Utilization of an NF-ATp Binding Promoter Element for EGR3 Expression in T Cells but Not Fibroblasts Provides a Molecular Model for the Lymphoid Cell-Specific Effect of Cyclosporin A

Abstract

Cyclosporin A (CsA) mainly exerts its immunosuppressive action by selectively inhibiting Ca²⁺/calcineurin-dependent gene transcription in lymphoid cells. A model explaining the tissue-specific effect of this drug on gene expression has not been established to date, since none of the known intracellular targets of CsA (e.g., cyclophilins, calcineurin, and NF-AT) is lymphoid cell specific. To investigate this issue, we performed a detailed comparative analysis of the promoter regulating the two-signal-dependent (Ca²⁺ ionophore plus phorbol myristate acetate [PMA]), CsA-sensitive expression of EGR3 in T cells and the one-signal-dependent (PMA), CsA-insensitive expression of EGR3 in fibroblasts. As a result, we identified a 27-bp promoter element functionally interacting with transcription factors NF-ATp and NF-ATc that is crucial for the CsA-sensitive expression of the EGR3 gene in T cells. In contrast, the same element was without function in fibroblasts, and other, CsA-insensitive promoter regions were found to be responsible for EGR3 gene expression in these cells. The inactivity of the 27-bp element in fibroblasts was apparently due to insufficient expression levels of NF-ATp, since overexpression of NF-ATp, but not NF-ATc, restored the two-signal phenotype and CsA sensitivity of EGR3 promoter induction in these cells. The differential usage of an NF-AT binding site explains the selective effect of CsA on EGR3 gene expression in T cells versus fibroblasts and may represent one of the basic mechanisms underlying the tissue specificity of CsA.

Cyclosporin A (CsA), since its discovery in the early 1970s, has gained widespread use in clinical medicine due to its selective modulatory effect on cells of the immune system. This selectivity also made CsA a valuable research tool in the delineation of signal transduction events in lymphoid cells. In recent years it has been shown that CsA exerts its immunosuppressive action by inhibiting a Ca²⁺-dependent pathway involved in the initiation of gene transcription in lymphoid cells and mast cells, whereas gene expression in other cell types was found to be largely CsA resistant. In spite of detailed studies on the effect of CsA on single elements of the signaling machinery in T cells, the overall mechanism responsible for the tissue specificity of CsA has not been identified to date.

In T cells, all genes known to be fully suppressed by CsA encode cytokines (e.g., interleukin 2 [IL-2], gamma interferon, granulocyte-macrophage colony-stimulating factor, IL-8, and activation-induced T-cell-derived and chemokine-related molecule [ATAC]) (29, 42, 46, 47) or cytokine-related proteins (e.g., TRAP/CD40 ligand and Fas ligand) (2, 6). All of these genes are dependent on two signals for induction: a Ca²⁺ signal, which can be provided by Ca²⁺ ionophores, and a protein kinase C-mediated signal, which can be delivered by phorbol esters. CsA blocks the Ca²⁺ pathway and thus fully abrogates the induction of these key genes in T cells. At the molecular level, CsA is found intracellularly complexed with members of the cyclophilin family (12, 19). The CsA-cyclophilin complexes bind to and inhibit the Ca²⁺/calmodulin-dependent phosphatase calcineurin and thus interrupt Ca²⁺/calcineurin-dependent gene transcription (3, 20, 32). In recent years, several transcription factors (TF) have been identified as downstream elements of the CsA-sensitive Ca²⁺/calcineurin signaling pathway. The most prominent among them is the nuclear factor of activated T cells (NF-AT; reviewed in references 36 and 39); less-characterized CsA-sensitive target proteins are AP-1 and NF-κB (5, 45). None of the components involved in the CsA-sensitive, Ca²⁺-dependent signaling pathway characterized so far is lymphoid cell specific. Cyclophilins, calcineurin, AP-1, and NF-κB are ubiquitous proteins, and at least one member of the NF-AT family of TF has been found to be expressed in a great variety of tissues (1, 7, 13, 15, 16, 40). The existing data thus do not provide an explanation for the tissue specificity of CsA.

