Identification of three NFAT binding motifs in the 5'-upstream region of the human CD3gamma gene that differentially bind NFATc1, NFATc2, and NF-kappa B p50.

Human immunodeficiency virus, type 1 (HIV-1) infection of CD4(+) T cells progressively abrogates T cell receptor (TCR).CD3 function and surface expression by specifically interfering with CD3gamma gene transcription. Our data show that the loss of CD3gamma transcripts begins very early after infection and accumulates to a >90% deficiency before a significant effect on surface receptor density is apparent. Blocking TCR.CD3-directed NFAT activation with cyclosporin A provokes a partial re-expression of CD3gamma gene transcripts and surface complexes in a time- and dose-dependent manner. We have identified three NFAT consensus sequences (5'-GGAAA-3') in the 5'-upstream region of the human CD3gamma gene at: -124 to -120 (NFAT(gamma1)), -384 to -380 (NFAT(gamma2)), and +450 to +454 (NFAT(gamma3)) from the first transcription initiation site. Using electrophoretic mobility shift and supershift assays, we show that NFATc2 alone binds to the NFAT(gamma2) motif; however, complexes containing either NFATc2 or NFATc1 plus NF-kappaB p50 bind to the NFAT(gamma1) and NFAT(gamma3) sites. We further demonstrate that NFATc1 and NF-kappaB p50 bind in the same protein.DNA complex and that a fourth Ala added to the core sequence (5'-GGAAAA-3') in NFAT(gamma1), and NFAT(gamma3) is critical for their binding. Finally, we have shown that an increase in the binding of nuclear NFATc2, NFATc1, and NF-kappaB p50 to these three motifs is correlated with a progressive loss of CD3gamma transcripts after HIV-1 infection.

T cell receptor (TCR) 1 ⅐CD3 cell surface density has been linked with the ability of the cell to elicit an effective signal, suggesting that T cells regulate their responsiveness to antigen-induced activation by increasing or decreasing the number of cell surface complexes (1)(2)(3)(4). The quantity of TCR⅐CD3 complexes present on the surface at any given time is a result of the balance between receptor internalization, leading to intracellular degradation or recycling to the surface, coupled with the synthesis, processing, and exportation of newly formed receptors (reviewed in Ref. 5). It is currently thought that two pathways regulate antigen-induced TCR⅐CD3 down-regulation from the cell surface: phosphorylation of the immunoreceptor tyrosine-based activation motifs present in the cytoplasmic tails of CD3, CD3␥, CD3␦, and CD3⑀ (6,7) and protein kinase C (PKC)-mediated serine phosphorylation of the di-leucine endocytosis motif in CD3␥ (8,9). A recent study has shown that the di-leucine motif in CD3␥ increases ligand-induced receptor internalization and degradation 3-to 10-fold, indicating that this chain plays a major role in TCR⅐CD3 down-modulation (10).
Defects in TCR⅐CD3 surface expression and function are increasingly being reported in an expanding range of clinical conditions, including both peripheral blood and tumor-infiltrating T cells in a wide variety of cancer patients (reviewed in Refs. 11 and 12) and after viral infection of CD4 ϩ T cells (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25). A common denominator for TCR⅐CD3 down-modulation by the CD4 ϩ T cell tropic viruses that has emerged from in vitro (15, 20 -22) and in vivo studies (14,(23)(24)(25) is their ability to interfere with expression of one or more of the CD3 genes. We have demonstrated that human immunodeficiency virus (HIV-1 (15,16) and HIV-2 (20)) infection of the human IL-2dependent CD4 ϩ T cell line, WE17/10, progressively abrogates TCR⅐CD3 function and surface expression by specifically interfering with transcription of the CD3␥ gene. Our data have shown that, when intracellular conditions favor expression of the viral regulatory genes tat and/or nef in the absence of rev, CD3␥ mRNA and TCR⅐CD3 surface density are down-regulated and TCR⅐CD3-mediated immune activities are diminished (26).
Nef is a multifaceted viral regulatory protein that is capable of a variety of different, independent functions, some of which have been linked with TCR⅐CD3-controlled events. It has been shown to directly associate with CD3 and lead to its downmodulation from the cell surface (27,28). Nef has also been shown to play a role in the post-transcriptional down-modulation of CD4 via a di-leucine motif in this receptor's membrane proximal cytoplasmic domain (29). This CD4 domain is strikingly similar to the di-leucine motif in CD3␥ (10, 30 -32) and thus conditions favoring Nef expression could potentially enhance the activity of the CD3␥ di-leucine motif.
The viral transcriptional transactivator protein Tat is also thought to play an important role in the immune suppression observed after infection by activating and suppressing the expression of a variety of cellular immune response genes (33)(34)(35)(36)(37). The transcriptional control elements for CD3␥ have remained elusive (the 5Ј-upstream region of this gene lacks a typical TATA or CAAT box), despite the identification of promoter and enhancer sequences for the other TCR⅐CD3 genes: TCR␣ (38,39), TCR␤ (40,41), TCR␥ (42), TCR␦ (43), CD3⑀ (44), CD3 (45,46), and the highly homologous CD3␦ (47)(48)(49). However, the recurring defect in CD3␥ gene transcripts observed after infection with a wide variety of HIV-1 and HIV-2 isolates suggests that transcription of this cellular gene might be controlled by a mechanism similar to the virus.
The primary function of HIV-1 Tat is to promote transcription by recruiting a kinase complex known as TAK (Tat-associated kinase) to the transactivation response RNA element present at the 5Ј-ends of all nascent HIV-1 transcripts and subsequently act in concert with cellular transcription factors bound to the long terminal repeat (LTR) (reviewed in Refs. 50 and 51). Among the many regulatory elements in the HIV-1 LTR, there are two adjacent NF-B binding sites that have been shown to be a major cis-acting element for viral gene expression (52). The NF-B/Rel family of transcription factors (p50, p65, RelB, c-Rel, and p52) are induced in response to T cell activation signals to bind to the NF-B consensus sequence (5Ј-GGGACTTTCC-3Ј) (53) and activate viral transcription (54). Members of the NFAT family of transcription factors (NFATc1 (NFATc, NFAT2); NFATc2 (NFATp, NFAT1); NFATc3 (NFATx, NFAT4); NFATc4 (NFAT3); and NFATc5 (TonEBP); approved UCL/HGNC/HUGO Human Gene Nomenclature) share a common architecture with the NF-B/Rel family and bind to a five-nucleotide core sequence (5Ј-GGAAA-3Ј), which, in addition to being found alone (55), is also contained within each NF-B consensus sequence (5Ј-GGGACTTTCC-3Ј). The promoter-enhancer regions of several activation-associated genes, some of which have been shown to be activated or suppressed after HIV-1 infection, possess NFAT binding sites, including those encoding IL-2, IL-3, IL-4, IL-5, IL-8, granulocyte-macrophage-colony-stimulating factor, tumor necrosis factor-␣, as well as cell surface receptors such as FasL and CD40L (reviewed in Ref. 56)).
