|Home | About | Journals | Submit | Contact Us | Français|
Human immunodeficiency virus type 1 (HIV-1) establishes a persistent, nonproductive state within a small population of memory CD4+ cells. The transcription factor LSF binds to sequences within the HIV-1 long terminal repeat (LTR) initiation region and recruits a second factor, YY1, to the LTR. These factors then cooperatively recruit histone deacetylase 1 to the LTR, resulting in inhibition of transcription. This appears to be one mechanism contributing to HIV persistence within resting CD4+ T cells. We sought to further detail LSF binding to the HIV-1 LTR and factors that regulate LSF occupancy. We find that LSF binds the LTR as a tetramer and that binding is regulated by phosphorylation mediated by mitogen-activated protein kinases (MAPKs). In vitro, phosphorylation of LSF by Erk decreases binding to the LTR, while binding is increased by p38 phosphorylation. LSF occupancy at LTR chromatin is increased by the p38 agonist anisomycin and decreased by specific p38 inhibition. p38 inhibition also results in increased acetylation of histone H4 at the LTR nucleosome adjacent to the LSF binding site. p38 inhibition also blocked the ability of YY1 to inhibit activation of the integrated HIV promoter. Finally, HIV was recovered from the resting CD4+ T cells of aviremic, HIV-infected donors upon treatment of these cells with specific inhibitor of p38. These data suggest that the MAPK pathway regulates LSF binding to the LTR and thereby one aspect of the regulation of HIV expression. This mechanism could be exploited as a novel therapeutic target to disrupt latent HIV infection.
Human immunodeficiency virus type 1 (HIV-1) infection is characterized by cycles of virus production and reinfection within activated CD4+ T lymphocytes. The efficiency of HIV-1 replication diminishes as cells enter the resting state, but this lentivirus can replicate and persist within resting cell populations (55, 56). Although suppressive antiretroviral therapy is often clinically successful, eradication of HIV infection is currently unattainable due to the persistent, quiescent infection of resting CD4 memory T cells (31).
Global activation of T cells appears to be an untenable therapeutic strategy to disrupt latency of HIV (45), and recently induction of HIV expression without full activation of T cells has been proposed as an alternative approach. Evidence suggests that signaling pathways exist that may reactivate virus expression in primary cells (2, 22, 24), and these pathways may serve as targets for future therapies. Consequently, it is important to understand in detail the mechanisms that regulate the transition from latent to productive infection.
Two human transcription factors, YY1 and LSF, play a role in the establishment or maintenance of HIV latency via changes in local chromatin architecture surrounding the HIV promoter (8, 9, 35, 53). LSF (late simian virus 40 [SV40] transcription factor) is a member of the LSF/CP2 subfamily of transcription factors (41, 47). LSF was initially identified as a factor that binds to and stimulates transcription from the viral SV40 major late promoter (20). LSF is involved in the regulation of several genes, including those for interleukin 4 (IL-4) (4), human α-globin (25), thymidylate synthase (33), and the HIV long terminal repeat (LTR) (19, 35). YY1 is a highly conserved GLI-Kruppel transcription factor that can function as activator, repressor, or initiator of transcription depending on the context of its binding site within a promoter and its interaction with other proteins. Of note, YY1 can downregulate transcription through recruitment of histone deacetylases (34, 51, 52).
Recent data suggest that this mechanism of regulation of HIV expression is relevant in latent infection of resting CD4+ T cells. Krishnan and Zeichner have found that YY1 expression is upregulated in latently infected cells prior to activation and expression of HIV (21). We have shown that histone deacetylase (HDAC) inhibition potently induces expression of latent HIV from the resting CD4+ T cells of HIV-infected patients (54).
LSF and YY1 form a complex at the HIV LTR that recruits HDAC1 and represses LTR transcription. Coexpression of YY1 and LSF synergistically inhibits LTR-directed gene expression and HIV virus production. Three binding motifs have been identified within the HIV promoter at a region from −10 to +27 surrounding the transcriptional start site designated the repressor complex sequence (RCS). LSF binds the RCS, but YY1 is required to repress LTR transcription. In contrast, YY1 does not form a specific protein complex with the RCS, and an LSF mutant unable to bind DNA blocks repression of the LTR by YY1. In turn, YY1 mutants unable to bind HDAC lack the ability to repress the LTR (8, 28, 35).
Mitogen-activated protein kinase (MAPK) pathways regulate entry into the cell cycle (48). The three central MAPKs, Erk, Jnk, and p38, are phosphorylated on both threonine and tyrosine residues by dual specificity by MAP kinase kinases (MEKs), which in turn are activated by serine/threonine phosphorylation by MAP kinase kinase kinase. Mammalian Erk1 and Erk2 and their upstream activators MEK1 and MEK2 are stimulated by signaling of growth and differentiation factors, such as epidermal growth factor, platelet-derived growth factor, and nerve growth factor, through receptor tyrosine kinases, heterotrimeric G protein-coupled receptors, or cytokine receptors. The mammalian Jnks and p38 are implicated in responses to cellular stress, inflammation, and apoptosis (16).
In response to extracellular stimuli, MAPKs are activated in the cytoplasm and can rapidly be transported into the nucleus. Once activated, MAPKs induce phosphorylation of transcription factors present in the cytoplasm or nucleus. The phosphorylation of transcription factors by MAPK can change their affinity for DNA or change their affinity for other transcription factors regulating the expression of specific genes (40).