In a previous study, when analyzing a collection of T-cell activation genes for “two-signal genes,” we identified PILOT (23), which turned out to be identical to EGR3 (33), a member of the early-growth response family of TF genes (reviewed in reference 8). EGR3, unlike most other two-signal genes, is expressed in a great variety of cell types. In T cells, induction of the EGR3 gene requires concomitant signaling by phorbol myristate acetate (PMA) and a Ca²⁺ ionophore (two signals) and is CsA sensitive (23). In nonlymphoid cells, e.g., fibroblasts, EGR3 is induced by PMA alone (one signal), and this induction is CsA insensitive (23). In the present report, we used the differential expression characteristics of the EGR3 gene to investigate the mechanism responsible for the tissue specificity of CsA.

MATERIALS AND METHODS

Isolation of the human EGR3 gene.

Genomic EGR3 clones were obtained by screening a human placenta genomic library (Lambda Fix II; Stratagene, La Jolla, Calif.) with various 5′ and 3′ EGR3 cDNA probes by standard techniques (38). Genomic DNA fragments encompassing the EGR3 gene 5′ regulatory region were cloned into pBluescript II SK(+) (Stratagene), and the nucleotide sequences of both strands were determined with the Sequenase kit from U.S. Biochemicals, Cleveland, Ohio.

Primer extension analysis.

Primer extension analysis was performed with a [γ-³²P]ATP end-labeled oligonucleotide primer complementary to nucleotides +27 to +46, which was hybridized to 10 μg of total RNA (obtained from stimulated peripheral blood T cells) for 30 min on ice. The extension reaction was run in reverse transcription buffer (50 mM Tris-HCl [pH 8.3 at room temperature], 75 mM KCl, 3 mM MgCl₂, 10 mM dithiothreitol [DTT], 0.5 mM [each] dNTP) with 200 U of Moloney murine leukemia virus H⁻ reverse transcriptase (Gibco-BRL, Gaithersburg, Md.) at 50°C for 60 min.

Cell culture.

Jurkat T cells were cultured as described previously (22). The human fibrosarcoma cell line Hs913T (American Type Culture Collection) was cultured in Dulbecco’s modified Eagle’s medium (Gibco-BRL) with 4.5 g of glucose per liter–10% heat-inactivated fetal calf serum–862 mg of l-alanyl-l-glutamine per liter–100 U of penicillin per ml–100 μg of streptomycin per ml–50 μM 2-mercaptoethanol.

RNA isolation and Northern analysis.

Isolation of RNA and Northern blot analysis were performed as described previously (22).

Generation of luciferase reporter constructs.

A genomic RsaI fragment containing the EGR3 sequence from −2952 to +615 was subcloned into the SmaI site of pBluescript II SK(+). The 3′ end was deleted to nucleotide position +86 by exonuclease III digestion, resulting in plasmid pBS-Rsa. To generate Rsa-luc, the EGR3 sequence from −2952 to +86 was excised from pBS-Rsa by using the polylinker restriction sites XbaI (blunt ended with Klenow polymerase) and KpnI and subcloned into the SmaI and KpnI sites of the pGL2basic luciferase vector (Promega, Madison, Wis.). 5′ and internal deletion constructs were generated by standard procedures using the various restriction sites indicated in Fig. 3. Rsa-luc mut-rep was created by PCR mutagenesis as described previously (14), by replacing the sequence from −127 to −122, CCATTG, with AGTCCA. To construct 4x(−134 to −95)SV40-luc, a double-stranded oligonucleotide with XhoI/SalI overhanging ends was cloned in four copies upstream of the simian virus 40 (SV40) promoter into the XhoI site of pGL2prom (Promega). Multimerization of the sequences from −122 to −95 and from −134 to −108 was performed as described previously (41). The multimerized sequences were subcloned into the SacI and XhoI sites of pGL2prom, resulting in the constructs 4x(−122 to −95)SV40-luc and 4x(−134 to −108)SV40-luc. All constructs contained the multiple copies in the forward orientation.

FIG. 3.

