Combinatorial Signals from CD28 Differentially Regulate Human Immunodeficiency Virus Transcription in T Cells*

Activation through the T-cell receptor and the costimulatory receptor CD28 supports efficient HIV transcription as well as reactivation of latent provirus. To characterize critical signals associated with CD28 that regulate HIV-1 transcription, we generated a library of chimeric CD28 receptors that harbored different combinations of key tyrosine residues in the cytoplasmic tail, Tyr-173, Tyr-188, Tyr-191, and Tyr-200. We found that Tyr-191 and Tyr-200 induce HIV-1 transcription via the activation of NF-κB and its recruitment to the HIV-long terminal repeat. Tyr-188 modifies positive and negative signals associated with CD28. Importantly, signaling through Tyr-188, Tyr-191, and Tyr-200 is required to overcome the inhibition posed by Tyr-173. CD28 also regulates P-TEFb activity, which is necessary for HIV-1 transcription processivity, by limiting the release of P-TEFb from the HEXIM1–7SK inhibitory complex in response to T-cell receptor signaling. Our studies reveal that CD28 regulates HIV-1 provirus transcription through a complex interplay of positive and negative signals that may be manipulated to control HIV-1 transcription and replication.

One of the major blocks to eradicating human immunodeficiency virus (HIV) 2 infections with highly active anti-retroviral therapy has been the inability of this treatment to eliminate cellular reservoirs harboring latent provirus. (1)(2)(3). T cells are a major target for HIV-1 infection, and T-cell signal transduction has been demonstrated to impact multiple steps of HIV-1 replication, including provirus transcription (4 -7). Characterizing T-cell signaling regulatory networks that govern T-cell function and HIV-1 transcription is critical for understanding the molecular mechanisms that directly contribute to the establishment, maintenance, and breaking of proviral transcription latency (8,9).
HIV provirus transcription is controlled by the upstream long terminal repeat (LTR), which includes cis-elements that are recognized by cellular transcription factors, including NF-B, AP-1, and NFAT that are induced in response to T-cell receptor (TCR)/CD28 engagement (10,11). These transcription factors recruit coactivators, including histone acetyltransferases and the ATP-dependent chromatin-remodeling Swi/Snf complexes that influence the chromatin structure of integrated provirus (12)(13)(14)(15)(16). Furthermore, the LTR forms an RNA stem loop structure, TAR, which the HIV transactivator, Tat, binds. Tat enhances RNA polymerase II (pol II) processivity by recruiting P-TEFb to the HIV-LTR (17,18). The availability of P-TEFb, which is negatively regulated through association with the HEXIM1/7SK RNA particle, is also controlled by cellular signals (19,20). Therefore, it may be possible to manipulate specific signaling cascades to control HIV transcription and improve the efficacy of current anti-viral regimens.
We have previously shown that signaling associated with CD28 positively and negatively regulates HIV-1 provirus transcription. Specifically, we demonstrated that Tyr-200 positively regulated HIV transcription by initiating Vav-1 and NF-B signaling, whereas recruitment of PI3K to the Tyr-173 residue inhibited the ability of Tat to bind P-TEFb and HIV-1 transcription (30,33). How these apparently opposing signals are coordinated to lead to induction of HIV-1 transcription, as well as the role of the other tyrosines in modulating HIV-1 transcription in response to CD28, has not been extensively investigated. Using chimeric CD28 receptors harboring mutations in different key tyrosines in the cytoplasmic domain, we show that CD28 induces HIV transcription through distinct but cooperative activities associated with the individual tyrosines.

EXPERIMENTAL PROCEDURES
Cell Lines and Primary Cell-Jurkat E6.1 T cells were purchased from American Type Culture Collection (ATCC, Manassas, VA) and grown in RPMI 1640 medium supplemented with 5% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 0.2 M L-glutamine. Human embryonic kidney 293T cells were also obtained from ATCC and cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Peripheral blood mononuclear cells were isolated from whole blood by Ficoll/histopaque gradient (Sigma-Aldrich), and CD4 ϩ T cells were positively selected using the Dynal isolation kit (Invitrogen, 113.21D).
