Transcriptional Repression of Human Immunodeficiency Virus Type 1 by AP-4*

Elucidation of the mechanism of transcriptional silencing of human immunodeficiency virus type 1 (HIV-1) provirus in latently infected cells is crucial to understand the pathophysiology of HIV-1 infection and to develop novel therapies. Here we demonstrate that AP-4 is responsible for the transcriptional repression of HIV-1. We found that AP-4 site within the viral long terminal repeat (LTR) is well conserved in the majority of HIV-1 subtypes and that AP-4 represses HIV-1 gene expression by recruiting histone deacetylase (HDAC) 1 as well as by masking TATA-binding protein to TATA box. AP-4-mediated transcriptional repression was inhibited by an HDAC inhibitor, tricostatin A, and could be exerted even at distant locations from the TATA box. In addition, AP-4 interacted with HDAC1 both in vivo and in vitro. Moreover, chromatin immunoprecipitation assays have revealed that AP-4 and HDAC1 are present in the HIV-1 LTR promoter in latently infected ACH2 and U1 cells, and they are dissociated from the promoter concomitantly with the association of acetylated histone H3, TBP, and RNA polymerase II upon TNF-α stimulation of HIV-1 replication. Furthermore, when AP-4 is knocked down by siRNA, HIV-1 production was greatly augmented in cells transfected with a full-length HIV-1 clone. These results suggest that AP-4 may be responsible for transcriptional quiescence of latent HIV-1 provirus and give a molecular basis to the reported efficacy of combination therapy of conventional anti-HIV drugs with an HDAC inhibitor in accelerating the clearance of HIV-1 from individuals infected with the virus.

Human immunodeficiency virus type 1 (HIV-1) 2 is a cytopathic retrovirus and the primary etiological agent of acquired immunodeficiency syndrome (AIDS) and related disorders. Among the various steps of viral life cycle, the step of transcription from HIV-1 provirus is conceived to be crucial for viral replication since amplification of the viral genetic information is attainable only through transcription. HIV-1 transcription is directed by the promoter located in the 5Ј long terminal repeat (LTR) of the integrated provirus and is controlled by cellular factors that bind to the multiple cis-regulatory elements located in the LTR as well as the virally encoded Tat protein (reviewed in Refs. 1 and 2). In cells chronically infected with HIV-1, activation of nuclear factor-B (NF-B) by external stimuli such as tumor necrosis factor (TNF)-␣ and its binding to LTR triggers the initiation of transcription of viral genes including Tat, which results in explosive HIV-1 replication (reviewed in Refs. 3 and 4). However, little is known how transcription from HIV-1 provirus remains silent during the viral latency.
There are multiple mechanisms known to be involved in the negative regulation of HIV transcription including elimination of transcriptional activator TATA-binding protein (TBP) transcriptional factor IID (TFIID) and the initiator protein complex by leader-binding protein (LBP)-1 (5) and YY-1 (6) that recruits histone deacetylase (HDAC) (7,8), and actions of transcription factors that interact with the negative regulatory element (NRE) located from Ϫ340 to Ϫ184 of HIV-1 LTR (2,9). Regarding the action of NRE, the mechanism by which NRE exerts its negative effect on transcription remains unknown because most of the transcription factors that interact with NRE are transcriptional activators. In addition, an in vitro study has revealed a potential role of activator protein (AP)-4 in blocking the TBP binding to TATA box (10). However, biological significance of this finding has not been clarified although the sequence comparison has revealed conservation of AP-4 sites in the majority of HIV-1 isolates (Fig. 1A) (11)(12)(13).
The HIV-1 LTR TATA box is located at Ϫ27 to Ϫ23 relative to the transcription initiation site (2,(11)(12)(13)(14). TFIID interacts with TATA box and is crucial for HIV-1 gene expression (9, 14 -16). TFIID contains the 38-kDa TBP as the major component and induces transcriptional initiation by interacting with other general transcription factors and recruiting RNA polymerase II (RNAPII) (17). TBP (TFIID) also serves as the target of DNA-binding factors binding to the cis-regulatory elements within HIV-1 LTR in both positive and negative fashions and thus determines its promoter activity (reviewed in Refs. 16, 18, and 19).
AP-4 is a ubiquitously expressed transcription factor of the basic helix-loop-helix leucine-zipper (bHLH-Zip) subgroup of bHLH proteins and binds to the symmetrical DNA sequence 5Ј-CAGCTG-3Ј (20,21). AP-4 site is found adjacent (Ϫ21/Ϫ16) to the HIV-1 TATA box (Ϫ27/Ϫ23) (10). Although AP-4 was initially identified as a cellular protein that binds to the simian virus 40 (SV40) enhancer and activates the viral late gene transcription (21), transcriptional repression by AP-4 was reported in a number of other genes including angiotensinogen (21) and E7 oncoprotein of human papillomavirus type 16 (23). In addition, we recently found that AP-4 negatively regulates transcription of 8-oxo guanine DNA glycosylase 1 (OGG1) gene (24). However, because AP-4 sites are not located adjacent to the TATA box in these promoters, the molecular mechanism of its repressive action is yet to be clarified.
