Glucocorticoids Inhibit Calcium- and Calcineurin-Dependent Activation of the Human IL-4 Promoter¹

Glucocorticoids (GC)³ are the most potent immunosuppressive agents available for inflammatory diseases and work largely by inhibiting the production of proinflammatory molecules such as cytokines (1). Recently, considerable progress has been made in identifying the molecular mechanisms by which GC exert their effects. GC bind the cytosolic GC receptor (GR), which then translocates to the nucleus and inhibits the transcriptional activation of target genes (2). In general, the GR mediates transcriptional repression through: 1) interfering with the function of transacting factors (e.g., AP-1 and NF-κB) via protein-protein interactions; 2) direct DNA binding to poorly conserved negative GC response elements (GRE); or 3) inducing the expression of inhibitory factors such as IκBα (3, 4, 5, 6, 7, 8).

IL-4, a pleiotropic cytokine produced by activated T cells, basophils, and mast cells, regulates many cellular and humoral immune responses (9, 10). Dysregulated IL-4 gene expression has been linked with several inflammatory, allergic, and autoimmune diseases (11, 12, 13, 14, 15). GC readily inhibit IL-4 expression in vivo (16, 17, 18) and also suppress IL-4 production in isolated cell types including T cells, basophils, and mast cells (19, 20, 21, 22). The molecular mechanisms by which GC inhibit IL-4 gene expression have not been previously defined.

In T cells, IL-4 expression is tightly controlled at the level of transcription by multiple regulatory elements located within a proximal promoter region (23, 24, 25, 26, 27, 28, 29, 30). Regulatory elements outside of the promoter have also been detected (31, 32). NF-AT can enhance IL-4 transcription by binding up to five related sequences in the IL-4 promoter (termed the P elements P0–P4) (24). NF-AT is now known to comprise a group of related factors that recognize a common DNA motif (5′-GGAAAA-3′) (33). The precise roles of individual NF-AT proteins in regulating IL-4 gene expression are currently not clear. Experiments using NF-AT-deficient mice have demonstrated a critical role for NF-ATc in enhancing the differentiation of IL-4-secreting Th2 cells, whereas NF-ATp and NF-AT4 down-regulate this process (34, 35, 36, 37). In contrast, IL-4 expression in activated T cells appears to be due in part to NF-ATp-dependent promoter transactivation (38).

Unlike the IL-2 promoter, which requires costimulation of calcium- and protein kinase C (PKC)-mediated signaling pathways for full activation, the IL-4 promoter can be maximally induced by a calcium signal alone (26, 39, 40). A constitutively active mutant of the calcium-sensitive phosphatase calcineurin (CN) can efficiently substitute for this signal (40, 41, 42). Of the potential factors that are activated by CN in T cells (43, 44, 45, 46), NF-AT appears to be the most likely target in the IL-4 promoter. Currently, whether GC inhibit NF-AT-dependent transactivation is controversial. In one study, transcription driven by the IL-2 promoter-distal NF-AT site was partially inhibited by dexamethasone (Dex) in transfected Jurkat T cells (47). Because the IL-2 promoter requires cooperative interactions between NF-AT and AP-1 proteins for maximal induction, it was not possible to separate the repressive effects of GC on NF-AT from their known inhibition of AP-1 activity (3). Additionally, NF-AT activity was not inhibited by GC in other studies of the IL-2 promoter (48, 49).

The IL-4 promoter provides a unique opportunity to study the regulation of NF-AT-dependent transactivation by GC in T cells. In this paper we report that transcription driven by the intact human IL-4 promoter is strongly inhibited by GC in Jurkat T cells in a GR-dependent manner. We show that the repressive effects of Dex map downstream of CN activation and by EMSA identify the P1 NF-AT element as the site of an activation-induced and GC-inhibited nuclear protein complex that contains NF-ATc.