After mitogenic stimulation of human peripheral blood mononuclear cells (PBMCs), LSF is rapidly phosphorylated by Erk1/Erk2. In primary T cells, Erk signaling is correlated with a rapid and transient increase in the ability of LSF to bind DNA (30, 43, 44). However, Erk phosphorylation in vitro is not sufficient to increase its binding to the LSF site. We sought to further understand the LSF binding to the HIV LTR and the role of phosphorylation in regulating this event.
The HeLa-CD4-LTR-CAT cell line (6) was maintained in Dulbecco modified eagle medium with 10% fetal bovine serum (FBS) and 1% penicillin, streptomycin, and l-glutamine. Cells (2 × 106) were plated and cultured in 0.5% FBS overnight. The plates were then washed and treated for 1.5 to 2 h with 10 μM of p38 inhibitor, SB203580 (4-(4-fluorophenyl)-2-(4-methylsulfinylphenyl)-5-(4-pyridyl)-1H-imidazol) (Sigma, St. Louis, MO), SB239063 (kind gift Glaxo SmithKline, Research Triangle Park, NC), MEK1/2 inhibitor, U0126 (1,4-diamino-2,3-dicyano-1,4-bis[2-aminophenylthio]butadiene) (New England Biolabs, Beverly, MA), or anisomycin (2-p-methoxyphenylmethyl-3-acetoxy-4-hydroxypyrrolidine), (Sigma, St. Louis, MO) in the absence of FBS. FBS was then added to a final concentration of 20%, and cells were incubated for an additional 1 to 2 h. Following incubation with inhibitors, cells were washed and treated with trypsin and nuclear extracts were prepared.
Cells (3 × 106) were washed in phosphate-buffered saline (PBS), and the cell pellet was resuspended in 500 μl of lysis buffer (10 mM HEPES [pH 8], 10 mM KCl, 1.5 mM MgCl2, 0.5% NP-40, 20 μl of protease inhibitor cocktail [Sigma, St. Louis, MO]) for 1 min on ice. After lysis, the samples were spun at 5,500 rpm for 3 min at 4°C, and the cytosolic extract was discarded. The pellet was resuspended in 100 μl of extraction buffer (20 mM HEPES [pH 8], 420 mM NaCl, 25% glycerol, 0.2 mM EDTA, 10 μl of protease inhibitor cocktail) and incubated for 15 min on ice. After centrifugation at 10,000 rpm for 10 min at 4°C, the supernatant was recovered.
The maltose-binding protein (MBP)-LSF fusion protein and histidine-tagged LSF (His-LSF) were produced as described previously (29). Briefly, MBP-LSF and His-LSF were expressed in Escherichia coli strain JM 109. MBP-LSF and His-LSF were purified by use of an affinity column according to the manufacturer′s instructions (NEB or QIAGEN, respectively). His-LSF (2.5 μg) was phosphorylated using 50 units of p38, Jnk (Upstate Biotechnology, Lake Placid, NY), and/or Erk2 (New England Biolabs, Beverly, MA). The kinase reaction buffer was composed of 20 mM morpholinepropanesulfonic acid (pH 7.2), 25 mM β-glycerol phosphate, 5 mM EGTA, 1 mM sodium orthovanadate, 1 mM dithiothreitol (DTT), 100 μM ATP, 15 mM MgCl2, and 6.7 pmol [γ-32P]ATP. Phosphorylation was carried out at 30°C or 4°C in 50 μl for up to 10 min. To verify the efficiency of phosphorylation, 2.5 μl of reaction mixture was spotted on p81 filter paper. Once dried, the filter was washed three times in 250 μl of 0.75% of phosphoric acid and once in acetone, and filter-bound radioactivity was quantitated. For MBP-LSF and His-LSF291, 1 μg of purified protein was phosphorylated with 50 units of p38 and Erk as described above (43). The phosphorylated product was isolated on a 10% sodium dodecyl sulfate (SDS)-polyacrylamide gel and visualized by autoradiography.
Electrophoretic mobility shift assay (EMSA) was performed using the following double-stranded oligonucleotides corresponding to the RCS and oligonucleotides encoding mutations in one to three of the five base-pair motifs (shown 5′ to 3′, altered bases in bold) previously shown to be required for LSF binding, as shown in Fig. Fig.1A1A (19): WT, TGC CTG TAC TGG GTC TCT CTG GTT AGA CCA GAT CTG A; M1, TGC CTG TAG ATG GTC TCT CTG GTT AGA CCA GAT CTG A; M2, TGC CTG TAC TGG GTC TCT AGT CTT AGA CCA GAT CTG A; M3, TGC CTG TAC TGG GTC TCT CTG GTT AGA AAT GCT CTG A; M1 + 2, TGC CTG TAG ATG GTC TCT AGT CTT AGA CCA GAT CTG A; M2 + 3, TGC CTG TAC TGG GTC TCT AGT CTT AGA AAT GCT CTG A; M1 + 3, TGC CTG TAG ATG GTC TCT CTG GTT AGA AAT GCT CTG A; M1 + 2+3, TGC CTG TAG ATG GTC TCT AGT CTT AGA AAT GCT CTG A.