In this study we investigated the role of AP-4 in HIV-1 gene expression. Here we show that AP-4 represses HIV-1 transcription by recruiting HDAC1 as well as by masking the TBP to the HIV-1 TATA box. Biological and therapeutic implications are discussed.

EXPERIMENTAL PROCEDURES
Cell Culture-CEM, HL-60, Jurkat, ACH2, and U1 cells were maintained at 37°C in RPMI 1640 (Sigma) with 10% fetal bovine serum (Sigma), penicillin (100 units/ml), and streptomycin (100 g/ml). To maintain the latency of the HIV-1 in ACH2 and U1, 20 M AZT was added in the culture medium and was excluded prior to experiments. Human embryonic kidney 293 cells were grown at 37°C in Dulbecco's modified Eagle's medium (Sigma) with 10% heat-inactivated fetal bovine serum, penicillin, and streptomycin.
Recombinant Protein and Purification-pGEX expression vector (Amersham Biosciences) was utilized to express glutathione S-transferase (GST) fusion proteins in bacteria . To generate pGEX-AP-4-expressing GST-AP-4, the AP-4 cDNA was amplified by PCR using pMyc-AP-4  as a template with oligonucleotide primer pairs: forward, 5Ј-CGG GAT  CCC GGA GTA TTT CAT GGT GCC CAC TCA G-3Ј, containing an  BamHI site; reverse, 5Ј-GGA ATT CCT CAG GGA AGC TCC CCG  TCC CCC G-3Ј, containing an EcoRI site. This product was digested with BamHI and EcoRI, and subcloned in-frame into pGEX-5X-3 vector at the BamHI/EcoRI sites. pGEX-AP-4 was transformed in Escherichiacoli strain DH5 and expression of recombinant GST-AP-4 protein was induced by 0.1 mM isopropyl-1-thio-␤-D-galactopyranoside at 25°C for 6 h. Recombinant GST proteins were purified by affinity chromatography on glutathione-agarose beads as described previously (26).
Electrophoretic Mobility Shift Assay (EMSA)-The experimental procedure was carried out as described previously (24). Purified recombinant TBP and GST-TBP proteins were purchased from Promega and Santa Cruz Biotechnology, respectively. The double-stranded DNA oligonucleotides corresponding to Ϫ42/ϩ4 of HIV-1 LTR (CD12) containing the binding sites of TBP and AP-4 and their mutants were synthesized. The wild-type and mutant oligonucleotide sequences (sense strand) were the following: wild-type (5Ј-CCC TCA GAT CCT GCA TAT AAG CAG CTG CTT TTT GCC TGT A-3Ј) and mutants (the underlined AP-4 site has been changed to TGACGG (m1), TAGCTC (m2), CAGCGC (m3), and CAGTCG (m4)) (Fig. 2B). These oligonucleotides were labeled using the 5Ј-end-labeling kit (Takara, Otsu, Shiga, Japan) in the presence of [␥-32 P]dATP (Amersham Biosciences). DNA binding reactions were performed at 30°C for 30 min for TBP and room temperature for 20 min for AP-4. Analysis of protein-DNA complexes was performed by electrophoresis in 6% native polyacrylamide gels with 0.5ϫ Tris borate-EDTA buffer at a constant voltage of 125 V at 4°C, followed by autoradiography. The specificity of DNA binding was assessed by preincubating with purified GST-AP-4, GST-TBP, or control GST proteins with specific antibodies or competitors for 20 min prior to the addition of the probe.
Anti-AP-4 Antibody-Anti-AP-4 antibody was obtained by immunizing rabbits with GST-AP-4 fusion protein as no immunoprecipitable anti-AP-4 antibody was currently available from any commercial source. The immunized rabbit anti-AP-4 sera were affinity-purified by passing through affinity columns, and the lack of immunoreactivity with GST and other E. coli components was confirmed.
Immunoprecipitation and Immunoblot Assays-The experimental procedures for immunoprecipitation and immunoblotting were performed as described (24,27). Briefly, cells were harvested with lysis buffer (25 mM HEPES-NaOH, pH 7.9, 150 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.3% Nonidet P-40, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). The lysates were cleared by centrifugation, and the supernatants were incubated with anti-AP-4 antibody overnight at 4°C. Immune complexes were washed three times with 1 ml of lysis buffer and antibody-bound proteins were dissolved by boiling in 2ϫ Laemmli sample buffer. After centrifugation, the supernatant proteins were separated by SDS-PAGE and transferred to nitrocellulose membrane (Hybond-C, Amersham Biosciences). The membrane was probed with anti-AP-4 antibody, and immunoreactive proteins were visualized by enhanced chemiluminescence (SuperSignal, Pierce). To evaluate the level of AP-4 protein, cells were similarly treated with the lysis buffer, and the cell lysates were analyzed by immunoblotting using anti-AP-4 or V5 antibody (Invitrogen).