Deletion analysis of the upstream region of the EGR3 gene. 5′ deletions (A) or internal deletions (B) of the 5′ flanking region of the EGR3 gene were fused to a luciferase reporter and were cotransfected with an SV40 β-galactosidase control vector into Jurkat T cells or Hs913T fibroblasts. The cells were left unstimulated (unst) or were stimulated with a Ca²⁺ ionophore (Iono) and/or PMA for 3 h in the presence or absence of CsA. The results were normalized for transfection efficiency and are expressed as percentages of the Rsa-luc activity induced by PMA plus the Ca²⁺ ionophore. They represent the means of three independent transfection experiments ± the standard errors of the means. SI values (defined in the legend for Fig. 2) are shown. The diagram at the top displays schematically the EGR3 5′ flanking region (tss, transcription start site). The restriction enzymes used for generating the deletion constructs are indicated. n.d., not determined.

Cell transfections and stimulation.

A total of 15 × 10⁶ Jurkat T cells were transiently transfected with 30 μg of luciferase reporter construct and 30 μg of pSV40 β-galactosidase (Promega) or 5 μg of cytomegalovirus (CMV) β-galactosidase expression vector (Stratagene) by electroporation at 240 V and 960 μF. The cells were split equally among five wells. Next day, the cells were left unstimulated or were stimulated with PMA (20 ng/ml; Sigma), Ca²⁺ ionophore A23187 (125 ng/ml; Sigma), or PMA plus Ca²⁺ ionophore A23187 in the presence or absence of CsA (1 μg/ml; Sandoz). When used, CsA was added 30 min before cell stimulation. After a 3-h stimulation period the cells were harvested, lysed in 0.2% Triton X-100 buffer, and assayed for luciferase and β-galactosidase activity. Hs913T fibroblasts were transfected by the calcium phosphate precipitation method as described previously (38). Briefly, 3 × 10⁶ cells were seeded into 10-cm-diameter petri dishes (10⁶ cells/dish) and transfected for 16 h with 20 μg of luciferase reporter construct and 20 μg of pSV40 β-galactosidase or 5 μg of CMV β-galactosidase expression vector. Next day, the adherent cells were trypsinized, distributed, and stimulated as described for Jurkat T cells. Cotransfection studies were performed in the presence of 10 μg of expression plasmid pLGPmNF-AT1-C (containing the murine NF-ATp C isoform) (21) or pRSV NF-ATc (containing the human NF-ATc α isoform; kindly provided by E. Serfling).

Nuclear extract preparation and EMSA.

Nuclear extracts were prepared by a modification of the method of Dignam et al. (4). Briefly, cells were lysed in a buffer containing 10 mM HEPES-KOH (pH 7.8), 15 mM KCl, 2 mM MgCl₂, 1 mM DTT, 0.1 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 10 μg of aprotinin per ml, 25 μM leupeptin, 2 μM pepstatin A, and 0.2% Nonidet P-40 at 4°C. The nuclei were centrifuged, and nuclear proteins were extracted under high-salt-level conditions in a solution containing 20 mM HEPES-KOH (pH 7.8), 0.42 M NaCl, 1.5 mM MgCl₂, 0.5 mM DTT, 0.2 mM EDTA, 0.5 mM PMSF, 10 μg of aprotinin per ml, 25 μM leupeptin, 2 μM pepstatin A, and 25% (vol/vol) glycerol for 30 min at 4°C. After centrifugation at 125,000 × g for 30 min, the amount of protein in the supernatant was determined with the Bio-Rad protein assay kit. For the electrophoretic mobility shift assay (EMSA), 4 μg of nuclear proteins was preincubated for 10 min at room temperature in a 20-μl volume in a buffer containing 10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 0.5 mM EDTA, 8% Ficoll 400, 1 mM DTT, and 50 ng of poly(dI-dC) per ml. Specific competitors were added in 200-fold molar excess to the preincubation mixture. When indicated (see the legend for Fig. 7), the nuclear extracts were preincubated with 1 μl of a peptide-specific NF-ATp antiserum (9) or 1 μl of a monoclonal antibody raised against recombinant NF-ATc (Alexis Corporation, San Diego, Calif.). Radiolabeled double-stranded oligonucleotides (0.2 to 0.4 ng; approximately 30,000 cpm) were added last, and the whole reaction mixture was incubated for a further 30 min at room temperature. The protein complexes were separated on a 4% nondenaturating polyacrylamide gel in 0.5× Tris-borate-EDTA.

