Transcription Factor Binding and Histone Modifications on the Integrated Proviral Promoter in Human T-cell Leukemia Virus-I-infected T-cells*

The human T-cell leukemia virus (HTLV-I)-encoded Tax protein is a potent transcriptional activator that stimulates expression of the integrated provirus. Biochemical studies indicate that Tax, together with cellular transcription factors, interacts with viral cAMP-response element enhancer elements to recruit the pleiotropic coactivators CREB-binding protein and p300. Histone acetylation by these coactivators has been shown to play a major role in activating HTLV-I transcription from chromatin templates in vitro. However, the extent of histone modification and the precise identity of the cellular regulatory proteins bound at the HTLV-I promoter in vivo is not known. Chromatin immunoprecipitation analysis was used to investigate factor binding and histone modification at the integrated HTLV-I provirus in infected T-cells (SLB-1). These studies reveal the presence of Tax, a variety of ATF/CREB and AP-1 family members (CREB, CREB-2, ATF-1, ATF-2, c-Fos, and c-Jun), and both p300 and CREB-binding protein at the HTLV-I promoter. Consistent with the binding of these coactivators, we observed histone H3 and H4 acetylation at three regions within the proviral genome. Histone deacetylases were also present at the viral promoter and, following their inhibition, we observe an increase in histone H4 acetylation on the HTLV-I promoter and a concomitant increase in viral RNA. Together, these results suggest that a variety of transcriptional activators, coactivators, and histone deacetylases participate in the regulation of HTLV-I transcription in infected T-cells.

Human T-cell leukemia virus type-I (HTLV-I) 1 is a complex retrovirus etiologically linked to an aggressive and often fatal malignancy called adult T-cell leukemia (1,2). Following T-cell infection, HTLV-I integrates randomly into the host chromosomal DNA (3). Expression of the virally encoded oncoprotein Tax leads to both clonal expansion of the infected cell and efficient expression of the viral genome (4). To activate HTLV-I transcription, Tax interacts with enhancer elements in the transcriptional control region of the virus in a complex that contains members of the ATF/CREB family of cellular transcription factors (reviewed in Ref. 5). Three conserved 21-base pair enhancer elements are critical to Tax-activated transcription (6 -8). These elements, referred to as viral cyclic AMPresponse element (viral CREs), carry a central CRE that serves as a binding site for members of the ATF/CREB family of transcription factors. This octanucleotide CRE sequence is immediately flanked by conserved GC-rich DNA sequences. Tax associates with the viral CREs through protein-DNA interactions with the minor groove of the GC-rich sequences (9 -11) and protein-protein interactions with the CRE-bound cellular transcription factors (12,13). Although the precise ATF/CREB proteins that are responsible for mediating Tax transactivation in vivo are not known, a significant number of studies have demonstrated a prominent role for CREB in HTLV-I transcription and Tax transactivation (12)(13)(14)(15)(16)(17)(18).
The formation of the Tax/CREB (or other ATF/CREB protein) complex on the HTLV-I promoter appears critical for the recruitment of the multifunctional cellular co-activators CBP and p300 (18 -22). CBP and p300 are very large structurally and functionally homologous proteins that are central mediators of gene expression in metazoans (reviewed in Ref. 23). Transcription factor binding to CBP/p300 brings the coactivators to target promoters, resulting in covalent modification of the promoter-associated nucleosomes. CBP/p300 have been shown to directly acetylate lysine residues present within the amino-terminal tails of all four core histones (24). Enhanced acetylation of the core histone tails strongly correlates with gene activation (25). In support of this, several recent studies have shown that Tax recruitment of p300 to the HTLV-I promoter enhances the level of Tax transactivation in vitro and is directly correlated with histone tail acetylation (26,27). 2 In the last few years there has been significant progress in the identification of transcriptional regulatory proteins that can interact with sequences in the HTLV-I transcriptional control region in vitro. However, very little is known about which activators and/or other ancillary factors interact with the viral promoter in HTLV-I infected T-cells. For example, the molecular interactions between Tax and CREB are well established in vitro; however, whether CREB and/or other ATF/CREB family members participate in HTLV-I transcription in living cells is unknown. Other proteins, such as CREB-2 (29 -32), ATF-1 (14,15,33), ATF-2 (15,34), CREM (20,33,35), and the AP-1 proteins (36 -39), have also been implicated in binding to the viral CREs and mediating HTLV-I gene expression. This is not surprising, as members of the ATF/CREB and AP-1 families recognize DNA sequences identical or similar to the viral CRE (40). Consistent with this, previous in vivo footprinting studies have shown occupancy of the viral CREs (41,42). However, the identities of these CRE-binding proteins and their roles in the regulation of the chromosomally integrated HTLV-I provirus remain unknown.
