cis-elements that recruit both cellular and viral factors. The HIV-1 LTR is often divided into functional elements: the Tat-activating region (TAR), the promoter, the enhancer, and the negative/modulatory regulatory element. The promoter, enhancer, and modulatory elements recruit host transcription factors, such as Sp-1, NF-B and CCAAT/enhancer-binding protein, necessary to initiate transcription (1,2). These factors also recruit coactivators, including histone acetyltransferases and SWI/SNF complexes, that influence the chromatin structure of integrated provirus (3)(4)(5)(6)(7). For example, the 5Ј-untranslated leader region located downstream of the transcriptional start site is associated with a nucleosome in latent HIV proviruses, which is displaced upon induction of virus transcription. Furthermore, agents that inhibit histone deacetylases, such as trapoxin and trichostatin A, activate HIV provirus transcription, suggesting a critical role for chromatin remodeling in the repression of HIV transcription (7,8).
Transcription elongation has also been demonstrated to be a limiting step for HIV expression. HIV encodes a transcriptional activator (Tat) that binds the RNA stem-loop structure formed by TAR, and by recruiting P-TEFb to the LTR, HIV enhances processive transcription. P-TEFb, which is composed of cyclin T1 and CDK9, modifies RNA polymerase II (pol II) activity by hyperphosphorylating the C-terminal domain of pol II. In the absence of Tat, transcription elongation by RNA pol II from the HIV promoter is very inefficient (9). In vitro transcription analyses in the absence of chromatin revealed that the majority of elongation complexes initiating from the HIV-1 promoter prematurely terminate transcription within 500 bp of the transcriptional start site (10). In vivo, it has been observed that cells infected with HIV accumulate short transcripts of ϳ60 bases in the cytoplasm. The accumulation of these short transcripts and the identification of a site in this region where purified pol II strongly pauses elongation have led to speculation that transcription of the integrated provirus is repressed by premature termination (11,12). The presence of Tat substantially increases the production of long transcripts (9,11,13).
Analysis of numerous cellular genes has revealed many cases in which pol II initiates transcription but pauses in the region 20 -50 nucleotides downstream from the transcriptional start site. Pausing of pol II on the hsp70 heat shock gene in Drosophila requires two proteins called DSIF and NELF (14,15). NELF and DSIF associate with the elongation complex and inhibit elongation (16,17). The inhibitory action of these two proteins is overcome by P-TEFb (17,18). In addition, NELF and DSIF have been implicated in regulating HIV transcription elongation. The E subunit of NELF associates with the TAR element (19), and ectopic expression of this NELF subunit inhibits tran-* This work was supported by Pennsylvania State Tobacco Formula Funds and National Institutes of Health Grants AI62467 (to A. J. H.) and GM47477 (to D. S. G.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 Both authors contributed equally to this work. 2  sient basal transcription from the HIV LTR (20). Both DSIF and NELF contribute to Tat-dependent activation of HIV transcription and are phosphorylated by P-TEFb (20,21). Phosphorylation alters DSIF so that it stimulates elongation. NELF dissociates from the elongation complex during activation, but it is unclear if this dissociation is caused by phosphorylation of NELF (15). The role that NELF plays in repressing transcription of the HIV provirus has not been investigated. Transient expression assays and RNA binding assays led to the conclusion that NELF represses HIV transcription by associating with TAR (20). This would require pol II to transcribe at least to ϩ59. In contrast, NELF-mediated repression of several cellular genes appears to occur within 20 -50 nucleotides of the start site (17,22). If pol II pausing on the HIV LTR is similar to that of other genes, then NELF would repress transcription before synthesis of TAR is complete. In this study, we have investigated whether NELF repression of the HIV provirus might occur at a stage earlier than suggested by transient expression studies. Moreover, we provide the first detailed analysis of the behavior of pol II in the promoter proximal region of the latent HIV provirus.