FIG. 7.

Supershift experiments demonstrate the binding of NF-ATp and NF-ATc to the 27-bp element. (A) Nuclear extracts from Jurkat T cells and Hs913T fibroblasts stimulated with PMA plus a Ca²⁺ ionophore for 30 min were preincubated without antibodies (lane 2), with a peptide-specific NF-ATp antiserum (lanes 3 and 7), with a monoclonal anti-NF-ATc antibody (lanes 4 and 8), or with both antibodies (lanes 5 and 9). EMSAs were performed with an oligonucleotide spanning the 27-bp element as a probe. The NF-AT proteins present in each complex are indicated on the left (p, NF-ATp; c, NF-ATc). Supershifted complexes are indicated by arrows. (B) A longer exposure of lanes 6 to 9 of panel A.

Nucleotide sequence accession number.

The entire EGR3 promoter sequence was deposited in the EMBL database under accession no. Y07558.

RESULTS

Essential regulatory elements required for CsA-sensitive, two-signal induction in T cells and one-signal induction in fibroblasts are located within the 5′ flanking region of the EGR3 gene.

For studies on the transcriptional control of EGR3, we isolated the EGR3 gene and sequenced approximately 3.6 kb upstream of the putative translation initiation codon (a part of the sequence is shown in Fig. 1). Primer extension analysis (data not shown) indicated that EGR3 gene transcription is initiated predominantly at a guanosine (assigned position +1), which is preceded by a putative TATA box at nucleotide −28 (Fig. 1). To investigate the contribution of the 5′ flanking region (−2952 to +86) to the activation of the EGR3 gene, a chimeric EGR3/luciferase reporter construct (Rsa-luc) was analyzed for inducibility in Jurkat T cells and Hs913T fibroblasts, since these cell lines exhibit virtually the same signal requirements for EGR3 gene expression as do the respective primary cells (Fig. 2A) (23). In Jurkat T cells, the Rsa-luc construct was only fully induced by PMA plus a Ca²⁺ ionophore and was very poorly induced by PMA or a Ca²⁺ ionophore alone, whereas in fibroblasts, PMA alone was sufficient for optimal activation (Fig. 2B). In Jurkat T cells, CsA inhibited luciferase activity induced by PMA plus Ca²⁺ ionophore by about 70% (to the level induced by PMA alone), whereas in fibroblasts CsA was without effect (Fig. 2B). These results indicated that both the CsA-sensitive, two-signal induction of EGR3 in T cells and the CsA-insensitive, one-signal induction in fibroblasts are determined by the isolated 5′ flanking region of the EGR3 gene.

FIG. 1.

Nucleotide sequence of the immediate upstream region of the EGR3 gene. The numbers on the left refer to the nucleotide sequence, with the predominant transcription start site as nucleotide +1. Putative binding sites for known TF are boxed. A sequence resembling an NF-κB-binding site is marked by arrows. A direct repeat 11 nucleotides in length is underlined. Restriction enzymes (BsmI, AatII, and SmaI) used for the subcloning of EGR3 promoter regions are indicated. TCF-1, T-cell-specific transcription factor 1; PEA3, polyomavirus enhancer A3; SRE, serum response element.

FIG. 2.

The 5′ flanking region is responsive to the signals that induce endogenous EGR3 gene expression in T cells and fibroblasts. (A) Northern blot analysis of EGR3 gene expression. Jurkat T cells and Hs913T fibroblasts were stimulated with a Ca²⁺ ionophore (I), PMA (P), or PMA plus a Ca²⁺ ionophore for 3 h in the presence of cycloheximide (10 μg/ml). When used, CsA was added to the culture 30 min prior to cell activation. ∅, unstimulated. (B) Jurkat T cells and Hs913T fibroblasts were transiently transfected with Rsa-luc (containing the EGR3 sequence from −2952 to +86), displayed schematically at the top (tss, transcription start site). After transfection, the cells were left unstimulated (unst) or were stimulated with a Ca²⁺ ionophore (I) and/or PMA (P) for 3 h in the presence or absence of CsA. The luciferase activity of cells stimulated with PMA plus the Ca²⁺ ionophore was set at 100%. Bars represent the means ± standard errors of the means of three independent experiments. Each SI (SI = luciferase activity in the stimulated culture/luciferase activity in the unstimulated culture) is the average of three independent experiments.