In this study, we used the chromatin immunoprecipitation (ChIP) assay to identify transcription factor and cellular coactivator binding at the proviral promoter in HTLV-I-infected, Tax-expressing human T-cells (SLB-1). We find that, in addition to Tax and CREB, a number of ATF/CREB and AP-1 family members associate with the HTLV-I transcriptional control region in vivo. We also demonstrate the association of both CBP and p300 and the TFIID component TAF II 250. Consistent with the activator/coactivator interactions, we find that the nucleosomal histones H3 and H4 are acetylated at the HTLV-I promoter, as well as within the gag and envelope genes of the provirus. Furthermore, we find histone H3 lysine 4 methylation within these same regions. Both of these histone aminoterminal tail modifications are found within transcriptionally active chromatin. Interestingly, we also find the presence of histone deacetylases (HDAC-1-3) at the viral promoter. Treatment of HTLV-I-infected cells with the histone deacetylase inhibitors trichostatin A or sodium butyrate increases H4 acetylation at the HTLV-I promoter and enhances viral transcription, further supporting a direct role for histone acetylation in HTLV-I gene expression.

MATERIALS AND METHODS
Cell Culture-HTLV-I-transformed T-cells (SLB-1 and MT-2) were cultured in Iscove's modified Dulbecco's medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and penicillin/streptomycin. Trichostatin A and sodium butyrate (Upstate Biotechnology, Inc., Lake Placid, NY) were used at final concentrations of 400 nM and 5 mM, respectively.
Plasmids-The pHTLV-1563 plasmid used in real time PCR was prepared by standard PCR cloning techniques and carries Ϫ345 to ϩ1189 of the HTLV-I genome cloned into pUC19.
ChIP Assay in Vivo-This assay was performed essentially as described (Upstate Biotechnology protocol and Ref. 44). Formaldehyde cross-linked chromatin from 10 6 or 10 7 cells/antibody was used for immunoprecipitation analyzing histone modifications or transcriptional activators, respectively. Cross-linking reactions were quenched with 125 mM glycine, cells were lysed, and chromatin was sonicated to obtain an average DNA length of 500 bp. Following centrifugation, the chromatin was diluted 10-fold and precleared with a protein A-agarose slurry containing salmon sperm DNA and bovine serum albumin (Upstate Biotechnology). Precleared chromatin (1 ml) was incubated with 1-5 g of antibody overnight at 4°C, followed by immunoprecipitation with protein A-agarose. Protein A-agarose was precoated with the appropriate secondary antibody when monoclonal antibodies were used. Immunoprecipitated complexes were washed and eluted twice with 200 l of elution buffer. The protein-DNA cross-links were reversed by heating at 65°C overnight, and 10% of the recovered DNA was used for PCR amplification (27-30 cycles).
