The roles of nuclear factor of activated T cells and ying-yang 1 in activation-induced expression of the interferon-gamma promoter in T cells.

Nuclear factor of activated T cells (NFAT) plays an important role in expression of many cytokine genes including interleukin-2 and interleukin-4. However, its role in interferon-gamma (IFN-gamma) expression is not well understood. In the current studies, two strong NFAT-binding sites in the IFN-gamma promoter were identified by DNase I footprint analysis at positions -280 to -270 and -163 to -155. NFATp bound independently to both sites and was required for the formation of a composite element with AP-1 spanning position -163 to -147. In Jurkat T cells and primary lymphocytes, activation-induced expression of IFN-gamma reporter constructs containing point mutations in either NFAT site or the AP-1 component of the composite site was decreased by approximately 40-65%. Despite elimination of both strong NFAT-binding sites, the IFN-gamma promoter remained completely sensitive to inhibition by cyclosporin. This suggests that other elements in the IFN-gamma promoter, such as the IFN-gamma proximal element, are sufficient for cyclosporin sensitivity of this gene. Ying-Yang 1 (YY1), a potential inhibitor of IFN-gamma expression, binds to sites located between the two NFAT sites. Mutation of the YY1 sites alone had little effect on IFN-gamma promoter activity. However, mutation of both the NFAT and YY1-binding sites abolished activation-induced expression in primary murine splenocytes but not in Jurkat T cells. This suggests that under some conditions, YY1 may play a positive role in activation-induced transcription of IFN-gamma.

Originally described as an essential transcription factor for IL-2 1 gene expression in T cells (1)(2)(3), nuclear factor of activated T cells (NFAT) is thought to play a major role in coordinating transcription of a number of cytokine genes, which are sensitive to the inhibitory effects of cyclosporin (4). In addition to binding to the IL-2 promoter, NFAT binds to sites in the regulatory regions of a number of other cytokine genes, including TNF-␣ (5, 6), IL-3, IL-4 (7-9), IL-5 (10,11), and GM-CSF (12)(13)(14). In T cells, NFAT sites commonly bind two components, a pre-existing cytoplasmic component designated NFATp or NFATc (6) and a nuclear component comprised of members of the Fos and Jun (AP-1) protein families (15)(16)(17). Recent studies have demonstrated that the cytoplasmic NFAT components belong to a large family of regulatory transcription factors comprised of at least four members, NFATp (NFAT1), NFATc (NFAT2), NFAT3, and NFAT4 (NFATx), which are differentially expressed in lymphoid and nonlymphoid cells (18 -22). In T cells, NFATp is expressed in both unstimulated and stimulated cells, whereas NFATc is expressed primarily in activated cells (18,19,23,24).
The interferon-␥ (IFN-␥) gene is cyclosporin-sensitive, but the mechanism for cyclosporin sensitivity of this gene has not been fully elucidated. Previous studies in our laboratory have demonstrated that the region between Ϫ108 and Ϫ40 base pairs (bp) upstream of the transcription start site in the IFN-␥ promoter is able to confer activation-specific, cyclosporin-sensitive expression in T cells (25). Within this region, one target of the inhibitory effect of cyclosporin is the IFN-␥ proximal element (Ϫ73 to Ϫ48 bp), which binds multiple transcription factors but not NFAT (26). Further upstream the IFN-␥ promoter contains several sequences that are homologous to NFAT-binding sites in the IL-4 gene (27). NFATp has been shown to bind to one of these sites, and it was proposed that calcineurin-inducible transcriptional factors act at this site to enhance IFN-␥ expression (28). The role of NFAT in the function of the IFN-␥ promoter was more fully explored in the current studies. NFAT bound to additional sites in the IFN-␥ promoter, and the two strong binding sites were required for maximal expression of an IFN-␥ reporter construct containing 538 bp of the IFN-␥ promoter. However, these two NFATbinding sites were not required for cyclosporin sensitivity. In contrast to results seen with the IL-2 and IL-4 promoters, transient cotransfection studies suggested that NFAT was not a limiting factor for IFN-␥ gene expression in Jurkat T cells. In primary murine splenocytes, which have lower levels of endogenous NFAT protein, overexpression of NFATp or NFATc enhanced expression of the IFN-␥ promoter.
YY1 is a multifunctional DNA-binding protein, which can activate, repress, or initiate transcription depending on the context in which its binds (29). YY1 is constitutively expressed and is known to interact with other proteins involved in transcriptional regulation such as c-Myc (30), Sp-1 (31,32), ATF/ CREB (33), TATA-binding protein, TFIIB (34,35), and NF-B (36). Previous studies suggested that YY1 inhibits basal IFN-␥ expression through two mechanisms: binding to an AP-2-like repressor protein, and competition with AP-1 for DNA binding (37,38). In this work, the YY1-binding sites, which are located between the two strong NFAT-binding sites in the IFN-␥ promoter, did not mediate inhibition of basal expression of IFN-␥. Furthermore, in contrast to its putative role as an inhibitor of IFN-␥ expression, YY1 appeared to play a positive role in initiating IFN-␥ gene expression in primary murine splenocytes.