Transfection and Luciferase Assay-293 cells cultured in 12-well plates were transfected using Fugene-6 transfection reagent (Roche Applied Science) as described previously (26,27). CEM and HL60 cells were transiently transfected by electroporation as reported (24). Briefly, 2 ϫ 10 7 cells/ml were electroporated with 2 g of CD12-luc together with 2 g of pCMV-Tat and indicated amounts of Myc-AP-4 in 400 l of serum-free RPMI using the Electro Cell Manipulator 600 (BTX Electroporation System) apparatus at 260 V/1050 farads. For the internal control, we employed pRL-TK, expressing Renilla luciferase under the control of the thymidine kinase promoter not containing the AP-4 site. The transfected cells were harvested, and the extracts were subjected to the luciferase assay using the Luciferase Assay System TM (Promega). All the experiments were carried out in triplicates, and the data were presented as the fold increase in luciferase activities (means Ϯ S.D.) relative to the control for three independent transfections.
In Vitro Binding Assay-An in vitro protein-protein interaction assay was carried out as described previously (26). Briefly, AP-4 and luciferase proteins were labeled with [ 35 S]methionine in vitro transcription/translation using the TNT wheat germ extract-coupled system (Promega) according to the manufacturer's protocol. Approximately 20 g of GST fusion proteins were immobilized on 20 l of glutathione-Sepharose beads and washed two times with 1 ml of modified HEMNK buffer (20 mM HEPES-KOH, pH 7.5, 100 mM KCl, 12.5 mM MgCl 2 , 0.2 mM EDTA, 0.3% Nonidet P-40, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride). After the final wash, 0.6 ml of beads suspension was incubated with radiolabeled proteins for 12 h at 4°C. The beads were then washed two times with 1 ml of HENMK buffer and two times with HENMK buffer containing 150 mM KCl. Bound radiolabeled proteins were eluted with 30 l of Laemmli sample buffer, boiled for 3 min, and resolved by 10% SDS-PAGE.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assay was performed according to the provider's protocol (Upstate Biotechnology) with some modifications as previously described (24). Briefly, cells were cross-linked with 1% formaldehyde for 10 min at room temperature, washed twice with ice-cold phosphate-buffered saline, and lysed for 10 min at 2 ϫ 10 6 cells in 200 l of SDS lysis buffer. The cross-linked chromatin was sheared by sonication 13 times for 10 s at one-third of the maximum power of microson XL sonicator (Wakenyaku, Co., LTD., Kyoto, Japan) with 20 s of cooling on ice between each pulse. Crosslinked and released chromatin fractions were precleared with salmon sperm DNA and protein A-agarose beads for 1 h, followed by immunoprecipitation with the desired antibodies overnight at 4°C. The immunoprecipitates were sequentially washed once with lysis buffer, twice with high salt buffer, twice with low salt buffer, and twice with TE buffer. After the wash, immune complexes were collected with salmon sperm DNA and protein A-agarose beads at room temperature for 1 h and extracted with 1% SDS, 0.1 M NaHCO 3 . The eluted samples were reverse cross-linked by proteinase K at 45°C for 1 h and treated with RNase at 37°C for 1 h. DNA was recovered by phenol/chloroform and chloroform extractions, and ethanol precipitation. Finally, DNA was dissolved in 30 l of TE buffer and subjected to PCR. The primer sequences used for PCR were the following: HIV-1 LTR (Ϫ109 to ϩ79): forward (5ЈTAC AAG GGA CTT TCC GCT GG-3Ј) and reverse (5Ј-TTG AGG CTT AAG CAG TGG G-3Ј); ␤-actin promoter (Ϫ980 to Ϫ915) (as a control): forward (5Ј-TGC ACT GTG CGG CGA AGC-3Ј) and reverse (5Ј-TCG AGC CAT AAA AGG CAA-3Ј). The number of PCR cycles was as the following: 33 PCR cycles for all the ChIP experiments and 24 PCR cycles for the input samples, in which PCR amplification was obtained under the linear range of AP-4 binding to the HIV-1 LTR DNA. For each reaction, 10% of cross-linked released chromatin was saved and reversed by proteinase K digestion at 45°C for 1 h followed by DNA extraction, and the recovered DNA was used as input control.
Antiviral Assay and Measurement of Viral p24 Antigen-Antiviral activity of AP-4 was evaluated based on the extent of inhibition of viral antigen expression in the culture supernatants of Jurkat or 293 cells transfected with a full-length HIV-1 molecular clone (pNL4-3) or mutant pNL4-3, in which AP-4 site is mutated. 293 cells were transfected with 0.1 g of pNL4-3, together with various amounts of plasmids encoding wild-type AP-4 or AP-4 mutants with Fugene-6 transfection reagent. For siRNA studies, 100 nM siRNAs were introduced with 0.1 g of pNL4-3 using Lipofectamine 2000 reagent. Jurkat cells were transfected by Nucleofector TM kit V for Jurkat cell (Amaxa Biosystems) according to the manufacturer's protocol. Briefly, 3 ϫ 10 6 cells were mixed with 0.2 g of wild-type or mutant pNL4-3 together with indicated amounts of FLAG-AP-4 in 100 l of Nucleofector TM solution V. These samples were transferred into a transfection cuvette and subjected to electroporation using program T-14. The transfected cells were incubated in culture flasks with a complete media for 36 h. Then, cells were incubated for an additional 24 h in the presence or absence of TNF-␣ (3 ng/ml). The p24 antigen level in the cell culture supernatant was measured by p24 antigen capture ELISA assay using a commercial kit (RETRO-TEK HIV-1 p24 Antigen ELISA kit; Zepto Metrix Corp., Buffalo, NY) as described previously (25).