Human IL-4 promoter constructs were amplified from genomic DNA using the PCR. Twenty-five base pair primers annealing 372, 265, 225, 95, 65, and 35 bp upstream from the transcription start site (according to Otsuka et al. (50)) were used with a 25-bp primer ending at position +65. PCR products were ligated into the SrfI site of pCR-Script (Stratagene, La Jolla, CA) and sequenced to confirm accurate replication. pLuc 372 was synthesized by ligating the KpnI and SacI restriction fragment from pCR-Script 372 into compatible sites in pGL3 (Promega, Madison, WI). pCAT plasmids were synthesized by ligating the HindIII and XbaI restriction fragments from the corresponding pCR-Script plasmids (generated using PCR primers with those restriction sites included) into compatible sites in pCAT Basic (Promega). The IL-2 promoter reporter construct IL-2.15ΔCX has been described (51) and was a gift of Dr. Gerald Crabtree (Stanford University, Stanford, CA). The respiratory syncytial virus long terminal repeat-driven full-length human GRα expression vector (pRS hGRα, see Ref. 52) was a kind gift of Dr. Ron Evans (Salk Institute, La Jolla, CA). The plasmid ΔCaMAI contains a constitutively active fragment of CN, lacking the calmodulin-binding and autoinhibitory domains (53), and was kindly donated by Dr. Randall Kincaid (Veritas, Rockville, MD). Construction of the pREP4-based NF-ATc expression vector (provided by Dr. Timothy Hoey, Tularik, South San Francisco, CA) has been described (54).

Three different lines of human Jurkat T cells were used in these experiments. One line deficient in GR expression was identified on the basis of minimal immunoreactive GR using nuclear extracts from Dex-stimulated cells in EMSA with a radiolabeled GRE (Promega; not shown). The generation of a stably transfected GR-expressing subline of Jurkat T cells, derived from a parental GR-negative line, has been described (55). These cells (A11 cells), expressing roughly 70,000 copies/cell of the GR under control of the β-actin promoter (55), and a subline transfected with empty vector (β-actin cells), were both kindly donated by Dr. Michael Karin (University of California at San Diego, La Jolla, CA). Cells were maintained in RPMI 1640 supplemented with 10% FCS (Life Technologies, Gaithersburg, MD), 50 μg/ml gentamicin (Life Technologies), and, in the case of A11 and β-actin cells, 1.5 mg/ml G418 (60% active compound). Cells (5 × 10⁶) were washed and resuspended in RPMI 1640 containing 5% charcoal-filtered FCS (Gemini Bioproducts, Calabasas, CA) for 3 h and then transfected in duplicate using the Superfect method (Qiagen, Chatsworth, CA) with 1 μg reporter, the indicated amounts of expression vector, and empty vector to keep total amount of DNA constant with 3 μl Superfect per μg plasmid DNA. Cells were stimulated with calcium ionophore (A23187, 0.5 μM; Calbiochem, La Jolla, CA) with or without PMA (20 ng/ml, Calbiochem) in the presence or absence of the indicated concentrations of Dex (Sigma, St. Louis, MO) or DMSO control for 18 h before lysis by three freeze-thaw cycles. Reporter gene expression was determined by measuring CAT enzyme levels using a sensitive ELISA (Boehringer Mannheim, Indianapolis, IN) or using a Monolight 3010C luminometer and luciferase assay kit (Analytical Bioluminescence, Gaithersburg, MD). Cell extracts were normalized for protein content (Bio-Rad, Richmond, CA) before assays for reporter gene expression. To minimize intraexperimental variability in transfection efficiency, the following variables were controlled for in each experiment: cells passage number, plasmid purity, and total DNA amount per sample using empty expression vector to normalize. Under these conditions, transfection efficiency is almost identical within a given experiment (data not shown).