EXPERIMENTAL PROCEDURES
Cell Lines and Reagents-A line of Jurkat T cells, which expresses IFN-␥ following activation, was maintained as described previously (25). Splenocytes obtained from B10.BR or C57Bl/6 mice, which were originally purchased from The Jackson Laboratories (Bar Harbor, ME) and bred in our specific pathogen-free facility, were depleted of erythrocytes with an ammonium chloride lysis solution (0.15 M NH 4 Cl, 1 mM NaHCO 3 , 0.1 mM EDTA). The remaining mononuclear cells were cultured overnight in RPMI 1640 supplemented with 10% fetal calf serum (HyClone Laboratories, Logan, UT), GlutaMAX, 100 units/ml penicillin, 100 g/ml streptomycin, and 20 g/ml gentamicin plus 3 g/ml concanavalin A (ConA) prior to electroporation.
Plasmid Constructs-␤-Galactosidase reporter constructs containing various amount of 5Ј-flanking sequence of the IFN-␥ promoter (basal promoter pIFN-39 and full-length promoter pIFN-538), a dimer of the IFN-␥ proximal element (Ϫ71 to Ϫ44) placed upstream of the pIFN-39 basal promoter (Prox dimer), and the IL-2 promoter (pIL2-568) have been previously described (25,26). Point mutations in the NFAT-, AP-1-, and YY1-binding sites were made in the context of the full-length IFN-␥ promoter (pIFN-538) by polymerase chain reaction-site directed mutagenesis. The NFAT-binding site mutations contained a GG to TT substitution, which disrupts NFAT binding. The AP-1-binding site mutation (a CT to AA substitution at Ϫ151, Ϫ150) disrupts AP-1 binding but not NFAT binding to the NFAT Ϫ160 site. For the NFAT Ϫ160 mutation (N160) and the AP-1 Ϫ160 mutation (AP160), a 280-bp BglII/ Kpn fragment containing the substitution was subcloned into pIFN-538⌬Ϫ177/Ϫ109. A 360-bp Xba/BglII fragment with the NFAT Ϫ280 substitution was subcloned into pIFN-538⌬Ϫ214/Ϫ178 for the NFAT Ϫ280 mutation (N280), and into the N160 construct for the double NFAT mutation (N160/N280). Point mutations in the Y1 site alone (a T to A substitution at Ϫ201) or in this site and the Y2 site (a GG to CA substitution at Ϫ218, Ϫ217) are known to disrupt YY1 binding at these sites (38). The YY1 mutations were then inserted into the wild-type pIFN-538 (Y1 or Y1Y2) or double NFAT mutant Ϫ538 construct (NY1 or NY1Y2). The sequence of each construct was confirmed by dye terminator cycle sequencing (Perkin-Elmer). All of the IFN-␥ promoter constructs mentioned above were subcloned into a firefly luciferase reporter plasmid, pGL3, from Promega (Madison, WI).
Transient Transfection Assays-Jurkat T cells were transiently transfected with 1 g/10 6 cells of an IFN-␥ ␤-galactosidase plasmid and 0.54 g/10 6 cells of control ␤-actin chloramphenicol acetyltransferase plasmid in 0.5 ml of RPMI 1640 by electroporation (ECM600, BTX Inc., San Diego, CA) as described previously (25,26). In cotransfection experiments, the total amount of plasmid per cuvette was kept constant by addition of control pREP4 plasmid. After electroporation, the cells were either not stimulated or stimulated with ionomycin (1.5 M) and phorbol myristate acetate (PMA, 25 ng/ml), with or without pretreatment with cyclosporin A (CSA, 500 ng/ml) for 20 min as described previously (25,26). Cell lysates were analyzed for ␤-galactosidase activity using chlorophenol red as substrate (40) and corrected for transcription efficacy by normalizing to chloramphenicol acetyltransferase content as described previously (25,26).
In the transfection experiments involving primary lymphoid cells, murine splenocytes were harvested and treated with 3 g/ml ConA 21 h prior to transfection to allow uptake of DNA (41). 5 ϫ 10 6 splenocytes were transiently transfected with 10 g of IFN-␥ firefly luciferase reporter plasmid and 1 g of control plasmid containing the ␤-actin promoter driving Renilla luciferase in 0.25 ml of complete media (RPMI 1640, 10% fetal calf serum, glutamine, penicillin/streptomycin) by electroporation (Bio-Rad, 250 V, 960 microfarads). After electroporation, the cells were rested for 2 h and then stimulated with 25 ng/ml PMA and 1.5 M ionomycin or media alone. In certain experiments, some cells were treated with cyclosporin A (1 g/ml) for 35-45 min prior to stimulation. Lysates were harvested at 5 h and assayed for firefly and Renilla luciferase activity utilizing a Dual Luciferase Reporter kit (Promega, Madison, WI). Values are reported as relative light units after correction for transfection efficiency by normalization to the ␤-actin promoter driving Renilla luciferase. In the cotransfection experiments, the splenocytes were transiently transfected with 10 g of IFN-␥ firefly luciferase reporter and 10 g of the appropriate NFAT expression vector without addition of the ␤-actin promoter Renilla luciferase plasmid. Values are reported as absolute light units without correction for transfection efficiency by normalization to the ␤-actin promoter, since overexpression of NFAT proteins in murine splenocytes increased ␤-actin promoter expression compared with the control pREP4 vector (data not shown).