AP-4 Competes with TBP for Binding to the HIV-1 TATA Element-
The HIV-1 TATA is located at nucleotide position from Ϫ27 to Ϫ23 relative to the transcription initiation site. The consensus AP-4 site, CAGCTG, is located Ϫ21 to Ϫ16 nucleotides immediately downstream of the TATA box. The AP-4 binding site in the HIV-1 LTR appears to be conserved in the majority of HIV-1 isolates (11)(12)(13). As shown in Fig.  1A, a majority of HIV-1 clones contain typical AP-4 binding sequence, CAGCTG, whereas it is mutated to CAGCCG in HIV-1 subtypes F1-F2, G, O, and 01-AE.
Because the AP-4 binding site is located close to the TATA box, we first examined the effect of AP-4 on the binding activity of TBP to the TATA box in vitro. To address this issue, recombinant AP-4 protein was produced and purified ( Fig. 1, B and C). As shown in Fig. 1D (left panel), EMSA analysis using a DNA probe (Ϫ42/ϩ4) containing both the TATA box and AP-4 site, showed that AP-4 blocked the TBP binding to TATA box in a dose-dependent manner (lanes 6 and 7). The control GST proteins did not alter the DNA binding activity of TBP (Fig. 1D, right panel). These results were consistent with a previous study by Ou et al. (10).
Repression of HIV-1 LTR Gene Expression by AP-4-Because AP-4 masks the TBP binding to the HIV-1 TATA box in vitro, we examined the effect of AP-4 on transcription from HIV-1 LTR. The luciferase reporter plasmid containing the HIV-LTR (CD12-luc) was cotransfected with an AP-4 expression vector (pMyc-AP-4) into CEM, HL-60, and 293 cell lines. As shown in Fig. 2A, the basal transcriptional level from HIV-1 LTR was inhibited by AP-4 in a dose-dependent manner in all the cell lines tested. Upon stimulation of HIV-1 promoter by TNF-␣, a physiological inducer of NF-B, AP-4 could similarly exert its negative effect. In addition, AP-4 also inhibited the Tat-induced HIV-1 gene expression in these cells.
To address whether the inhibitory effect of AP-4 depends on the presence of AP-4 site, we have created HIV-1 LTR mutants where the AP-4 binding site was mutated (Fig. 2B). As shown in Fig. 2C, EMSA confirmed that these mutants lost AP-4 binding. Although the inhibitory effect of AP-4 on wild-type LTR was clearly observed ( Fig. 2A), it was abolished when basal, TNF-␣-stimulated, and Tat-stimulated gene expression were assessed with mutant HIV-1 LTR reporter constructs (Fig. 2D). To further address whether the inhibitory effect of AP-4 depends on the presence of AP-4 site, we repeated similar experiments using other luciferase reporter plasmids including 5ϫ B-TATA-luc and 4ϫ CRE-TATA-luc (26), in which no AP-4 site is present. AP-4 did not inhibit gene expression from 5ϫ B-TATA-luc nor 4ϫ CRE-TATA-luc (data not shown). These results demonstrated for the first time that AP-4 exhibits repressive action on HIV-1 gene expression in cultured cells in vivo.

Overexpression of TBP Overcomes the Inhibitory Effect of AP-4-To
confirm that the repressive effect of AP-4 is through masking the TBP binding to the HIV-1 TATA box, we examined the effect of TBP overexpression on the action of AP-4. As shown in Fig. 3A, the inhibitory effect of AP-4 on basal gene expression was abrogated by TBP overexpression in a dose-dependent manner. Similarly, AP-4-mediated repression of the TNF-␣-stimulated HIV-1 gene expression was abolished by TBP overexpression (Fig. 3B). The Effect of AP-4 Knock-down-To examine the effect of endogenous AP-4, we adopted siRNA technique to specifically knock-down AP-4 mRNA and examined the HIV-1 gene expression when the endogenous AP-4 was depleted. Transduction of AP-4 siRNA caused the depletion of AP-4 protein (Fig. 4A), which resulted in significant increase in the basal transcriptional level from HIV-1 LTR (8.2-fold as compared with control siRNA) (Fig. 4B). In addition, TNF-␣-stimulated LTR gene expression was greatly elevated by AP-4 depletion (5.7-fold).
These results indicate that endogenous AP-4 acts as a negative regulator of HIV-1 gene expression.
