The SWI/SNF Chromatin-remodeling Complex Is a Cofactor for Tat Transactivation of the HIV Promoter*

Tat is a critical viral transactivator essential for human immunodeficiency virus (HIV) gene expression. Activation involves binding to an RNA stem-loop structure and recruitment of the positive transcription elongation factor b. Tat also induces the remodeling of a single nucleosome in the HIV promoter. However, the mechanism of this remodeling has remained unclear. Knockdown of INI-1 and BRG-1, two components of the SWI/SNF chromatin-remodeling complex, suppressed Tat-mediated transactivation. Cells lacking INI-1 (G401 and MON) or BRG-1 (C33A) exhibited defective transactivation by Tat that was restored upon INI-1 and BRG-1 expression, respectively. Tat was co-immunoprecipitated with several SWI/SNF subunits, including INI-1, BRG-1, and β-actin. The SWI/SNF complex interacted with the integrated HIV promoter in a Tat-dependent manner. We also found that INI-1 and BRG-1 synergized with the p300 acetyltransferase to activate the HIV promoter. This synergism depended on the acetyltransferase activity of p300 and on Tat Lys50 and Lys51. In conclusion, Tat-mediated activation of the HIV promoter requires the SWI/SNF complex in synergy with the coactivator p300.

Infection by human immunodeficiency virus (HIV) 2 sets in motion a complex series of actions that result in the efficient transcription of the viral genome. Once the virus is integrated into the host genome, nucleosomes are deposited at specific positions within the HIV promoter region (1). A large nucleosome-free region is present between nucleosome (nuc)-0 and nuc-1 and contains binding sites for transcription factors such as NF-B and Sp1 and other basal transcription factors. Tran-scription is initiated within this nucleosome-free region. Transcriptional activation of the HIV promoter is associated with the remodeling of nuc-1, which is positioned immediately downstream of the transcription start site (1).
In the early phase of HIV infection, cellular transcription factors activate transcription from the HIV promoter. However, the basal HIV promoter shows a striking elongation defect, resulting in the accumulation of short transcripts corresponding to the first ϳ50 transcribed nucleotides. This elongation defect is presumed to occur because of deficient loading of the transcription elongation complex pTEFb at the HIV promoter. We propose that the presence of nuc-1 (immediately downstream of the transcription start site) accentuates the elongation defect of the polymerase complex assembled at the HIV promoter.
However, the elongation defect of RNA polymerase II assembled at the HIV promoter is not absolute, and basal transcription leads to the accumulation of the viral Tat protein, a potent transactivator. Tat binds to TAR (an RNA stem-loop in the nascent viral RNA) and recruits pTEFb (which contains CDK9 and cyclin T 1 ). The recruitment of pTEFb leads to phosphorylation of the C-terminal domain of RNA polymerase II and increased transcriptional elongation of the HIV promoter. Efficient transcription elongation of the HIV genome in response to Tat leads to more Tat synthesis and generates a Tat-dependent positive feedback loop (2).
Tat expression also leads to the remodeling of nuc-1 (1,3). This remodeling is thought to remove an obstacle to RNA polymerase II elongation. Both Tat activities (pTEFb recruitment and nuc-1 remodeling) are thought to synergize in enhancing the ability of RNA polymerase II to elongate. The molecular mechanism of this Tat-induced nucleosome-remodeling event has remained unclear.
Chromatin-modifying complexes are classified into two main groups. The first contains factors that mediate covalent modifications of histones. The N-terminal tails of histone proteins are subject to extensive post-transcriptional modifications, including acetylation, phosphorylation, and methylation. The interaction of Tat with a number of histone acetyltransferase complexes such as p300/CBP, p300/CBP-associated factor (PCAF), and human GCN5 and their relevance to Tat-mediated activation of the HIV promoter have been established (4 -11). The complexes acetylate the N-terminal tails of histones of nucleosomes at the HIV promoter, inducing destabilization of histone-DNA contacts and thus facilitating transcrip-tion. In addition, Tat itself is subject to modification by acetyltransferases (6, 10 -12).