Primers and PCRs-The HTLV-I primer sets were as follows: Ϫ321/ Ϫ31 (viral CREs), 5Ј-TTCCGAGAAACAGAAGTCTG/5Ј-CTCCTGCTA-GTTTATTGAGC; Ϫ50/ϩ142 (TATA), 5Ј-GCTCAATAAACTAGCAGGA-G/5Ј-TCTCGACCTGAGCTTTAAAC; ϩ5405/ϩ5649 (env), 5Ј-ACTCTA-ACCTAGACCACATC/5Ј-GGTCCAGTTAAATGGTAACG. The ␤-globin promoter primers were 5Ј-GTAACCAGATCTCCCAATGTG/5Ј-ATATG-TGGATCTGGAGCTCAG (45). To ensure that amplification was within a linear range, control PCRs were performed with decreasing amounts of templates. For real time PCR, immunoprecipitated DNA was analyzed with an iCycler and the optical assembly unit (Bio-Rad). Reactions were done using the Platinum Quantitative Supermix-UDG (Invitrogen) and SYBR-green (Molecular Probes, Inc., Eugene, OR). Realtime PCR primer sets were as follows: Ϫ117/Ϫ18 (promoter), 5Ј-TTTGTCAAGCCGTCCTCAGG/5Ј-CGCTTTTATAGACTCCTGCTAGT; ϩ468/ϩ581 (gag), 5Ј-GTAGCGCTAGCCCTATTCC/5Ј-GTGGAAATCG-TAACTGGAGG; ϩ5561/ϩ5665 (env), 5Ј-GTCTCTGTTCCATCCTCT-TCTT/5Ј-GGGGTCAAAGCAGTGGGT. PCRs with each primer set were optimized using genomic DNA from SLB-1 cells so that primer sets produced nearly 100% amplification efficiencies in a range of at least 400 to 2.6 ϫ 10 5 copies of the pHTLV-1563 plasmid. Copies of the provirus in coimmunoprecipitated DNA samples were found to be within this range in the PCRs. Each primer set was found to produce only the predicted product and a single melt curve peak (Ϫd[RFU]/dt versus temperature). For each experimental set, 2.5% of the input and 5% of the coimmunoprecipitated DNA were used in 25-l reactions. Reactions were done in triplicate, and the mean cycle thresholds were determined. Real time PCR results were analyzed using the iCyclerIQ optical system software version 3.0a (Bio-Rad) and were quantified relative to the input as described (46). S1 Nuclease Assay-RNA was extracted from 10 7 SLB-1 cell nuclei using Rnazole, as described by the manufacturer (Tel-Test, Inc.). Endlabeling of oligonucleotide probes, hybridization (50 g of RNA), and S1 nuclease processing of reactions has been described (47). The HTLV-I S1 probe anneals within the env/pX region of the viral RNA (5Ј-CCCA-TGGTGTTGGTGGTCTTTTTCTTTGGGATCGGCGGGGAAGAATCA). The ␤-actin probe has been described and is adapted to the human gene sequence (48): 5Ј-CACCATCACGCCCTGGTGCCTGGGGCGCCCCAC-GATGGAGGGGAATCATTAA). Reactions were resolved on a 6% denaturing polyacrylamide gel. The TATA sequence, transcription start site at the U3/R junction (ϩ1), and the three viral CREs (boxes), are also indicated. The viral CREs are located at approximately Ϫ100, Ϫ200, and Ϫ250 relative to the transcriptional start site. The PCR primer sets used for standard PCRs are indicated below each line, and the real time PCR primers are indicated above each line. Primers are denoted according to the nucleotide boundaries of the amplicons, relative to the transcription start site.

Multiple Transcription Factors from the ATF/CREB and AP-1 Families Bind to the HTLV-I Promoter in Vivo-
pected to be constitutively active. To characterize the binding of ATF/CREB family members at the HTLV-I promoter, we selected antibodies specific for ATF-1, ATF-2, CREB (CREB-1), and CREB-2. Cross-linked SLB-1 chromatin was immunoprecipitated with the relevant ATF/CREB antibody, and the purified genomic DNA was amplified using PCR primers that bracketed the HTLV-I promoter region (Fig. 1).