The sequence of the canonical LSF binding site (LSF-280) from the SV40 promoter was 5′-CAG CTG GTT CTT CCG CCT C-3′. The RCS and LSF-280 probes were labeled using polynucleotide kinase (Roche Molecular Biochemicals, Indianapolis IN) and [γ-32P]ATP following the manufacturer's instructions. One hundred thirty nanograms of nuclear extract or 96 ng of purified His-LSF was used for binding reactions in a 20-μl final volume as described previously (35). After 15 min at 4°C, EMSA was performed using a 4% nondenaturing polyacrylamide gel.
For EMSA with MBP-LSF and His-LSF, 1 μg of each of the purified proteins was phosphorylated with ATP as described, allowed to bind with the labeled RCS probe for 15 min at 4°C, and run on a 4% nondenaturing polyacrylamide gel. Phosphorylation of LSF was confirmed in a parallel reaction using [γ-32P]ATP.
Each chromatin immunoprecipitation (ChIP) assay used 2 × 106 cells. Cells were fixed in 1% formaldehyde at 37°C for 8 min. After being cross-linked, cells were rinsed twice with PBS. Then, 100 μl of lysis buffer (Upstate Biotechnology, Lake Placid, N.Y.), together with 5 μl of protease inhibitor cocktail (Sigma, St. Louis, Mo.), was added to the cell pellets, followed by incubation at 4°C for 10 min. Lysates were resuspended with 1 ml of dilution buffer (Upstate Biotechnology) in a 15-ml conical tube, subjected to sonication for five 30-s pulses with 15-s pauses in a microtip ultrasonicator, and transferred to a 1.5-ml microcentrifuge tube. Soluble chromatin was collected as a supernatant after a 10-min centrifugation at 13,000 rpm and 4°C. Appropriate chromatin fragmentation (300 to 1,000 bp) was confirmed by agarose gel electrophoresis. Next, 50 μg of soluble chromatin was incubated on a rotating platform with 4 μl of anti-acetyl-histone H4 (Upstate Biotechnology), anti-HDAC1 (Upstate Biotechnology), anti-LSF (gift of M. Sheffery and S. Swendenmann), or rabbit preimmune immunoglobulin G serum (Sigma), as appropriate, overnight at 4°C. Immunoprecipitates were incubated with 40 μl of salmon sperm DNA-protein A-agarose beads (Upstate Biotechnology) for 1 h at 4°C. Agarose beads were recovered by centrifugation and washed sequentially for 5 min with 1 ml of each of the following five buffers (Upstate Biotechnology): ChIP dilution buffer, low-salt wash buffer, high-salt wash buffer, LiCl wash buffer, and Tris-EDTA buffer. Immunoprecipitated DNA was eluted with 500 μl of elution buffer (1% SDS, 0.1 M NaHCO3). Reversal of DNA cross-linking was performed by incubating 50 μg of soluble chromatin fraction with 19 μg of proteinase K (PCR grade; Boehringer, Mannheim, Germany) at 56°C for 1 h. DNA was extracted in phenol-chloroform-isoamyl alcohol, precipitated in ethanol, washed, and resuspended in 50 μl of water.
Immunoprecipitated DNA was quantitated by PCR using the following sets of primers: LTR-109F (5′-TAC AAG GGA CTT TCC GCT GG-3′) and LTR + 82R (5′-AGC TTT ATT GAG GCT TAA GC-3′); and for β-actin promoter, P-β-Actin-F (5′-TGC ACT GTG CGG CGA AGC-3′) and P-β-Actin-R (5′-TCG AGC CAT AAA AGG CAA-3′).
One hundred nanograms of glutathione S-transferase (GST)-Erk (Upstate Biotech, Lake Placid, NY) was incubated with 1 μg of His-LSF for 15 min at 30°C in 20 μl kinase buffer containing 20 mM MOPS (pH 7.2), 25 mM β-glycerophosphate, 25 mM EGTA, 1 mM Na3VO4, 1 mM DTT, 15 mM MgCl2, 5 μg insulin, and 1 μM ATP (with 10 μCi [γ-32P]ATP). Phosphorylation by Jnk or p38 was performed following purification and activation of GST-Jnk or GST-p38 as previously described (39). One microgram of GST-Jnk or GST-p38 was incubated with 1 μg of His-LSF for 20 min at 30°C in 20 μl kinase buffer containing 20 mM HEPES (pH 7.2), 20 mM MgCl2, 2 mM DTT, and 1 μM ATP (with 10 μCi [γ-32P]ATP).
Samples were separated by 7.5% SDS-polyacrylamide gel electrophoresis (PAGE) and transferred to nitrocellulose membranes. Nitrocellulose strips containing His-LSF were excised from the blot and incubated in 500 μl buffer containing 0.5% polyvinylpyrrolidone in 100 mM acetic acid at 37°C for 30 min. The nitrocellulose strips were rinsed three times (1 ml each) in dH2O and two times (1 ml each) in 50 mM ammonium bicarbonate, and incubated in 250 μl 50 mM ammonium bicarbonate with 10 μg chymotrypsin (Boehringer Mannheim, Indianapolis, IN) for 2 h at 37°C followed by the addition of another 10 μg of chymotrypsin and incubation for an additional 2 h. Digested protein was washed and lyophilized as described previously (1). Chymotryptic peptides were separated on thin-layer cellulose plates by electrophoresis at pH 1.9 (88% acetic acid:formic acid:H2O [78:25:897]) for 25 min at 1.0 kV in the first dimension followed by ascending chromatography in buffer containing (butanol:pyridine:acetic acid:H2O [75:50:15:60]) for 12 h in the second dimension. Labeled chymotryptic peptides were detected using a PhosphorImager.