The second group of chromatin-modifying complexes comprises proteins that use the energy from ATP hydrolysis to change the location or conformation of nucleosomes, resulting in increased DNA accessibility within a nucleosomal array. One family of remodeling complexes, the SWI/SNF family, has BRM, or the closely related BRG-1, as the catalytic subunit (13)(14)(15)(16). In addition to BRG-1 or BRM, the complexes include core subunits, which appear to be common to all SWI/SNF complexes, as well as specific subunits. Some subunits appear to be expressed in a tissue-specific manner (14,17).
The dramatic chromatin remodeling of nuc-1 that occurs at the HIV promoter in response to Tat suggests that Tat may recruit an ATP-dependent chromatin-remodeling complex to the HIV long terminal repeat (LTR) to facilitate transcription. An attractive candidate for the regulation of HIV transcription and chromatin remodeling is the host cell factor INI-1 (integrase interactor-1), also known as SNF5 or BAF47, a core subunit of the SWI/SNF chromatin-remodeling complex (13). INI-1 binds specifically to HIV integrase and is found packaged in the infecting viral particle (18,19). After HIV infection, cellular INI-1 rapidly translocates from the nucleus to the cytoplasm, where it interacts with the HIV pre-integration complex (20). Here, we report that Tat recruits the SWI/SNF complex to the HIV promoter and is necessary for Tat-mediated activation of the HIV promoter.
Establishment of Stable Cell Lines-To create polyclonal cell lines that contained an integrated LTR-luciferase-GFP reporter, G401 and MON cell lines were infected with viral LTR-luciferase-GFP. Vesicular stomatitis virus G-pseudotyped particles were produced as described (25) using vesicular stomatitis virus envelope, the NL4-3 packaging vector, and the retroviral vector LTR-EGFP-IRES-luciferase (pEV677). Fortyeight hours after infection, the cells were transfected with CMV-Tat to stimulate GFP expression in cells containing the integrated LTR-luciferase-GFP reporter. After 24 h, GFP-positive cells were sorted by flow cytometry and maintained in culture until they became GFP-negative upon dilution of the CMV-Tat expression vector. G401 and MON cells containing integrated LTR-luciferase-GFP were then used in transient transfection experiments.
HCT116 cell lines expressing doxycycline-inducible short hairpin RNA against the INI-1 gene were generated essentially as described (26). Briefly, a monoclonal Tet repressor-expressing HCT116 cell line (HCT116TR) was first generated in accordance with the manufacturer's instructions with pcDNA 6 TR (Invitrogen) and blasticidin selection. HCT116TR cells were then cotransfected with four pTER-INI-1 (for Tetinducible RNA interference against INI-1) vectors generated by cloning gene-specific oligonucleotides against the INI-1 gene into pTER (26). Zeocin-resistant doxycycline-inducible INI-1 knockdown HCT116 cell lines were then tested for their ability to knock down INI-1 by Western blotting.
For immunoprecipitation of the INI-1, BRG-1, and PKD-1 complexes, 3 g of polyclonal antibody was incubated overnight with 5 mg of Jurkat cell lysate at 4°C and bound to 60 l of protein A-agarose beads for 2 h at 4°C, followed by extensive washes with immunoprecipitation buffer. The BRG-1, INI-1, and PKD-1 complex-coated beads were incubated for 2 h with reticulocyte lysate-expressed 35 S-labeled Tat, washed extensively with immunoprecipitation buffer, and analyzed by SDS-PAGE and autoradiography.
For Tat immunoprecipitation with pTEFb and SWI/SNF, 293T cells were transfected with control pcDNA3.1 or N-terminally FLAG-tagged wild-type Tat (pEV280) or mutant Tat(K50R/K51R) (pEV537) in the presence or absence of p300 expression vector; 36 h after transfection, cells were stimulated with 1 M TSA and 5 mM nicotinamide for 6 h. Cells were then harvested and lysed in phosphate lysis buffer (PLB; phosphatebuffered saline, 2 mM EDTA, 1% Triton X-100, 0.5 mM dithiothreitol, 1 M TSA, 5 mM nicotinamide, and protease inhibitor mixture). Lysates were centrifuged, and 2 mg of protein lysate was incubated overnight with 20 l of M2-agarose beads at 4°C on a rotator. After extensive washing with PLB, the beads were resuspended in SDS loading buffer containing 2-mercaptoethanol and separated on an SDS-polyacrylamide gel. Coprecipitated proteins were identified by immunoblotting with the indicated antibodies.