Interestingly, we observed enrichment of the HTLV-I promoter sequences with immunoprecipitates prepared with all four ATF/CREB antibodies tested (CREB, CREB-2, ATF-1, ATF-2), as compared with the control samples ( Fig. 2A, lanes  3-9). In contrast, we did not detect significant binding of the ATF/CREB proteins at the envelope (env) gene of the provirus or at the ␤-globin promoter (␤-globin is not expressed in T-cells) ( Fig. 2A).
Although the viral CREs do not contain strictly conserved AP-1 sites, previous studies have shown that AP-1 proteins bind to these sequences and may potentiate transactivation by Tax (37,39). Furthermore, certain members of the AP-1 family, such as c-Jun and c-Fos, are known to dimerize with ATF/ CREB proteins (reviewed in Ref. 40). To determine whether c-Fos and/or c-Jun bind to the HTLV-I promoter in vivo, we performed ChIP analysis using antibodies directed against these two proteins. Fig. 2B shows the presence of both c-Fos and c-Jun at the HTLV-I promoter in vivo.
Since Tax is a major regulator of HTLV-I transcription and it participates in complex formation with ATF/CREB proteins on the viral CRE, we were interested in determining whether we could detect Tax at the HTLV-I promoter in vivo. Compared with the control antibody (IgG), we consistently observed enrichment of the HTLV-I promoter sequences with immunoprecipitates prepared with our anti-Tax antibody, in agreement with a previous study (27). However, as shown in Fig. 2C, the PCR signal was weak. This observation is not completely unexpected, as the participation of Tax in elaborate protein-DNA complexes at the viral CREs may obscure the carboxyl-terminal epitope that is recognized by the anti-Tax antibody. Additionally, we have evidence that this polyclonal antibody possesses only weak avidity for the native Tax protein, immunoprecipitating less than 25% of Tax protein in SLB-1 nuclear extracts (data not shown).
Cellular Histone Acetyltransferases Bind the HTLV-I Pro- moter in Vivo-Several previous studies have demonstrated that Tax, in complex with CREB (and perhaps other ATF/ CREB proteins) and the viral CRE DNA, serves as a high affinity binding platform for the recruitment of CBP/p300 (18,19,21,22). This recruitment has been shown to be critical in Tax transactivation in vitro (26,27). 2 Furthermore, a recent study indicates that the functional contribution of CBP/p300 to Tax transactivation resides primarily, or exclusively, in the intrinsic acetyltransferase activity of the coactivators. 2 Therefore, we were interested in determining whether p300 and/or CBP associate with the integrated HTLV-I promoter in vivo. Using antibodies specific for each of these two coactivators, we detected the binding of both CBP and p300 at the HTLV-I promoter (Fig. 3A, lanes 6 and 7). We were also interested in testing whether additional HAT-containing transcriptional regulatory proteins were present at the HTLV-I promoter. Although we were able to detect the TFIID component TAF II 250, we were unable to detect the p300/CBP-associated factor PCAF (Fig. 3B, lanes 6 and 7). The absence of PCAF was unexpected, since two previous studies have shown that PCAF interacts with Tax and stimulates viral transcription in transfection assays (50,51). Therefore, we performed immunoprecipitations using two additional anti-PCAF antibodies that have been successfully used in the past for ChIP analysis (52,53). In all cases, we were unable to immunoprecipitate HTLV-I promoter sequences (Fig. 3B, lane 6, and data not shown). This lack of detection may stem from inaccessibility of PCAF to antibodies when it is incorporated into a promoter-bound complex. In support of this notion, we found that, whereas antibodies rec-ognizing an amino-terminal epitope of p300 were able to coimmunoprecipitate the viral promoter, antibodies against a carboxyl-terminal epitope were not.