Chromatin Immunoprecipitation Assays-U1 cells (25 ml, 2 ϫ 10 6 cells/ml) were either left untreated or stimulated with 10 ng/ml phorbol 12-myristate 13-acetate (PMA) for 4 h, and 675 l of 37% formaldehyde stock solution was added directly to the culture medium to give a final concentration of 1%. Cells were incubated at room temperature for 10 min, transferred to centrifuge tubes on ice, collected by centrifugation at 700 ϫ g for 10 min, and washed once with cold phosphate-buffered saline. Subsequent steps were done according to the chromatin immunoprecipitation procedure described by Upstate Biotechnology to yield ϳ3 ml of soluble chromatin. Ten g of soluble chromatin (DNA) was used for each immunoprecipitation along with 5 g of each antibody. The DNA was dissolved in 30 l of TE buffer (10 mM Tris-HCl (pH 8.0) and 1 mM EDTA). To prepare DNA representing input DNA, 10 g of soluble chromatin was combined with elution buffer. Reversal of formaldehyde cross-links and isolation of the DNA were done as for DNA eluted from the immunoprecipitates. The final DNA was dissolved in 30 l of TE buffer and further diluted with TE buffer as needed. Two l of immunoprecipitated DNA or input DNA was subjected to PCR in a final volume of 50 l containing 10 pmol of each primer, 200 M dNTP, 2 or 4 Ci of [␣-32 P]dCTP, 2.5 units of Taq polymerase, and buffer (Roche Diagnostics). Each sample was subjected to 20 cycles at 94°C for 1 min 57°C for 55 s, and 72°C for 50 s, followed by a final extension at 72°C for 10 min. The following primers were used to amplify different regions of the HIV-1 gene: LTR Ϫ155 to ϩ186, 5Ј-CCGAGAGCTGCATCCGGAGT-3Ј and 5Ј-ACTG-CTAGAGATTTTCCACACT-3Ј; LTR ϩ45 to ϩ248, 5Ј-GGG-AGCTCTCTGGCTAAC-3Ј and 5Ј-AGTCCTGCGTCGAGA-GAG-3Ј; and HIV ϩ2415 to 2690, 5Ј-GTAACAGTACTGGA-TGTGGGTGATG-3Ј and 5Ј-CTGCCCTATTTCTAAGTCA-GATCC-3Ј.
Ten l of each PCR was run on a 6% nondenaturing acrylamide gel. Gels were fixed in 10% acetic acid for 20 min and then dried. Radioactivity was detected in the dried gel using a phosphoimager (GE Healthcare). The intensity of each band was quantified by volume analysis using ImageQuant software.
Reverse Transcription-PCR-NELF-B small interfering RNA (siRNA) or control siRNA (250 nM) was delivered to 1 ϫ 10 6 U1 cells using an Amaxa Biosystems Nucleofector TM II. Fortyeight h post-transfection, cells were stimulated with 10 ng/ml PMA or not for 4 h. Total RNA was isolated using TRIzol (Invitrogen) according to the manufacturer's instructions. The final total RNA was dissolved in 30 l of TE buffer , and 4 l was used for each reverse transcription reaction. Reverse transcription reactions were performed with either the LTR or Tat region primers in the presence or absence of reverse transcriptase in a 10-l volume. cDNA was amplified by PCR for initiated short transcripts (ϩ11 to ϩ61) and elongated transcripts (ϳ5 kb downstream of the LTR) as described (23). U1 cells treated with PMA were a positive control for induction of HIV transcription. Two l of cDNA was subjected to a PCR in a final volume of 50 l containing 10 pmol of each primer, 200 M dNTP, 2 or 4 Ci of [␣-32 P]dCTP, 2.5 units of Taq polymerase, and buffer (Roche Diagnostics). Each sample was subjected to 20 cycles at 94°C for 1 min, 55°C for 55 s, and 72°C for 50 s, followed by a final extension at 72°C for 10 min. The following primers were used to amplify different regions of the HIV-1 gene: initiated short transcripts, 5Ј-GTTAGACCAGATCTG-AGCCT-3Ј and 5Ј-GTGGGTTCCCTAGTTAGCCA-3Ј; and elongated transcripts, 5Ј-ACTCGACAGAGGAGAGCAAG-3Ј and 5Ј-GAGTCTGACTGTTCTGATGA-3Ј.