Several groups have investigated the potential role of NFAT in HIV-1 replication and the interaction between Tat and NFAT and concluded that the NFAT family of proteins may have distinct effects on HIV-1 replication. NFATc2 is thought to negatively regulate the LTR by competing with the NF-B for its binding sites, whereas NFATc1 has been shown to positively regulate HIV-1 LTR through the NF-B binding sites (55). Recent studies suggest that virally induced immune suppression may be due to the interaction of Tat with several transcription factors, including Oct, Sp1, and NFAT (57, 58) as well as through indirect effects on the transcriptional activity of NF-B and AP-1 (59).
The data presented in this paper show that very early after HIV-1 infection in an IL-2-dependent T cell line, the majority (Ͼ90%) of CD3␥ transcripts are lost, and this occurs before significant TCR⅐CD3 down-modulation from the surface is apparent. Furthermore, treatment with the immunosuppressive drug, cyclosporin A (CsA), which acts by blocking translocation of NFAT proteins to the nucleus, partially reverses this CD3␥ transcription defect. We located three NFAT binding motifs (5Ј-GGAAA-3Ј) in the 5Ј-upstream region of the CD3␥ gene (NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 ) and found that increased nuclear translocation and binding of NFAT family proteins to these three sites parallels the loss of CD3␥ gene transcripts. Electrophoretic mobility shift assays (EMSA) show that the NFAT ␥1 and NFAT ␥3 motifs bind complexes containing either NFATc2 or NFATc1 plus NF-B p50, whereas the NFAT ␥2 motif binds NFATc2 containing complexes only.

EXPERIMENTAL PROCEDURES
Cell Culture Conditions and Reagents-The WE17/10 cell line is a human interleukin 2 (IL-2)-dependent CD4 ϩ T cell line (15,60) that was established and is maintained in RPMI 1640 containing 20% fetal bovine serum, 1.25 mM L-glutamine, 0.55 mM L-arginine, 0.24 mM Lasparagine, and 100 units of recombinant human IL-2 per ml (Cetus Corp., Emeryville, CA). WE17/10 cells infected with the HIV-1 isolate LAI (61) or the molecular clone HXB2 (62) were used in previous experiments (15,60). The human B lymphocyte line, Raji, was obtained from the American Type Culture Collection (Rockville, MD) and maintained in RPMI 1640 supplemented with 10% fetal bovine serum.
Flow Cytometry-Cells were analyzed for CD3 surface expression by flow cytometry as previously described (26). Briefly, cells were labeled with the murine monoclonal antibody OKT.3 (directed to CD3⑀) in a two-step process (using 1 g/ml of antibody to ensure saturation binding) followed by the manufacturer's recommended dilution of fluorescein-conjugated goat anti-mouse immunoglobulin (BD Biosciences, Erembodegen, Belgium). The labeled cells were fixed in 2% paraformaldehyde, and fluorescence was analyzed on a FACSCalibur (BD Biosciences).
Electrophoretic Mobility Shift Assay-Nuclear extracts were prepared from 2 ϫ 10 7 cells according to a modified version of the method described by Osborn (63). All buffers contained a mixture of protease inhibitors (Complete, Roche Diagnostics, Brussels, Belgium) to minimize proteolysis. The cellular pellet was washed with ice-cold phosphate-buffered saline and then resuspended twice with 1 ml of ice-cold buffer A (10 mM HEPES buffer, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl). Cells were collected by centrifugation (600 ϫ g for 10 min), resuspended, and incubated for 10 min with 40 l of ice-cold lysis buffer A containing 0.2% Nonidet P-40 (this step was repeated twice). The pellet (nuclear fraction) was incubated with 30 l of ice-cold extraction buffer C (20 mM HEPES buffer, pH 7.9, 25% glycerol, 1.5 mM MgCl 2 , 420 mM NaCl, 0.2 mM EDTA) for 20 min at 4°C and then centrifuged at 20,800 ϫ g for 10 min at 4°C. The nuclear supernatants were diluted with 150 l of buffer D (20 mM HEPES buffer, pH 7.9, 20% glycerol, 50 mM KCl, 0.2 mM EDTA) and stored frozen at Ϫ80°C. Protein concentrations were determined by the Bradford method (64).
EMSAs were performed as described by Van Lint et al. (65) with some modifications. Single-stranded oligonucleotides were 5Ј-end-labeled with [␥-32 P]ATP (Ͼ5000 Ci/mmol, Amersham Biosciences, AT Roosendal, Netherlands) using T4-polynucleotide kinase, annealed, isolated on a polyacrylamide gel, and extracted from the gel using the QIAXE II kit (Westburg, AE Leusden, Netherlands) prior to their use in EMSA experiments. Nuclear extracts (10 g of protein) were preincubated for 10 min in a reaction mixture containing 10 g of bovine serum albumin (Sigma-Aldrich, Bornem, Belgium), 1.5 g of the nonspecific competitor DNA poly(dI-dC) (Amersham Biosciences), 50 M ZnCl 2 , 0.25 mM dithiothreitol, 20 mM Tris-HCl, pH 7.5, 60 mM KCl, 1 mM MgCl 2 , 0.1 mM EDTA, and 10% (v/v) glycerol. 15,000 cpm of the 32 Plabeled probe was subsequently added, and the mixture (final volume, 20 l) was incubated for a further 20 min at room temperature before being loaded onto a 6% non-denaturing polyacrylamide gel (1ϫ Trisglycine-EDTA buffer, migrated at 50 V overnight). The radiolabeled proteins were detected by autoradiography on Biomax MR film (Amersham Biosciences).
Electrophoretic Mobility Shift Assay-Antibodies directed to the NFAT family proteins NFATc1 (SC-7294X) and NFATc2 (SC-7295X), the NF-B family proteins p50 (SC-1190X), p65 (SC-109X), c-Rel (SC-6955X), Rel-B (SC-226X), and p52 (SC-7386X), and the AP-1 family proteins c-Jun (SC-1694X) and c-Fos (SC-52) (all from Santa Cruz Biotechnology) were preincubated with nuclear extracts for 1 h on ice prior to the addition of the radiolabeled probe for the supershift assay. In the super-supershift and double-supershift assays, the first antibody was preincubated with the nuclear extract for 45 min on ice followed by a subsequent incubation with the second antibody for an additional 45 min on ice before a final 20-min incubation with the radiolabeled probe at room temperature.
Five micrograms of total RNA from uninfected and HIV-1-infected WE17/10 cells at various stages of TCR⅐CD3 down-modulation was reverse-transcribed into cDNA in the presence of Moloney murine leukemia virus reverse transcriptase (2.5 units/l; Roche Diagnostics, Brussels, Belgium), 0.5 mM of each dNTP, 1 unit/l RNase inhibitor, 30 pmol of the forward primers for CD3␥ or CD3␦, 0.01 M dithiothreitol, 20 l of 5ϫ first-strand buffer (250 mM Tris-HCl, 200 mM KCl, 25 mM MgCl 2 , 2.5% Tween 20 (v/v), pH 8.3) in a total volume of 100 l. The RT mix was incubated at 30°C for 10 min and 42°C for 45 min.