Acetylation of Histone H3 and H4 Tails on the HTLV-I Provirus-Since p300-dependent Tax transactivation from chromatin templates in vitro has been shown to directly correlate with histone tail acetylation (26,27), we were interested in examining the acetylation state of nucleosomes on the HTLV-I promoter in vivo. We used antibodies directed against the acetylated H3 and H4 amino-terminal tails and PCR primers that encompassed the viral CREs, TATA region, and env gene ( Fig. 4A; see Fig. 1 for primer locations). We detected strong H3 and H4 acetylation at the HTLV-I promoter region and within the env gene (Fig. 4A). By comparison, histone tail acetylation was not detected at the inactive ␤-globin promoter (Fig. 4A).
Other histone modifications associated with active genes include H3 phosphorylation of serine 10 and methylation of lysine 4. H3 serine 10 phosphorylation has been correlated with enhanced acetylation at lysine 14, and together these modifications are proposed to play a role in transcriptional activation (54,55). Similarly, H3 lysine 4 methylation has been shown to antagonize the repressive effect of methylation on lysine 9 and to facilitate acetylation by p300, supporting a role for H3 lysine 4 methylation in transcriptional activation (56,57). Using methylation-specific anti-H3 antibodies, we detected H3 lysine 4 methylation (Fig. 4B) but little or no H3 serine 10 phosphorylation at the promoter (data not shown).
Since ChIP assays with standard PCR provide only a qualitative assessment of the modification state of nucleosomes within the provirus, real-time PCR was used to quantify differences in histone acetylation and methylation at three positions within the provirus. We examined the levels of histone modification on proviruses in SLB-1 cells, and for comparison, we examined the levels of histone modifications on proviruses in MT-2 cells. We have determined that SLB-1 cells carry about five copies of the integrated provirus, whereas MT-2 cells carry about two copies of the provirus (data not shown) (58). Primer pairs were selected that amplify regions within the promoter, the gag gene, and the env gene (Fig. 1). We found H3 and H4 acetylation and H3 methylation at all three of the regions examined with a spike of H4 acetylation within the gag gene in both cell lines (Fig. 4C).
Increased Acetylation of Histone H4 at the HTLV-I Promoter Augments Transcription in Vivo-In contrast to the association of acetylated histones with active genes, deacetylation by histone deacetylases (HDACs) has been found to generally play a role in transcriptional repression (59). To determine whether HDACs might function in regulating HTLV-I transcription, we used the ChIP assay to test for their presence at the HTLV-I promoter in SLB-1 cells. Fig. 5A shows that each of the class I HDACs (HDAC-1, -2, and -3) was detected at the viral promoter.
The presence of HDACs at the HTLV-I promoter in vivo allowed us to explore the functional significance of histone acetylation in viral transcription. Following treatment of infected cells with the HDAC inhibitors trichostatin A (TSA) or sodium butyrate (NaBT), we looked for changes in histone acetylation and viral transcription. Western blot analysis indicated that TSA or NaBT treatment of SLB-1 cells produced an increase in global histone H3 and H4 acetylation, with a 2.4-fold increase in H4 acetylation and a 1.4-fold increase in H3 acetylation (Fig. 5B). We next examined changes in histone acetylation at the HTLV-I promoter following NaBT treatment using real time PCR. We observed a ϳ2-fold increase in H4 acetylation levels, whereas H3 acetylation levels remained basically unchanged (Fig. 5C). There- FIG. 3. p300, CBP, and TAF II 250 bind to the HTLV-I promoter in SLB-1 cells. A, coactivator binding at the HTLV-I and ␤-globin promoters were analyzed by ChIP as described in Fig. 2. PCR results from input, mock antibody, preimmune serum (IgG), irrelevant antibody (␣-BRCA-2), and antibodies against p300 and CBP are shown. B, PCAF and TAF II 250 binding at the HTLV-I promoter were analyzed by ChIP as described in the legend to Fig. 2. PCR results from input, mock antibody, preimmune serum (IgG), and antibodies against p300, CBP, PCAF, and TAF II 250 are shown. For greater sensitivity of PCAF, radioactive PCR was performed, and reactions were resolved on a polyacrylamide gel. fore, with both global and HTLV-I promoter-specific histone acetylation, HDAC inhibitors produced a more pronounced increase in acetylated H4.