The ␤-actin internal standard was included in all the PCRs. The ␤-actin gene was amplified with primers 5Ј-GTCGACAA-CGGCTCCGGC-3Ј and 5Ј-GGTGTGGTGCCAGATTTTCT-3Ј. Ten l of each PCR was run on a 6% nondenaturing acrylamide gel. Gels were fixed in 10% acetic acid for 20 min and then dried. Radioactivity was detected in the dried gel using a phosphoimager. The intensity of each band was quantified by volume analysis using ImageQuant software.
Permanganate Footprinting-U1 or ACH-2 cells (2 ϫ 10 6 ) were washed with phosphate-buffered saline and resuspended in 100 l of phosphate-buffered saline. The cells were treated with permanganate by adding 100 l of ice-cold 20 mM KMnO 4 , which was prepared by dissolving solid KMnO 4 in phosphatebuffered saline. The permanganate reaction was incubated on ice for 1 min and stopped by the addition of 200 l of stop solution (20 mM Tris-HCl (pH 7.5), 20 mM NaCl, 40 mM EDTA, 1% SDS, and 400 mM 2-mercaptoethanol). The solution was vigorously shaken until all coloration had vanished. Each sample was treated with 50 g of proteinase K for at least 1 h and then extracted with a sequence of liquefied phenol, phenol/ chloroform/isoamyl alcohol (49.5:49.5:1), and chloroform. Nucleic acid was precipitated with 0.6 M sodium acetate (pH 6.0) and ethanol. The DNA pellets were washed with 75% eth-anol and dissolved in 20 l of TE buffer (pH 7.5). To determine the pattern of permanganate reactivity, 500 ng of each DNA sample was diluted to 15 l of TE buffer (pH 7.5). Seventy-five l of H 2 O and 10 l of piperidine were added, and each sample was incubated at 90°C for 30 min. Three-hundred l of H 2 O was added to each sample, and the samples were then extracted three times with 700 l of isobutyl alcohol and one time with ether. The volume of the DNA was adjusted to 100 l with H 2 O, followed by ethanol precipitation. The DNA was dissolved in 10 l of TE buffer (pH 7.5), transferred to a fresh siliconized tube, and used in a primer extension.
Restriction Enzyme Accessibility Assays-Restriction enzyme accessibility assays were performed as described previously (4). Nuclei from 2 ϫ 10 6 U1 cells were used for each assay.
Depletion of NELF-B from U1 Cells with siRNA-Depletion of NELF-B with siRNA was done according to Aiyar et al. (24). NELF-B (COBRA1) siRNA (250 nM; Dharmacon) was transfected into 1 ϳ 2 ϫ 10 6 U1 cells using the Amaxa Biosystems Nucleofector TM II and solutions and recovered in RPMI 1640 medium supplemented with 10% fetal bovine serum. Twentyfour h post-transfection, fresh RPMI 1640 medium was added, and cells were harvested at 48 h post-transfection.

Characterization of RNA pol II and Associated Factors on the
HIV LTR-U1 cells, a cell line derived from U937 promonocytic cells chronically infected with HIV-1, have been used as a model for HIV-1 latency (25). Provirus expression in U1 cells is poor in part because of defective Tat function (26), but is highly inducible following treatments with various cytokines or phorbol esters (25,27). U1 cells have been used to examine the recruitment of transcription factors to the LTR as well as the changes in chromatin organization following induction of HIV-1 provirus transcription. Initially, chromatin immunoprecipitation (ChIP) was employed to characterize transcription factor occupancy at the HIV LTR in unstimulated and stimulated U1 cells. ChIPs were performed with antibodies against NF-B subunits p50 and p65 (which have been demonstrated to be necessary for induction of HIV expression), TBP, pol II, and the pol II-associated factors NELF and Spt5 (a DSIF subunit). As shown in Fig. 1, in unstimulated U1 cells, all of these factors were bound to the LTR. Furthermore, in the absence of U1 stimulation, no significant binding of pol II or the associated proteins NELF or Spt5 with downstream HIV genomic sequences was observed, as would be predicted if transcription elongation were being repressed.