The following 30-bp oligonucleotides and their complements were synthesized: 5′-ATCTGGTGTAACGAAAATTTCCAATGTAAAC-3′ (P1/OAP −92 to −60), 5′-TGTAACGAAAATTTCCAATGTAAAC-3′ (P −86 to −60), and 5′-ATTGCTGAAACCGAGGGAAAATGAGTTTACATTG-3′ (P0 −69 to −36). A canonical GRE oligonucleotide was purchased (Promega). The −35 to +65 promoter fragment used for competition EMSA was generated by gel purifying the HindIII-XbaI restriction fragment from pCR Script 35. Nuclear extracts were obtained from 5 × 10⁶ Jurkat cells using the method of Schreiber et al. (56). EMSAs were performed using 5 μg nuclear protein, 0.8 μg poly(dG-dC) (Pharmacia, Piscataway, NJ), and 5 pg [γ-³²P] end-labeled probe in a final volume of 10 μl. Free probes and protein-DNA complexes were resolved by 5% PAGE with 0.5 × Tris-borate-EDTA (TBE). In some experiments, extracts were incubated with 1 μl of the following antisera for 30 min at 4°C before addition of probe: anti-NF-ATp (Upstate Biotechnology, Lake Placid, NY) and anti-NF-ATc (7A6, Affinity BioReagents, Golden, CO), each non-cross-reactive with other NF-AT family members; anti-c-Fos and anti-c-Jun, both broadly reactive with other Fos and Jun family members, respectively (Santa Cruz Biotechnology, Santa Cruz, CA); and isotype- and species-matched control IgG (Sigma).

A recombinant fragment of murine NF-ATp (including 298 aa of the DNA binding domain (DBD), highly conserved among different NF-AT family members) was expressed as a hexahistidine-tagged protein and extracted as described (57). A recombinant fragment of the rat GR (containing aa 440–552 of the DBD), which binds DNA independent of ligand (58), was kindly donated by Dr. Leonard Freedman (Memorial Sloan-Kettering Cancer Center, New York, NY). A transcriptionally active preparation of the full-length GR was provided courtesy of Dr. Barbour Warren (Laboratory of Receptor Biology and Gene Expression, National Institutes of Health, Bethesda, MD; Ref. 59).

Fifty micrograms of nuclear protein per condition (extracted from A11 cells using the method of Schreiber et al. (56)) were adjusted to a final concentration of 84 mM KCl and 15 mM Tris-HCl. To immunoprecipitate NF-ATc, lysates were precleared with 1 μg of mouse IgG (Sigma) and 20 μl protein G-coated Sepharose beads (protein G PLUS, Santa Cruz Biotechnology), and were then incubated with 8 μl of monoclonal anti-NF-ATc (7A6, Affinity BioReagents) and 20 μl protein G PLUS for 18 h under constant rotation at 4°C. Equal amounts of species-matched control antisera were used for each condition. After centrifugation, beads were extensively washed, and immunoprecipitated proteins were collected by boiling in 40 μl SDS sample buffer (2% SDS, 50 mM Tris (pH 6.8), 10 mM DTT, 0.1% bromophenol blue, and 10% glycerol). Forty-microliter aliquots were separated by 6% SDS-PAGE then transferred to Trans-Blot transfer medium polyvinylidene difluoride membrane (Bio-Rad). After blocking in PBS-5% BSA/0.1% Tween 20 for 1 h, membranes were probed with anti-NF-ATc (7A6, 1:200), or anti-GR (E20, 1:200) for 1 h. After two washes (5 min each) with PBS/0.1% Tween, membranes were incubated for 1 h with the secondary Ab (HRP-conjugated goat anti-mouse or anti-rabbit (1:3000; Amersham Pharmacia Biotech, Piscataway, NJ)), before a final washing step. Immunoreactive bands were visualized by enhanced chemiluminescence and autoradiography using the ECL Western blotting detection kit according to the manufacturer’s directions (Amersham Pharmacia Biotech). To detect coprecipitated proteins, membranes were stripped by submerging in stripping buffer (100 mM 2-ME, 2% SDS, and 62.5 mM Tris-HCl (pH 6.7)) for 45 min at 55°C with constant shaking. Membranes were then washed three times with large volumes of PBS/0.1% Tween, blocked for 1 h, and reprobed with a different primary Ab as indicated.