HeLa-CD4-LTR-chloramphenicol acetyltransferase (CAT) cells were transfected with plasmids encoding CMV-Tat and CMV-YY1 (35) using Lipofectamine (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol in serum-free medium. Each transfection used 5 μg of total DNA. Four hours after transfection, some cultures were treated with 10 μM of SB203580 for 30 min, and all cultures were then incubated for 24 h with 10% FBS in a 5%-CO2 incubator. Cells were then washed three times with PBS, treated with trypsin, collected in 500 μl of PBS, and spun at 14,000 rpm for 1 min at 4°C. The cell pellet was resuspended in 150 μl of 0.25 M Tris-HCl and subjected to three rapid freeze/thaw cycles with vigorous vortexing after each thaw cycle. The lysate was then incubated at 60°C for 10 min and spun at 14,000 rpm at room temperature for 2 min, and the supernatant was collected. Protein content was measured by Bradford assay (Bio-Rad, Hercules, CA). The CAT assay was performed by using the CAT assay kit (Promega, Madison, WI) according to the manufacturer's instructions; 10 to 50 μl of cell lysate normalized for protein content was used per reaction.
PBMC from healthy donors were phenotyped immediately after isolation. PBMCs were incubated with 20 U/ml IL-2 (Chiron, Emeryville, California), 2 μg/ml phytohemagglutinin (PHA) (Murex Biotech, United Kingdom), and IL-2, 1 μM SB239063 and IL-2, or 20 μM SB239063 and IL-2 for 72 h; conditions were identical to those used for resting-cell HIV outgrowth assays. PBMCs were washed and phenotyped using a cocktail of monoclonal antibodies for 30 min at 4°C: HLA-DR-fluoroscein isothiocyanate, CD25-phycoerythrin (PE), CD69-PE, CD38-PE, CD3-peridinin-chlorophyll, a complex protein, and CD4-allophycocyanin (Becton Dickinson, Palo Alto, California). The cells to be stained with Ki67-fluoroscein isothiocyanate were fixed, permeabilized with the Fix and Perm kit (Caltag, Burlingame, CA), and stained with the monoclonal antibody. Stained cells were washed in PBS containing 2% FBS and subjected to cytofluorimetric analysis performed on a FACSCalibur (Becton Dickinson) equipped with Cell Quest (Macintosh, Cupertino, California) software. For the analysis, the total lymphocyte population was identified and then gated by forward and side scatter. Cells were then gated for CD4 expression. A total of 10,000 gated events were collected for each sample.
PBMC obtained from HIV-seronegative donor buffy coats were treated with 20 U/ml IL-2, 2 μg/ml PHA and IL-2, 1 μM SB239063 and IL-2, or 20 μM SB239063 and IL-2 for 3 days. PBMC (4 × 106 cells) were then infected for 12 h with HIVLAI (10 ng p24) in the presence of 5 μg/ml polybrene. PBMCs were washed and fed media with IL-2; samples for p24 enzyme-linked immunosorbent assay were taken daily, and media were replaced.
Leukopheresis of five stably treated, aviremic HIV-positive volunteers (plasma HIV-1 RNA at <50 copies/ml for more than 6 months; CD4 cell counts of >350 × 106 cells/liter) was performed on six occasions following informed consent. Resting CD4 T cells were negatively selected from leukopheresis samples using magnetic-bead depletion, and HIV outgrowth assays were performed as previously described (54).
In limiting-dilution format, 5.0 × 106 to 6.25 × 105 resting CD4 cells were (i) activated with 2 μg/ml PHA, 6 × 106 allogeneic irradiated PBMC from an HIV-seronegative donor, and 100 U/ml IL-2, (ii) treated with 1 μM SB239063 and 20 U/ml IL-2, or (iii) cultured in 20 U/ml IL-2 alone. After 72 h, the cultures were expanded with the addition of 106 PHA-activated CD8-depleted PBMC from seronegative donors and carried for up to 21 days as described previously (54).
LSF has been reported to regulate transcription of a number of cellular and viral genes by interacting with a sequence, CNRG-N6-CNR(G/C), composed of two direct 4-bp repeats separated by a 6-bp linker, (4, 26). For example, in the α-globin promoter, the 4-bp half-motif is repeated four times (26), but in the Sα region responsible for immunoglobulin isotope switching, as many as eight potential repetitive half-LSF binding sites appear (11).
The LSF binding site in the HIV LTR is unusual in that it is composed of three 4-bp half-sites (CTGG), with the third motif located on the reverse strand (Fig. (Fig.1A)1A) (19, 27). To analyze the importance of these three binding sites, we studied the interaction of recombinant LSF with DNA by EMSA, using P32-radiolabeled oligonucleotides corresponding to the sequence of RCS (Fig. (Fig.1B,1B, lane 2). Binding to the native RCS was compared to that when RCS was mutated within the first (M1), second (M2), or third (M3) LSF binding motif. As shown in Fig. Fig.1B,1B, LSF binding to M1, M2, or M3 is diminished compared to native RCS binding (lanes 2 to 5); however, there was not a striking difference in LSF binding between these three mutated oligonucleotides. This result suggests that mutation of one of three sites destabilizes LSF binding but does not eliminate it completely.