CD8/28 Chimeric Receptor Mutants-The 8WT, YFFF, FFYF, FFFY, YFFF, YFYY, and YYFY expression vectors have been described previously (25,30,33,34), and these key residues are shown in Fig. 1. To generate receptors, FYYF and YFFY, the plasmids corresponding to pMHneo FFFY and pMHneo YFFF were digested with ApaI and HindIII (New England Biolabs). Two fragments were generated, a 400-bp fragment containing CD8␣ and the nucleotide sequence coding for tyrosine 173 of the cytoplasmic tail of CD28, and a 6.8-kb fragment containing the rest of CD28 and the pMHneo backbone. The 400-bp fragment from pMHneo FFFY and 6.8-kb fragment from pMHneo YFFF were gel-purified and ligated to generate pMHneo FYYF using T4 DNA ligase (Invitrogen). The pMHneo YFFY was generated similarly by ligating the 400-bp fragment isolated from pMHneo YYYF and the 6.8-kb fragment from pMHneo FFFY vectors. Additional mutants were generated by site-directed mutagenesis. Primers were designed (Table 1) to mutate key tyrosines to phenylalanines. Primers are listed in Table 1. For PCR reactions template plasmid DNA and appropriate primers were amplified using Vent polymerase (New England Biolabs) following standard protocols. The PCR products were digested with DpnI to eliminate donor plasmid and transformed into competent Escherichia coli DH5␣ cells. Positive clones were confirmed by sequencing.
Generation of CD8/CD28 Cell Lines-CD8/28 expression constructs were introduced in Jurkat E6.1 T cells using electroporation. 3 ϫ 10 7 cells were washed and resuspended in 750 l of serum-free RPMI containing 20 mM HEPES. 15 g of plasmid DNA was then added to these cells and electroporated using a T280 square electroporation system (BTX, San Diego, CA). Cells were given 1 pulse for 65 ms at 215 V in a 4-mm cuvette and then recovered in complete RPMI. 48 h post-transfections cells were put on selection by including 1 mg/ml G418 in the growth media. After 3 weeks cells expressing the chimeric receptors were positively selected for CD8 using the Dynal isolation kit (Invitrogen) to generate a polyclonal pool of cells.
Several independent pools for each receptor were generated to assure that there was no bias from an individual transfection and selection protocol. We also generated clonal cell lines and these behaved identically to the CD8/CD28 pooled cell lines (Refs. 30, 33 and data not shown). The expression of the chimeric receptors was verified by Western blot (data not shown) and flow cytometry.
Flow Cytometry-For flow cytometry 2 ϫ 10 6 cells were washed and resuspended in 100 l of staining media (phosphate-buffered saline containing 2% serum). Cells were incubated with 2 l of anti-CD8␣-phycoerythrin (BD 555635) and anti-CD28-FITC (BD 555728) for 45 min on ice. Cells were washed three times with staining media and fixed with 2% paraformaldehyde. Fluorescence was measured using a BD Biosciences FACScan at the Flow Core Facility at Boston Medical Center.
Generation of HIV-1 Infectious Titers and Infections-0.5 ϫ 10 6 293T cells were plated in a 6-well plate 24 h prior to calcium phosphate transfections, which were performed using 15 g of pNL4-3-Luc(ϩ) Env(Ϫ) Nef(Ϫ) (35) or pHXB-PLAP-Env Nef(ϩ) (36) (obtained from the National Institutes of Health AIDS Research and Reference Reagent Program) and 3 g of RSV-Rev, 3 g of LTR VSV-G. 293T transfection efficiency for pNL4-3-Luc was assessed by determining luciferase activity using a luciferase kit (Promega Madison, WI), whereas p24 enzyme-linked immunosorbent assays were performed for the pHXB-PLAP virus. Supernatants were collected and filtered through a 0.45-m disc prior to infection. Jurkat cells were infected with this virus for 12-16 h. Cells were then recovered and cultured in complete RPMI.
Activation of T Cells-Jurkat T cells were washed and resuspended in 5% fetal calf serum RPMI. 1 ϫ 10 6 cells were plated in each well of a 24-well plate. Cells were either left unactivated, or activated with 0.1 g/ml anti-human CD3 alone (BD 555336), anti-CD3, and 1.0 g/ml anti-human CD28 (BD 555725) or 1.0 g/ml anti-human CD8␣ antibodies (BD 555630) for 30 min. 5 g/ml of goat anti-mouse antibody (Sigma M 4280) was added to cross-link the receptors. Following 8 h of stimulation, Jurkat cells were harvested, and luciferase activity was measured. In experiments using cyclosporin A (CsA), infected Jurkat T cells were recovered and activated in the presence of 500 ng/ml CsA or vehicle control.