For GST pulldown experiments, GST fusion proteins were expressed in bacteria and purified on glutathione beads. GST-BRG-1 deletion peptides immobilized on beads were then incubated with synthetic acetylated or unacetylated biotinylated Tat in PLB for 2 h at 4°C on a rotator. The beads were washed extensively with PLB and wash buffer (25 mM Hepes (pH 7.9), 400 mM KCl, 1 mM EDTA, 5 mM MgCl 2 , 5% glycerol, 1% Nonidet P-40, 0.5 mM dithiothreitol, 1 M TSA, 1 mM nicotinamide, and protease inhibitor mixture) before addition of SDS loading buffer and electrophoresis. Acetylated and unacetylated biotinylated Tat proteins were detected by Western blotting using horseradish peroxidase-conjugated anti-streptavidin antibody.
Transient Transfection and Luciferase Assays-G401, C33A, MON, and doxycycline-inducible INI-1 knockdown HCT116 cells were seeded at a density of 5 ϫ 10 5 cells/35-mm plate and transfected the next day with FuGENE 6 reagent (Roche Applied Science) according to the manufacturer's instructions. Typically, transient transfections were carried out with 200 ng of LTR-luciferase reporter plasmid and expression vectors CMV-FLAG-Tat (5-30 ng), Rous sarcoma virus-FLAG-INI-1 (50 -500 ng), CMV-BRG-1 (50 ng), CMV-p300 (50 ng), and CMV-⌬p300 (50 ng) as indicated. The total DNA amount was adjusted using the corresponding empty vectors. Transfection efficiencies were Ͼ50% for all assays. Transfection efficiency was normalized using the cotransfected Renilla luciferase activity as an internal control. The cells were lysed after 24 h using luciferase lysis buffer (Promega Corp.), and luciferase activities were measured using the Dual-Luciferase reporter assay system with a Lumat LB 9501 luminometer (Berthold Technologies GmbH, Bad Wildbad, Germany). The data shown are from representative experiments conducted a minimum of three times and in triplicate.
AMAXA Nucleofection-Nucleofection of Jurkat cell clones was conducted using Nucleofector Kit R and program O28. Jurkat cell clones A72 and A2 were split to 300,000/ml 24 h before AMAXA nucleofection. Cells (5 ϫ 10 6 ) were spun at 1000 rpm for 10 min at room temperature, resuspended in 100 l of solution R, and nucleofected with 2 g of either short interfering RNA (siRNA) or expression plasmid using program O28. Nucleofected cells were resuspended in 500 l of prewarmed serum-free RPMI 1640 medium lacking antibiotics and allowed to recover at 37°C in a 5% CO 2 incubator for 15 min, and 4 ml of prewarmed complete RPMI 1640 medium was added to the cells.
Flow Cytometry-Samples were analyzed on a FACSCalibur flow cytometer using the CellQuest program (BD Biosciences). Forward versus side scatter profiles were used to define the live population. Cells were further gated using forward scatter versus FL1 to differentiate between GFP-positive and GFP-negative cells.
RNA Interference-Pre-designed Dharmacon siRNA pools targeting transcripts of the human SNF5 (INI-1) and BRG-1 genes, as well as a control siRNA pool, were used to knock down the respective genes in Jurkat A72 cells. siRNA was delivered into Jurkat cell clones by nucleofection (AMAXA). siRNA (2 g) was used to nucleofect 5 million cells; and 24, 48, and 72 h after nucleofection, protein levels were examined by Western blot analysis.

RNA Interference-mediated Depletion of INI-1 and BRG-1
Compromises Tat Transactivation-To examine whether INI-1-containing chromatin-remodeling complexes are involved in the transcriptional activation of the HIV promoter, we used a cell line containing a doxycycline-inducible short hairpin RNA against the INI-1 gene. Cells were transfected with the HIV LTR-luciferase reporter construct in the presence or absence of Tat. Induction of the short hairpin RNA against INI-1 by doxycycline led to nearly complete suppression of INI-1 expression (Fig. 1A). Under these conditions, Tat-mediated transactivation of the HIV promoter was compromised, whereas basal HIV promoter activity was unchanged (Fig. 1A).