When motifs 1 and 2 were mutated, the binding of LSF remained comparable to that seen in the presence of one mutated motif (Fig. (Fig.1B,1B, lane 6). LSF binding was almost eliminated when the third motif was mutated (Fig. (Fig.1B,1B, lanes 7 and 8). An oligonucleotide in which all LSF binding motifs are mutated (Fig. (Fig.1B,1B, lane 9), cannot bind LSF in EMSA. These results suggest that the atypically oriented motif 3 contributes most to the binding of LSF to HIV RCS.
Oligomerization of sequence-specific transcription factors contributes to the stability of protein-DNA interactions. LSF binds as a homo-oligomer to viral and cellular promoters, with some exceptions (17). Although LSF is a dimer in solution, it binds to DNA at two half-sites as a tetramer (29, 37, 38). To determine the oligomerization state of LSF when bound to an unusual HIV LTR site, we used two LSF derivatives expressed and harvested from Escherichia coli: MBP-LSF and His-tagged LSF. Both are competent to bind DNA, and each forms a complex with distinguishable mobility with RCS oligonucleotides (Fig. (Fig.2A2A).
To normalize quantities of functional LSF fusion proteins compared in these experiments, LSF was added according to quantitated gel shift units (gsu). Thus, 1 gsu of His-LSF forms a gel shift band with the same density as that formed by 1 gsu of MBP-LSF. As shown in Fig. Fig.2A,2A, His-LSF homo-oligomers form the highest-mobility band (lane 2), whereas the band of lowest mobility corresponds to oligomers containing only the larger MBP-LSF (lane 7). In lanes 2 through 6, a constant concentration of His-LSF was mixed with an increasing proportion of MBP-LSF. In these titration experiments, we observed five unique complexes. As each motif can bind one LSF dimer, the RCS oligonucleotide might accommodate two to six LSF molecules, depending on how many of the motifs are functional.
Three intermediate complexes are observed found between the pure His-LSF (lane 2) and pure MBP-LSF (lane 7) complexes. By deduction these represent, in order of decreasing mobility, heterotetramers containing one, two, and three MBP-LSF subunits (Fig. (Fig.2B).2B). These results indicate that LSF binds to the HIV LTR as a tetramer.
It has been previously determined that a central region of LSF, including four serine-proline sequences at amino acid positions 278, 289, 291, and 309, was both a target of the Erk kinase in vitro and a target of a growth-regulated signal transduction cascade in vivo. The serine at position 291 is, in particular, the major target of phosphorylation by Erk and by growth induction in vivo (43). To test whether Jnk or p38 also phosphorylates LSF, purified, recombinant His-LSF was incubated in vitro with activated GST-Jnk or GST-p38 in the presence of [γ-32P]ATP, and the radiolabeled products were then separated by SDS-PAGE. Analysis of the gel revealed that both GST-Jnk and GST-p38 phosphorylated LSF in vitro (Fig. (Fig.3).3). In order to determine whether Jnk and p38 target the same serine-proline sequences in LSF as does Erk, we subjected the radiolabeled LSF from these reactions to chymotryptic digestion and two-dimensional phosphopeptide mapping and compared these maps to that derived from LSF phosphorylated by GST-Erk in vitro (Fig. (Fig.3).3). Spot A represents an incompletely digested product of the peptide containing both serine residues at positions 289 and 291, spot C is the chymotryptic phosphopeptide containing a singly phosphorylated serine residue at either position 289 or 291, and spot B is an alternatively migrating configuration of the singly phosphorylated peptide in spot C that is primarily phosphorylated on serine 289 (44). Spot X is an unidentified peptide whose intensity is variable from digest to digest. As expected, in vitro phosphorylation by Erk of a recombinant His-LSF, LSF291 mutant containing a serine-to-alanine substitution at amino acid position 291 resulted in a chymotryptic phosphopeptide map in which spot A remained while the intensity of spot C was greatly diminished, indicating a loss of phosphorylation at amino acid 291 (data not shown). Interestingly, the intensity of spot B was enhanced in this mutant, suggesting that serine 289 is preferentially phosphorylated when serine 291 is unavailable for phosphorylation. Finally, the chymotryptic phosphopeptide map of a serine-to-alanine double-substitution mutant at positions 289 and 291 lacked all three major spots (data not shown). We conclude from these data that the major in vitro sites of phosphorylation on LSF by Erk are the serine residues at positions 289 and 291, with a preference for serine 291.
Like phosphorylation by Erk, chymotryptic phosphopeptide analysis of His-LSF phosphorylated in vitro with activated Jnk revealed that Jnk also targeted the serine residues at position 289 and 291 (Fig. (Fig.3).3). However, the presence of additional spots (Fig. (Fig.3,3, Jnk spots α, β, and γ) indicated that Jnk also targets other site(s) in LSF. These additional phosphopeptides are also generated by p38 in vitro (Fig. (Fig.3).3). Interestingly, the serine residues at amino acid position 289 or 291 were not phosphorylated by p38 kinase, as indicated by absence of phosphopeptides A, B, and C (Fig. (Fig.3).3). Thus, all of the MAP kinase family members phosphorylate LSF in vitro, but with various site specifications.