Identification of a T-cell-specific regulatory region (−226 to −99) within the EGR3 promoter/enhancer.

To identify the region conferring T-cell-specific two-signal induction and CsA sensitivity to the EGR3 gene, successive 5′ deletion constructs were generated and functionally compared in Jurkat T cells and Hs913T fibroblasts (Fig. 3A). In T cells, deletion of a large fragment between −2952 and −777 reduced overall luciferase activity by about 60% but did not have major effects on the two-signal inducibility and CsA sensitivity of the EGR3 promoter (Fig. 3A, left panel). Successive deletions of the promoter regions from −777 to −226 resulted in minor changes of inducible luciferase activity. In contrast, deletion of the promoter sequence from −226 to −99 resulted in a dramatic loss of both constitutive and inducible activity (stimulation index [SI] for Bsm-luc = 7.8; SI for Aat-luc = 2.6). By further deleting the EGR3 sequence to base pair −37, all of the residual luciferase activity was lost. As expected, stimulation of transfected cells with a Ca²⁺ ionophore or PMA alone did not markedly influence the inducibility of the constructs in T cells.

In contrast to the results obtained with Jurkat T cells, the deletion of nucleotides −2952 to −895 dramatically reduced PMA-inducible promoter activity in fibroblasts (Fig. 3A, right panel). On the other hand, deletion of EGR3 sequences from −226 to −99, which led to a substantial loss in inducible promoter function in T cells (see above), had virtually no effect in fibroblasts. As predicted, none of the constructs in fibroblasts was influenced by the Ca²⁺ ionophore, and CsA had no effect.

Comparative transfection studies with internal deletion constructs of the EGR3 gene demonstrated that deletion of the sequences from −229 to −37 (ΔBsm-Sma) reduced promoter activity by about 90% both in T cells and fibroblasts (Fig. 3B). A similar result was obtained by deleting the subfragment from −104 to −37 (ΔAat-Sma), indicating an important role for this sequence in EGR3 gene expression in both cell types. Interestingly, upon the deletion of the subfragment from −229 to −99 (ΔBsm-Aat), promoter activity was retained in fibroblasts but not in T cells, where a dramatic decrease in activity was observed (about 90%). The data were thus in agreement with the results obtained with the 5′ deletion constructs (Fig. 3A) and underscored the importance of the −226 to −99 subregion in the T-cell-specific, two-signal induction of the EGR3 gene. No negative effect on promoter function was observed when sequences from −611 to −226 (ΔEco-Bsm) or from −896 to −614 (ΔMsc-Eco) were deleted (Fig. 3B).

Identification of a T-cell-specific, CsA-sensitive 27-bp regulatory element.

To identify the critical elements within the region from −226 to −37, we first searched this sequence for homology to binding sites for known TF (Fig. 1). The sequence between −134 and −95 contained a conspicuous 11-bp direct repeat and was reminiscent of a tumor necrosis factor alpha (TNF-α) promoter element, in which adjacent NF-κB- and cyclic AMP response element (CRE)-like sites participated in the activation and CsA-sensitive induction of the TNF-α gene (44). When a construct with four copies of the sequence from −134 to −95 in front of an SV40 promoter was analyzed in transient transfection experiments with Jurkat T cells, it was found that luciferase activity was induced more than eightfold by stimulation with PMA plus a Ca²⁺ ionophore and was fully blocked by CsA (Fig. 4). In contrast to what was found with the TNF-α model, a multimerized subregion from −122 to −95 containing both the NF-κB-like element and the CRE failed to induce promoter activity (Fig. 4). However, the construct with four copies of the subregion from −134 to −108 lacking the CRE but including the upstream half of the direct repeat was significantly induced after the transfected cells were stimulated with PMA plus a Ca²⁺ ionophore (Fig. 4). This result indicated that nucleotides critical for two-signal induction and CsA sensitivity are located between positions −134 and −122, upstream of the putative NF-κB-like element. Consistent with a T-cell-specific role for the sequence from −134 to −95 in EGR3 gene induction, neither of the stimuli applied, either alone or in combination, was able to activate the heterologous promoter in fibroblasts (Fig. 4).