To determine whether there is a correlation between HDAC inhibition, histone acetylation, and HTLV-I transcription, we measured relative viral RNA levels in SLB-1 cells by an S1 nuclease protection assay. RNA extracted from nuclei following treatment with TSA or NaBT was hybridized with a radiolabeled DNA probe specific for either HTLV-I RNA or ␤-actin RNA and digested with S1 nuclease. A 2.4-fold increase in viral RNA was observed after NaBT treatment, and a 1.6-fold in-

FIG. 4. Histone H3 and H4 amino-terminal tail modifications within the HTLV-I provirus in SLB-1 and MT-2 cells.
A, histone acetylation at the HTLV-I upstream promoter, start site region (TATA), env gene, and the ␤-globin promoter was analyzed by ChIP in SLB-1 cells, as described in the legend to Fig. 2. PCR results from input, mock antibody, preimmune serum (IgG), and antibodies against acetylated H3 (acH3) and H4 (acH4) are shown. B, histone methylation at the HTLV-I promoter was analyzed by ChIP in SLB-1 cells. PCR results from input, mock antibody, and antibodies against acetylated H3 (acH3) and dimethylated lysine 4 of H3 (H3 mK4) are shown. C, quantification of relative histone modification levels at three regions of the provirus. Real time PCR was used to quantify anti-acetyl H3, antiacetyl H4, and anti-H3 mK4 immunoprecipitations from SLB-1 cells (left panel) and MT-2 cells (right panel). The HTLV-I promoter, gag gene, and env gene were analyzed (see Fig. 1). Histone modification levels were normalized to the signal observed at the promoter region (set to 1). The SLB-1 and MT-2 graphs show data averaged from four and two independent ChIP experiments, respectively. Histone acetylation levels were normalized to the signal observed with DNA samples from untreated cells (set to 1). D, increase in HTLV-I RNA following TSA or NaBT treatment. S1 nuclease analysis was used to quantify levels of HTLV-I (upper panel) and ␤-actin RNA (lower panel) in the nuclei of untreated SLB-1 cells and cells treated with 400 nM TSA or 5 mM NaBT (24 h). E, graphical representation of S1 nuclease data. Results of the S1 nuclease assays from three independent experiments were quantified by ImageQuaNT software (Amersham Biosciences) and are represented graphically (normalized to the untreated samples). crease was observed following TSA treatment (Fig. 5, D and E). These data suggest that HDAC inhibition produces an increase in HTLV-I transcription, probably due to the concomitant elevation in H4 acetylation. We observed a slight drop in the levels of ␤-actin RNA levels, possibly attributable to restricted cell growth caused by the HDAC inhibitors, as previously shown (60 -62) (Fig. 5, D and E). These results link enhanced HTLV-I transcription to hyperacetylation of H4 in vivo. DISCUSSION In living T-cells infected with HTLV-I, the specific cellular transcription factors that participate in viral transcription are unknown. Prior studies have relied on in vitro binding and transient transfection assays to determine the transcription factors that interact with the HTLV-I transcriptional control region. Although these studies have implicated a variety of transcription factors, their direct binding to the integrated proviral promoter has never been demonstrated. In this study, we used the ChIP assay to analyze transcription factor and coactivator binding as well as histone modifications at the HTLV-I promoter in infected human T-cells. This approach allows direct detection of essentially any protein at its genomic binding site in vivo, including associated proteins that are not in direct contact with the proviral DNA.