DNA Binding Assays-The DNA template for the footprinting experiments was made by end labeling a human (Ϫ350 to ϩ50) IFN-␥ promoter fragment at either end. The binding reactions and DNase I digestions were carried out as described previously (8).
Recombinant Nuclear Proteins-The Rel domain of NF-ATp (amino acids 185-537) (18) was expressed by the T7 polymerase expression system and purified as described previously (20). c-Jun protein was expressed in Escherichia coli and purified from the insoluble portion of the extract (42). Fra-1 protein was expressed in E. coli and purified from the soluble fraction to approximately 80% homogeneity on heparin-Sepharose. ATF-2BR (DNA binding domain, amino acids 350 -505), obtained from boiling lysis preparations of bacteria without further purification (ϳ90% pure), was kindly provided by J. Hoeffler (Invitrogen, San Diego, CA) (43).

Identification of NFAT-binding Sites in the IFN-␥ Promot-
er-Previous studies have identified two elements located within 108 bp of the 5Ј-flank of the IFN-␥ gene which are sufficient for T cell and activation-specific expression by the IFN-␥ promoter (25,26,49,50). Although expression mediated by this region of the promoter and by the proximal element (25,26,49) is cyclosporin-sensitive, neither element appears to contain binding sites for NFAT proteins. This suggests that unlike other cyclosporin-sensitive promoters of lymphokines expressed in T cells, cyclosporin sensitivity of the IFN-␥ promoter may not be due to the inhibitory effects of this drug on NFAT activation. However, it is possible that NFAT proteins bind at other sites within the full-length promoter (27,28) and thereby contribute to the cyclosporin sensitivity of this gene.
To address this possibility, DNase I footprint analysis of the IFN-␥ promoter was performed using recombinant NFATp, AP-1 (c-Jun and Fra-1) and ATF-2 proteins (Figs. 1 and 2). AP-1 and ATF-2 proteins were used because they are known to bind to the IFN-␥ promoter (26), and AP-1 can form a composite element with NFAT (4,12,22). Recombinant NFATp bound strongly and independently to two regions protecting nucleotides spanning positions Ϫ280 to Ϫ265 and Ϫ168 to Ϫ153 bp relative to the transcription start site (Figs. 1A, lane 12, and B, lane 3, and 2B, lane 2). The most distal NFAT site, which is referred to as the NFAT Ϫ280 site, contains a consensus NFAT-binding sequence at position Ϫ280 to Ϫ270 and is identical to the recently described C3-3P site that binds both NFAT and NF-kB proteins (28). The second NFAT site contains a consensus NFAT-binding sequence at position Ϫ163 to Ϫ155 and is referred to as the NFAT Ϫ160 site. The ability of NFATp to bind the NFAT Ϫ160 site is a novel finding. AP-1 (c-Jun plus Fra-1) bound independently to two previously defined sites (Figs. 1B, lane 4, and 2B, lane 3) as follows: the distal element at position Ϫ98 to Ϫ72 (25) and a second site at position Ϫ196 to Ϫ183 (38, 51). AP-1 also bound to the NFAT Ϫ160 site but did so only in the presence of NFATp (Fig. 1B, lane 5). NFATp also protected a region spanning position Ϫ106 to Ϫ91 just 5Ј to the distal element but bound weakly and only in the presence of AP-1 (Fig. 1B, lane 5; Fig. 2B, lane 4). NFATp did not bind to more proximal regions of the IFN-␥ promoter, such as the proximal element (Fig. 2, A and B). Consistent with previous data, recombinant ATF-2 protected a large region spanning position Ϫ63 to Ϫ43 of the proximal element alone or in combination with c-Jun ( Fig. 2A). AP-1 protected a smaller region within the proximal element (Fig. 2B).