Following PMA treatment and induction of HIV transcription, p65 binding increased 3-fold, and p50 binding remained unchanged (Fig. 1, A and B). These results are in agreement with those reported previously (7), thus validating our ChIP analysis of the provirus in U1 cells. Following PMA treatment, ϳ3-fold more pol II, TBP, and Spt5 were associated with the HIV LTR, whereas there was a Ͻ2-fold change in the level of NELF (Fig. 1, A and B). This suggests that a lower proportion of the pol II molecules associates with NELF under induced conditions. When factors binding downstream proviral sequences were examined, only pol II and Spt5 were detected (Fig. 1, C and  D). The amount of pol II complexes associated with the LTR and genomic sequences was approximately equal in cells stimulated with PMA, suggesting that pol II is evenly distributed over the HIV genome, in contrast to unstimulated cells, in which the majority of pol II was found on the LTR. NELF was notably absent, suggesting that NELF does not travel with the elongation complex but dissociates from the elongation complex during transcriptional activation. Treating U1 cells with interleukin-1, tumor necrosis factor-␣, and interferon-␥ resulted in similar changes in the occupancy of factors on the LTR and downstream proviral sequences, indicating that the changes were not unique to PMA induction but correlated with general events required for activation of HIV transcription (data not shown).
A more detailed analysis of the behavior of pol II on the LTR in the context of U1 cells was done using permanganate genomic footprinting (Fig. 2). Permanganate footprinting detects transcriptionally engaged pol II molecules by reacting ChIPs were performed using 10 g of soluble chromatin plus 5 g of specific antibodies or nonspecific isotype control antibody (N.S.). DNA was detected using primers that amplify Ϫ155 to ϩ186 (A) or downstream proviral sequences spanning ϩ2415 to ϩ2690 (C). The 10% and 4% lanes are products generated from input DNA that had not been precipitated. Densitometry was used to quantify LTR-associated bands (B) or downstream proviral sequences (D), and the data are normalized to input DNA. Data are representative of four independent experiments.
preferentially with single-stranded thymines located in the transcription bubble (28,29). Unlike ChIP, which permits only the absence or presence of pol II on the LTR to be determined, permanganate footprinting allows the distribution of pol II to be mapped on DNA in vivo at a resolution of ϳ10 bp. Fig. 2A shows the permanganate reactivity of thymines in unstimulated U1 cells. Comparison of this pattern with that observed with purified DNA reveals very reactive thymines at ϩ41 and ϩ45 and more modestly reactive thymines in the region extending from ϩ49 to ϩ85. This suggests that pol II is concentrated in the region 40 -45 nucleotides downstream from the transcriptional start site, with a low level of pol II at ϩ65, ϩ75, ϩ85, and ϩ95 in unstimulated cells. In contrast, the pol II footprint generated in PMA-stimulated cells begins at ϩ13 and extends through TAR to at least ϩ111 ( Fig. 2A). To confirm that this footprint is not unique to U1 cells, we examined the position of pol II on the HIV LTR in the chronically infected ACH-2 T-cell line. ACH-2 cells are similar to U1 cells in that transcription is not processive unless induced with cytokines or PMA. The defect in transcription has been mapped to a mutation in the TAR element (30). Comparable with the pattern observe for U1 cells, the reactivity for thymines at ϩ41 and ϩ45 was high relative to naked DNA, indicating that pol II was present in this region. The reactivity between ϩ49 and ϩ85 was low, whereas treatment of ACH-2 cells with PMA increased the sensitivity of thymines within this region (Fig. 2B). These data suggest that HIV transcription in U1 and ACH-2 cells is limited in part by paused pol II in the region proximal to the promoter.