An internal standard for use in the competitive RT-PCR assay was constructed from a full-length cDNA sequence of the human CD3␥ gene subcloned from pJ6T3␥-2 (68) into the EcoRI site of pUC18 (Invitrogen, Merelbeke, Belgium), and the resulting plasmid was called pUC18␥. This recombinant plasmid was then used to construct a competitor by cutting a 1071-bp XhoI fragment from pV344 (69) and ligating it into XhoI-digested pUC18␥, producing the plasmid pUC18␥c. The competitor copy number was calculated using the concentration measured by absorbance at 260 nm and the molecular weight of pUC18␥c (i.e. 1 mol of the full-length pUC18␥c DNA is equal to 4557 bp ϫ 700 Da (the average molecular mass of a deoxynucleotide base pair) ϭ 3.1899 ϫ 10 6 ).
Human CD3␥ gene expression was measured in a quantitative competitive RT-PCR assay, where the target cDNA was co-amplified with the same stock dilution series of the pUC18␥c competitor in all experiments. For each target sequence, 20 sequential dilutions of the pUC18␥c competitor DNA (from a minimum of 3.3 ϫ 10 3 to a maximum of 6.6 ϫ 10 9 copies) were co-amplified with 100 ng of cDNA, 1 unit of Taq polymerase (Amersham Biosciences), 0.2 mM dNTP, 0.4 M of each primer, 15 mM MgCl 2 in a final volume of 50 l in Taq DNA polymerase buffer (Amersham Biosciences). Amplification of CD3␥ was performed with an initial denaturation step of 5 min at 94°C followed by 35 cycles of amplification: 10 cycles of denaturation at 94°C for 35 s, annealing at 50°C for 20 s, and extension at 72°C for 30 s followed by 25 cycles of denaturation at 94°C for 30 s, annealing at 65°C for 20 s, and extension at 72°C for 45 s with a 1-s/cycle automatic prolongation of the extension period. Amplification of CD3␦ was performed with an initial denaturation step of 5 min at 94°C followed by 40 cycles of denaturation at 94°C for 20 s, annealing at 58°C for 15 s, and extension at 72°C for 1 min. After amplification, the samples were incubated at 72°C for 7 min, separated on a 1% agarose gel, and stained for 10 min with a freshly prepared ethidium bromide solution (0.5 g/ml).

Measurement of the Relative Amounts of CD3␥ mRNA in Uninfected TCR⅐CD3 ϩ and HIV-1-infected Cells with Downmodulated TCR⅐CD3 Surface
Complexes-Our previous work, using dot and Northern blot hybridization analyses, suggested that the specific loss of CD3␥ transcripts after HIV-1 and HIV-2 infection does not parallel the down-regulation of TCR⅐CD3 complexes from the surface at a ratio of 1:1 (15,20). To better define the relationship between the number of CD3␥ gene transcripts and the density of TCR⅐CD3 complexes on the cell surface, we used quantitative competitive RT-PCR to examine transcript levels in uninfected and HIV-1-infected WE17/10 cells. RNA was extracted from cells at different stages in the progression from TCR⅐CD3 hi 3 TCR⅐CD3 lo 3 TCR⅐CD3 Ϫ (previously described in Ref. 16; in this report the uninfected cells designated as 100% TCR⅐CD3 ϩ are all TCR⅐CD3 hi ; whereas, the HIV-1-infected cells described as 90% TCR⅐CD3 ϩ (for example) are 10% TCR⅐CD3 Ϫ and 90% TCR⅐CD3 lo ). cDNAs, reverse-transcribed from the native RNA preparation, were co-amplified with serial dilutions of a competitor specific for the human CD3␥ gene (pUC18␥c), which had been engineered to produce a larger PCR product ( Fig. 1, upper band) than the cellular CD3␥ RNA (Fig. 1, lower band).
Representative results comparing the relative amounts of RT-PCR products from uninfected and HIV-1-infected cells expressing various levels of TCR⅐CD3 surface receptors are shown in Fig. 1A. In the uninfected 100% TCR⅐CD3 ϩ cells, the competitor was initially detected when 3.3 ϫ 10 6 molecules were added to the reaction mixture, followed by a corresponding decrease in native CD3␥ transcripts until they are no longer detectable in the presence of Ͼ6.6 ϫ 10 8 molecules of the competitor. The competitor was detected earlier (at 6.6 ϫ 10 5 molecules) in RNA amplified from 100% TCR⅐CD3 ϩ HIV-1infected cells (the mean fluorescence revealed that these cells were actually 100% TCR⅐CD3 lo with a receptor density equal to 85% of the uninfected control cells analyzed in parallel; data not shown) and indicated that these TCR⅐CD3 lo cells had already lost Ϯ80% of their CD3␥ gene transcripts. Amplification of RNA from 90% TCR⅐CD3 ϩ HIV-1-infected cells initially detected the competitor at a concentration of 1 ϫ 10 5 molecules, revealing a further decline equivalent to a total loss of Ͼ90% of CD3␥ gene transcripts. This extensive loss of transcripts prior to significant TCR⅐CD3 down-modulation was consistent for cells infected with a wide variety of viral variants. RNA extracted from HIV-1-infected cell lines expressing 60 -89% TCR⅐CD3 ϩ (64% is shown in Fig. 1A) were competed at essentially the same concentrations as the 90% TCR⅐CD3 ϩ cells, most likely due to the limited sensitivity of this series of competitor concentrations once transcript numbers are low. Because the cells have lost more than 90% of their CD3␥ gene transcripts before substantial numbers of TCR⅐CD3 Ϫ cells are detectable, any changes in the remaining transcript levels (only 10% of normal levels) would have a magnified effect on the number of surface receptor complexes. The erosion of CD3␥ transcripts (represented graphically in Fig. 1B) continues in 25 and 5% TCR⅐CD3 ϩ cells (Fig. 1A) and were completely undetectable in HIV-1-infected TCR⅐CD3 Ϫ cells and the B cell line Raji (Fig. 1C). Under the same standardized RT-PCR conditions, transcript levels for the highly homologous CD3␦ gene were unchanged in all of the RNA preparations (Fig. 1D). These data demonstrate that the loss of CD3␥ gene transcripts in HIV-1-infected cells begins very early after infection and that a substantial drop in transcript levels (Ͼ90% of the normal number) must occur before a significant effect is observed on receptor surface density.
Cyclosporin A Partially Restores TCR⅐CD3 Expression on the Surface of HIV-1-infected Cells-We next asked whether activators or inhibitors known to affect various steps in the TCR⅐CD3 activation pathway could arrest or reverse the loss of CD3␥ gene transcripts after infection and thereby partially or completely restore receptor surface expression. Uninfected and HIV-1-infected WE17/10 cells at different stages of receptor down-modulation were treated with the PKC activator, PMA, the calcium ionophores, A23187 and ionomycin (which can induce phosphorylation of CD3␥ on Ser-126 without activation of PKC), a combination of PMA plus ionophore (PMAϩIono), as well as immobilized anti-CD3 antibody to mimic antigen-induced activation. Cells were also treated with the calcium channel blocker EGTA and its membrane-permeant derivative BAPTA/AM, the tyrosine-protein kinase inhibitor herbimycin A, the PKC inhibitor staurosporine, and the immunosuppressive agent cyclosporin A (CsA).