The IFN-␥ NFAT Ϫ280 and NFAT Ϫ160 Elements Bind NFAT Proteins in T Cell Nuclear Extracts-To determine if NFAT proteins present in T cell nuclear extracts can bind to the sites identified by the DNase I footprint analysis using recombinant proteins, the NFAT Ϫ280 and NFAT Ϫ160 sites were tested in electrophoretic mobility gel shift assays (EM-SAs) with nuclear extracts from Jurkat T cells (Fig. 3). Similar results were obtained using nuclear extracts from a murine Th1 clone, CD6 (data not shown). As shown in Fig. 3A, a low basal level of binding to the NFAT Ϫ280 site was seen in nuclear extracts from unstimulated Jurkat T cells (lane 4). Binding to the NFAT Ϫ280 site was greatly induced by treatment of the cells with ionomycin and PMA and was blocked by cyclosporin A (lanes 5 and 6). When used as competitors, unlabeled NFAT Ϫ280 and NFAT Ϫ160 oligonucleotides inhibited formation of the NFAT Ϫ280 complex (Fig. 3B, lanes 13, 15, and 16). NFAT Ϫ280 and NFAT Ϫ160 oligonucleotides containing a GG to TT mutation (280m and 160m), which disrupts the consensus NFAT-binding sequence, did not block formation of the complex nor did oligonucleotides containing consensus AP-1, YY1, GATA-1, and Oct-1 elements (Fig. 3B, lanes 14, 17, and 18 and data not shown). In EMSA experiments using antisera to specific transcription factors, rabbit polyclonal antiserum against NFATp supershifted the NFAT Ϫ280 complex (Fig. 3C, lane 8), whereas antisera directed against Jun, YY1, and Oct-1 proteins had no effect (Fig. 3C, lane 9, and data not shown). A supershift was also observed with antisera directed against p50, p65, and c-Rel (Fig. 3C, lanes 10 -12) as has been described previously (28), indicating that members of the NF-B family can also bind to the NFAT Ϫ280 site.
There were similarities and differences in factors binding to the NFAT Ϫ160 site compared with the NFAT Ϫ280 site. As seen in Fig. 3A, no binding to the NFAT Ϫ160 probe was seen in unstimulated Jurkat T cells (lane 1). The NFAT Ϫ160 complex was induced upon stimulation with ionomycin and PMA and blocked by cyclosporin (Fig. 3A, lanes 2 and 3). In competition experiments, unlabeled NFAT Ϫ160 and NFAT Ϫ280 oligonucleotides strongly inhibited formation of the NFAT Ϫ160 complex, whereas the mutated NFAT Ϫ160 and NFAT Ϫ280 oligonucleotides (160m and 280m) did not (Fig. 3B, lanes  2-6). The NFAT Ϫ160 complex appeared to have a somewhat lower affinity than the NFAT Ϫ280 site, since the NFAT Ϫ160 complex was completely competed by only 5 ng of the unlabeled NFAT Ϫ280 oligonucleotide (Fig. 3B, lane 4), whereas a higher concentration of unlabeled NFAT Ϫ160 oligonucleotide was required to completely compete the NFAT Ϫ280 complex (Fig.  3B, lanes 15 and 16). In some experiments, an AP-1 consensus oligonucleotide partially competed binding of the NFAT Ϫ160 complex, but this was not a consistent finding (Fig. 3B, lane 7). Consensus YY1, GATA-1, and Oct-1 oligonucleotides failed to compete for binding to the NFAT Ϫ160 complex (data not shown). Antiserum to NFATp supershifted the NFAT-160 complex (Fig. 3C, lane 2), but antisera directed against YY1 and Oct-1 had no effect (data not shown). In contrast to the results with the NFAT Ϫ280 complex, antiserum against Jun proteins (p-Jun) partially blocked the NFAT Ϫ160 complex.
Since the Ϫ161 to Ϫ153 region of the IFN-␥ NFAT Ϫ160 site has sequence homology with the CD28 response element (CD28RE) of the IL-2 promoter, which is a composite NFAT/ AP-1 site (52) and binds members of the NF-B family including c-Rel, p50, and p65 (53), the ability of antisera to NF-B proteins to interact with the NFAT Ϫ160 complex was tested.
In contrast to the results seen with the NFAT Ϫ280 element (Fig. 3C, lanes 10 -12), antisera to NF-B family proteins p50, p65, and c-Rel did not supershift the NFAT Ϫ160 complex (Fig.  3C, lanes 4 -6). This analysis was performed with nuclear extracts from Jurkat T cells that were stimulated with ionomycin and PMA. Previous studies have demonstrated that the CD28RE complex of the IL-2 promoter is only weakly evident with PMA treatment and required both a T cell receptor signal (e.g. ionomycin and PMA or anti-CD3) and an anti-CD28 signal for maximal induction (54 -56). Thus, it remains possible that
Although the GG to TT point mutation in the NFAT Ϫ160 site disrupts the NFAT binding consensus sequence, it should be noted that the NFAT Ϫ160 oligonucleotide containing this mutation (160m) bound a single complex which differed from the complex seen with the wild-type NFAT Ϫ160 probe (Fig.  3B, lane 8 versus lane 1). This new complex was only competed by itself (lane 10) and not by the NFAT Ϫ160 or AP-1 oligonucleotides (lanes 9 and 11). The GG to TT mutation of the NFAT Ϫ160-binding motif generated the sequence GTCTAAAttAAA which is homologous to the Octamer consensus sequence ATG(C/T)AAAT. Binding of Oct-1 to the NFAT Ϫ160 mutant and lack of binding of Oct-1 to the wild-type NFAT Ϫ160 oligonucleotide was confirmed by EMSA using Oct-1-specific antibody (data not shown).