The DNA Binding Activity of AP-4 Is Essential for the Repression of HIV-1 Gene Expression-AP-4 contains three functional domains, a basic HLH (bHLH) motif (48 -99 aa) and two distinct leucine repeat elements, leucine repeat (LR) 1 (99 -120 aa) and LR2 (151-179 aa) (Fig.  5A). A previous study (20) showed that the HLH motif and an adjacent basic domain are necessary and sufficient to direct sequence-specific DNA binding to its target DNA. Unlike other HLH proteins, AP-4 contains two additional protein dimerization motifs LR1 and LR2. Although both LRs contribute to the formation of AP-4 homodimers, AP-4 requires LR2 to form a stable homodimer (20). The C-terminal half of AP-4 contains a Gln/Pro-rich domain and an acidic region.
To investigate the role of functional domains of AP-4 in down-regulating HIV-1 gene expression, we examined the effects of deletion mutants of AP-4 (shown in Fig. 5A). Deletion of the N-terminal regions (⌬N100, ⌬N143, and ⌬N180) of AP-4 abolished the repressive action on both basal and TNF-␣-stimulated HIV-1 expression (Fig. 5B). In contrast, deletion of the C-terminal region of AP-4 (⌬C179, retaining the bHLH and two LR domains) repressed HIV-1 gene expression similarly to the full-length AP-4. These results indicate that bHLH domain is indispensable for the repression of HIV-1 gene expression. Because inhibitory effect of ⌬C130 (excluding LR2 domain from the ⌬C179) was weaker than ⌬C179, AP-4 dimerization is important for its effect through stabilization of AP-4 homodimer (20).

Effect of the Location of AP-4 Site Within HIV-1 LTR on the Repressive Effect of AP-4-To further examine whether the repressive effect of AP-4 depends on its location relative to the TATA box within HIV-1 LTR, we created mutant HIV-1 LTR reporter constructs in which AP-4 binding sites were inserted into various positions of the HIV-1 LTR.
Thus, AP-4 binding site were inserted at nucleotide positions Ϫ400, Ϫ15, Ϫ79, and ϩ55 (Fig. 6A) within HIV-1 LTR into CD12-luc-m2 in which the authentic AP-4 site was mutated. Basal promoter activities of these promoter constructs were not significantly changed as compared with the original construct (data not shown). As shown in Fig. 6B, even when AP-4 sites were distantly located from TATA box, AP-4 could still exert repressive action irrespective of the stimulation by TNF-␣. The greatest repressive effect of AP-4 was observed with CD12-luc-m2(Ϫ79) although it was less than that with the wild-type promoter. Other AP-4 site mutants exhibited less susceptibility to AP-4-mediated transcriptional repression. These findings indicate that AP-4 could repress HIV-1 transcription even from the distant locations from TATA box although the maximal repressive effect of AP-4 was observed when AP-4 was located in close proximity to the TATA box.
Interaction of HDAC with AP-4-Cumulative evidence has demonstrated that chromatin modification by HDAC complex plays a significant role in transcriptional repression (reviewed in Refs. 10 and 48) and many transcriptional repressors, such as YY-1 (8), silencing mediator of retinoic acid and thyroid hormone receptor (28), nuclear receptor corepressor (29), and special AT-rich sequence-binding protein 1 (30) have been shown to tether HDACs to the promoter. Because AP-4 could still exert repressive action even when AP-4 sites were distantly located from TATA box (Fig.  6B), we examined whether HDAC was involved in the HIV-1 gene repression by AP-4. To address this possibility we first examined whether AP-4  interacts with HDACs in culture cells. In Fig. 7, 293 cells were transfected with FLAG-or Myc-(used as a negative control for immunoprecipitation) tagged AP-4 expression plasmids, and cell lysates were subjected to immunoprecipitation with anti-FLAG antibody. Immune complexes were collected and subjected to SDS-PAGE followed by immunoblotting for detection of the class I HDACs using antibodies to HDAC1, 2, and 3. As shown in Fig. 7A, AP-4 interacted with HDAC1, and to a much lesser extent HDAC2, but not detectably with HDAC3. In Fig. 7B, similar experiments with untransfected cells showed the interaction of endogenous AP-4 with endogenous HDAC1. The interaction of AP-4 with HDAC2 was observed but much less than that with HDAC1. No interaction between AP-4 and HDAC3 was observed (data not shown). To examine whether AP-4 directly binds to HDAC1, we performed in vitro protein-protein interaction assay using GST-AP-4 fusion protein, radiolabeled HDAC1, and luciferase (as a control). The radiolabeled HDAC1 directly bound GST-AP-4 but not GST and no binding was observed between GST-AP-4 and luciferase (Fig. 7C). These results demonstrated that AP-4 directly interacts with HDAC1. To further confirm the involvement of HDACs in transcriptional repression of HIV-1 by AP-4, trichostatin A (TSA), a specific inhibitor of HDACs, was added to the cells transfected with AP-4, and the luciferase assay was performed. As shown in Fig. 7D, TSA abrogated the repressive effect of AP-4 on HIV-1 gene expression in a dose-dependent manner. These findings suggest that HDACs are involved in the AP-4-mediated repression of HIV-1 gene expression.