We studied the effects of the synthetic hormone Dex on transcription driven by a comprehensive panel of human IL-4 promoter deletion constructs transiently transfected into GR-deficient Jurkat T cells (see Materials and Methods). Fig. 1 shows the results of experiments using a full-length promoter construct that contains 372 bp of the IL-4 promoter linked to the firefly luciferase gene (pLuc 372). As we and others have previously observed with different IL-4 promoter constructs (26, 39), pLuc 372 was maximally inducible with a calcium signal alone. Promoter activity was not affected by Dex alone, consistent with the negligible expression of nuclear GR protein in these cells. We next used a full-length expression vector encoding the wild-type human GR (pRShGRα) in transient cotransfection experiments. As shown in Fig. 1, Dex strongly inhibited calcium-induced pLuc 372 activity in a GR-dependent manner.

NF-κB is a well-established molecular target of GC (7, 8). We have previously shown that PMA-induced NF-κB proteins down-regulate the human IL-4 promoter by competing with NF-AT for binding to the P1 element (39). Therefore, GC might be expected to relieve NF-κB-mediated repression of the IL-4 promoter induced by PMA. However, as shown in Fig. 1, pLuc 372 activity in Jurkat cells costimulated with calcium ionophore and PMA was further inhibited by Dex in GR-cotransfected cells. This observation suggests that GC inhibit another factor that activates the proximal IL-4 promoter in our system.

CN is a critical phosphatase that mediates calcium-induced gene expression in T cells (43), and activated CN alone is sufficient to maximally induce the IL-4 promoter in Jurkat cells (42). Unlike other potential targets of CN in T cells that require concomitant stimulation with PMA for full induction (44, 45), NF-AT can be exclusively activated by CN alone (46). To better characterize the inhibition of calcium ionophore-induced IL-4 transcription by GC, we studied the repressive effect of Dex on IL-4 promoter activity induced by cotransfecting cells with a constitutively active form of CN (ΔCaMAI, see Materials and Methods). Fig. 2 shows that cotransfection of ΔCaMAI was sufficient to strongly induce pLuc 372 activity. Interestingly, this induction was inhibited by Dex in a GR-dependent manner. These results map the repressive effect of Dex on calcium-induced IL-4 promoter activity downstream of CN activation.

To map the Dex-responsive element(s) in the IL-4 promoter, we next used a series of promoter deletion constructs in additional transfection experiments. The calcium-induced activation of each construct examined was inhibited by Dex in the presence of the GR (data not shown). Fig. 3 shows the results of experiments using a minimal promoter construct containing 95 bp upstream from the transcription start site linked to the chloramphenicol acetyltransferase (CAT) gene (pCAT 95). In these experiments we used the A11 subline of Jurkat cells, which was stably transfected with a GR expression vector under the control of the β-actin promoter, to ensure constant GR protein expression (55). Dex completely inhibited calcium-induced pCAT 95 activity in A11 cells and further inhibited the promoter in combination with PMA. Importantly, Dex did not inhibit pCAT 95 or other promoter constructs in control cells stably transfected with empty vector alone (data not shown).

These experiments localized the repressive effect of Dex to the proximal 95 bp of the IL-4 promoter. There are two P elements within this region. First, the P1 site is a well-characterized NF-AT response element that is critical for maximal activation of the IL-4 promoter (23, 24, 25, 29) and is the site of an activation-induced nuclear protein complex in EMSA (29, 60). Second, the P0 element is located just downstream and has been shown to interact with NF-AT, c-Maf, and CCAAT/enhancer-binding protein (28, 61). To better map potential Dex-responsive elements in this region of the IL-4 promoter, we synthesized a minimal promoter construct containing only the P0 element and downstream sequences (pLuc 65, see Materials and Methods). In these experiments, we found that pLuc 65 was uninducible by calcium ionophore in transiently transfected Jurkat T cells (0.95 ± 0.11-fold induction, n = 4, data not shown). These data are consistent with the observation that multiple P elements are needed for maximal promoter inducibility (24) and strongly suggest that the P1 element harbors a Dex-sensitive regulatory element.