NFAT Is Required for Maximal Expression of IFN-␥ Promoter Constructs but Is Not Necessary for Cyclosporin Sensitivity-
To determine the influence of NFAT on IFN-␥ gene expression and its role in the cyclosporin sensitivity of this gene, specific point mutations (GG to TT) were made in the two strong NFAT-binding sites in the context of the full-length IFN-␥ promoter reporter construct pIFN-538 by polymerase chain reaction site-directed mutagenesis. These mutations disrupted NFAT binding determined by DNase I footprint analysis and by EMSA (Fig. 3B and data not shown). As represented in Fig. 4A, point mutations in the NFAT Ϫ160 site (N160), NFAT Ϫ280 site (N280), or both sites (N160/N280) decreased activation-induced expression by an average of 48, 41, and 49%, respectively, compared with the wild-type pIFN-538 construct tested concomitantly (p Ͻ 0.003 by two-tailed Student's t test, cumulative results of 7 experiments). Inducible expression by these constructs remained completely sensitive to cyclosporin, as was expression by a reporter construct containing a dimer of the IFN-␥ proximal element (26), which does not bind NFATp.
To examine whether NFAT plays a similar role in IFN-␥ expression in primary lymphocytes, constructs containing the IFN-␥ promoter driving a luciferase reporter were transiently transfected into murine splenocytes (Fig. 5) (41). Results paralleled those with Jurkat T cells compared with the wild-type pIFN-538 construct; expression of the N160, N280, and N160/ N280 constructs was reduced by 45, 52, and 66% (p Ͻ 0.03, n ϭ 3), respectively, in response to ionomycin and PMA and by 39, 44, and 61% (p Ͻ 0.04, n ϭ 2), in response to anti-CD3. The N160/N280 construct with point mutations in both NFAT sites remained sensitive to cyclosporin.

AP-1 Is Also Required for Maximal Expression of IFN-␥ Promoter Constructs-
To determine whether the interaction of AP-1 at the composite NFAT/AP-1 Ϫ160 element influences IFN-␥ gene expression, a point mutation (CT to AA) was made in the AP-1 site in this composite element. As determined by DNase I footprint analysis, this mutation disrupted the binding of AP-1 to this element in the presence of NFAT but did not affect NFAT binding (data not shown). Compared with the wild-type pIFN-538 construct, expression of the construct containing the AP-1 Ϫ160 point mutation (AP160) was decreased by 47% (p Ͻ 0.003, n ϭ 6) in Jurkat T cells and by 68% (p Ͻ 0.007, n ϭ 4) in murine splenocytes in response to ionomycin and PMA (Fig. 6).
NFAT Enhances IFN-␥ Expression in Murine Splenocytes but Not Jurkat T Cells-Overexpression of NFATp in Jurkat T cells enhances transcription of reporter genes driven by the IL-2, IL-4, TNF-␣, and GM-CSF promoters (57). To determine whether NFATp plays a similar role in IFN-␥ expression, Jurkat T cells were transfected with a NFATp or NFATc expression vector and reporter plasmids containing various amounts of 5Ј-flanking sequence of the IFN-␥ promoter. As seen in Fig.  7A, overexpression of NFATp enhanced expression of a control IL-2 promoter construct but failed to enhance expression of the full-length IFN-␥ promoter construct (pIFN-538). Cotransfection with expression vectors encoding other members of the NFAT family (NFATc, NFAT3, and NFAT4) also did not enhance IFN-␥ reporter expression (Fig. 7A and data not shown).
Since expression of NFAT proteins can vary in different T

FIG. 4. Effects of cyclosporin and NFAT point mutations on expression of IFN-␥ promoter constructs in Jurkat T cells.
Constructs containing the full-length pIFN-538 promoter without or with point mutations in the NFAT Ϫ160 site (N160), the NFAT Ϫ280 site (N280), both NFAT sites (N160/N280), the first YY1-binding site (Y1), both YY1-binding sites (Y1Y2), combined double NFAT and YY1-binding site mutations (NY1 and NY1Y2), or a dimer of the proximal element (Ϫ71 to Ϫ44) placed upstream of the basal IFN-39 promoter (Prox Dimer) were transiently transfected into Jurkat cells and assayed for ␤-galactosidase activity with normalization to a ␤-actin promoter as described previously (26). A, the Jurkat T cells were either not stimulated (uns, black bars), stimulated with 25 ng/ml PMA and 1.5 M ionomycin (IϩP, gray bars), or pretreated with 500 ng/ml cyclosporin A (IϩP/CSA, white bars) as indicated. B, the cells were either not stimulated or stimulated with PMA and ionomycin. Values are the means Ϯ S.E. of two experiments in A and three experiments in B, in which all constructs shown were tested in parallel.
cell populations, and Jurkat T cells express a higher level of endogenous NFAT proteins compared with peripheral T cells (22,58), murine splenocytes were also cotransfected with the NFAT expression vectors and the IFN-␥ reporter plasmids. In contrast to the results seen with the Jurkat T cells (Fig. 7A), expression of the full-length IFN-␥ promoter construct (pIFN-538) was enhanced 1.5-fold (p Ͻ 0.004 by two-tailed Student's t test, n ϭ 3) by NFATp overexpression and 1.9-fold (p Ͻ 0.02) by NFATc overexpression (Fig. 7B). In comparison, overexpression of NFATp enhanced expression of the IL-2 promoter construct (pIL2-568) by 2.0-fold (p Ͻ 0.006). Overexpression of NFATc did not significantly enhance IL-2 promoter-dependent expression (p ϭ 0.25).