Previously, a significant number of in vitro studies have focused on the transcription factor CREB, and based on this work, CREB has been implicated as the primary player in both basal and Tax-activated HTLV-I transcription (12)(13)(14)(15)(16)(17)(18)33). However, a number of additional bZIP proteins have been suggested to participate in HTLV-I transcription (5). The ChIP studies presented herein reveal that, in addition to CREB, several other bZIP proteins bind specifically to the HTLV-I transcriptional control region in living SLB-1 cells. These proteins include ATF-1, ATF-2, and CREB-2 as well as the AP-1 family members c-Fos and c-Jun. Significantly, our data provide the first direct demonstration of an interaction between these bZIP proteins and the HTLV-I promoter under physiological conditions. These observations substantiate many of the previous studies using DNA binding and transient transfection assays. Furthermore, they suggest that a closer examination of the role of these other bZIP proteins in HTLV-I transcription and Tax transactivation is warranted.
The fact that multiple factors are found at the HTLV-I promoter is surprising and may indicate that they collectively form a higher order complex, such as an enhanceosome, required for basal transcription and Tax transactivation. Such a complex could augment CBP/p300 recruitment through interactions with multiple DNA-bound bZIP proteins, explaining why at least two viral CREs are required for efficient transactivation (6,63,64). This type of mechanism has been proposed in other systems (reviewed in Ref. 23). Furthermore, it has recently been shown that the binding of a single activator may be insufficient for CBP recruitment in vivo (65). Alternatively, distinct bZIP proteins (or distinct bZIP complexes) may assemble on individual proviral promoters, with each having a unique role in HTLV-I transcription. This scenario permits enhanced flexibility in the protein-DNA complexes that may participate in HTLV-I transcriptional regulation. However, which of these individual proteins or protein complexes mediate Tax transactivation is not known.
An essential role for CBP/p300 in activating HTLV-I transcription from chromatin templates has recently been shown in vitro (26,27). 2 Our ChIP assay reveals the presence of both p300 and CBP at the HTLV-I promoter in living T-cells. This observation is intriguing and raises the question as to whether they play distinct or redundant roles in HTLV-I transactivation. Consistent with the binding of CBP and p300, we found histone H3 and H4 acetylation at the HTLV-I promoter and within the proviral genome. Surprisingly, we did not observe a spike in histone acetylation at the viral promoter that would be expected to coincide with the binding of CBP and p300. Instead, we observed enhanced H4 acetylation centered about 525 bp downstream of the transcriptional start site. Histone acetylation has previously been observed within the coding regions of transcriptionally active genes (66). These results are in contrast to a previously reported in vitro ChIP assay that showed H3 and H4 histone acetylation localized to the HTLV-I promoter (27). In addition to acetylation, we found histone H3 lysine 4 methylation throughout the provirus, another modification associated with transcriptionally active genes (56,57,67). It should be noted that in our ChIP assays, analysis of the HTLV-I promoter reveals histone modifications at both the 5Ј and 3Ј long terminal repeats. If the 3Ј long terminal repeat is inactive, then our results represent the average of histone modifications at both active and silenced promoter regions of the provirus. Future studies will be aimed at distinguishing factor binding and histone modifications at the individual long terminal repeats.
The regulation of transcriptional activity through histone acetylation is also influenced by HDACs, which serve to remove acetyl groups from the histone tails. We tested for the presence of HDACs on the HTLV-I promoter and found that HDAC-1 to -3 were each associated with at least a subset of HTLV-I promoters in living cells. HDACs may be directly recruited to the HTLV-I promoter by Tax, as has recently been suggested (28). By inhibiting HDAC function, we show an increase primarily in H4 acetylation at the HTLV-I promoter, concomitant with an increase in viral RNA levels. This study shows that HDACs are localized to the HTLV-I promoter and are involved in regulating HTLV-I transcription in vivo, suggesting that a dynamic balance between HDAC and HAT activities may dictate the overall level of HTLV-I transcription in vivo.