In keeping with the uninducibility of pLuc 65, we found that the P0 element binds a constitutive, but not activation-induced, complex in EMSA (T.F.B., R.C., and S.N.G., unpublished observations). Therefore, we next studied the regulation of nuclear protein complex formation on the P1 element by Dex. Fig. 4 shows that a broad complex formed using nuclear extracts isolated from calcium ionophore-stimulated, but not resting, A11 cells (Fig. 4,A, complex I, lane 3) which was strikingly prevented by Dex (Fig. 4,A, lane 4). We have previously observed a factor that binds constitutively in this region (Fig. 4,A, complex II; Ref. 39), which was not affected by Dex treatment. To determine the identity of complex I, we used specific antisera in additional EMSA. Fig. 4,B shows that complex I contains predominantly NF-ATc because its formation was largely inhibited by a specific anti-NF-ATc antiserum. The formation of complex I was not affected by anti-NF-ATp Abs (Fig. 4 B) or by antisera that were broadly reactive for c-Fos, c-Jun, and related AP-1 family members (data not shown).

We wondered whether the inhibition of NF-AT binding to the IL-4 promoter in Dex-treated cells was due to a noncanonical GR binding site overlapping the P1 element. To test this hypothesis, we analyzed the ability of the GR DBD to interact with the P1 element in EMSA. Using concentrations of the GR DBD that formed strong monomeric and dimeric complexes on a canonical GRE (Fig. 5,A, lane 3), we found that the IL-4 P1 element supported only weak monomeric binding of the GR (Fig. 5,A, lane 1). However, even high concentrations of this factor were unable to displace the NF-AT DBD from its cognate site in this region (Fig. 5,B). In competition experiments, up to 50-fold molar excess of oligonucleotides containing the P1 element were unable to compete for binding of the GR to a canonical GRE in EMSA (Fig. 6, lanes 8 and 9). Taken together, these results suggest that the GR interacts only weakly with the P1 element and that inhibition of NF-AT binding to the P1 element by Dex is not due to competitive binding of the GR to an overlapping monomeric binding site.

To further rule out the possibility that the GR interacted directly with promoter elements located downstream of P1, we first studied the ability of the GR to bind to an oligonucleotide probe containing the P0 element (−69 to −36). No binding to this element was detected in EMSA (Fig. 5,C). We next studied the ability of a 100-bp promoter fragment (containing nucleotides −35 to +65, see Materials and Methods) to compete for binding of the GR to a canonical GRE. No competition was observed under these conditions (Fig. 6, lanes 6 and 10). Thus, the contiguous 134 bp downstream of P1 do not appear to contain a high-affinity GR binding site.

We next analyzed nuclear extracts isolated from resting and activated Jurkat A11 cells for the expression of NF-AT proteins and the GR using immunoprecipitation and Western blot analysis. To detect potential protein-protein interactions between these factors, these experiments were performed under nondenaturing conditions (see Materials and Methods). As shown in Fig. 7,A, multiple isoforms of NF-ATc were immunoprecipitated under these conditions. A predominant band of ∼86 kDa was observed (Fig. 7,A, band III) together with two more slowly migrating species corresponding to the ∼110-kDa (Fig. 7,A, band II) and ∼140-kDa (Fig. 7,A, band I) NF-ATc isoforms recently reported by Lyakh et al. (62). Interestingly, a quickly migrating, NF-ATc-specific band was also detected just above the Ig heavy chain (Fig. 7,A, band IV), possibly corresponding to a ∼56-kDa NFATc isoform previously detected in murine mast cells (63). The nuclear expression of all isoforms was increased by cell stimulation with calcium ionophore (compare Fig. 7,A, lanes 2 and 4), but was not noticeably inhibited by Dex (Fig. 7,A, lane 6). When the membrane was stripped and reanalyzed for expression of the GR (Fig. 7,B), a single band of ∼94 kDa was immunoprecipitated by the anti-GR Ab from Dex-treated nuclear extracts (Fig. 7,B, lane 3). This corresponds to the known stable expression of the GRα isoform in these cells (55). Interestingly, the GRα coprecipitated with NF-ATc using nuclear extracts from calcium ionophore- and Dex-treated cells (Fig. 7,B, lane 6). Importantly, this factor was not immunoprecipitated using control antisera and nuclear extracts from cells treated with either calcium ionophore or Dex alone (Fig. 7 B, lane 5; and data not shown). These results are suggestive of direct protein-protein interactions between the GRα and NF-ATc within the nuclear environment.