NFAT and YY1 Play a Positive Role in Expression of the IFN-␥ Promoter-Based on the results of the transient trans-fection experiments in Figs. 4 and 5, NFAT appeared to play a positive role in expression of the IFN-␥ promoter. However, it is possible that the decreased expression of the constructs containing the NFAT point mutations is secondary to an overriding or unbalanced effect of a silencer element within the IFN-␥ promoter located between the NFAT Ϫ280 and NFAT Ϫ160binding sites. In this region, several potential YY1-binding sites have been identified by homology to a consensus YY1binding sequence, and YY1 has been shown by gel shift analysis to bind to two sites named Y1 and Y2 (37,38). In those studies, when regions containing the YY1-binding sites (Y1 region Ϫ246 to Ϫ211; Y2 region Ϫ211 to Ϫ186) were placed upstream of an IFN-␥ reporter construct (pIFN-108), point mutations of the YY1-binding sites resulted in increased ex- FIG. 5. Expression of IFN-␥ promoter reporter constructs in fresh splenocytes. The constructs described in Fig. 4 were subcloned into a firefly luciferase reporter plasmid pGL3 and transiently transfected into murine splenocytes which were treated with 3 g/ml ConA 21 h prior to transfection to allow for uptake of DNA. After electroporation, the cells were rested for 2 h and then either not stimulated or stimulated with 25 ng/ml PMA and 1.5 M ionomycin (top panel, A) or with an ␣CD3 antibody 145-2C11 (bottom panel, B) in the presence or absence of cyclosporin A (1 g/ml). Lysates were harvested at 5 h, assayed for firefly luciferase activity, and corrected for transfection efficiency by normalization to a ␤-actin promoter driving Renilla luciferase. Values are the means Ϯ S.E. of three experiments in A and two experiments in B.

FIG. 6. Effect of AP-1 point mutation on expression of IFN-␥ promoter constructs in Jurkat T cells and fresh splenocytes.
A, constructs containing the full-length pIFN-538 promoter without or with point mutations in the AP-1 (AP-160)-or NFAT (N160)-binding sites of the composite NFAT/AP-1 Ϫ160 element were transiently transfected into Jurkat T cells, stimulated and assayed as described in Fig.  4. Values are the means Ϯ S.E. of six experiments for the AP-1 construct and three experiments for all other constructs. B, the corresponding luciferase constructs were transfected into murine splenocytes as described in Fig. 5. The cells were either not stimulated (uns) or stimulated with 25 ng/ml PMA (P) and 1.5 M ionomycin (I). Values are the means Ϯ S.E. of four experiments for the AP-1 construct and three experiments for all other constructs. pression (Y1) or loss of suppression (Y2) of the pIFN-108 reporter construct (37,38). To examine the role of the NFATbinding sites on IFN-␥ expression in the absence of the silencer element, point mutations of the YY1-binding sites (Y1 at position Ϫ201 and Y2 at position Ϫ218 to Ϫ217) were made in the context of the full-length pIFN-538 promoter, with or without the double NFAT point mutations (N160/N280), and tested by transient transfection in both Jurkat T cells and murine splenocytes (Figs. 4B and 5).
Although it has been suggested that the two YY1-binding site may play a role in suppressing basal transcription of the IFN-␥ gene, the full-length pIFN-538 reporter constructs with point mutations in one (Y1) or both YY1 sites (Y1Y2) were not expressed in unstimulated Jurkat T cells and showed no increase in expression compared with wild-type pIFN-538 in stimulated cells (Fig. 4B). Mutations in the YY1-binding sites also did not increase expression of constructs containing point mutations in the two NFAT-binding sites (Fig. 4B, compare N160/N280 to NY1 and NY1Y2).
When transfected into murine splenocytes, the reporter construct with the single YY1 point mutation (Y1) showed no significant difference in expression compared with the pIFN-538 construct in both unstimulated and stimulated cells (Fig.  5). The reporter construct with the double YY1 point mutation (Y1Y2) showed a slightly higher level of basal expression (1.7fold increase compared with pIFN-538, p Ͻ 0.002) in unstimulated cells but no significant difference in stimulated cells. In contrast to the results seen with the Jurkat T cells (Fig. 4B), combined mutation of the NFAT and one or both YY1 sites (NY1 and NY1Y2) abolished inducible expression and reduced basal expression in murine splenocytes (Fig. 5). The slight increase in basal expression seen in the Y1Y2 construct was also completely abolished in the presence of the NFAT mutations (NY1Y2).