Previous studies have shown that phosphorylation of LSF by MAP kinases, specifically Erk1/Erk2, modulates its DNA binding activity in primary T cells (30, 43). We wanted to analyze the effect of the MAPKs, Erk, Jnk, and p38, in the regulation of LSF binding to DNA. We thus studied LSF binding ability in vitro after phosphorylation by the above MAPKs. Consistent with the findings of Volker et al. (43), after phosphorylation by Erk, Jnk, or p38, LSF binding to the canonical LSF-280 site from SV40 was unaffected (Fig. (Fig.4A4A).
However, MAPK phosphorylation does affect LSF binding in EMSA to the RCS within the HIV LTR. When phosphorylated by Erk, LSF binding was significantly decreased after 1 min of phosphorylation in vitro and completely ablated after 10 min of phosphorylation in vitro (Fig. (Fig.4B,4B, compare lane 2 with lanes 3, 4, and 5). To exclude the possibility that LSF lost its binding ability due to degradation, LSF was incubated without enzyme at 30°C or at 4°C for up to 10 min. Incubation in the absence of Erk did not affect LSF binding. Incubation with denatured Erk also had no effect on RCS binding (not shown). Phosphorylation by Erk is therefore required to diminish LSF binding to the RSC in EMSA (Fig. (Fig.4B,4B, compare lanes 1 and 2).
Surprisingly, when LSF is phosphorylated by p38, the opposite effect is observed. LSF binding to the RCS in EMSA increases quantitatively in a time-dependent fashion following in vitro phosphorylation by p38 (Fig. (Fig.4C).4C). In vitro phosphorylation by Jnk induces a marginal increase of LSF binding to the RCS in EMSA (Fig. (Fig.4D4D).
Taken together, these results demonstrate that in vitro, phosphorylation of LSF by MAPKs modulates the binding of LSF to the RCS of the HIV LTR, a region encoding two parallel and one antiparallel LSF motif. Of note, MAPKs have no effect on in vitro LSF binding to the canonical SV40 site, encoding two parallel motifs.
LSF has multiple phosphorylation sites, both unique and common to Jnk, p38, and Erk (42, 43) (Fig. (Fig.3).3). We therefore asked if phosphorylation by multiple MAPKs could have synergistic or competitive effects on LSF binding to the HIV RCS. We treated LSF with Erk and p38 (Fig. (Fig.4E)4E) and observed decreased LSF binding to RCS. This effect is comparable to the effect obtained after phosphorylation by Erk alone, suggesting that the effect of Erk is predominant.
We then tested the LTR binding of LSF291, which is mutated at the site which is phosphorylated by Erk. LSF and LSF291 were phosphorylated with both Erk and P38 in the presence of [γ-32P]ATP. As expected, p38 mediated phosphorylation of both LSF and LSF291, whereas Erk only phosphorylated LSF (Fig. (Fig.5A).5A). EMSAs were performed using phosphorylated LSF and LSF291. Binding of LSF291 phosphorylated with Erk did not decrease, whereas binding of the phosphorylated LSF291 with p38 increased (Fig. (Fig.5B5B).
Specific and selective inhibitors have been used to differentiate the effects of MAPK enzymes in cellular regulation by extracellular stimuli. In order to test the hypothesis that p38 and Erk uniquely modulate LSF binding to the HIV LTR in vivo, we used a cell line containing a single copy of the HIV LTR (6). Cells were treated with SB203580, a highly specific inhibitor of p38 (10, 46), or U0126, a specific inhibitor of MEK1 and MEK2 (13), and nuclear extracts were prepared to test the effect of MAPK inhibition on the ability of LSF to bind HIV LTR RCS oligonucleotide. When cells were treated with the Erk inhibitor U0126, the binding of LSF present within nuclear extract increased threefold compared with untreated cell controls (Fig. (Fig.6A,6A, compare lanes 1 and 2). Consistent with in vitro studies, when cells were treated with the p38 inhibitor SB203580, the binding of LSF decreased significantly compared to untreated cell controls (Fig. (Fig.6A,6A, compare lane 1 and 3).
In comparison, just as in vitro phosphorylation of LSF binding has no effect on LSF binding to the canonical SV40 site, LSF present in inhibitor-treated nuclear extract binds well to LSF-280 in EMSA. We saw no effect of MAPK inhibition on LSF binding to the LSF-280 oligonucleotides (Fig. (Fig.6B).6B). Further, no effect was seen when subsaturating concentrations of LSF were added (not shown).
We then performed ChIP assays to extend these findings in the natural environment of an integrated HIV promoter within cellular chromatin. HeLa-CD4-LTR-CAT cells were treated with or without the inhibitors U0126 and SB203580, fixed, and cross-linked with formaldehyde, and ChIP was performed. The p38 inhibitor SB203580 decreased LSF occupancy in the region of the HIV RCS (Fig. (Fig.7A).7A). PCR using β-actin promoter primers showed no effect of SB203580 treatment, confirming the specificity of the SB203580 effect on LSF occupancy. LSF occupancy at the HIV LTR binding to the RCS is therefore regulated by p38. A significant change in LSF occupancy following treatment with the Erk inhibitor U0126 was not observed in HeLa-CD4-LTR-CAT cells (data not shown).
To confirm the effect of p38 on LSF occupancy, we treated HeLa-CD4-LTR-CAT cells with anisomycin, which strongly activates the MAP kinases Jnk and p38 but not Erk (3, 13). The occupancy of LSF at the LTR was analyzed by ChIP assay after 2 h of exposure to 10 μM anisomycin. LSF occupancy increased significantly compared with untreated cell controls, without change in the β-actin PCR product (Fig. (Fig.7B7B).