If the GR was interfering with the ability of NF-ATc to transactivate the IL-4 promoter, we reasoned that overexpression of this factor would restore promoter inducibility in the presence of Dex. To test this hypothesis, Jurkat cells were cotransfected with an expression vector encoding the GR and full-length NF-ATc, and promoter activity was analyzed in calcium ionophore-activated cells treated with Dex. As shown in Fig. 8, overexpression of NF-ATc modestly but consistently increased the calcium-dependent activation of pLuc 372. However, note that under these conditions transcriptional activity was rendered resistant to inhibition by Dex in cells cotransfected with the GR.

GC are known to interfere with the expression and/or function of several transcription factors (2). Based on the experiments described in this report, we conclude that calcium- and CN-dependent IL-4 promoter activity is inhibited by GC. This involves antagonism of NF-ATc-driven promoter activity by the GRα, which may alter the DNA-binding ability of NF-ATc via protein-protein interactions. The observation that NF-AT itself is a target of the GR in T cells is a novel finding and was facilitated by the unique calcium inducibility of the IL-4 promoter. This allowed us to more precisely analyze transcriptional activation mediated by NF-AT without the concurrent activation of PKC-induced cofactors. Concurrent stimulation of calcium- and PKC-mediated pathways resulted in lower IL-4 promoter activity compared with calcium activation alone (Fig. 1), which we previously found was due to inhibition of NF-AT binding to the P1 element by PMA-induced NF-κB heterodimers (39). This observation suggested that inhibition of NF-κB by the GR is not responsible for the down-regulation of IL-4 promoter activity by Dex.

In conjunction with previous reports (40, 41, 42), we found that activated CN can efficiently substitute for the calcium-mediated signal necessary for IL-4 promoter activity. Unlike other potential downstream targets of CN that require concomitant activation with PMA, such as c-Jun N-terminal kinase (45) and IκB (44), NF-AT can be activated by CN alone (33). Importantly, CN-driven IL-4 promoter activity was efficiently inhibited by Dex in our experiments. Paliogianni et al. (64) reported that the phosphatase activity of CN was not inhibited by Dex in T cells. Consistent with this observation, we found that Dex did not inhibit calcium ionophore-induced NF-ATc nuclear accumulation in A11 cells as determined by immunoprecipitation and Western blot analysis (Fig. 8) or by immunofluorescence (our unpublished observations). These data suggest that GC might directly interfere with NF-AT-dependent promoter transactivation. Studies to date of the effects of GC on NF-AT-driven IL-2 promoter activity have not reached a consensus. In one study, transcription driven by the IL-2 promoter-distal NF-AT site was partially inhibited by Dex in transfected Jurkat T cells (47). The same investigators also found that Dex inhibited the binding of nuclear factors from Jurkat and primary T cells to this element in EMSA (65). Because the IL-2 promoter requires cooperative interactions between NF-AT and AP-1 proteins for maximal induction, it was not possible to separate the repressive effects of GC on NF-AT from their known inhibition of AP-1 activity in these studies. In contrast, other investigators found that the inhibition of IL-2 transcription by GC did not involve NF-AT (48, 49). Reasons for these discrepancies are currently unknown, but may be due to differences in experimental conditions or cell lines used.