Cells, treated for various lengths of time and at a variety of different drug concentrations, were screened by flow cytometry for modulation of surface CD3, and representative data are shown in Fig. 2. As expected, activation by PMA, PMAϩIono, or anti-CD3 resulted in further down-modulation of receptors on TCR⅐CD3 hi uninfected or TCR⅐CD3 lo HIV-1-infected cells but had no effect on the TCR⅐CD3 Ϫ -infected cells (Fig. 2, A and B; histograms for anti-CD3 are not shown but were similar to those shown for PMA or PMAϩIono). Staurosporine and herbimycin A had a deleterious effect on cell viability after 48 h, but they had no discernable positive or negative effect on receptor surface density after treatment for 18 -24 h where viability was not affected (the histogram profiles shown in Fig. 2, A and B, for staurosporine are identical to those for herbimycin A). Cells treated with BAPTA/AM, but not EGTA, exhibited a slight but consistent down-modulation of TCR⅐CD3 complexes on both uninfected and HIV-1-infected cells, particularly noticeable as an increased number of cells in the TCR⅐CD3 lo range (Fig. 2, A and B), but this intracellular calcium chelator also had a deleterious effect on cell growth and viability.
The most consistent positive effect was observed after CsA treatment of HIV-1-infected cells, which partially restored TCR⅐CD3 complexes on the cell surface of HIV-1-infected cells in a time-and dose-dependent manner. This effect is shown graphically as an increase in the percentage of TCR⅐CD3 ϩ cells after treatment with 0.1-1.0 g of CsA for 3 or 7 days (Fig. 2C), as well as by histograms that illustrate the movement of cells from the negative to positive phenotype after 5 days of treatment with 0.1 g of CsA (Fig. 2D). CsA also provoked a slight down-modulation of CD3 density on the surface of uninfected cells (Fig. 2D), which was augmented with increased time and drug concentrations (Fig. 2C). No cytotoxicity was observed in any of the CsA-treated cell cultures likely due to the fact that WE17/10 cells were grown in the presence of an excess of exogenously added IL-2 (70).
RNA from CsA-treated HIV-1-infected cells (85% TCR⅐CD3 ϩ ) was analyzed by quantitative competitive RT-PCR (to increase  (Fig. 2E). CD3␥ transcripts were initially detected at a competitor concentration of 8.3 ϫ 10 4 for the CsA-treated cells compared with 4.9 ϫ 10 4 for the untreated cells, which represents an approximate 2-fold increase in transcript numbers. Taken together with the fluorescenceactivated cell sorting data, these results suggest that CsA treatment has a net positive effect on CD3␥ gene transcription in HIV-1-infected cells, resulting in the formation of more complete TCR⅐CD3 complexes that can then be processed to the cell surface. The cellular target of CsA is the calcium-regulated phosphatase calcineurin, which controls nuclear translocation of the NFAT family of transcription factors. Translocation of NFAT to the nucleus, induced in response to antigen activation, is essential for immune response-directed cytokine gene expression, and it is via this pathway that CsA exerts its immunosuppressive activity.
Identification of Three NFAT Consensus Sequences in the Human CD3␥ Gene-Consequent to the up-regulation of CD3␥ transcripts observed after cyclosporin A treatment, we asked whether there were any potential NFAT binding motifs in the 5Ј-upstream sequence of the human CD3␥ gene. We identified three NFAT consensus sequences (5Ј-GGAAA-3Ј) at Ϫ124 to Ϫ120 (NFAT ␥1 ), Ϫ384 to Ϫ380 (NFAT ␥2 ), and ϩ450 to ϩ454 (NFAT ␥3 ) from the first transcription initiation site (Fig. 3A, based on the published sequence NCB accession number X06026 (71)). We further asked whether alignment of the 5Ј-upstream region of CD3␥ gene with the 5Ј-LTRs of HIV-1 (Strain HXB2, NCB accession number K03455) and HIV-2 (Strain BEN, NCB accession number M30502) would expose regions of sequence homology. This analysis revealed that the second motif, NFAT ␥2 (5Ј-TTTCC-3Ј), is nested in a region (Ϫ412 to Ϫ372) that shares sequence similarity with the functional NF-B cis-acting sequences located upstream from the SP1 binding sites and the TATA promoter in both the HIV-1 and HIV-2 LTRs (Fig. 3B). However, the first NF-B consensus sequence in the HIV-1 and HIV-2 LTRs varies from the potential site in CD3␥ by two nucleotides (GGGACTTTCC in HIV compared with GTGGCTTTCC in CD3␥) of which the first three Gs are thought to be critical for NF-B binding (72,73).
Nuclear Protein Complexes Bind to the NFAT ␥1 Motif in CD3␥-An oligonucleotide probe extending from Ϫ132 to Ϫ113 (underlined in Fig. 3A) was used to examine the in vitro binding of nuclear proteins to the NFAT ␥1 motif by EMSA. Nuclear extracts of unstimulated WE17/10 cells (100% TCR⅐CD3 ϩ ), PMAϩIono-stimulated WE17/10 cells (100% TCR⅐CD3 lo ), and receptor negative HIV-1-infected WE17/10 cells (TCR⅐CD3 Ϫ ) were analyzed in parallel. At least four bands (Fig. 4, A-D), representing DNA⅐protein complexes with different electro-  5), an oligonucleotide containing the NFAT consensus sequence in the IL-2 promoter (lanes 6 and 7), an oligonucleotide containing the IL-2 promoter NFAT consensus sequence mutated to abrogate NFAT binding (76) (lanes 8 and 9), and an oligonucleotide containing the NFAT ␥1 sequence mutated from GGAA to CCTT (lanes 10 and 11). Bands A-D indicate the four different protein⅐DNA complexes that specifically bind to the NFAT ␥1 probe. B, binding of proteins from the same nuclear extracts shown in A to a 32 P-labeled Oct-1 probe in an EMSA performed as a control (lanes 1-3).
phoretic mobility levels bind to the NFAT ␥1 probe. Nuclear extracts from uninfected, unstimulated cells contain only nominal amounts of the lower molecular weight bands C and D (lane 1). Stimulation for 18 h with PMAϩIono (lane 2) both down-regulated TCR⅐CD3 surface complexes (Fig. 2) and induced binding of B and C and to a lesser extent A (but no D) to the NFAT ␥1 probe. A similar binding profile was observed for the TCR⅐CD3 Ϫ HIV-1-infected cells (Fig. 4, lane 3). The differential binding observed between nuclear extracts from uninfected/unstimulated TCR⅐CD3 ϩ cells and TCR⅐CD3 lo PMAϩIono-stimulated cells or TCR⅐CD3 Ϫ HIV-1-infected cells was reproducible among different preparations of nuclear extracts and specific, because binding of the constitutively expressed Oct-1 transcription factor to its consensus sequence did not vary (Fig. 4B, lanes 1-3).