To confirm this result, we used INI-1-deficient G401 cells. These cells were transfected with the LTR-luciferase reporter vector in the presence or absence of expression vectors for Tat and INI-1. In the absence of Tat, INI-1 had no effect on basal HIV promoter activity (Fig. 1B). Expression of INI-1 enhanced the HIV LTR activation mediated by Tat (Fig. 1B). Expression of both Tat and INI-1 was confirmed by Western blotting (Fig.  1B). These results indicate that INI-1 is necessary for optimal Tat activation of the HIV promoter.
Next, we studied the role of INI-1 and BRG-1 in HIV promoter activity in the Jurkat A72 T cell line. This clonal cell line was generated by infection of Jurkat cells with viral particles containing an HIV retroviral vector lacking the tat gene. In this vector, the HIV promoter drives expression of GFP (25). siR-NAs specific for INI-1 or BRG-1 were transfected by nucleofection, resulting in transfection of ϳ80% of the cells (data not shown) and leading to the specific depletion of either INI-1 or BRG-1 (Fig. 1C ). Notably, depletion of INI-1 did not affect the expression of BRG-1, and conversely, depletion of BRG-1 did not deplete INI-1 (Fig. 1C ). Cells were transfected with an expression vector for Tat or the control empty vector, and Tat-mediated transactivation was assessed by measuring GFP expression by flow cytometry. Depletion of either INI-1 or BRG-1 suppressed Tat-mediated transactivation of the HIV promoter as indicated by a decrease in the percentage of GFPpositive cells (Fig. 1D). Similar results were also obtained when the mean fluorescence intensity was measured (siRNA control, mean fluorescence intensity ϭ 92.9; siRNA for BRG-1, mean fluorescence intensity ϭ 49.7; and siRNA for INI-1, mean fluorescence intensity ϭ 58.6), indicating that depletion of INI-1 and BRG-1 decreases Tat-mediated transcriptional activation of the LTR. To ensure that the observed decrease in transactivation was not due to lowered activity of the CMV promoter and therefore lower Tat expression, we transfected Jurkat A72 cells lacking either INI-1 or BRG-1 with a CMVluciferase reporter vector. Luciferase values in cells with or without INI-1 and BRG-1 were similar (data not shown). Together, these results demonstrate that Tat activation of the LTR depends on INI-1 and BRG-1. In addition, the reduced transactivation activity of Tat in the absence of BRG-1 suggests that INI-1 plays a role in the transactivation as a subunit of the BRG-1 complex.
Tat Interacts with Subunits of the SWI/SNF Chromatin-remodeling Complex-Our results above showed that Tat requires INI-1 and BRG-1 for transcriptional activation. To determine whether Tat interacts with SWI/SNF components in cells, we used Jurkat A2 cells containing a latently integrated LTR-Tat-IRES-GFP virus (24). This cell line expresses detectable levels of FLAG-tagged Tat protein under the control of the HIV promoter only after stimulation of these cells with PMA ( Fig. 2A, upper panel ). We immunoprecipitated Tat from Jurkat A2 cell extracts after PMA stimulation and probed for association of Tat with endogenous components of SWI/SNF. The immunoprecipitations showed that Tat specifically associated with the core components of SWI/SNF: INI-1, BRG-1, and ␤-actin ( Fig. 2A). Interestingly, Tat co-immunoprecipitated with BRG-1 but not BRM ( Fig. 2A), indicating that Tat interacts specifically with the SWI/SNF complex containing BRG-1 as its catalytic subunit. We used the unrelated proteins 14-3-3 and PKD-1 as controls, and they did not co-immunoprecipitate with Tat.
We also examined the interaction between Tat and SWI/ SNF by immunoprecipitating the endogenous INI-1 or BRG-1 complexes using antibodies specific for each protein. Western blot analysis of the immunoprecipitated proteins confirmed that BRG-1 co-immunoprecipitated with INI-1 and vice versa (Fig. 2B, upper panel). Immunoprecipitated complexes bound to protein A-Sepharose beads were incubated with in vitro translated 35 S-labeled Tat, washed, and analyzed by autoradiography after SDS-PAGE. In vitro 35 S-labeled Tat efficiently bound to both the BRG-1 and INI-1 complexes, but not to control or PKD-1-coated beads (Fig. 2B, lower panel ).