Immunoprecipitation and Immunoblots-Jurkat T cells were serum-starved for 12-16 h, activated with antibodies as described above for 5 min prior to preparing protein extracts with lysis buffer (10 mM Tris-Cl (pH 7.4), 150 mM NaCl, 1.0 mM EDTA (pH 8.0), 2.0 mM sodium vanadate, 10 mM sodium fluoride, 10 mM sodium pyrophosphate, 1% Triton X-100, 1.0 mM phenylmethylsulfonyl fluoride, and protease inhibitor mixture III (Calbiochem). Lysates were precleared by incubating with protein A/G beads (Santa Cruz Biotechnology, sc-2003) for 30 min at 4°C before incubating with primary anti-Vav (Santa Cruz Biotechnology, sc-132). Protein A/G beads were added to the antibody-lysate mix for 1 h at 4°C, and beads were washed three times with lysis buffer and then suspended in SDS-PAGE loading buffer. The samples were heated for 5 min at 100°C before being loaded onto a 10% SDS-PAGE gel. Proteins were transferred to a polyvinylidene difluoride membrane (Millipore) by electroblotting. Western blot analysis was carried out using a phosphotyrosine antibody (BD Transduction Laboratories, 610024). The blot was stripped and reprobed with a Vav antibody. CyclinT1 (Santa Cruz Biotechnology, sc-8127) and HEXIM1 (Abcam, ab28016) immunoprecipitations were also carried out with the same protocol; however, nuclear extracts (described in the next section) instead of total protein extracts were used. In the immunoprecipitation experiments done with the PI3K inhibitor, 50 M LY294002 (Promega) was introduced 30 min before activation to the culture. Western blots were quantified by densitometry. The ratio of CyclinT1 over HEXIM was calculated for all samples in Fig. 3A, and the ratio of HEXIM over CyclinT1 was calculated for Fig. 3 (B-D). The numbers depicted in the figures represent ratio of immunoprecipitation in each lane versus immunoprecipitation from the unactivated lane.
Transfection of LTR Reporter Constructs-Fifteen micrograms of pGL2 LTR luc and pGL2-mB-LTR luc constructs (kindly provided by Dr. Suryaram Gummuluru (37)) were electroporated into 20 ϫ 10 6 Jurkat E6.1 cells using the T280 BTX electroporator. The cells were recovered in 5% fetal calf serum RPMI for 16 h. 1 ϫ 10 6 cells were either left untreated or activated with 0.1 g/ml anti-human CD3 or 0.1 g/ml anti-human CD28 and 1.0 g/ml anti-human CD28, and luciferase assays were performed 6 h post activation as described above.
Statistical Analysis-Statistical analysis was carried out using Student t test. A two-tailed distribution was performed on paired samples, comparing CD3 responses to CD3 plus CD8 responses. Values of Ͻ0.01 were considered significant.

Specific Tyrosines within the CD28 Cytoplasmic Tail Regulate
HIV-1 Transcription-To study the role of CD28 in regulating HIV transcription, we employed a strategy that was previously described (25,30,33), in which we generated CD8/28 chimeric receptors with mutations in key tyrosine residues. The chimeric receptors were designed such that the cytoplasmic domain of CD28 was fused to the transmembrane and extracellular domain of CD8␣, which we refer to henceforth as CD8/CD28. This chimeric receptor forms a dimer similar to CD28 and functions identical to endogenous CD28 (25). In addition, the generation and expression of these CD8/CD28 chimeras allow their expression in cells along with endogenous CD28, and the direct comparison of mutant and WT CD28 signals in the same cells (25,30,33). As shown in Fig. 1A, we changed individual or different combinations of tyrosines in the cytoplasmic tail of CD28 to phenylalanine (F). Polyclonal populations of the Jurkat E6.1 T cell line expressing these chimeric receptors were generated, and stable expression of CD8/CD28, as well as endogenous CD28 receptors, was evaluated by flow cytometric analysis (Fig. 1B).
This panel of Jurkat T cells expressing CD8/CD28 chimeras was infected with NL4-3 luciferase virus to evaluate the function of different tyrosines in HIV-1 transcription. Using the HIV-luciferase clone, which lacks envelope and sup-ports only a single round of infection, allowed us to focus on CD28 signaling and HIV-1 transcription, rather than potential effects CD28 signaling has on virus replication and spread. Infected Jurkat T cells were stimulated through the TCR using anti-CD3 antibodies and either the chimeric CD8/CD28 receptor (using anti-CD8␣ antibodies) or the endogenous CD28 receptor (using anti-CD28 antibodies), which served as a control to ensure that the different T cell lines were capable of being activated. Controls included stimulating cells through the CD28, CD8/ CD28, and CD3 receptors alone. Consistent with our previous studies the chimeric receptor CD8/ CD28 WT, which retains all functional tyrosines, leads to transcriptional activation similar to that mediated by the endogenous CD28 receptor (25,30,33). Receptors lacking the cytoplasmic domain, del167, and the All F mutant, where all four tyrosines in the cytoplasmic tail of CD28 were mutated to phenylalanines, did not induce HIV-1 transcription confirming an indispensable role for these tyrosines in the cytoplasmic tail of CD28 (Fig. 2) (25,30,33).