The specificity of the complexes bound to the NFAT ␥1 probe was further investigated by competition experiments using the homologous oligonucleotide (NFAT ␥1 ; lanes 4 and 5), an oligonucleotide containing the NFAT consensus sequence in the human IL-2 promoter (74, 75) (NFAT-IL-2 wt ; lanes 6 and 7) or versions of NFAT-IL-2 wt and NFAT ␥1 mutated to abrogate binding (76) (GGAA 3 CCTT; NFAT-IL-2 mut , lanes 8 and 9; NFAT ␥1mut , lanes 10 and 11). The homologous and the NFAT IL-2 wt probes efficiently compete for binding, whereas the NFAT IL-2 mut and the NFAT ␥1mut probes were unable to compete. Furthermore, oligonucleotides containing the HIV-1 LTR NF-B consensus sequence, either wild type (Fig. 3B) or mutated (72) (GGG 3 CTC, known to abrogate NF-B but not NFAT binding), both efficiently compete for binding (data not shown). These experiments indicate that the nuclear protein complexes binding to the NFAT ␥1 probe in PMAϩIono-induced and HIV-1-infected cells are specific for the NFAT but not the NF-B consensus sequence.
The Nuclear Protein Complexes Bound to NFAT ␥1 Contain NFATc1, NFATc2, and NF-B p50 -Identification of some of the proteins present in the complexes bound to the NFAT ␥1 probe was achieved using antibodies to the NFAT family members, NFATc1 and NFATc2, the NF-B family members, p50, p65, c-Rel, Rel B, and p52, and the AP-1 family members, c-Fos and c-Jun, with nuclear extracts from TCR⅐CD3 Ϫ HIV-1-infected cells in a supershift assay (Fig. 5A) anti-NFATc1 or the anti-p50 antibody, although the electrophoretic mobility of the anti-p50-supershifted complex (upper A1) was slower than the anti-NFATc1-supershifted complex (lower A1). The B and C complexes were both supershifted to a similar electrophoretic mobility with the anti-NFATc2 antibody only (B1ϩC1). The C and D complexes expressed at low levels in unstimulated, uninfected WE17/10 cells could be supershifted entirely and exclusively with the anti-NFATc2 antibody indicating that these lower molecular weight complexes contain NFATc2 but not NFATc1 or NF-B p50 (data not shown). These data suggested that there were at least three different nuclear complexes bound to the NFAT ␥1 probe in activated or infected cells, one containing NFATc1 and NF-B p50 (band A; present at lower concentrations) and the other two containing NFATc2 (bands B and C; present at higher concentrations; the low molecular weight NFATc2 containing band D was detected only in the unstimulated, uninfected cells).
Confirmation of this observation was achieved by developing a modified supershift assay where combinations of the anti-NFATc1, anti-NFATc2, and anti-p50 antibodies were added sequentially to the binding reaction. These experiments lead to two distinct results: 1) a double-supershift where each antibody binds to a separate complex and individually supershifts the band(s) and 2) a super-supershift where the two antibodies bind to the same complex and their synergy further increases its molecular weight thereby reducing its electrophoretic mobility. Combining the anti-NFATc1 and anti-NFATc2 antibodies (Fig. 5A, lanes 7 and 10) or the anti-NFATc2 and anti-p50 antibodies (lanes 6 and 9) in either order produced doublesupershifts where the A, B, and C complexes were all supershifted (A1 and B1ϩC1), migrating with the same electrophoretic mobility as with the individual antibody alone (lanes 2-4). Alternatively, both combinations of anti-NFATc1 ϩ anti-p50 (lanes 5 and 8) produced a super-supershift where the electrophoretic mobility of the A complex (A11) was consistently further retarded compared with the anti-p50 antibody alone (upper A1, lane 4). A second A band, migrating with the same electrophoretic mobility as with the anti-NFATc1 antibody alone (lower A1, lane 2), was also detected when the anti-p50 and anti-NFATc1 antibodies were used together. This suggests that although some of the A complexes contain NFATc1 and NF-B p50 others contain NFATc1 alone. Alternatively, NF-B p50 could be present but inaccessible to the antibody in some of the A complexes. Thus, as many as four different complexes present in nuclear extracts from activated or infected cells bind to the NFAT ␥1 probe, including: two abundant complexes that contain NFATc2 (bands B and C) but not NFATc1 or NF-B p50 and two low concentration complexes (band A) devoid of NFATc2, one which contains NFATc1 and NF-B p50 and the other either NFATc1 alone or NFATc1 and an inaccessible NF-B p50.
Uninfected TCR⅐CD3 ϩ and HIV-1-infected TCR⅐CD3 Ϫ cells were treated with CsA and then stimulated with PMAϩIono to achieve the maximum potential induction of nuclear NFAT in the presence of CsA. In all cases, there was a Ͼ90% inhibition of nuclear protein binding to the NFAT ␥1 probe in EMSA binding studies (data not shown), which is in agreement with the ability of CsA to block T cell activation (77). These extracts were also used in a supershift assay with the NFAT ␥1 probe and anti-NFATc1, anti-NFATc2, and anti-NF-B p50 antibodies (Fig. 5B). Binding of the NFATc1 and NF-B p50 containing A complex was totally inhibited by CsA treatment (overexposure of the gels did not detect the A complex either in the presence or absence of the anti-NFATc1 and anti-p50 antibodies). The NFATc2-containing B and C complexes were both largely inhibited by CsA, and although a faint B complex could be detected in longer exposures, the normally weaker C complex was readily detectable in lower exposures of the gels (Fig.  5B). This shift in the relative abundance of these two complexes after treatment with CsA suggests that the higher molecular weight B complex is more sensitive to CsA than the lower molecular weight complex and may thus contain a second CsAsensitive component.
The Quantity of Nuclear NFATc1, NFATc2, and NF-B p50 Is Negatively Correlated with TCR⅐CD3 Surface Expression in HIV-1-infected Cells-The relationship between the presence of NFATc1, NFATc2, and/or NF-B p50 in the nucleus and the concentration of CD3␥ gene transcripts was assessed by examining differential binding to the NFAT ␥1 probe of nuclear extracts during the progression of HIV-1-infected cells from TCR⅐CD3 hi 3 TCR⅐CD3 lo 3 TCR⅐CD3 Ϫ (Fig. 6A). Characteristically, only low levels of the NFATc2-containing complexes (bands C and D) were detectable in the uninfected and unstimulated 100% TCR⅐CD3 ϩ cells (lane 1). Alternatively, increased binding of the NFATc1/NF-B p50-containing complex (band A) and NFATc2-containing complexes (bands B and C) to NFAT ␥1 occurs in parallel with a decrease in surface TCR⅐CD3 expression from 98% (lane 2) to 87% (lane 3) to 39% (lane 4) to 0% (lane 5) of normal receptor levels. A nonspecific band (indicated as NS) was also detectable in these nuclear extracts, but this band could neither be supershifted with the anti-NFATc1, anti-NFATc2, or anti-p50 antibodies nor could it be competed for with the homologous oligonucleotide (data not shown). This escalation in binding to the NFAT ␥1 probe is specific, because similar amounts of the constitutively expressed Oct-1 protein from each extract bound to an Oct-1 sequence-specific probe (Fig. 6B). These results suggest that a correlation exists between the quantity of NFATc2, and to a lesser extent, NFATc1 and NF-B p50, in the nucleus and down-modulation of CD3␥ transcripts and TCR⅐CD3 complexes after HIV-1 infection.