Tat has been reported to stimulate the assembly of transcription complexes (31) and to promote pol II processivity (9,11,13). Therefore, we wanted to determine whether Tat can influence the behavior of the paused pol II complexes. Because we were concerned about low transfection efficiencies for U1 cells, cells were cultured with 10 g/ml recombinant Tat (obtained from the National Institutes of Health AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, National Institutes of Health). Recombinant Tat is rapidly taken up by cells and transactivates HIV expression in U1 cells (32,33). pol II pausing in the absence and presence of Tat was examined by permanganate footprinting. Fig. 2C shows that treating cells with Tat increased the permanganate reactivity of thymines downstream of ϩ49, generating a pol II footprint that was similar to that observed in U1 cells cultured in PMA. In summary, these data suggest that when Tat is limiting, early HIV proviral transcription is repressed in part by a paused pol II complex located ϳ40 -45 nucleotides downstream from the transcriptional start site.
Depleting NELF-B with siRNA Stimulates Transcription of HIV-NELF is a key factor in establishing a paused polymerase on the hsp70 gene in Drosophila and the junB gene in humans (15,34), and its association with the HIV LTR (but not downstream HIV sequences) suggests that it could be actively repressing HIV transcription elongation in U1 cells. To test this possibility, we used siRNA to deplete NELF-B, one of the four subunits of NELF. Successful depletion of NELF-B in cells was confirmed by immunoblots (Fig. 3A). More important, 70% less NELF-B was associated with the HIV LTR following siRNA treatment of U1 cells as determined by ChIP (Fig. 3, B and E). Diminishing NELF-B did not alter the binding of other transcription factors, including TBP, p65, and p50 (Fig. 3, B and E). Furthermore, reducing NELF-B had no effect on recruitment of Spt5 (a subunit of DSIF) or pol II to the HIV LTR (Fig. 3, B and E), demonstrating that diminishing NELF-B does not interfere with the association of pol II with the LTR. However, knocking down NELF-B did increase the amount of pol II associated with the HIV sequences 2415 bp downstream of the HIV-1 promoter by Ͼ5-fold in the absence of PMA stimulation (Fig. 3, C and D), indicating that NELF-B negatively regulates HIV transcription elongation.
The pol II ChIP data indicate that depletion of NELF-B could cause HIV transcription to increase. We tested this prediction by measuring the level of initiated HIV transcripts versus elongated transcripts in cells in which NELF-B was depleted. HIV transcription initiation and elongation were assayed in unstimulated U1 cells treated with control siRNA and NELF-B siRNA by reverse transcription-PCR using primers to TAR (ϩ11 to ϩ61) and to Tat exon 1 (ϳ5 kb downstream of the LTR), respectively, as described by Williams et al. (23). Although there were no differences observed between control and NELF-B-depleted cells in their ability to initiate short transcripts, there was an increase in elongated transcripts in cells treated with NELF-B siRNA (Fig. 4, A and B). Consistent with this increase in transcription, we observed by p24 enzymelinked immunosorbent assay a 2.5-fold induction in HIV replication in unstimulated cells that had reduced NELF-B (Fig. 4C). Therefore, NELF establishes a paused pol II complex at the HIV LTR that negatively regulates HIV-1 transcription elongation and HIV-1 replication.