Previous studies have shown that the Tyr-173 residue negatively regulates HIV-1 transcription, whereas signaling through Tyr-200 was necessary for HIV-1 transcription (30,33). We were interested in determining the functional interplay between these two apparently opposite activities mediated through Tyr-200 and Tyr-173, as well as the integration of signals downstream of other tyrosines within the CD28 cytoplasmic domain. We initially examined the ability of CD28 receptors with one functional tyrosine, Tyr-173 (YFFF), Tyr-188 (FYFF), Tyr-191 (FFYF), or Tyr-200 (FFFY), to support HIV-1 transcription. As expected, the YFFF receptor was unable to activate HIV transcription, consistent with our previous report that implicated Tyr-173 signaling as inhibitory (33). Furthermore, the FYFF receptor did not support HIV-1 transcription indicating that signals downstream of Tyr-188 were not sufficient to induce HIV-1 transcription. However, the FFYF and FFFY receptors activated HIV-1 transcription to levels comparable to the wild-type CD8/CD28 or the endogenous CD28   unstimulated, or stimulated with 0.1 g/ml anti-CD3 plus 1.0 g/ml anti-CD8␣, or with 1.0 g/ml anti-CD28, followed by 5 g/ml secondary antibody (goat anti-mouse) to cross-link the receptors. Eight hours post-activation the cells were lysed, and luciferase activity was measured. The luciferase activity is shown as a percentage of endogenous response, wherein the CD3 plus CD28 response of each cell line was set to a 100%. The dotted line represents the luciferase activity corresponding to activation through CD3 alone. The data shown were done in triplicate and represent one experiment of a total of three independent infections. A two-tailed t test was performed on paired samples comparing CD3 responses to CD3 plus CD8 responses. *, p Ͻ 0.01; **, p Ͻ 0.005. Signaling through YFFY and YFYF did not support activation of HIV-1 transcription, suggesting that positive signals downstream of Tyr-191 and Tyr-200 were not sufficient to overcome the repressive activity associated with Tyr-173. Surprisingly, Tyr-188, which was not capable of inducing HIV-1 transcription when it was the only tyrosine within the CD28 cytoplasmic tail was able to partially overcome the repressive activity of Tyr-173 resulting in 70% activity compared with the endogenous CD28 receptor (Fig. 2). Therefore, none of the tyrosines individually were able to completely overcome the negative signals associated with Tyr-173, implying that the ability of CD28 to activate HIV-1 transcription is a combinatorial event that requires cooperative activities of Tyr-188, Tyr-191, and Tyr-200. These data also suggest that Tyr-188 has a modulatory role in CD28 costimulation by interacting and modifying the activity of other tyrosines within the cytoplasmic tail, including Tyr-173.
Our previous data examining the YYYF receptor showed that Tyr-188 and Tyr-191 were not able to cooperate to activate HIV-1 transcription in the presence of Tyr-173 (33). To investigate what combination of tyrosine residues in CD28 could overcome the inhibition posed by Tyr-173, we studied the functions of the receptors containing the combination of Tyr-173, Tyr-188, and Tyr-200 (YYFY) and the combination of Tyr-173, Tyr-191, and Tyr-200 (YFYY). YYFY was compromised in its ability to activate HIV-1 transcription as it led to only 55% of the endogenous CD28 response. Activation of cells via YFYY receptor did not activate HIV-1 transcription (40% of endogenous receptor activation, similar to stimulation via the TCR alone) despite Tyr-191 and Tyr-200 being shown to be sufficient to activate HIV-1 transcription when present by themselves in the receptor. Even though Tyr-188 does not appear to directly activate HIV-1 transcription, this residue is indispensable for coordinating signals that induce HIV-1 transcription.