Differential Binding of NFATc1, NFATc2, and NF-B p50 to the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 Motifs-We next asked whether members of the NFAT and/or NF-B protein families could also bind to the NFAT ␥2 and/or the NFAT ␥3 motifs. EMSA experiments using the NFAT ␥2 probe (Ϫ392 to Ϫ372, underlined in Fig. 3) and extracts from unstimulated cells, PMAϩIono-stimulated cells, and TCR⅐CD3 Ϫ HIV-1-infected cells bound in a similar pattern to NFAT ␥1 (Fig. 4) except that only the NFATc2-containing complexes (bands B, C and D) but not the NFATc1/NF-B p50-containing complex (band A) were bound (data not shown). Alternatively, binding to the NFAT ␥3 probe (ϩ447 to ϩ466, underlined in Fig. 3) was identical to NFAT ␥1 with all four of the complexes bound (A-D; data not shown). An experiment using the NFAT ␥2 and NFAT ␥3 probes in competition with the homologous or the NFAT ␥1 and NFAT-IL-2 wild type and mutated probes revealed that the binding of bands A-D to these three sequences was highly specific (data not shown). The differential binding of NFATc1, NFATc2, and NF-B p50 to the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 sequences was confirmed in a supershift assay using antibodies to the NFAT, AP-1, and NF-B family members and nuclear extracts from TCR⅐CD3Ϫ HIV-1-infected cells (Fig. 7). Only the anti-NFATc2 antibody specifically shifted the complex bound to the NFAT ␥2 probe (Fig. 7A, lane 2, bands B and C) (band A, lane 4). We compared the relative binding of the NFATc1 plus NF-B p50-and NFATc2-containing complexes to the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 motifs (Fig. 7C) and found that NFAT ␥1 binds significantly more of these protein complexes compared with NFAT ␥2 and NFAT ␥3 , with binding to the NFAT ␥3 probe the weakest among the three motifs. Furthermore, there did not appear to be cooperative recruitment of c-Jun and c-Fos in any of the complexes bound to NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 .
Sequence Variation Is Responsible for the Differential Binding of NFATc1, NFATc2, and NF-B p50 to the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 Motifs-In an effort to understand the basis for the qualitative and quantitative differences in binding to the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 motifs, a series of mutant probes were constructed and used in EMSA experiments (the mutations are listed with a summary of the results in Table I, and the gels are shown in Fig. 8). We noted that the nucleotides bordering the core 5Ј-GGAAA-3Ј sequence differed by an AA immediately following the core sequence in NFAT ␥1 and NFAT ␥3 in contrast to a GC in NFAT ␥2 , suggesting that these  4) antibodies. C, the relative quantity of proteins bound to the 32 P-labeled NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 probes was examined using nuclear extracts from uninfected 100% TCR⅐CD3 ϩ (lanes 1, 3, and 5) and TCR⅐CD3 Ϫ HIV-1-infected WE17/10 cells (lanes 2, 4, and 6).
nucleotides could potentially play a role in the binding of NFATc1 and NF-B p50. Alternatively, a T rather than an A preceding the core sequence is thought to facilitate stronger binding of NFAT family proteins (56), and this nucleotide was C, T, or A in the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 sequences, respectively. Our rationale was that if these three nucleotides do play an important role in binding, then successively mutating the NFAT ␥1 sequence to look like the NFAT ␥2 sequence and vise versa should alter binding accordingly.
Mutation of the first A following the core sequence in NFAT ␥1 to a G (NFAT ␥1mut1 ; Fig. 8A, lane 2) completely abrogated binding of NFATc1 and NF-B p50 (band A), significantly decreased the binding of NFATc2 (bands B and C) compared with the wild type sequence (NFAT ␥1wt , lane 1) and provided a pattern similar to that of wild type NFAT ␥2 (NFAT ␥2wt , lane 6). Additionally mutating the second A to a C in NFAT ␥1 (NFAT ␥1mut2 , lane 3) changed the 3Ј sequence to that of NFAT ␥2 and reduced NFATc2 binding even further. Mutation of the outside A only in the AA pair of NFAT ␥1 (NFAT ␥1mut3 , lane 4) had a less dramatic effect on the quantity of NFATc2 bound compared with the inside A (lane 2) and did not abrogate binding of NFATc1 and NF-B p50, although quantitatively all of the complexes were significantly reduced. Mutation of the C preceding the core sequence in NFAT ␥1 (NFAT ␥1mut4 , lane 5) to a T, creating the sequence 5Ј-TG-GAAAAA-3Ј, greatly enhanced the amount of NFATc1, NFATc2, and NF-B p50 bound to this probe, providing better binding than that observed with any of the wild type sequences.
Alternatively, the reverse mutations in NFAT ␥2 converted the binding profile of this probe to one similar to NFAT ␥1 with increased binding of NFATc2 (bands B and C) and de novo binding of NFATc1 and NF-B p50 (band A) achieved by simply changing the 3Ј G (NFAT ␥2wt , lane 6) to an A (NFAT ␥2mut1 , lane 7). Adding a second A 3Ј of the core sequence in NFAT ␥2 (NFAT ␥2mut2 , lane 8) further increased the binding of all three complexes (A, B, and C). However, substituting the C for an A in the outside 3Ј position did not confer binding of NFATc1 and NF-B p50, although it did increase the binding of NFATc2 (NFAT ␥2mut3 , lane 9). Finally, mutation of the T preceding the core sequence to a C, creating the 5Ј-CGGAAAGC-3Ј, completely abrogated all binding (NFAT ␥2mut4 , lane 10). Confirmation that the specific binding of NFATc1 and NF-B p50 was conferred by adding a fourth A to the NFAT core sequence (5Ј-GGAAAA-3Ј) was demonstrated by a supershift assay using the NFAT ␥2mut2 probe (Fig. 8B). This experiment clearly shows that a simple G 3 A substitution 3Ј of the core sequence in NFAT ␥2 is sufficient to confer binding of NFATc1 and NF-B p50. Finally, binding to the wild type NFAT ␥3 sequence is normally weak, and mutation of the A following the core se-quence to G completely abrogated binding (NFAT ␥3mut1 , Fig.  8C, lane 2) compared with the wild type (NFAT ␥3wt , lane 1).