We wanted to determine whether NELF acts on pol II at a point prior to completion of the TAR element as would be predicted based on where NELF affects pol II on cellular genes. Depletion of NELF-B in unstimulated U1 cells led to changes in the pattern of permanganate reactivity on the LTR that were consistent with release of pol II from the region where it appeared to pause in control cells. As is readily apparent by inspection of the band intensities in lanes 4 and 5 in Fig. 4D and the densitometric trace of these two lanes (Fig. 4E), depletion of NELF-B caused the intensity of bands around the ϩ41/ϩ45 region to decrease and the intensity of bands downstream of ϩ85 to increase. These results suggest that NELF inhibits elongation at the HIV LTR in U1 cells by imposing a block to elongation in the region 40 -45 bp downstream from the transcriptional start site, well before the end of the TAR element and the strong intrinsic pause sites that have been identified by in vitro transcription assays. Depletion of NELF removes this block, resulting in less pol II being present around ϩ45 and more pol II present farther downstream. The net result is that there is little change in the level of pol II associated with the LTR, in agreement with the ChIP result (Fig. 3B), but significant change in the distribution of pol II located within the LTR.

Transcription Caused by Depletion of NELF Induces Changes in
Chromatin Structure-A key checkpoint for transcription regulation is at the level of promoter accessibility and chromatin structure. For example, repression of HIV transcription has been associated with a nucleosome positioned at the transcriptional start site. Induction of HIV transcription correlates with histone modifications and displacement of this positioned nucleosome (7,8,35). To gain additional insights into the coordinate regulation of transcription elongation and chromatin structure, we assessed the chromatin organization at the HIV LTR in the control and NELF-B-depleted cells by restriction enzyme accessibility assays and ChIPs. If HIV transcription is repressed in part by a nucleosome positioned at the transcriptional start site, then it is anticipated that the chromatinbound DNA will be resistant to cutting by HindIII but sensitive to cutting by PvuII, as reported previously (35). To measure restriction enzyme accessibility, nuclei were isolated from normal and NELF-depleted U1 cells. The nuclei were treated with the enzymes shown in Fig. 5A, and the degree of digestion was measured by ligation-mediated PCR. As expected, both the control and NELF-depleted cell lines were equally susceptible to digestion by PvuII, which lies outside the location of the potential positioned nucleosome. However, the level of cutting by HindIII was increased when NELF-B was knocked down in U1 cells, as shown in Fig. 5B. Although this increase in restriction enzyme accessibility suggests the displacement of a positioned nucleosome, we cannot rule out the possibility that the changes observed in HindIII cutting reflect the position of paused and released pol II in control and NELF-depleted cells, respectively.
ChIP assays were performed to determine whether potential histone post-translational modifications, such as acetylation, are influenced by depletion of NELF. The data in Fig. 5 (C and  D) show that acetylation of histone H4 increased following NELF-B knockdown. The observed changes in restriction accessibility and histone modification suggest that transcription elongation is coupled to chromatin organization and that inducing transcription elongation by depleting NELF leads to overall changes in promoter accessibility.

DISCUSSION
Control of transcription elongation is a critical checkpoint in the regulation of a number of genes, including c-myc, c-fms, hsp70, and HIV (22,36). One basis for this checkpoint is thought to be the interplay between the negative elongation factors NELF and DSIF and a positive elongation factor, a kinase called P-TEFb (37). NELF

. Diminishing NELF increases pol II associated with downstream HIV proviral sequences.
A, Western blot analysis showed that siRNA against NELF-B decreased the level of NELF-B in cells. 1 ϫ 10 6 U1 cells were nucleofected with 250 nM NELF-B siRNA or nonspecific control siRNA. Forty-eight h post-transfection, cell lysates were prepared, and NELF-B protein was detected by immunoblotting with anti-NELF-B antibody. Blots were stripped and reprobed for ␤-actin as a loading control. B-E, ChIP assay was performed to detect factors binding to the LTR or downstream proviral sequences following NELF-B depletion. Using a specific antibody or a nonspecific isotype control antibody (N.S.), ChIPs were performed on cells treated for 48 h with control siRNA or NELF-B siRNA. DNA was detected using primers that amplify Ϫ155 to ϩ186 (B and E) or downstream proviral sequences spanning ϩ2415 to ϩ2690 (C and D). B and C show representative data, and the 5% and 1% lanes are products generated from input DNA that had not been precipitated. The graphs in D and E are based on densitometric analysis of band intensities from three independent experiments. and DSIF associate with the early elongation complex and inhibit elongation by pol II, possibly by interfering with the extrusion of the nascent transcript from the elongation complex (16). P-TEFb alleviates repression by phosphorylating one or more of the components in this complex. P-TEFb phosphorylates the C-terminal domain of pol II at serine 2, and this correlates with pol II being actively engaged in elongation (38). Recently, phosphorylation of DSIF was shown to convert DSIF from a negative to a positive elongation factor (37). NELF is also phosphorylated by P-TEFb, and this reduces the ability of NELF to associate with RNA (20). Notably, NELF dissociates from the elongation complex when the complex is transcribing the DNA in vivo, suggesting that NELF functions primarily as an inhibitor of elongation (15).