Whether Tyr-188 also modifies the activity of Tyr-191 and Tyr-200 was examined with chimeric receptors that had two functional tyrosines in the absence of the inhibitory signals associated with Tyr-173: Tyr-188 and Tyr-191 (FYYF), Tyr-188 and Tyr-200 (FYFY), and Tyr-191 and Tyr-200 (FFYY). The FYYF receptor was able to fully activate HIV-1 transcription upon stimulation through the TCR and the CD8/CD28 receptor. FYFY was also able to activate HIV-1 transcription; however, it was reduced ϳ20% compared with the FYYF, CD8/CD28 wild-type, or the endogenous receptor. Surprisingly, the FFYY receptor that includes Tyr-191 and Tyr-200, which alone are sufficient to induce HIV-1 transcription, does not activate HIV-1 transcription (Fig. 2). This suggests that Tyr-191 and Tyr-200 do not cooperate to activate HIV-1 transcription. Furthermore, these data are consistent with Tyr-188 modifying signals downstream of the other tyrosines in the intracellular domain of CD28. . TCR signals negatively regulate CyclinT1-HEXIM1 interaction. A, Jurkat cells stably expressing the indicated CD8/28 chimeric receptors were serum-starved for 12-16 h prior to activating with 0.1 g/ml anti-CD3 and 1.0 g/ml anti-CD8, whereas wild-type Jurkat T cells were activated with 0.1 g/ml anti-CD3 and 1.0 g/ml anti-CD28 for 30 min before adding 5 g/ml goat-anti-mouse for 4 h. Nuclear extracts were prepared, and CyclinT1 immunoprecipitation was carried out. The samples were run on a SDS-PAGE gel, Western blot analysis assessed the amount to CyclinT1, and an anti-HEXIM1 Western blot was done to determine the amount of HEXIM1 associated with CyclinT1. Blots were quantified in the panel on the right. B, CD4 ϩ T cells were isolated from human peripheral blood and were left unstimulated or activated with 0.1 g/ml anti-CD3, 0.1 g/ml anti-CD3, or 1.0 g/ml anti-CD28 alone. Receptors were cross-linked by adding 5 g/ml goat-antimouse for 4 h. Nuclear extracts were prepared, and an anti-HEXIM1 immunoprecipitation was carried out. Western blot analysis determined the amount of HEXIM1 immunoprecipitated and the amount of CyclinT1 associated with HEXIM1. Blots were quantified in the lower panel. C, nuclear extracts were prepared from Jurkat cells activated through the CD3 and CD28 receptors as described above, and an anti-HEXIM1 immunoprecipitation was carried out. Western blot analysis determined the amount of HEXIM1 immobilized as well as CyclinT1 associated with HEXIM1 (quantified in the panel below). D, CD4 ϩ T cells were treated with 50 M LY294002 for 4 h prior to preparing nuclear extracts. HEXIM1 was immunoprecipitated, and Western blot analysis determined the levels of HEXIM1 and CyclinT1 (quantified in the panel below). For all panels, nuclear extracts (NE) were probed with the indicated antibodies as an input control. Data shown are from a single experiment, and quantification was performed from at least two independent experiments. CD28 Signaling Does Not Activate P-TEFb-HIV-1 transcription is in part regulated by RNA pol II processivity (38,39). In the absence of Tat, there is an accumulation of short initiated transcripts as a result of RNA pol II pausing and premature termination that occurs due to the lack of P-TEFb (40,41). Tat enhances transcription processivity by binding the TAR element and recruiting P-TEFb to the HIV-LTR. P-TEFb is comprised of two subunits, CyclinT1 and Cdk9, and its availability to bind promoters is regulated by its association with the Hexim/7SK RNA particle complex (7SK RNP). Release of P-TEFb from the repressive 7SK RNP permits P-TEFb to target and activate RNA pol II complexes (19,20). We examined if CD28 signaling regulates P-TEFb availability and activity and determined whether this influences HIV transcription. We assessed the ability of the different cell lines expressing chimeric receptors to modulate CyclinT1-HEXIM interactions. Of the chimeric receptors examined, FFYF, FYFF, and FFFY, mediated release of P-TEFb from HEXIM (Fig. 3, A and B). Furthermore, the All F receptor and activation with anti-CD3 alone induced release of P-TEFb. These data indicate that TCR signaling is sufficient for releasing P-TEFb from the 7SK RNP. This was also observed in primary CD4 ϩ T cells isolated from peripheral blood, which showed a decrease in HEXIM1-associated CyclinT1 upon activation through the TCR (Fig. 3B).
PI3K is recruited to the YMNM motif in CD28, which includes Tyr-173. We have previously shown that PI3K inhibits P-TEFb-Tat interactions and HIV-1 transcription (33). Although CD28 signaling did not influence P-TEFb release from the HEXIM-7SK RNP complex, we wanted to determine if CD28 might negatively regulate P-TEFb by promoting the formation of P-TEFb-7SK RNP. Treating Jurkat T cells with the PI3K inhibitor LY294002 decreased CyclinT1-HEXIM interactions both before and after CD3 plus CD28 stimulation (Fig. 3C). Because Jurkat T cells have a defect in PI3K signaling due to a defect in phosphatase and tensin homolog (PTEN), we also assessed the effect of inhibiting PI3K in CD4 ϩ T cells (42). Inhibition of PI3K in primary T cells decreased the association of CyclinT1 and HEXIM (Fig. 3D). Our data indicate that active PI3K signaling promotes interactions between CyclinT1 and HEXIM, thus sequestering P-TEFb from Tat and inhibiting HIV-1 transcription. These data suggest that CD28, through the recruitment of PI3K to Tyr-173, limits HIV-1 transcription by inhibiting the release of P-TEFb.