Taken altogether, these mutation experiments demonstrate that a fourth A added to the NFAT core sequence (5Ј-GGAAAA-3Ј) is vital for NFATc1 and NF-B p50 binding and important for the quantity of NFATc2 that binds. They further illustrate the important role that the T preceding the NFAT core sequence (5Ј-TGGAAAA-3Ј) plays in the quantity or stability of the bound complexes, including both those containing NFATc2 and those containing NFATc1 alone or in association with NF-B p50. DISCUSSION We have previously demonstrated that T cell receptor downmodulation, due to a defect in CD3␥ gene transcription (15,20), occurs in a two-phase progression after HIV-1 or HIV-2 infection and can be summarized by the formula TCR⅐CD3 hi u TCR⅐CD3 lo u TCR⅐CD3 Ϫ in which the forward progression is markedly favored (16). The TCR⅐CD3 hi to TCR⅐CD3 lo phase is characterized by a steady decrease in receptor density on all cells from 100% to 50% of control values, prior to the subsequent conversion of individual cells to the TCR⅐CD3 Ϫ phenotype (16). The RT-PCR data presented in this study provide further insight into the molecular events generating this progression by showing that the initial conversion from TCR⅐CD3 hi to TCR⅐CD3 lo involves a substantial (80 -90%) decrease in the number of CD3␥ gene transcripts.
These data answer a fundamental question of why the progression, viewed from the cell surface, appears to be very slow by showing that transcriptional down-modulation is actually initiated very early (or most likely immediately) after infection with a considerable and rapid erosion of transcripts until a threshold is reached where the normal number of complete TCR⅐CD3 complexes can no longer be assembled and exported to the cell surface (78). The individual TCR⅐CD3 proteins have been shown to be synthesized in great excess, followed by rapid degradation if they are not stabilized through incorporation into partial or complete complexes (79). The CD3␥ protein forms a stable complex with CD3⑀ (80) and thus can persist both in complete TCR⅐CD3 complexes, which are continuously recycled to the cell surface in the absence of antigen stimulation, as well as in partially formed complexes in the endoplasmic reticulum. Thus, recycling and partial complex formation precludes an immediate and deleterious effect on surface receptor expression during the initial stages of CD3␥ transcript loss.
Our earlier studies examining TCR⅐CD3 expression over time post-infection found a minor modulation of receptor density immediately following the acute phase of infection (first 4 -6 weeks) (15,16,20). These studies also revealed that an initial 4-to 5-fold drop in p24 antigen levels in the culture supernatant occurred coincident with down-modulation from TCR⅐CD3 hi 3 TCR⅐CD3 lo , with a further 4-to 5-fold reduction accompanying the transition from TCR⅐CD3 lo 3 TCR⅐CD3 Ϫ (16). However, a subsequent extensive examination of productively infected cells did not reveal a direct relationship between intracellular p24 antigen levels and TCR⅐CD3 surface density (26). Furthermore, non-productively infected cells expressing the multiply spliced, virally encoded tat, nef, and rev regulatory gene transcripts also demonstrated the same progressive loss of surface TCR⅐CD3 complexes (26). Treatment of productively infected cells with antisense oligonucleotides targeted to tat, nef, and rev revealed that the relative level of tat and nef gene transcripts could be directly correlated with a loss of CD3␥ transcripts (26). Antisense oligonucleotides directed to the splice acceptor of the tat gene were particularly efficient in provoking a coordinate down-regulation of virus expression in concert with an up-regulation of surface TCR⅐CD3 complexes (26). One interpre- Differential NFAT Binding in the Human CD3␥ Chain Gene tation of our previous data in light of the RT-PCR results presented here is that Tat-dependent viral gene expression and the availability of Tat and/or Tat-dependent cellular transcription factors (81) plays an important role in initiating and maintaining the escalating CD3␥ transcription defect.
HIV-1 is known to activate its CD4 ϩ T cell host and trigger the expression of a variety of antigen-induced immune response genes as a means of facilitating virus integration, replication, and expression (82,83). CD3␥ plays an important role in both tyrosine-and PKC-mediated TCR⅐CD3 down-modulation, and it seems likely that HIV-1 could exert its effect on receptor expression via these normal immune pathways. We asked whether it was possible to restore CD3␥ transcription in HIV-1-infected cells by activating or inhibiting steps in the TCR⅐CD3-directed activation pathway and found that the immunosuppressive drug cyclosporin A could partially restore TCR⅐CD3 surface expression on infected cells. CsA inhibits the calcium-regulated phosphatase calcineurin, which dephosphorylates NFAT family proteins in response to antigen activation. Dephosphorylation of NFAT proteins is a prerequisite for their translocation to the nucleus, where they function as major players in the transcriptional activation of a wide array of cytokine genes possessing NFAT binding motifs (5Ј-GGAAA-3Ј) (56). Four NFAT sites are located in the HIV-1 LTR, one within each of the two NF-B consensus sequences, and an additional two in the negative regulatory element (52). The HIV-1 B sequences have been shown to play an important role in the transcriptional regulation of viral gene expression (84,85) and to competitively bind NF-B and NFAT family proteins (55,58,83). The up-regulation of TCR⅐CD3 surface expression observed on CsA-treated HIV-1-infected cells suggested that NFAT might also be directly or indirectly involved in the elusive transcriptional control mechanisms that regulate expression of the CD3␥ gene.
A search of the 5Ј-upstream region and exon 1 of the human CD3␥ gene revealed three potential binding motifs for NFAT family proteins (NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 ). The NFAT ␥1 motif is located in a DNase I-hypersensitive site that has been designated as the putative promoter for CD3␥ (71,86,87), whereas the NFAT ␥2 motif is nested in a region with sequence homology to the HIV-1 B elements. Three different molecular weight complexes (A, B, and C) could be induced by PMAϩIono or HIV-1 infection to specifically but differentially bind to these motifs in the CD3␥ gene. NFATc2 was shown to be present in both the B and C complexes, as well as in the low abundance D complex found in unstimulated cells. The different electrophoretic mobilities of the three complexes could be correlated with the binding of NFATc2 as a monomer or dimer (56,88) and/or the presence of other currently unidentified factors, potentially including an additional CsA-sensitive protein in the B complex. The B complex might be the active complex, with the C complex an intermediate stage in assembly and the D complex representing the low level of NFATc2 known to be present in the nucleus of resting T cells (89). Alternatively, the C complex could be a positive transcription complex and the additional protein(s) bound in the B complex could provide a negative signal.
The highest molecular weight A complex was found to contain NFATc1 and NF-B p50 (but not NFATc2). To determine whether NFATc1 and NF-B p50 were present in the same protein⅐DNA complex, we designed a modified supershift assay whose purpose was to reduce the molecular mobility of one or more complexes containing both proteins by the sequential addition of the two different antibodies (referred to as a supersupershift assay). This experiment demonstrated that some of the A complexes contain both NFATc1 and NF-B p50, whereas others contain either NFATc1 alone or an inaccessible NF-B p50. The relatively small impact on the molecular mobility afforded by the additional binding of the anti-NFATc1 antibody in the super-supershift over the band in the anti-p50 antibody simple supershift can be explained by the nature of these antibodies. The anti-NFATc1 used was a mouse monoclonal antibody, whereas the anti-p50 employed was a goat polyclonal antibody. Therefore, the single isotype of the anti-NFATc1 antibody directed to only one epitope of this protein in combination with the repertoire of anti-p50 antibody molecules po- FIG. 8. Mutation analysis of the NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 probes. A, EMSAs were performed using nuclear extracts from TCR⅐CD3 Ϫ HIV-1-infected WE17/10 cells and the 32 P-labeled NFAT ␥1 and NFAT ␥2 probes mutated as shown in Table I ( lanes 1-10). B, binding to the 32 P-labeled NFAT ␥2mut1 probe was examined in a supershift assay using nuclear extracts from TCR⅐CD3 Ϫ HIV-1-infected WE17/10 cells and anti-NFATc1 tentially bound to NF-B p50 contributed relatively little additional weight to this already extremely high molecular mass protein⅐DNA complex, thereby slightly but consistently decreasing its electrophoretic mobility.