INI-1 Synergizes with Tat and the Transcriptional Coactivator p300-Previous studies showed that p300 is a cofactor in the
Tat-dependent activation of the HIV LTR (4 -6, 9). To test whether INI-1 and p300 act synergistically, we transfected the INI-1-negative G401 and MON cells with the HIV LTR-luciferase reporter plasmid with or without expression vectors for Tat, INI-1, p300, and a mutant p300 protein with a defective histone acetyltransferase domain (p300⌬HAT) (Fig. 3A). In the absence of Tat, INI-1, p300, and p300⌬HAT did not significantly affect basal reporter activity. Expression of Tat alone mediated relatively weak (ϳ10-fold) activation of transcription in both the G401 and MON cell lines. Expression of INI-1 or p300 individually with Tat present at limiting concentrations resulted in a slight increase in transcription compared with Tat alone. Strikingly, concomitant expression of INI-1, p300, and Tat strongly activated the HIV LTR to 450-fold in G401 cells and to 100-fold in MON cells (Fig. 3A). The synergistic activation of the HIV promoter by Tat, INI-1, and p300 was abolished when p300⌬HAT was used (Fig. 3A). The cooperative effects on transcription were specific for the HIV promoter, as the effect was not observed with the CMV promoter (Fig. 3A).
During the HIV life cycle, activation of the HIV LTR takes place after integration of the HIV genome in the host cell genome. Therefore, we examined the effect of INI-1 on the regulation of the integrated HIV promoter. We generated viral particles containing the retroviral vector LTR-EGFP-IRES-luciferase-LTR (Fig. 3B) and infected the INI-1-deficient G401 and MON cell lines. Polyclonal cell lines containing an integrated LTR-luciferase reporter were obtained and transiently transfected with the expression vectors for Tat, INI-1, p300, and p300⌬HAT. In this system, the same synergy was noted between INI-1, p300, and Tat (Fig. 3B). This synergy was also dependent on the histone acetyltransferase domain of p300 (Fig. 3B). In conclusion, these results reveal a striking cooperation between the ATP-dependent chromatin-remodeling subunit INI-1 and the acetyltransferase p300 during Tat-directed transcription of the HIV LTR.
Tat Lys 50 and Lys 51 Are Necessary for the Synergy between Tat, p300, and SWI/SNF-The p300 transcriptional coactivator acetylates Tat at Lys 50 (6,11,12). This acetylation mediates the dissociation of Tat from TAR (11,12,27). Because our results indicated that the synergism between INI-1 and p300 in Tat activation of the LTR depends on the acetyltransferase activity of p300, we tested the role of Tat Lys 50 in SWI/SNF recruitment and LTR activation. Although Lys 50 is the primary target of acetylation in Tat, its mutation to alanine leads to the secondary acetylation of Lys 51 (6). Introduction of both mutations is therefore necessary to abrogate acetylation. Accordingly, we tested the ability of a mutant Tat protein in which both Lys 50 and Lys 51 were substituted with arginine (Tat(K50R/K51R)) to recruit SWI/SNF. Notably, the synergistic activation by p300 and INI-1 was abolished in both G401 and MON cells when coexpressed with Tat(K50R/K51R) (Fig. 4A).
We also used the BRG-1-deficient C33A cell line to examine the synergism between Tat, BRG-1, and p300 (Fig. 4B). The cells exhibited low LTR activation by Tat that was increased upon BRG-1 expression (Fig. 4B). Concomitant expression of BRG-1, p300, and Tat strongly activated the LTR to 70-fold. This synergism was abolished in the presence of the p300 catalytic mutant (p300⌬HAT) as well as the Tat(K50R/K51R) mutant (Fig. 4B).