Acetylation of histones by histone acetyl transferases reduces their net positive charge, weakens internucleosome interaction, and is generally associated with transcriptional activation. Conversely, hypoacetylated chromatin is associated with transcriptional repression (14, 18). Therefore, as LSF binds the LTR and recruits YY1, which in turn recruits HDAC1 (8), decreased occupancy of LSF at the LTR after treatment with SB203580 was predicted to increase acetylation of histone H4 (15).
Representative results from several independent ChIP assays using nuclear extract from cells treated or not with SB203580 are shown in Fig. Fig.7C.7C. Acetylated histone H4 is easily detected at Nuc 1 in untreated cells, but acetylation is significantly increased upon exposure to SB203580. This functional consequence could facilitate HIV LTR transcription.
In cooperation with cellular LSF, overexpression of YY1 can increase HDAC recruitment to the LTR, blocking Tat activation (8, 15). DNA-bound LSF is required for recruitment of YY1 and thereby HDAC1 to the HIV promoter. Therefore, SB203580 would be predicted to block the ability of YY1 to inhibit Tat activation. HeLa-CD4-LTR-CAT cells transfected with CMV-YY1 and/or CMV-Tat were treated with 10 μM SB203580 for 30 min after transfection. As expected, Tat activated expression of the integrated HIV LTR (mean increase of 75% ± 4%; n = 5 independent transfections [Fig. [Fig.8,8, lane 2]), and this increase was blunted by cotransfection of YY1 (fivefold decline of mean activation to 15% ± 5% [Fig. [Fig.8,8, lane 3]). However, YY1-mediated repression was ablated by SB203580 (Fig. (Fig.8,8, lane 4). In the absence of YY1, SB203580 had no effect on LTR expression (Fig. (Fig.8,8, lanes 1, 5, and 6).
Ideally, a reagent that induces latent HIV expression should not simultaneously increase the likelihood of de novo infection. Cell surface phenotype analysis and de novo infection in the presence of SB239063 were examined to explore the effect of SB239063 on the ability of PBMC to support HIV infection (Fig. (Fig.99).
Primary PBMC were cultured in the presence of IL-2, PHA plus IL-2, or SB239063 plus IL-2 for 3 days. Pretreated PBMC were then infected with a CXCR4-tropic clone (HIVLAI), and p24 was assayed over 2 weeks. Infection of seronegative donor cells in the presence of 1 to 20 μM SB239063 and IL-2 did not result in a change in production of HIV compared with cells cultured with IL-2 alone (Fig. (Fig.1010).
A major latent reservoir of HIV infection resides in resting memory CD4 T cells, which harbor integrated, functional, but quiescent proviral HIV genomes. This reservoir can be quantified and studied by negative selection of resting CD4 T cells on the basis of surface markers, incubation of cells with an activator of T-cell proliferation or signaling, and amplification of the resultant output virus by the addition of CD4 lymphoblasts from uninfected donors (4, 19, 25, 28, 33, 43). The selection of HIV-infected donors with prolonged suppression of viremia for such studies minimizes the potential for the outgrowth of actively replicating HIV in cultures of resting cells.
Lymphocytes were obtained by leukopheresis of five volunteers with CD4 cell counts ranging from 369 × 106 to 1,090 × 106 cells/liter. These subjects had been aviremic (<50 copies/ml HIV-1 RNA) for 24 to 42 months. More than 110 × 106 resting cells from each subject were assayed. As a positive control, cells were exposed to PHA and activated, irradiated allogeneic PBMCs. As a negative control for nonspecific stimuli ex vivo, cells were maintained in culture in the presence of 20 U/ml IL-2. To test the ability of p38 inhibition to induce outgrowth of HIV, resting CD4 cells were cultured in the presence of 20 U/ml IL-2 and SB239063 at concentrations of 1 μM and 20 μM.
Exposure to SB239063 resulted in outgrowth of HIV from cultures of three of five of these patients' cells (Table (Table11).
LSF represents a novel family of homo-oligomeric transcription factors that bind direct DNA repeats. In the human α-globin and γ-fibrinogen sites and chicken α-crystallin gene, LSF binds preferentially as a homotetramer (29, 38). Our findings suggest a model in which LSF binds to the RCS as a homotetramer, recognizing two of the LSF motifs within the RCS, including the most 3′ motif. We find that LSF binding to the HIV LTR is regulated by phosphorylation of LSF.
DNase I footprinting and gel shift experiments show that LSF binds with high affinity to the LTR (19, 35, 49). The LSF binding site is characterized by a 4-bp motif separated by a linker sequence composed of five or six nucleotides (CTGG-N5/6-CTGG) (26). This motif is repeated in the LTR three times, within the −10 to +27 sequences.
Using bacterially expressed LSF and radiolabeled oligonucleotides, we characterized LSF binding to the LTR in greater detail. Mutation of any two of the three 4-bp binding sequences reduces but does not abolish LSF binding. Interestingly, the mutant oligonucleotide M2, which has the two CTGG motifs separated by 15 bp, can bind LSF to the same extent as the mutants M1 or M3, in which the CTGG motifs are separated by five or six nucleotides. This suggests that the altered linker length destabilizes binding but does not eliminate the ability of LSF to bind its cognate sequence. Moreover, the binding of LSF to single site 3 seems to be as efficient as LSF binding to any combination of two 4-bp motifs. It is possible that LSF binds first to site 3, and this interaction allows the protein to recognize the second and first binding 4-bp motif, possibly because site 3 in this context is more accessible to LSF.