FIG. 4.

Characterization of a minimal regulatory element conferring inducibility upon an unrelated promoter in T cells but not in fibroblasts. The sequences from the EGR3 promoter region displayed at the bottom of the figure were cloned in multiple copies upstream of an SV40/luciferase reporter gene as described in Materials and Methods. After transfection of the constructs into Jurkat T cells or Hs913T fibroblasts, cells were left unstimulated (unst) or were stimulated with a Ca²⁺ ionophore (Iono) and/or PMA for 3 h in the absence or presence of CsA. The results were corrected for transfection efficiency and are expressed as fold induction (luciferase activity in the stimulated culture/luciferase activity in the unstimulated control). All transfections were repeated three times, and results are shown as means ± standard errors of the means. The CRE is boxed; the 11-bp direct repeat is underlined; and the NF-κB-like motif and the site cleaved by restriction enzyme AatII are indicated.

Proteins binding to the 27-bp regulatory element in T cells are missing in fibroblasts.

When EMSAs were performed with nuclear extracts from Jurkat T cells and Hs913T fibroblasts, the inducible region from −134 to −108 (27-bp element) formed several distinct complexes with nuclear proteins from unstimulated Jurkat T cells (Fig. 5). Interestingly, complexes II and III were significantly enhanced after Jurkat T cells were stimulated with PMA plus a Ca²⁺ ionophore, and the formation of complexes I, II, and III was strongly inhibited by treatment of the cells with CsA (Fig. 5, lanes 2 to 4). In contrast to the results obtained with Jurkat T cells, inducible complexes II and III could not be detected in nuclear extracts of fibroblasts (Fig. 5, lanes 5 to 7). The lack of these two complexes appears not to be a consequence of incomplete stimulation of the fibroblasts, since NF-κB was efficiently induced (Fig. 5, lanes 11 and 12). CsA-sensitive complex I was present in fibroblasts as it was in Jurkat T cells, albeit at significantly lower levels (Fig. 5, lanes 5 to 7; see also Fig. 6 and 7).

FIG. 5.

Comparative EMSAs to demonstrate the formation of T-cell-specific DNA-protein complexes with the 27-bp regulatory element. EMSAs were performed with nuclear extracts from Jurkat T cells (lanes 2 to 4, 9, and 10) or Hs913T fibroblasts (lanes 5 to 7, 11, and 12), which were left unstimulated (−) or were stimulated (+) with PMA plus a Ca²⁺ ionophore for 30 min in the presence or absence of CsA. Probes used: oligonucleotide spanning the 27-bp element (EGR3 sequence from −134 to −108) (lanes 1 to 7) and murine Igκ enhancer NF-κB site, 5′ TCGAGGGGACTTTCCGAG 3′ (the NF-κB-binding motif is underlined) (lanes 8 to 12). The specific complexes are numbered on the left. The fastest-migrating complex (ns) appears to be nonspecific since, as shown by the competition study (Fig. 6), it was only partially inhibited by the 27-bp element itself; neither of the competitors used affected its formation.

FIG. 6.

Binding specificity of the DNA-protein complexes formed in T cells and fibroblasts analyzed by competition experiments. EMSAs were performed with nuclear extracts from T cells and fibroblasts stimulated with PMA plus a Ca²⁺-ionophore for 30 min in the absence of a competitor (lanes 2 and 9) or in the presence of 200-fold molar excesses of the specific competitors indicated (lanes 3 to 8 and 10 to 15), with the 27-bp element as a probe. Competitors used (binding motifs of TF are underlined): oligonucleotide spanning the 27-bp element of the EGR3 gene (27-bp; self-competition), the human distal IL-2 NF-AT site (5′ TCGAGGAGGAAAAACTGTTTCATACAGAAGGCG 3′; NF-AT), the human metallothionein AP-1 site (5′ TCGAGTGACTCAGCGCGG 3′; AP-1), the rat somatostatin CRE site (5′ TCGAGGCTGACGTCAGAGAG 3′; CRE), the murine Igκ enhancer NF-κB site (Ig κB; see Fig. 5), and the murine IL-2 NF-κB site (5′ TCGAGAGGGATTTCACCTG 3′; IL-2 κB). ns, nonspecific complex.