By using siRNA to deplete a subunit of NELF and ChIP and permanganate genomic footprinting to monitor the behavior of pol II, we provide the first direct evidence that NELF inhibits transcription of an integrated HIV provirus. Diminishing NELF expression results in increased pol II associated with HIV genomic sequences, more elongated transcripts, and enhanced virus replication in the absence of activation by PMA or cytokines. Furthermore, permanganate footprinting, which allows us to examine the behavior of pol II at a resolution beyond that afforded by ChIP, demonstrates that pol II preferentially accumulates at approximately ϩ41 to ϩ45 and that reducing NELF redistributes pol II throughout the LTR. It has been shown that overexpression of one of the four subunits of NELF and one of the two subunits of DSIF inhibits expression of an HIV LTR reporter gene in a transient expression assay (20); however, it is noteworthy that the promoter proximal pausing we detected on the HIV LTR has not been demonstrated to occur on transiently transfected DNA, suggesting that these previous work measured aspects of pol II elongation downstream from the promoter proximal region. Our results show that NELF acts on pol II before it has completed transcription of the TAR element and that this interaction has a role in repressing transcription.
Our data suggest that, in the presence of NELF, pausing of pol II on the HIV LTR is similar to the promoter proximal pausing that has been observed on other cellular genes wherein the pol II pauses 20 -50 nucleotides downstream from the start site (17,22). This is before intrinsic pause sites within the LTR defined by in vitro transcription reactions and short of the distance required to generate the TAR structure in the nascent transcript (39), suggesting that these elements have a limited role in repressing HIV transcription. Indeed, the only evidence that suggests that these two features of the LTR inhibit elongation by pol II in vivo is the finding that cells accumulate short transcripts of ϳ60 nucleotides in length. These short transcripts may result from the combined action of premature termination downstream of the TAR element and exonucleolytic degradation of released transcripts back to the base of the stem-loop structure constituting TAR. The presence of these short transcripts suggests that the block to elongation imposed by NELF in the ϩ40 region of the LTR is not complete. In accord with this prediction, comparison of the permanganate patterns detected on unstimulated cells with the pattern detected on naked DNA provides evidence that there is a low level of pol II at ϩ65, ϩ75, ϩ85, and ϩ95 in unstimulated cells (Fig. 2).
If pol II pauses in the ϩ40 region before completing transcription of TAR, then how does Tat stimulate transcription as we observed in Fig. 2C? The incomplete block to HIV elongation may allow limited TAR formation and initial recruitment of Tat and P-TEFb to the LTR, which could mobilize transcription complexes (31) and release the paused pol II from proximal pausing (40). An alternative mechanism is that Tat is recruited to the HIV LTR by interacting with cellular transcription factors. Tat has been observed to associate with Sp1, pol II, and the pol II holoenzyme complex (41,42). These TAR-independent pathways might be necessary for the elongation complex to efficiently escape the pause in the ϩ45 region. In the absence of Tat, the pause at ϩ45 would effectively block reinitiation, thus contributing to repression of HIV replication.