CD28-mediated Induction of HIV Transcription Requires NF-B Activation-To gain a better understanding of the differential role of specific tyrosines in CD28 signaling, we examined if specific signaling events were associated with individual tyrosine residues. Vav, a guanine nucleotide exchange factor for the Rho GTPase, is known to be a target of T-cell activation and has been suggested to be downstream of CD28 signaling (43). We have shown that Vav is activated by the FFFY CD8/28 chimeric receptor and, in the absence of Tyr-200, Vav and Rac1, as well as NF-B, are not efficiently activated (30). These observations indicate that Vav and Rac are key regulatory events for inducing HIV-1 transcription. Jurkat cells expressing CD8/28 FYFF, FFYF, and FFFY receptors were stimulated by cross-linking TCR and CD28 receptors, and Vav activation and tyrosine phosphorylation were assessed by immunoblotting. Following T-cell activation, FFYF and FFFY, which fully support HIV transcription, induced Vav phosphorylation (Fig. 4A). However, FYFF, which is compromised in its ability to activate HIV transcription, also activated Vav (Fig. 4A), indicating that, although activation of Vav is downstream of CD28 signaling, it is not sufficient to induce HIV-1 transcription mediated by T-cell activation. We also inhibited NFAT signaling with CsA. As expected, inhibition of NFAT led to a decrease in overall HIV-1 transcription, but CD28 signaling still enhanced HIV-1 transcription, with an ϳ3-fold induction in both CsA-treated and untreated cells, indicating that the activation of NFAT is primarily regulated by the TCR (Fig. 4B). In addition, we examined the MAPK signaling pathways, ERK1/2, JNK, and p38, and did not detect differential activation of these molecules upon costimulation via the chimeric receptors FYFF, FFYF, and FFFY (data not shown).
A consequence of T cell signaling is the activation of transcription factors, which in turn bind regulatory cis-elements within promoters and enhancers to regulate transcription. The HIV LTR has binding sites for multiple transcription factors, including NF-B, NFAT, and AP-1 (44 -46). We wanted to examine if the activities of the chimeric receptors on HIV transcription reflected differential activation and recruitment of transcription factors to the HIV LTR. Because costimulation was only modestly affected by CsA, and MAPK pathways were not differentially activated by the CD28 mutants, which are upstream of NFAT and AP-1, respectively, we focused on the

. Vav activation is not necessary for induction of HIV transcription.
A, Jurkat T cells stably expressing the CD8/28 chimeric receptors FYFF, FFYF, and FFFY were activated as indicated with antibodies against CD3, CD28, and CD8, for 5 min after cross-linking the receptors. Vav was immunoprecipitated from cell lysates, and Vav phosphorylation was determined by anti-phosphotyrosine Western blot. Vav phosphorylation downstream the chimeric receptors (CD8 activation) were compared directly to the endogenous CD28 receptor. B, Jurkat T cells were infected with NL4-3 luc virus and activated with anti-CD3 and CD28 antibodies in the presence of CsA. Six hours post activation luciferase assay was performed. The unstimulated response was set as 1, and -fold activation has been plotted. Data shown are from a single experiment that is representative of three independent experiments. activation of NF-B, which has been shown to be indispensable for HIV-1 transcription (46). Electrophoretic mobility shift assays were performed using nuclear extracts isolated from activated Jurkat T cells stably expressing CD8/28 chimeric receptor mutants to measure NF-B binding activity following T-cell activation. Receptors that had Tyr-191 (FFYF) and Tyr-200 (FFFY) induced NF-B binding, whereas FYFF,YFFY, or the All F receptors were unable to activate NF-B, correlating with the ability of the former receptors to support HIV transcription (Fig. 5A). The specificity of NF-B binding was confirmed using unlabeled competitor oligonucleotides (Fig. 5A). We also verified the identity of the NF-B subunits by performing supershift assays in the presence of anti-p65 and anti-p50 antibodies (Fig. 5B). Anti-p50 antibody generated supershifted complexes and anti-p65 disrupted binding of NF-B consistent with the conclusion that the activated NF-B complexes are composed of p50 and p65 subunits. These data suggest that activation of NF-B is critical for CD28-mediated HIV-1 transcription.
To verify a role of NF-B in CD3 plus CD28-mediated HIV transcription, Jurkat T cells were transiently transfected with luciferase reporters under the control of either wild-type HIV LTR or a LTR harboring mB binding sites (mBLTR). LTR-luc was induced by ϳ4-fold when cells were activated through CD3 plus CD28 compared with cells either treated with anti-CD3 alone or not activated (Fig. 5C). Consistent with NF-B being necessary for transcriptional activation, CD3 plus CD28 signaling did not induce the mB-LTR (Fig. 5C). These data suggests that the activation and binding of NF-B to the HIV LTR is indispensable for HIV transcription.