ChIP Assays Detecting AP-4 and HDAC1 on HIV-1 LTR-Together with the results demonstrated above, it was suggested that AP-4-mediated HDAC recruitment to and elimination of TBP (TFIID) from the HIV-1 promoter might play a role in the cellular maintenance of HIV-1 latency. We thus examined the presence of AP-4 and HDAC1 on the HIV-1 promoter in latently infected cells, ACH2 (T-cell line latently infected with HIV-1) and U1 (promyelocytic cell line latently infected with HIV-1) (31). As shown in Fig. 8A, the newly raised anti-AP-4 antibody could specifically precipitate AP-4. Although the nature and/or the extent of chromatin formation on transiently transfected DNA templates likely differ from that of chromosomal genes, previous reports of ChIP assays using plasmids containing HIV-1 LTR have validated ChIP studies in transiently transfected cells (32)(33)(34)(35). In Fig. 8B, 293 cells were transiently transfected with CD12luc or its mutant CD12-luc-m2 followed by ChIP analysis using the anti-AP-4 antibody. We amplified the HIV-1 LTR DNA fragment (Ϫ109/ϩ79) containing binding sites for NF-B, Sp1, TBP, AP-4, LBP-1 (YY-1), TCF1, and CTF (Fig. 6A) in the AP-4 and HDAC1 immune complexes. We were able to detect the binding of both AP-4 and HDAC1 to the HIV-1 promoter by ChIP assay in 293 cells transfected with CD12-luc (Fig. 8B). The binding with AP-4 or HDAC1 was detectably reduced when CD12-luc-m2, in which AP-4 site was mutated, was transfected. The trace amount of HDAC1 recruitment was detected even with the mutant, presumably due to the presence of binding region (from Ϫ10 to ϩ27) of LBP-1 and YY-1, known to recruit HDAC1 (8).
In Fig. 8C, ChIP assays were similarly performed with ACH2 and U1 cells using antibodies to AP-4, HDAC1, acetylated histone H3 (Ac-H3), TBP, and RNAPII. AP-4 and HDAC1, but only traceable amounts of TBP (TFIID) or RNAP II, were detected on the HIV-1 promoter when these cells maintained the latency (without any stimulation). However, when ACH2 and U1 cells were treated with TNF-␣ to stimulate HIV-1 replication, AP-4 and HDAC1 were readily dissociated from the HIV-1 promoter, and TBP and RNAPII became clearly detectable on the HIV-1 promoter over time (Fig. 8C). Moreover, the disappearance of AP-4 from the HIV-1 LTR correlated with dissociation of HDAC1 and appearance of the acetylated form of histone H3 (Ac-H3). These results, together with the findings described above, suggest that AP-4 acts as a negative regulator of HIV-1 gene expression by recruitment of HDAC1 as well as by preventing the TBP (TFIID) binding to the TATA box in latently infected cells.

Repression of HIV-1 Production by AP-4-
To assess the biological relevance of the repressive action of AP-4, we examined the effect of AP-4 on HIV-1 production. 293 cells were transfected with a replication-competent full-length HIV-1 clone (pNL4-3) together with various amounts of AP-4expression plasmid pFLAG-AP-4, and virus production was evaluated by measuring HIV-1 p24 antigen levels in the culture supernatant. In Fig. 9A, transduction of AP-4 resulted in dose-dependent decrease in the HIV-1 p24 level by 3.7-fold (Fig. 9A, left panel). Inhibition of viral protein synthesis was also observed in these cells when AP-4 was overexpressed (Fig. 9A, right  panel). In Fig. 9B, the effect of AP-4 on the TNF-␣-stimulated HIV-1 production was examined. When pNL4-3 was transfected and cells were subsequently stimulated with TNF-␣, 7.4-fold increase of HIV-1 production was observed (Fig. 9B, right panel). When AP-4 was overexpressed, a dramatic inhibition of HIV-1 production was observed in a dose-dependent manner, almost to the basal unstimulated level (Fig. 9B). In Fig. 9C, effects of AP-4 mutants were examined. Although an AP-4 mutant ⌬C179, retaining the AP-4 DNA binding and dimerization domains, could suppress HIV-1 production as well as wild-type AP-4, another mutant ⌬N143, lacking these two functional domains, showed no suppressive effect on the HIV-1 pro- duction (Fig. 9C). These results indicate that AP-4 can inhibit HIV-1 production and that the DNA binding activity of AP-4 is crucial for its inhibitory action.
In Fig. 9D, we examined the effect of depleting endogenous AP-4 on HIV-1 production using siRNA for AP-4. AP-4 depletion resulted in significant increase in the basal HIV-1 production (3.2-fold as compared with control siRNA (GFP)). Similarly, TNF-␣-stimulated HIV-1 production was elevated by the treatment with AP-4 siRNA (5.7-fold as compared with control siRNA). We also observed the elevation of viral protein synthesis in the transfected cells by AP-4 depletion (Fig. 9D, right panel).