Of the four known NF-AT family members, NF-ATp and NF-ATc are thought to predominate in mature T cells (62). Multiple isoforms of these factors have been identified, likely due to alternative splicing of common gene products (66, 67). In both peripheral blood T and Jurkat cells, NF-ATp is constitutively expressed (62, 68). Both the levels and DNA-binding ability of this factor decrease upon continued cell activation (62, 69). In contrast, the expression of NF-ATc is significantly increased by cell activation, and a calcium signal alone can achieve this effect (62). Like other Jurkat lines (62, 68), the cells used in our experiments express both NF-ATp and NF-ATc (Fig. 6, and data not shown). Our observation that the calcium-induced, Dex-sensitive nuclear complex forming on the P1 element in EMSA contained predominantly NF-ATc is likely due to the delayed activation of this factor in calcium-stimulated T cells (62).

Although the nuclear accumulation of NF-ATc was not significantly affected by Dex, its DNA-binding ability was strongly inhibited under these conditions (Figs. 4 and 7). The observation that the GR DBD was unable to displace the highly conserved NF-AT DBD from this region in EMSA (Fig. 5) suggests that this inhibition does not involve competitive DNA binding of the GRα. In contrast, our finding that NF-ATc and the GRα coprecipitated in nuclear extracts of Dex-treated and calcium-stimulated cells supports a model in which the GR interferes with NF-ATc binding to the IL-4 promoter via direct protein-protein interactions (Fig. 9). Current experiments are exploring this possibility. Because we did not detect evidence of interactions between the NF-AT DBD and either the GR DBD (Fig. 5 B) or the full-length GR (data not shown) in EMSA, it is possible that interaction with the GRα involves other regions of NF-ATc.

NF-AT domains outside of the DBD are quite divergent, which likely explains the distinct phenotypes of mice rendered deficient for different NF-AT family members. These experiments demonstrated that both NF-ATp and NF-AT4 serve to down-regulate lymphoid responses and Th2 differentiation (34, 37), whereas NF-ATc is essential both during embryogenesis and for cytokine gene expression in Th2 cells (35, 36, 70). Our observation that NF-ATc is inhibited by GC is especially relevant in this regard. The molecular basis for these differential effects of NF-AT proteins are currently not well understood. Our observation that overexpressed NF-ATc increased calcium-driven IL-4 promoter activity is consistent with a direct role for this factor in enhancing IL-4 gene expression. The modest increase in transactivation in NF-ATc-overexpressing cells may be due to the lack of posttranslational modification of NF-ATc or of heteromeric cofactors in our system. Importantly, however, IL-4 promoter activity was rendered Dex-insensitive under these conditions.

In summary, we conclude that GC inhibit IL-4 gene expression by interfering directly with NF-ATc-driven promoter activity (Fig. 9). These observations help explain the potent inhibitory effects of GC on Th2-driven inflammatory responses such as allergic asthma. Because Dex and cyclosporin A appear to inhibit NF-AT by different mechanisms, it will be interesting to study possible synergistic effects of these agents on IL-4 expression in T cells. The expression of IL-4 in a subset of patients with steroid-resistant asthma is not sensitive to GC (71). Studying the molecular pathway described in this report in subjects with this syndrome should be insightful. Finally, our results also provide evidence for a novel protein target of the GR. Recent studies using mice expressing a mutant GR lacking the DBD suggest that protein-protein interactions are responsible for many of the transactivating effects of the GR (72). It will be interesting to determine whether repressive effects of GC on cytokine gene expression are also affected in these models.

We thank the following collaborators for their generous gifts of reagents: Drs. Michael Karin, Ron Evans, Leonard Freedman, Timothy Hoey, Randall Kincaid, and Anjana Rao. We also thank Dr. Robert Schleimer for helpful suggestions and advice.