Although these results were obtained using oligonucleotides and so do not fully represent the interactions of transcription factors and DNA in the native chromatinized context, it is interesting that our results echo those of previous studies. Using DNase footprinting assays, Jones et al. found that out of all the sets of clustered point mutations in each LSF binding motif at the LTR, only mutations in all three motifs eliminated the binding of LSF (19).
By analyzing the DNA-protein complexes from mixtures of LSF derivatives of different sizes, we conclude that LSF binds to the HIV LTR as a tetramer, in agreement with Shirra and Hansen (38). However, our results differ from those obtained by Zhong et al. (57). In this article, the authors suggest that LSF binds to the LTR as a dimer. This difference can be explained by the fact that the authors used bacterially produced LSF isolated by glycerol gradient sedimentation and that LSF as a dimer in solution can bind DNA. However, dimers formed in solution could form tetramers upon interaction with DNA, as we observed. In gel shift experiments using LSF and GST-LSF, the authors did not explain the presence of other complexes with intermediate mobility between LSF and GST-LSF-HIV DNA (57).
We have previously proposed that LSF plays a role in maintaining or establishing transcriptional repression of the LTR (8, 35). This effect is mediated by the formation of the repressor complex at the LTR, consisting of LSF binding to HIV DNA, YY1 interacting with LSF via its zinc fingers, and histone deacetylase interacting with YY1. We have also shown that repression is blocked when LSF binding to the LTR is inhibited (8). We therefore sought to understand the signaling pathways that regulate LSF binding to the LTR and consequently regulate formation of repressor complex.
LSF can be phosphorylated by the cellular kinases Erk, p38, and Jnk. The LSF residues phosphorylated by Erk have been mapped and appear to be distinct from those phosphorylated by p38; Jnk appears to act on both sets of residues. These findings suggest that the in vivo regulation of LSF binding may be quite complex (42, 43). We have shown in vitro that phosphorylation of recombinant LSF by the activated kinase Erk or p38 is sufficient to regulate its binding to the HIV LTR but not to the SV40 late promoter. Phosphorylated LSF using activated Erk decreased its binding to the RCS, and conversely, LSF binding to the RCS is increased by p38 phosphorylation. These findings suggest that LSF binding to the HIV RCS is subject to counterregulation by MAP kinases.
It has been shown that LSF is phosphorylated by Erk in vitro on the same residues that are phosphorylated in vivo, with serine 291 being the major site of phosphorylation (43). However, in chromatin immunoprecipitation assays, we were unable to observe a significant increase in LSF occupancy after Erk inhibition by U1026. Either Erk phosphorylation of LSF does not play a role in vivo or there is little phosphorylation at Erk-responsive residues in LSF within the model system that we studied and so no change in LSF occupancy can be detected upon Erk inhibition.
The in vitro effect of p38 on LSF binding to the LTR site was confirmed in chromatin IP experiments. The specific inhibition of p38 decreases LSF occupancy at the LTR, and an agonist of p38, anisomycin, increases LSF occupancy. This provides strong evidence that p38 regulates LSF binding to the HIV LTR. Moreover, this decrease of LSF occupancy is accompanied by increased histone acetylation at nucleosome 1 of the LTR, an event that is correlated with transcriptional activation (5) and consistent with the requirement for LSF for recruitment of HDAC1 to this site.
Further, phosphorylation of LSF regulates the ability of the LTR promoter to be repressed. The p38 antagonist ablated repression of Tat activation of the integrated LTR. These experiments confirmed other studies showing that LTR activity increases when LSF binding is blocked (54).
Several investigators have studied the role of MAPK, such as p38 and Erk, in the activation of LTR in different cells lines and in primary PBMCs (7, 12, 23, 36, 50). These studies, performed in the context of activation of LTR by cytokines and stress, found that p38 inhibition blocks the activation of the HIV LTR in HeLa-LTR indicator cells in response to IL-1, tumor necrosis factor, UV radiation, or osmotic stress (23). Similar results were reported for the monocytic cell line U1 and for PBMCs (36).
However, we found that specific inhibition of p38 activity induced the outgrowth of latent HIV when the resting CD4 cells of aviremic patients were exposed ex vivo. Maintaining these CD4+ lymphocytes ex vivo with IL-2 (20 U/ml) did not result in virus outgrowth. Global inhibition of p38 activity is likely to alter many gene expression programs within a cell. This complexity likely explains the contradictory findings observed in different experimental systems; the effect of a biological signal in vivo is likely to result from the integration of many such opposing and synergizing pathways.
Nevertheless, when exposed to the specific inhibitor of p38, CD4+ cells did not become activated or more permissive for viral replication. Therefore, p38 inhibitors or future reagents with greater specificity for the cascade of factors that regulates HDAC1 recruitment to the HIV LTR might be considered as future therapeutic candidates to disrupt viral latency in patients on suppressive antiviral therapy (22, 32).
This work was supported by NIH grants CA81157 (U.H.), AI 45297, and AI 046376 (D.M.M.) and an amFAR fellowship to L.Y.
We thank Frederick Scott and Ginger Lehrman for technical assistance, Ronald Bosch for statistical assistance, and J. Victor Garcia and Donald Sodora for advice and collegial support.