Finally, we examined the effect of AP-4 on HIV-1 replication in Jurkat CD4ϩT cells. To examine the effect of AP-4 binding site within HIV-1 LTR, we created a mutant pNL4-3 lacking the AP-4 binding and quantified the amounts of HIV-1 virions in the culture supernatant of Jurkat cells transfected with either the wild-type or the mutant pNL4-3. As shown in Fig. 9E, the amounts of HIV-1 production were not significantly changed over time between the wild-type and the mutant HIV-1 clones. In Fig. 9F, the repressive effect of AP-4 on HIV-1 production was examined with or without TNF-␣ stimulation. When the mutant pNL4-3, containing mutation in the AP-4 binding, was cotransfected with AP-4, the inhibitory effect of AP-4 was abolished irrespective of the TNF-␣ stimulation (Fig. 9F, right panel).

DISCUSSION
The extent of HIV-1 replication is controlled at the step of transcription and the level of transcription is governed by the coordinated actions of viral and cellular transcription factors acting on LTR both in positive and negative fashions (1,2,9). In contrast to the abundant literatures reporting actions of positive transcription factors, mostly NF-B (1, 2), little is known about the action of negative transcription regulators. Here, we demonstrate evidences demonstrating that AP-4 is a negative regulator of HIV-1 transcription and its production. Although a previous study reported that AP-4 blocks TBP binding to TATA box in vitro (10) and suggested that AP-4 negatively regulates HIV-1 transcription, no evidence has been presented supporting the repressive action of AP-4 on HIV-1 transcription and viral production. In this study, we were able to demonstrate that AP-4 acts as a negative transcription factor for HIV-1 gene expression by recruitment of HDAC1, as well as by preventing the TBP (TFIID) binding to the TATA box, and that AP-4 is actively involved in the transcriptional silencing of HIV-1 gene expression in latently infected cell lines.
The AP-4 binding site in the HIV-1 LTR is located immediately downstream of TATA box (2,10,13). Among various HIV/SIV isolates AP-4 sites are conserved among HIV-1 subtypes A, B, C, D, and CRF02-AG, HIV-2, and SIVcpz-gab. However, although most (92%) of the clinical HIV-1 isolates had functional AP-4 binding site (11), no clear cor- with CD12-luc or CD12-luc-m2, incubated for 48 h, and ChIP assay was performed. Cross-linked chromatin fragments were prepared, and the association of AP-4 and HDAC1 was analyzed by ChIP assay using antibodies to AP-4 and HDAC1. The recovered DNA was amplified by PCR with promoter-specific primers (spanning from Ϫ109 to ϩ79 of HIV-1 LTR) and analyzed on a 2% agarose gel. Input DNA represents total input chromatin (1%) while immunoprecipitation with no antibody (No Ab.) serves as negative control. C, dynamic association/dissociation of AP-4, HDAC1, acetylated histone H3, TBP, and RNAP II. ChIP assays were performed with latently infected cell lines. ACH2 and U1 cells were either untreated or treated with TNF-␣ (3 ng/ml) for indicated times and subjected to ChIP assays. Cross-linked chromatin fragments were prepared, and the association of AP-4, TBP, RNAPII, HDAC1, Ac-H3, and HIV-1 LTR DNA (Ϫ109/ϩ79) was analyzed by ChIP assay as described in B. The ␤-actin promoter DNA (Ϫ980/Ϫ915) was similarly analyzed as a control. The experiments were repeated performed with reproducible results, and the representative results are shown. relation with the clinical stage was observed. Thus, the biological significance of the presence of AP-4 site in HIV-1 LTR in determining the virulence awaits further in vivo investigations.
There are a number of transcription factors acting as repressors. However, the mechanisms of their actions are not uniform and multiple modes of action are reported including: 1) inhibition of transcriptional activators upon its DNA binding or interaction with co-activators (36 -38), 2) prevention of the binding of general transcription factors such as TBP (TFIID) and TFIIB to the promoter (39 -43), 3) direct repression of promoter activity by recruiting co-repressors (8,28,30), 4) alteration of chromatin structure (19,40), and 5) inhibition of transcriptional elongation (44,45). AP-4 appears to exert transcriptional repression of HIV-1 promoter through bimodal mechanisms: 1) masking the HIV-1 TATA element from TBP binding and 2) recruiting HDAC1. In this context, it is noted that HIV-1 does not appear to use the TATA box but instead use the CATA box motif located two nucleotides upstream of the conventional TATA box (46). It is possible that AP-4 may preclude the usage of TATA box by physical masking. However, since we observed that AP-4 blocked the TBP binding to CATATA box in EMSA (Fig. 1D), AP-4 may also block the CATA box.