DISCUSSION
The current studies further elucidate the molecular mechanisms by which IFN-␥ expression is regulated. A diverse group of both constitutive and inducible cis-regulatory elements appear to be involved (Fig. 8). In previous studies, the most proximal region of the IFN-␥ promoter between Ϫ108 and Ϫ40 bp relative to the transcription start site was found to be sufficient for activation-specific, cyclosporin-sensitive expression in T cells (25). Within this region, members of the CREB/ ATF, AP-1, octamer, and GATA families of transcription factors bind to the proximal (Ϫ70 to Ϫ47) and distal (Ϫ98 to Ϫ72) elements (25,26). These two elements are necessary for expression of IFN-␥ reporter constructs (25,26), and like the endogenous gene, constructs containing multimers of these elements are expressed on stimulation in memory, but not naïve T cells in transgenic mice (49). Further upstream, an AP-1 site at position Ϫ196 to Ϫ183 contributes to expression induced by mitogens or by treatment with IL-18 (51,59). This AP-1 site and a STAT4-binding site at position Ϫ238 (60) play a positive role in IL-12-dependent activation of IFN-␥ constructs in primary human CD4ϩ T cells (59). While the studies described in the current report were in progress, Sica et al. (28) demonstrated that NFAT and NF-B proteins could bind to position Ϫ278 to Ϫ268. Based on cotransfection of expression constructs encoding a constitutively activated form of calcineurin or p65, they proposed that the activity of this (C3-3P or NFAT-280) site is mediated by calcineurin-inducible factors, whereas enhancement by NF-B is primarily mediated via the IFN-␥ B-C3-1P tandem elements located further upstream at positions Ϫ786 to Ϫ776 and Ϫ772 to Ϫ763. In addition to these positive regulatory elements, two constitutive YY1-binding sites have been identified at positions Y1 (Ϫ199 to Ϫ203) and Y2 (Ϫ217 to Ϫ221) (37,38). These sites are postulated to play a role in inhibiting basal expression of IFN-␥ by competition for DNA binding with AP-1 at the Y1 position and by activation of an AP-2-like protein by YY1 binding at the Y2 position.
The current studies demonstrated that NFAT proteins bind to additional sites in the IFN-␥ promoter. In addition to the NFAT Ϫ280 site identified by Sica et al. (28) (Figs. 1-3), a composite NFAT/AP-1 site was found at position Ϫ160, to which recombinant proteins or proteins from activated, but not resting T cell nuclear extracts bound (Fig. 3). The NFAT Ϫ160 site is similar to composite NFAT/AP-1 elements found in the promoters of other cytokine genes (22) and is comprised of a relatively strong NFAT consensus sequence (in bold letters) adjacent to a weak AP-1 site (underlined): Ϫ171 GAGTCTA- AAGGAAACTCTAACTACAACACCCAAA Ϫ138. It has also been suggested that AP-1 might form a composite element with NFAT at the IFN-␥ NFAT Ϫ280 site (22); however, in the current studies, an interaction of AP-1 and NFAT at the NFAT Ϫ280 site was not detected by DNase footprinting or gel shift analysis (Figs. 1-3). In addition to being a composite NFAT/ AP-1 site, the NFAT Ϫ160 region may serve as a CD28 response element for the IFN-␥ gene (55). Although the NFAT Ϫ160 region has sequence homology with the CD28 response element of the IL-2 promoter which binds members of the NF-B family (53), binding of NF-B family members to this site was not detected in our Jurkat nuclear extracts by EMSA (Fig. 3C) or by DNase I footprint analysis using p50, p65, and p52 recombinant proteins (data not shown). This does not formally exclude other members of the NF-B family, such as c-Rel, from binding to this site under certain conditions (56). A third weak NFAT-binding site was identified at position Ϫ106 to Ϫ99 adjacent to the distal element. We postulate that this site is not of physiologic importance based on the following findings: 1) recombinant NFATp protein bound weakly and only in the presence of AP-1 by in vitro footprinting, 2) deletion of this site (pIFN-538⌬99 -109) did not affect expression of reporter constructs (25), and 3) this site is not conserved in the mouse IFN-␥ promoter. However, these results do not formally exclude the possibility of NFAT being recruited to this site in vivo.
As seen in Figs. 4 and 5, the strong NFAT-binding sites at positions Ϫ280 and Ϫ160 are required for maximum inducibility of the IFN-␥ promoter in Jurkat T cells and murine splenocytes, as demonstrated by decreased levels of expression of full-length IFN-␥ promoter constructs containing point mutations in one or both of the NFAT-binding sites. This result is similar to results seen with other cytokine promoters, including the IL-2 (52) and the IL-4 promoters (8,9). The mutation of the NFAT Ϫ160 site created a site to which Oct-1 bound in vitro ( Fig. 3B and data not shown). Although decreased expression of the IFN-␥ reporter construct containing this mutation could reflect a suppressive effect of Oct-1 on promoter activity (61), the more likely possibilities are that the binding of Oct-1 to the NFAT-160 site had no effect on expression, had a weak positive effect which was masked by the loss of NFAT binding, or had an inhibitory effect by blocking NFAT binding as seen in the IL-4 promoter (62). In either case, the results support the notion that NFAT plays a positive role at this site. As seen in Fig. 6, AP-1 also plays a positive role at the NFAT/AP-1 Ϫ160 composite site, as demonstrated by the reduced level of expression of a full-length IFN-␥ construct containing a point muta-tion that blocks AP-1, but not NFAT, binding.