These results raised the possibility that the interaction between SWI/SNF and Tat is disrupted if Tat cannot be acetylated. To determine whether the interaction of Tat with SWI/ SNF is modulated by acetylation, wild-type Tat or the Tat(K50R/K51R) mutant was cotransfected in 293T cells with or without p300. To prevent deacetylation of Tat, cells were further treated with nicotinamide, an inhibitor of class III histone deacetylases (28), and trichostatin A, an inhibitor of class I and II histone deacetylases (Fig. 4C). We found that Tat association with BRG-1 increased in the presence of p300 as shown by co-immunoprecipitation of BRG-1 with Tat (Fig. 4C ). The same treatment markedly increased Tat acetylation (Fig. 4C ). We also examined the effect of Tat acetylation on its interaction with subunits of the pTEFb complex. A concomitant decrease in the interaction of acetylated Tat with cyclin T 1 and CDK9 was observed (Fig. 4C ). Notably, the Tat(K50R/K51R) mutant did not display increased affinity for BRG-1 in response to p300. In agreement with these data, the affinity of the Tat(K50R/ K51R) mutant for pTEFb subunits was not decreased in response to p300 (Fig. 4C). Mutation of Tat residues Lys 50 and Lys 51 to arginine decreased Tat acetylation significantly, but not completely, consistent with the existence of other Tat acetylation sites (29).
A likely candidate subunit in the SWI/SNF complex to directly interact with acetylated Tat is BRG-1. BRG-1 contains a C-terminal bromodomain, a recognition motif for acetylated lysine-containing proteins. Acetylated Tat has been shown to specifically interact with another transcriptional coactivator, PCAF, via its bromodomain (7). According to structure-based sequence alignment of bromodomains, the BRG-1 bromodomain shares the highest sequence homology with the PCAF bromodomain, including conserved key amino acid residues important for Tat binding (7). To test this model, we used GST fusion proteins made to overlap 300-amino acid stretches of the BRG-1 protein (Fig.  4D, upper panel ) (23). The BRG-1 bromodomain is located within the C-terminal domain of BRG-1 (amino acids 1400 -1700). Fusion proteins were immobilized on glutathione beads and tested for binding to synthetic biotinylated Tat protein acetylated at Lys 50 or to unacetylated Tat protein (27). The BRG-1 bromodomain specifically bound acetylated Tat (Fig. 4D). This binding was critically dependent on Tat acetylation at Lys 50 , as unacetylated Tat did not bind to the BRG -1 bromodomain-containing fragment. An additional weaker interaction was also observed between BRG-1 amino acids 400 -700 and acetylated Tat. Together, these results are consistent with the model that acetylated Tat recruits the SWI/SNF complex to the HIV LTR via BRG-1.
Tat Mediates the Recruitment of the SWI/SNF Complex to the HIV Promoter in Vivo-To demonstrate that Tat mediates the recruitment of SWI/SNF to the HIV promoter in vivo, we performed chromatin immunoprecipitation assays. We stimulated the Jurkat cell line A2 containing a latently integrated LTR-Tat-IRES-GFP virus (24) with PMA. PMA stimulation of the A2 cells resulted in GFP expression in 86% of the cells (Fig.  5A). Tat was expressed to detectable levels in response to PMA stimulation at 30 min and peaked at 4 h post-stimulation (Fig.  5B). Chromatin was prepared from cells at 0, 0.5, 4, and 8 h post-stimulation and subjected to chromatin immunoprecipitation with antibodies specific for BRG-1, p300, and YY1. PCR analysis of the immunoprecipitated material with primers specific for the HIV promoter indicated that BRG-1 and p300, while initially absent from the LTR, were specifically recruited to the HIV promoter in response to PMA stimulation (Fig. 5B). In contrast and in agreement with published observations (30), the transcriptional repressor YY1 was bound to the HIV promoter under basal conditions and was displaced in response to PMA (Fig. 5B). These results suggest that the SWI/SNF complex is recruited to the LTR in response to Tat. However, in this experiment, we could not exclude the possibility that recruitment of SWI/SNF to the LTR occurs indirectly via other LTR activators in response to PMA.
To demonstrate that SWI/SNF recruitment occurs directly via Tat and in a Tat-dependent manner, we used Jurkat A72 cells containing an integrated LTR-GFP virus that lacks Tat (24) in chromatin immunoprecipitation experiments (Fig. 5, C  and D). We transfected the Jurkat A72 cells with an expression vector for Tat or the control empty vector and observed that 46% of the cells expressed GFP 16 h after Tat transfection (Fig.  5C ). Tat was expressed 4.5 h post-transfection (Fig. 5D). Chromatin was prepared from cells 4.5 h after introduction of the Tat expression vector and subjected to chromatin immunoprecipitation with antibodies specific for BRG-1 and YY1. PCR analysis of the immunoprecipitated material with primers specific for the HIV promoter indicated that BRG-1 was specifically recruited to the HIV promoter, whereas YY1 was displaced in response to Tat expression (Fig.  5D). These results demonstrate that the SWI/SNF complex is specifically recruited to the LTR by Tat in vivo.