Like other members of bHLH family to which AP-4 belongs, the AP-4 HLH motif and the adjacent basic domain are necessary to confer sitespecific DNA binding (20). Unlike other HLH proteins, AP-4 also contains two additional protein dimerization motifs consisting of leucine repeat elements LR1 and LR2, through which AP-4 forms a homodimer (20). We found that both bHLH motif and two LR elements were necessary to exert transcriptional repression, presumably by masking the TATA box. However, unlike HIV-1 LTR, most other promoters where AP-4 has negative role have AP-4 sites at distant locations from the TATA box and even some promoters are TATA-less (22)(23)(24) and the mechanism by which AP-4 represses transcription other than masking the TATA box has not been elucidated. In addition, we found that AP-4 could exert transcriptional repression of HIV-1 even when the AP-4 site was located distant from the TATA box. Subsequent experiments have revealed that AP-4 could recruit HDAC1 to the promoter and that the AP-4-mediated repression could be restored by the treatment with a histone deacetylase inhibitor TSA. These bimodal actions of AP-4 make this factor a strong negative regulator for HIV-1 transcription. However, further studies are needed such as to determine where AP-4 binds to HDAC and to clarify whether AP-4 binds other transcriptional reg- or treated (B) with TNF-␣ (3 ng/ml) and incubated for additional 24 h. The culture supernatants and cell lysates were collected and subjected to the determination of p24 antigen level by ELISA and detection of virus proteins by immunoblot with AIDS patient serum, respectively. C, effects of AP-4 mutants on HIV-1 production. 293 cells were transfected with pNL4-3 together with expression vectors for wild-type or mutant AP-4. After 24 h of transfection, cells were untreated or treated with TNF-␣ (3 ng/ml) and incubated for additional 24 h. The p24 antigen level in the culture supernatant was determined as in A. D, effects of AP-4 knockdown on HIV-1 production. 293 cells were transfected with pNL4-3, and AP-4 siRNA-1 (Fig. 4) or control siRNA (GFP). After 36 h of transfection, cells were untreated or treated with 3 ng/ml of TNF-␣ and incubated for an additional 24 h. The culture supernatants and cell lysates were collected and analyzed the p24 antigen level and viral protein expression, respectively, as in B. *, p Ͻ 0.01. E, HIV-1 production of the pNL4-3 mutant lacking the AP-4 sites within both LTR. Either the wild-type or the mutant pNL4-3 was transfected into Jurkat CD4ϩT cells, and the amounts of HIV-1 virion production were quantified. The culture supernatants were collected after 1, 2, and 3 days post-transfection, and the p24 antigen levels were measured as in A. F, repression of HIV-1 production by AP-4 in Jurkat CD4ϩT cells. Jurkat cells were transfected either with the wild-type (left panel) or the mutant (right panel) pNL4-3 lacking the AP-4 sites in both 5Ј-and 3Ј-LTR, together with various amounts of pFLAG-AP-4 (0.2 and 0.8 g per transfection). After 36 h of transfection, cells were either untreated or treated with TNF-␣ (3 ng/ml) and incubated for an additional 24 h. The culture supernatants were prepared, and the p24 antigen levels were measured. Experiments were repeated at least three times, and reproducible results were obtained. The representative data are shown.
ulators. Interestingly, AP-4 was reported to be a transcriptional activator of transforming growth factor ␤ (47), immunoglobulin chain (48), and SV40 (late promoter) (21) where AP-4 binding sites were found in the enhancer elements of these genes although no direct evidence is thus far available to show that AP-4 actually act as a transcriptional activator for these genes. Mermod et al. (21) showed that AP-4 acts in concert with AP-1, binding to the adjacent site of AP-4, in stimulating SV40 transcription in vitro. Thus, it is possible that the effect of AP-4 binding on transcription may be modified by other transcription factors recruited to the promoter and depend on the promoter context through combinatorial interaction with other transcription factors.
Our finding that AP-4 is constitutively present on the silent HIV-1 promoter in latently infected cells may have significant biological implications. We also found that TNF-␣ stimulation abrogated the AP-4mediated repression of HIV-1 promoter. It is conceivable that nuclear translocation of NF-B and its binding to the HIV-1 promoter might induce local chromatin remodeling, thus eliminating AP-4 and its repressor complex. It appears that the interplay among various transcription factors on the HIV-1 promoter determines the transcriptional competence of the latent HIV-1 provirus.
The ability of HIV-1 to establish a latent infection is considered crucial for the pathogenesis of AIDS (49,50). Whereas HIV-1 entry into activated CD4 ϩ lymphocytes leads to a productive infection, the virus remains latent in resting CD4 ϩ lymphocytes (51). For many HIV-infected patients, although current anti-HIV treatment can reduce viral loads to undetectable levels, infected cells persist in a long term and harbor integrated proviruses capable of reseeding virus production after cessation of therapy. Our observation of AP-4 in the negative regulation of HIV-1 gene expression could give us a clue to understand how the latency is maintained at least in cells. Moreover, it is proposed that breakdown of viral latency during the early clinical stage where potent anti-viral cytotoxic T lymphocyte is still present is considered to benefit the outcome of HIV-1 infection by eliminating the otherwise long lasting chronically and latently infected cells (52,53). Interestingly, Lehrman et al. (54) have recently reported a clinical study that combination therapy of conventional anti-HIV therapy with an HDAC inhibitor, valproic acid, could successfully accelerate the clearance of HIV-1 from resting CD4ϩT cells. It is plausible that HDAC inhibitors have clinical benefit in preventing the clinical development of AIDS. Further studies are needed to clarify the role of AP-4 and other repressor proteins in the maintenance of HIV-1 latency in vivo and to determine the clinical benefit of HDAC inhibitors.