To verify that CD28 signaling leads to changes in NF-B binding at the HIV-LTR of HIV-1 infected T cells, we performed chromatin immunoprecipitation experiments. Jurkat T cells expressing CD28 chimeric receptors were infected with pHXB.2 and activated using anti-CD3, anti-CD8, or anti-CD28 antibodies. Chromatin was prepared, and p65 binding at the HIV-LTR was measured. Chromatin immunoprecipitation results indicate that recruitment of p65 to the HIV-LTR is compromised when costimulatory signals are initiated by the FYFF receptor (Fig. 6A), whereas recruitment of p65 to the LTR was comparable to the endogenous CD28 receptor when cells were activated through the FFYF and FFFY chimeric receptors. We also examined the distribution of RNA pol II at the HIV-LTR versus ϩ5396 bp downstream of the transcriptional start site to determine RNA pol II processivity. RNA pol II was detected at both the LTR and downstream sequences in cells activated through the FFFY and FFYF receptors, whereas RNA pol II was not detected at the LTR and ϩ5396 in activated FYFF cells (Fig. 6, B and C). Overall, the levels of pol II associated with downstream HIV sequences directly reflected levels of pol II at the LTR indicating that pol II recruitment and not RNA pol II processivity are regulated by signaling through the chimeric receptors. Taken together, these data indicate that the ability of CD28 to enhance NF-B signaling and RNA pol II recruitment are primarily responsible for inducing HIV transcription in response to CD3 plus CD28 activation.

DISCUSSION
We have studied CD28 signaling and its impact on HIV transcription, with particular focus on the interplay between signals arising downstream from the four tyrosine residues present in its cytoplasmic domain. Our study reveals a complex interplay FIGURE 5. NF-B activation is indispensable for CD28-mediated HIV-1 transcription. A, Jurkat cells stably expressing the CD8/28 chimeric receptors FYFF, FFYF, FFFY, and YFFY were activated as described above. Nuclear extracts were prepared and incubated with radiolabeled NF-B probe for 20 min, and samples were run on a 6% polyacrylamide gel. Reactions were also performed in the presence of a 100-fold excess of specific competitor or nonspecific competitor. B, electrophoretic mobility shift assays were performed as in A in the presence of anti-p50 and anti-p65 antibodies. Gel-shifted complexes are denoted by arrowheads; supershifted complexes are indicated by bolded arrowheads. Data shown are from a single experiment that is representative of three independent experiments. C, Jurkat cells were electroporated with wild-type LTR-luciferase and mKB-LTR luciferase constructs. Cells were either left unstimulated, or stimulated with 0.1 g/ml anti-CD3, or 0.1 g/ml anti-CD3 and 1.0 g/ml anti-CD28, followed by 5 g/ml secondary antibody (goat anti-mouse) to cross-link the receptors. Eight hours post-activation the cells were lysed, and luciferase activity was measured. The luciferase activity is shown as -fold induction wherein luciferase units corresponding to the unstimulated samples were set to 1. of signals downstream of these tyrosine residues that are integrated to enhance HIV transcription.
Examining individual tyrosines we confirmed an inhibitory role for signals emanating from Tyr-173 and positive activity for Tyr-200 in CD28 signaling (25,30,33 PI3K is recruited to the YMNM motif of CD28, and mutation of the Tyr-173 to Phe-173 abolishes the recruitment and activation of PI3K (47). Our previous studies have shown that Tyr-173 inhibits HIV-1 transcription in a Tat-dependent manner by negatively regulating the Tat-P-TEFb complex (33). Our current study indicates that PI3K diminishes the availability of P-TEFb by stabilizing interactions between CyclinT1 and HEXIM1. Therefore, PI3K-mediated signals appear to favor the sequestration of P-TEFb in the 7SK RNP, decreasing the availability of P-TEFb for recruitment to the HIV-LTR. Furthermore, we observed a minimal role for CD28 in activating P-TEFb, with release of P-TEFb being primarily controlled by TCR signaling. This is consistent with previous reports indicating that Ca 2ϩ signaling released P-TEFb from 7SK RNP (48). Our data seem to be in conflict with recent data from Contreras et al. (49), who showed that PI3K/AKT increases active P-TEFb and HIV transcription induced by hexamethlene bisacetamide treatment. It is not apparent why there is a discrepancy between our data, although it could reflect differences in signals induced by hexamethlene bisacetamide versus CD28, cell model sys- FIGURE 6. CD28-mediated signals induce recruitment of NF-B to proviral LTR. A, Jurkat cells expressing the indicated CD8/28 chimeric receptors were infected with pHXB-PLAP virus. Infected cells were then activated for 6 h, and chromatin was isolated. Chromatin immunoprecipitations were performed