The super-supershift approach was designed to demonstrate the dual binding of NFATc1 and NF-B p50 in a single complex, because both NFAT and NF-B family proteins are translocated to the nucleus after PMAϩIono stimulation or HIV-1 infection where the preferential and most abundant binding partner for p50 would be another NF-B family member such as p65 (supershifts using a NF-B consensus sequence probe detected abundant amounts of NF-B p50 and p65 in these nuclear extracts, data not shown). In light of the relatively low levels of the NFATc1⅐NF-B p50 complex present, we thought it was important to provide the NFAT ␥1 DNA binding site in the reaction mixture to favor their coordinate binding. Further evidence in support of the dual binding of NFATc1 and NF-B p50 to NFAT ␥1 and NFAT ␥3 but not NFAT ␥2 was provided by the EMSA experiments using mutant oligonucleotides. Changing the fourth A in the NFAT ␥1 and NFAT ␥3 motifs (5Ј-GGAAAA-3Ј to 5Ј-GGAAAG-3Ј) completely abrogated binding of the NFATc1-and NF-B p50-containing complex, whereas adding a fourth A to the NFAT ␥2 motif (5Ј-GGAAAG-3Ј to 5Ј-GGAAAA-3Ј) conferred binding to this sequence. It seems unlikely that simply altering a single nucleotide would have such a dramatic effect on the concurrent binding of NFATc1 and NF-B p50 binding unless they were present in the same complex.
These data are the first demonstration of a NFAT family member and a NF-B family member binding together in the same protein⅐DNA complex. NFAT and NF-B normally compete for binding to the B site, and this has been demonstrated to be true for the HIV-1 LTR B sites (58). The NF-B/Rel family of transcription factors are defined by a ϳ300-amino acid region called the Rel homology domain, which contains the residues involved in nuclear translocation, DNA binding, and protein-protein interactions (53,90). NF-B p50 and p65 preferentially form a heterodimer, although they are also capable of forming p50/p50 or p65/p65 homodimers. The formation of homo-and heterodimers leading to dimerization is known to be required for binding of the NF-B family proteins to DNA (91). Crystal structures have shown that NF-B p50 optimally binds to the 5Ј-GGAAA-3Ј half site and p65 the 5Ј-GGAA-3Ј half site, which are separated by a non-contact base in the palindromic B sequence (92). Although not all of the known physiological targets have this 10-bp B consensus sequence, NF-B proteins are still capable of binding to these nonideal sequences with similar affinities (56).
A Rel homology domain, with about 20% sequence homology to the NF-B Rel domain, is also found in all of the NFAT proteins (93,94). Structural studies have shown that the minimal DNA binding domain of NFATc1 is essentially identical to the Nterminal specificity domain of NF-B p50, the region involved in the majority of its base specific contacts with DNA (93,95,96). NFAT proteins normally bind as monomers in cooperation with other transcription factors such as AP-1. However, they have also been shown to bind as dimers to certain NF-B/Rel sites (56), and the HIV-1 LTR B sites are an example of NFATc2 forming both monomeric and dimeric complexes (55,58,97). Other common features between the NFAT and NF-B proteins include their responsiveness to immune activation and their regulation by cytoplasmic to nuclear translocation.
The NFAT ␥1 and NFAT ␥3 probes do not contain a palindromic purine-rich sequence similar to those found in the HIV-1 B elements, which if present could potentially explain the dual binding of NFATc1 and NF-B p50. Furthermore, the supershift assay performed on the CsA-treated cells revealed that NF-B p50 does not bind to the NFAT ␥1 motif in the absence of NFATc1, suggesting that NF-B p50 binding is completely dependent upon the presence of NFATc1. It was quite intriguing to discover that proteins from these two different transcription factor families bind together to DNA sequences whose only common component is the presence of an extended NFAT binding motif where the fourth adenosine (5Ј-GGAAAAA-3Ј) was found to be crucial for their binding. This core motif is also the only component common between the NFAT ␥1 and NFAT ␥3 but not the NFAT ␥2 probes and thus emerges as the requisite sequence for binding of the NFATc1⅐NF-B p50 complex. Sites in which the 5Ј-GGAAA-3Ј core sequence is preceded by a T rather than an A bind NFAT proteins more strongly (56), and although this was found to be true for NFAT ␥1 by replacing the preceding C with a T, the low level and lack of NFATc1 and NF-B p50 binding to NFAT ␥2 (5Ј-TGGAAAG-3Ј) suggests that the fourth A plays the greatest role in qualitative binding.
The dimerization relationships between the different NF-B proteins and the combinatorial binding associated with the NFAT family proteins allows a relatively small number of transcription factors to establish an extraordinarily complex and extensive regulatory network with different biological consequences dependent upon selective binding controlled by the flanking sequences. This may be just one more example of how the NFAT family proteins gain specificity and regulatory function through their coordinate binding with other transcription factors. NF-B p50 could potentially partner with NFATc1 to provide the binding stability it needs and normally acquires through coordinate binding with other transcription factors such as AP-1. The flexibility of binding with different partner proteins may be fundamental to the ability of NFAT proteins to integrate distinct signals through cooperative binding with specific nuclear partners on divergent consensus sequences in diverse genes and different chromatin structures.
In this study, we have shown that a loss of CD3␥ gene transcripts is initiated early after HIV-1 infection and rapidly accumulates to a defect of Ͼ90% of normal transcript numbers, leading to a down-modulation of surface TCR⅐CD3 expression and function. We identified three NFAT binding motifs (NFAT ␥1 , NFAT ␥2 , and NFAT ␥3 ) in the upstream region of the CD3␥ gene and have shown that they differentially bind complexes containing NFATc2, NFATc1, and NF-B p50. Furthermore, we found that a significant and progressive increase in these protein⅐DNA complexes could be negatively correlated with CD3␥ gene transcript numbers. The NFAT ␥1 site binds the greatest abundance of these transcription factors, which, together with its location in a DNase I-hypersensitive site (86), suggests it may play an active role in CD3␥ gene transcription. Normal activation via the TCR⅐CD3 complex initiates a cascade of molecular events leading to multiple signaling pathways that are integrated to induce the expression of specific cytokine genes. A sustained signal also changes the normal balance in receptor expression, favoring TCR⅐CD3 internalization and degradation rather than recycling and de novo synthesis (5). Although the accumulation of NFAT family proteins in the nucleus has a positive influence on cytokine gene transcription, it also potentially negatively regulates CD3␥ gene transcription as a means of controlling continued TCR⅐CD3 directed signaling. Thus, HIV-1 may have acquired the ability to intercede in both the positive and negative downstream pathways triggered by the TCR⅐CD3 as a means of controlling viral gene expression and latency.