We propose that, for active transcription elongation, NELF must dissociate from pol II. The conclusion is based on the observation that NELF does not travel with the elongation complex. This appears to contradict previous studies in which DSIF and NELF were biochemically co-purified with pol II using HIV elongation complexes generated in vitro (43) but is more consistent with in vivo studies, including those performed with the Drosophila hsp70 heat shock gene, in which NELF has been shown to disengage from the elongation complex during transcriptional activation (15). The release of NELF could be regulated in part by post-translational modifications, such as phosphorylation by P-TEFb.
We were also surprised that reduction of NELF and the subsequent release of pol II from its pause correlated with changes in restriction enzyme accessibility and post-translational modification of a nucleosome on the HIV LTR, suggesting changes in chromatin organization. Although it was demonstrated previously that Tat-mediated chromatin changes are insensitive to the pol II inhibitor ␣-amanitin, suggesting a process independent of transcription elongation (35), recent data using an in vitro chromatin remodeling assay imply that active transcription is required for chromatin remodeling by Tat and SWI/SNF (44). Our studies are looking at basal HIV transcription that would precede the transcription of Tat. The role of chromatin in repressing HIV transcription has been suggested by experiments showing that deacetylase inhibitors elevate basal virus expression (7). Furthermore, chromatin structure represses elongation at the hsp70 promoter in mammalian cells (45), although NELF is sufficient for the induction of promoter proximal pol II pausing on hsp70 in a Drosophila in vitro transcription system (46). Our findings imply that transcription and nucleosome modification and reorganization may not be independently regulated but coupled events and that full repression requires NELF and repressive chromatin structure. pol II has been demonstrated to differentially interact with histone acetyltransferase proteins CBP and p300/CBP-associated factor as well as with SWI/SNF-like chromatin remodeling complexes (36).
On the basis of our findings, we propose the existence of a previously unidentified step in the repression of HIV transcription that contributes to the latency of the virus. This NELFmediated repression of elongation occurs before pol II has completed transcription of the TAR element. A nucleosome adjacent or even overlapping this paused pol II could serve to reinforce the repression. Initial activation of HIV transcription would require release of the paused pol II and displacement of Nuclei were isolated from untreated U1 cells, U1 cells treated with control siRNA or NELF-B siRNA, and U1 cells stimulated with PMA. Nuclei were digested with 20 units of HindIII. Genomic DNA was isolated and cut to completion with PvuII, and digested DNA products were then analyzed by ligation-mediated PCR. As a positive control (contr.), purified genomic DNA was digested with either PvuII or HindIII. C, ChIP for acetylated histone H4 (AcH4). Chromatin was prepared from U1 cells treated with control siRNA or NELF-B siRNA, and ChIP was performed using anti-acetylated histone H4 antibody or normal rabbit IgG (NS). Immunoprecipitated DNA was isolated and analyzed by PCR using primers specific for a region of the HIV LTR from ϩ45 to ϩ248. Input DNA was analyzed as a control of total DNA added to each reaction. D, intensity of bands as quantified by densitometry. Data are representative of four independent experiments. the positioned nucleosome. NF-B may be required for this initial phase of transcriptional activation because it has been shown to interact with P-TEFb (47) and/or transcription factor IIH (48), which includes the CDK7 kinase, thus providing a means for overcoming NELF-mediated repression occurring ϳ40 nucleotides downstream from the start site. If there is sufficient cellular activation, modest HIV transcription will lead to Tat expression. Tat then becomes the dominant regulator in transcription elongation by recruiting P-TEFb, which phosphorylates the C-terminal domain of pol II and possibly the Spt5 subunit of DSIF. Furthermore, Tat may directly regulate pol II pausing. NELF might still impose a delay to elongation when pol II completes transcription of TAR to allow time for recruitment of the P-TEFb⅐Tat complex to the TAR element for subsequent phosphorylation of pol II, DSIF, and NELF-B by P-TEFb (20). Transcription of HIV is further activated by changes in chromatin mediated by the association of Tat with chromatin remodeling complexes, including SWI/SNF and histone acetyltransferase proteins, to assure robust processive HIV transcription.