Notably, the IFN-␥ reporter construct containing point mutations in both NFAT-binding sites (Figs. 4A and 5A, N160/ N280) remained completely sensitive to the inhibitory effects of cyclosporin. Thus, although NFAT proteins play a positive role in the IFN-␥ promoter and may contribute to the cyclosporin sensitivity of this gene (Fig. 3A), other regulatory elements in the IFN-␥ promoter, such as the IFN-␥ proximal element, confer cyclosporin sensitivity in an NFAT-independent manner. Consistent with this, cyclosporin also inhibits the calcineurindependent activation of c-Jun amino-terminal/stress-activated protein kinases in T cells, which are required for phosphorylation and activation of c-Jun and ATF-2 (63,64). We have previously shown that c-Jun is essential for activation-induced transcription of IFN-␥ (26,50) and binds preferentially to the IFN-␥ proximal element as a heterodimer with ATF-2 (26).
In contrast to the results seen with several other cytokine promoters (IL-2, IL-4, GM-CSF, and TNF-␣) (57), overexpression of NFATp and other NFAT protein family members did not further enhance expression of the IFN-␥ promoter construct in Jurkat T cells (Fig. 7A), suggesting that the abundance of NFAT proteins in these cells is sufficient for full IFN-␥ gene promoter function. This may reflect differences in the organization and relative role of NFAT proteins in the IL-2 (52), IL-4 (8,9), and TNF-␣ promoters (65), all of which contain three to five NFAT-binding sites within a total length of 200 -300 bp, in comparison to the IFN-␥ promoter, which contains only two strong NFAT-binding sites. However, in primary murine splenocytes, which have lower levels of endogenous NFAT activity, the positive role of NFAT became apparent in that expression of the IFN-␥ promoter construct was enhanced by overexpression of both NFATp and NFATc, whereas expression of the IL-2 promoter was enhanced predominantly by NFATp (Fig. 7B). These results are consistent with recent data in mice deficient for NFAT proteins, which suggest overlapping but distinct roles of NFATp and NFATc in the regulation of T lymphocyte development and cytokine expression (66 -69).
Previous data suggested that YY1 suppresses IFN-␥ promoter function in Jurkat T cells by interacting at two regions within a silencer element located between the NFAT-binding sites (38). In contrast to these results, which were obtained using a shorter IFN-␥ promoter construct (pIFN-225) or constructs in which portions of the silencer region were placed out of context upstream of the pIFN-108 promoter (38), the current studies demonstrated no significant increase in expression when either or both YY1-binding sites were mutated in the  (60); the Y1 and Y2 silencer element which contains two YY1-binding sites (37,38); the Ϫ196 AP-1-binding site (51); the NFAT Ϫ160/AP-1 site (this paper); CD28RE, the proposed CD28 response element (55); the weak NFAT Ϫ100 site (this paper); DE, distal element (25); and PE, proximal element (25,26). Members of the various transcription factor families known to bind to these sites in the IFN-␥ promoter are shown.
context of the full-length pIFN-538 construct and transfected into Jurkat T cells or murine splenocytes (Figs. 4B and 5). It should be noted that the pIFN-225 construct lacks the NFAT Ϫ280-binding site and has greatly decreased expression compared with the longer pIFN-538 reporter constructs (38). Of interest was the finding that in primary murine splenocytes but not Jurkat T cells, expression was abolished in IFN-␥ constructs containing both the NFAT and YY1 mutations, despite the presence in these constructs of the distal and proximal elements and the AP-1-binding site at Ϫ200. This difference can not be accounted for by the type of reporter construct, since the NY1 and NY1Y2 luciferase constructs were expressed in Jurkat T cells (data not shown). These results suggest that proteins binding to both the NFAT and YY1 sites may serve to initiate expression of IFN-␥ in primary splenocytes. Although YY1 has been shown to interact with many other transcriptional regulators including NF-B (reviewed in Ref. 29), interaction with NFAT proteins has not been demonstrated. Little information is known about the mechanisms by which YY1 induces transcriptional activation of cellular genes such as c-myc (30) and how cellular proteins modulate the activating and repressive activities of YY1 (29). The basis for the effect of the YY1 mutations in the IFN-␥ promoter are not known, and the importance of YY1 is uncertain, since no effect was observed unless the NFAT sites were also mutated. Nonetheless, the context-dependent effects of the YY1 mutations are consistent with the notion that YY1 has both activating and repressing properties that are influenced by the position of binding sites in a gene promoter and its interactions with other transcriptional regulators (29).
In summary, NFAT plays a positive role in IFN-␥ expression, but elements besides the NFAT-binding sites contribute to the cyclosporin sensitivity of the IFN-␥ promoter. In addition, YY1 may play a complex and context-dependent role in IFN-␥ expression. These factors and their binding sites, along with those previously identified in the essential distal and proximal elements, likely form a cooperative transcriptional complex, as described for IL-2 and other promoters (70,71).