DISCUSSION
We have shown that the SWI/ SNF chromatin-remodeling complex is a cofactor for Tat activation of the HIV promoter. Knockdown of INI-1 and BRG-1, two critical components of mammalian chromatin-remodeling complexes, suppresses Tat-mediated transactivation. Similarly, cells without INI-1 or BRG-1 exhibit defective transactivation by Tat that can be rescued by INI-1 or BRG-1 expression. Tat specifically interacts with several SWI/SNF subunits, INI-1, BRG-1, and ␤-actin. Similarly, SWI/SNF interacts with the integrated HIV promoter in a Tat-dependent manner. In addition, we found that INI-1 and BRG-1 act synergistically with the p300 acetyltransferase to activate the HIV promoter. This synergism is critically dependent on the histone acetyltransferase activity of p300 and on Tat Lys 50 and Lys 51 .
Studies in several experimental systems have revealed that enzymes that post-translationally modify chromatin proteins and chromatin-remodeling complexes are recruited in a stepwise fashion to specific promoters (31)(32)(33)(34)(35). The combinatorial assembly of transcription factors and these chromatin-modifying proteins mediates a precise transcriptional response. However, chromatin-modulating factors do not appear in a set order at all genes. For example, the human SWI/SNF complex is recruited at the interferon-␤ promoter after the histone acetyltransferases CBP and PCAF, whereas the yeast SWI/SNF chromatinremodeling complex is recruited first and is required for the subsequent recruitment of the SAGA histone acetyltransferase complex at the yeast HO promoter (31,32).
We observed that Tat, p300, and SWI/SNF synergistically activate the HIV promoter. This synergy depends on the acetyltransferase activity of p300 and on Tat Lys 50 and Lys 51 . We (6) and others (11) have reported that p300 acetylates Tat at Lys 50 , a modification that plays a significant role in Tat transcriptional C, Tat co-immunoprecipitation with pTEFb and SWI/SNF is modulated by Tat acetylation. Tat (wild-type or K50R/L51R) was immunoprecipitated (IP) using anti-FLAG antibody, and the immunoprecipitated material was analyzed by Western blotting using antibodies specific for BRG-1, cyclin T 1 , and CDK9. Tat acetylation levels were assessed using anti-acetyllysine antibody. All proteins were expressed at similar levels under the different experimental conditions (Input). D, the BRG-1 bromodomain binds preferentially to Tat acetylated at Lys 50 . GST fusions to the N terminus of the indicated BRG-1 deletion fragments were expressed in bacteria. BRG-1 deletions were purified and immobilized on glutathione beads (upper panel ) and incubated with acetylated (Ac-Tat) or unacetylated biotinylated Tat. Beads were washed; bound proteins were separated by SDS-PAGE; and bound Tat was visualized using horseradish peroxidase-conjugated streptavidin. activity in the HIV promoter. We have proposed previously that Tat acetylation serves as a molecular switch that coordinates the recruitment of different cofactors to the HIV promoter (7,8,27,36). Early in the transcription cycle, unacetylated Tat binds to the RNA element TAR and recruits pTEFb, including CDK9 and cyclin T 1 , to the HIV promoter. Tat bound to the HIV promoter becomes acetylated by p300, leading to the dissociation of the ternary complex between Tat, TAR, and pTEFb.
The results presented here are consistent with the model that acetylated Tat facilitates the recruitment of the SWI/SNF complex to the HIV promoter, leading to nuc-1 remodeling. Acetylated Tat preferentially interacts with another histone acetyltransferase (PCAF) via its bromodomain (7,8,37). We cannot presently determine whether acetylated Tat recruits PCAF and SWI/SNF sequentially, in a mutually exclusive manner, or simultaneously. The orthologs of these two protein complexes in Saccharomyces cerevisiae (GCN5 and SWI/SNF) cooperate in the transcriptional activation of several promoters (38 -42). Evidence has been presented that the bromodomain of GCN5 stabilizes the SWI/SNF complex in an artificial promoter and is required for nucleosome remodeling and transcriptional activation (43). This observation indicates that, in some cases, both GCN5 and SWI/SNF may bind together to a given promoter (43). Because Tat can also interact with PCAF, it is possible that PCAF-mediated hyperacetylation of the HIV promoter further stabilizes SWI/SNF binding to the HIV promoter by creating a hyperacetylated chromatin environment. Indeed, PCAF targets histones for acetylation in the HIV promoter (44).