Characterization of the nuclear complexes formed with the 27 bp-element.

To identify the TF contained within the DNA-protein complexes formed with the 27-bp regulatory element, EMSAs with specific unlabeled competitor oligonucleotides were performed. In Jurkat T cells complexes I, II, and III were efficiently competed by the addition of a 200-fold molar excess of an oligonucleotide containing the human distal IL-2 NF-AT site (Fig. 6, lane 4). Competition was specific, since a 200-fold molar excess of an oligonucleotide containing an AP-1- or a CRE-binding site did not influence complex formation (Fig. 6, lanes 5 and 6). The murine immunoglobulin κ (IgGκ) enhancer NF-κB site, which had been shown to bind NF-AT factors via its NF-AT core binding sequence, TTCC (24), also competed all three complexes very efficiently, whereas the murine IL-2 NF-κB site lacking such a sequence was without effect (Fig. 6, lanes 7 and 8). Complex I obtained with nuclear extracts from fibroblasts showed the same competition pattern as that which it showed in Jurkat T cells (Fig. 6, lanes 9 to 15).

NF-ATp and NF-ATc are major components of CsA-sensitive complexes I, II, and III.

The competition studies implied that NF-AT proteins bind to the 27-bp element. We therefore performed EMSAs with antibodies reactive with NF-AT family members. A peptide-specific NF-ATp antiserum clearly supershifted complexes II and III but did not affect the migration of complex I (Fig. 7A, lane 3). The same result was obtained with two additional antisera reactive either with recombinant NF-ATp (25) or an amino-terminal peptide fragment 67.1 of NF-ATp (13) (data not shown). The addition of a monoclonal anti-NF-ATc antibody resulted in the disappearance of complex I and in an alteration of the mobility of complex II, whereas the formation of complex III was not affected (Fig. 7A, lane 4). The addition of both antibodies resulted in a nearly complete supershift of complexes I, II, and III, demonstrating that NF-AT proteins or closely related factors are absolutely required for the formation of all three complexes (Fig. 7A, lane 5). As in Jurkat T cells, the formation of complex I in fibroblasts was unaffected by the NF-ATp antiserum but was fully blocked by the monoclonal anti-NF-ATc antibody (Fig. 7A, lanes 6 to 9, and Fig. 7B, which shows a longer exposure of lanes 6 to 9). Control experiments demonstrated that the supershifts were specific, since the antibodies did not react with an NF-κB protein complex and did not unspecifically bind to the DNA probe (data not shown).

Determination of nucleotides critical for the generation of complexes I, II, and III with the 27-bp element.

In EMSA experiments performed with an oligonucleotide containing a 6-bp mutation within the upstream half of the direct repeat (mut-rep), neither complex II nor complex III could be detected and the formation of complex I was greatly diminished, indicating that at least some of the nucleotides mutated are absolutely required for formation of these complexes and hence for the binding of NF-ATp and NF-ATc (Fig. 8A; compare lanes 1 and 2). In contrast, an oligonucleotide containing a 7-bp mutation within the NF-κB-like element (mut-κB) was still able to form all the complexes seen with the wild-type 27-bp element, demonstrating that the mutated nucleotides are dispensable for factor binding (Fig. 8A; compare lanes 1 and 3). As in Jurkat T cells, formation of complex I in fibroblasts was abolished by mut-rep but was not affected by the κB mutation (Fig. 8A, lanes 4 to 6).

FIG. 8.