using p65 antibody. FLAG antibody was used as a nonspecific antibody control. Real-time PCR was carried out for the Ϫ150 to ϩ76 region of the HIV LTR, and the data are plotted as a percentage of input immunoprecipitated in each reaction. B, Jurkat cells expressing the indicated CD8/28 chimeric receptors were infected with pHXB-PLAP virus. Infected cells were then activated for 6 h, and chromatin was isolated. Chromatin immunoprecipitations were performed using RNA pol II antibody. FLAG antibody was used as a nonspecific antibody control. Real-time PCR was carried out for Ϫ150 to ϩ76 region of the HIV LTR, and the data are plotted as a percentage of input immunoprecipitated in each reaction. C, Jurkat cells expressing the indicated CD8/28 chimeric receptors were infected with pHXB-PLAP virus. Infected cells were then activated for 6 h, and chromatin was isolated. Chromatin immunoprecipitations were performed using RNA pol II antibody. FLAG antibody was used as a nonspecific antibody control. Real-time PCR was carried out for ϩ5396 to ϩ5531 region of the HIV, and the data are plotted as a percentage of input immunoprecipitated in each reaction. tems, or temporal differences in our assays. It should be noted that a recent paper by Chen et al. (48) has also suggested a minimal role for PI3K in P-TEFb activation.
Activation of NF-B is indispensable for induction of HIV transcription. Residues that lead to NF-B signaling such as Tyr-191 and Tyr-200 promote NF-B binding, as well as recruitment of NF-B and RNA pol II to the HIV-1 LTR, whereas Tyr-188 does not activate NF-B or HIV-1 transcription. Furthermore, activation of Vav is not sufficient for activation of NF-B, because Tyr-188 is capable of inducing Vav phosphorylation but not HIV-1 transcription. Our results would suggest that NFAT and MAPK pathways have a minimal role in CD28-mediated HIV transcription and would implicate roles for other signaling pathways. Protein kinase C (50 -53), CARMA1 (54,55), and Cot kinase (56) are downstream of CD28 and potential mediators of NF-B signaling. The role of these factors and how they are integrated by CD28 signaling require additional investigation.
Recent studies have used similar approaches to examine the role of CD28 signaling in regulating T-cell activation and, in particular induction of IL-2 transcription. In summary, these reports suggest that the Tyr-173 residue and PI3K are dispensable for IL-2 transcription (25,(57)(58)(59), whereas Tyr-191 and Tyr-200 positively regulate IL-2 (25), correlating with CD28-mediated NF-B activation. Furthermore, in general, mutations in the CD28 cytoplasmic tail that failed to activate IL-2 transcription did not support efficient HIV-1 transcription in response to CD3 plus CD28 signaling. However, IL-2 and HIV-1, despite being activated by overlapping signals in response to T-cell activation, do exhibit some unique responses to CD28 signaling. For example, the FYYF chimeric receptor does not robustly induce IL-2 transcription but is capable of fully activating HIV-1 transcription (data not shown). The responses of IL-2 and HIV-1 transcription to CD28 signaling may reflect differential requirements for gene activation, such as, a lower signaling threshold for induction HIV-1 transcription.
In summary, we show that CD28-mediated induction of HIV-1 transcription is the result of coordinated positive and negative signals that ultimately lead to NF-B activation. We propose a model (Fig. 7) wherein TCR signaling positively regulates P-TEFb, whereas CD28 signaling primarily leads to the activation of NF-B and recruitment of RNA pol II. In addition, HIV-1 transcription can also be limited by CD28 signaling, which promotes the sequestration of functional P-TEFb in the 7SK RNP via PI3K, suggesting a negative feedback mechanism for CD28 to control T-cell activation. Our data, showing that NF-B is limiting in HIV-1 transcription, and that it is a key target of CD3 plus CD28 signaling, is consistent with several groups that have demonstrated a role for NF-B in the reactivation of latent HIV-1 transcription in cell lines and primary cell models (8,9,57,60,61). It is important to note that a recent report has suggested that NF-B may have a minor role in reactivating HIV-1 expression in latently infected primary T cells that phenotypically resemble central memory cells, and that NFAT and p38 MAPK are the endpoints of the CD3 plus CD28 signaling cascade responsible for HIV-1 transcription (62). Whether these apparently contradic-tory data reflect multiple mechanisms of latency or cell-specific responses to CD3 plus CD28 signaling is unclear, but they do underscore the need to further characterize the biochemical events necessary for activation of HIV-1 transcription if strategies are to be developed to mobilize latent HIV-1 from latently infected T-cell populations.