Our observations represent the first example of recruitment of a chromatin-remodeling complex to a promoter via an RNAbinding protein. The HIV Tat protein represents a unique transcriptional activator targeted downstream of the transcription start site via its interaction with the TAR RNA element. No rationale has emerged thus far to explain why Tat has evolved as an RNA-binding transactivator instead of a more classical DNA-binding protein. It is intriguing that nuc-1, the nucleosome that is remodeled by SWI/SNF in a Tat-dependent manner, is located immediately downstream of the transcription start site. Our observations that Tat contributes to the recruitment of a chromatin-remodeling complex could provide a rationale for the need for Tat to function via RNA instead of DNA. SWI/SNF bound to TAR via Tat would be positioned immediately at the site of nuc-1 and could explain the selective remodeling of nuc-1 by Tat. We cannot exclude the possibility that SWI/SNF also interacts with Tat when the latter is bound to the elongating polymerase (45). However, such a model is inconsistent with the observations that a single nucleosome (nuc-1) is remodeled at the level of the HIV promoter and that remodeling of nuc-1 is insensitive to ␣-amanitin, a specific inhibitor of RNA polymerase II (1).
Recruitment of the SWI/SNF complex to the HIV promoter in response to Tat is accompanied by removal of the YY1 transcriptional repressor. Previous reports have documented that YY1 is recruited to the HIV promoter at the transcription start site via its specific interaction with LSF (30,46). YY1 specifically recruits HDAC-1 and is likely to contribute to histone hypoacetylation at the level of nuc-1 in the absence of Tat (30). It is not entirely clear how Tat leads to the displacement of YY1, but nuc-1 remodeling could lead to a change in the affinity of the DNA for YY1 and its dissociation from the HIV promoter. Alternatively, Tat may mediate the displacement of YY1 from the HIV promoter independently of nuc-1 remodeling. Irrespective of the mechanism, removal of YY1 from the HIV promoter and the resulting loss of HDAC-1 could further contribute to the hyperacetylation of the HIV promoter mediated by PCAF. Such a mechanism could contribute to the hyperacetylation of the HIV promoter that has been observed in response to Tat expression (4).
The identification of the SWI/SNF chromatin-remodeling complex as a Tat cofactor provides a mechanism for the longstanding observation that Tat leads to the selective remodeling Chromatin immunoprecipitation of BRG-1, p300, and YY1 with the nuc-1 region of the HIV LTR in vivo in response to PMA stimulation is also shown. Formaldehydecross-linked chromatin from Jurkat A2 cells at 0, 0.5, 4, and 8 h post-stimulation was immunoprecipitated with the specified antibodies and subjected to PCR as described under "Experimental Procedures." PCR products were run on a 2% agarose gel. Input represents 0.5% of the chromatin used in immunoprecipitation. Ab, antibody. C, Jurkat A72 cells containing an integrated LTR-GFP virus lacking Tat were nucleofected with either empty or Tat expression vectors and analyzed by fluorescence-activated cell sorting after 24 h. The percentage of GFP-positive cells in the absence or presence of Tat is shown. D, chromatin immunoprecipitation of BRG-1 and YY1 with the nuc-1 region of the HIV LTR in vivo in response to Tat expression is shown. Tat expression was visualized by Western blot analysis using anti-FLAG antibody. Chromatin from Jurkat A72 cells was cross-linked with formaldehyde 5 h after nucleofection with either the CMV-Tat or empty expression vector. Chromatin was immunoprecipitated with the specified antibodies and subjected to PCR, and the products were separated on a 2% agarose gel. Input represents 0.5% of the chromatin used in immunoprecipitation. SSC-H, side scatter-height.
of nuc-1. The recruitment of SWI/SNF via Tat and RNA represents a novel mechanism for the recruitment of a chromatinremodeling complex to a promoter. Further study of this process will contribute to providing an integrated understanding of HIV transcriptional regulation in the context of chromatin.