Correlation of NF-AT binding to the 27-bp element with EGR3 promoter activity in T cells and fibroblasts. (A) EMSAs were performed with nuclear extracts from T cells and fibroblasts stimulated with PMA plus a Ca²⁺ ionophore for 30 min by using radiolabeled oligonucleotides spanning the 27-bp element or mutated variants thereof (displayed at the bottom) as probes. Mutated bases are in lowercase. The direct repeat is underlined (the downstream half of the direct repeat is not complete in this sequence). The NF-κB-like binding sequence is marked by arrows. (B) Jurkat T cells and Hs913T fibroblasts were transfected with the wild-type EGR3 promoter construct (Rsa-luc) or a construct in which the NF-AT site had been mutated (Rsa-luc mut-rep). After stimulation the cells were assayed as described in the legend for Fig. 2B. unst, unstimulated; Iono, Ca²⁺ ionophore.

NF-AT binding to the 27-bp element is essential for inducible EGR3 gene expression in T cells but not in fibroblasts.

To determine the functional role of the NF-AT binding site for EGR3 gene regulation, mut-rep was introduced into the wild-type −2952 EGR3 promoter by site-directed mutagenesis. In Jurkat T cells, mutation of the NF-AT binding site reduced EGR3 promoter activity inducible by PMA plus a Ca²⁺ ionophore to the level observed with PMA alone, and this residual activity was insensitive to CsA (Fig. 8B). These experiments clearly demonstrated that the same nucleotides responsible for NF-AT binding in vitro are responsible for two-signal-dependent, CsA-sensitive EGR3 gene expression in vivo. In contrast to what was found for Jurkat T cells, mutation of the NF-AT binding site was without effect in fibroblasts, thus also demonstrating that the weak NF-ATc-containing complex I formed with the 27-bp element in fibroblasts does not significantly contribute to the inducible EGR3 gene expression in these cells (Fig. 8B).

Overexpression of NF-ATp but not NF-ATc renders EGR3 gene expression CsA sensitive in fibroblasts.

In order to test the hypothesis that insufficient NF-AT complex formation at the 27-bp element is responsible for the silence of this two-signal regulatory region in fibroblasts, we investigated the influence of ectopically expressed NF-ATp and NF-ATc on the inducibility of the wild-type and the mutated EGR3 promoters. Overexpression of NF-ATp in fibroblasts increased promoter activity in response to stimulation with PMA plus a Ca²⁺ ionophore more than threefold in a CsA-sensitive manner, whereas the responses to PMA or the Ca²⁺ ionophore alone were only slightly affected (Fig. 9A). Thus, ectopic expression of NF-ATp could restore the two-signal phenotype and CsA sensitivity of EGR3 expression in fibroblasts. These effects were specific and could clearly be attributed to the NF-AT binding site in the 27-bp element, since overexpressed NF-ATp was unable to enhance the activity of an EGR3 promoter with the NF-AT site mutated (Fig. 9A). In contrast, overexpression of NF-ATc failed to restore the two-signal response in fibroblasts (Fig. 9A). Instead, ectopic NF-ATc increased EGR3 promoter activity more than twofold after stimulation with PMA alone, and this increase was partially independent of the NF-AT binding site within the 27-bp element. In Jurkat T cells, overexpression of NF-ATp and NF-ATc resulted in increased promoter activity, by approximately 2.7-fold for NF-ATp and approximately 2-fold for NF-ATc. The enhancing effect of both overexpressed NF-ATp and NF-ATc was dependent on the NF-AT binding site, since it was not observed with the mutated EGR3 promoter (Fig. 9B).

FIG. 9.

Transactivation of the EGR3 promoter by NF-ATp or NF-ATc in fibroblasts and T cells. EGR3 promoter construct Rsa-luc or mutant variant Rsa-luc mut-rep was cotransfected into Hs913T fibroblasts (A) or Jurkat T cells (B) with an NF-ATp or an NF-ATc expression plasmid, as described in Materials and Methods. Control transfections were performed with the empty expression vector. After transfection of the constructs the cells were left unstimulated (unst) or were stimulated with a Ca²⁺ ionophore (Iono) and/or PMA for 3 h in the absence or presence of CsA. After correction for transfection efficiency, data were standardized to the unstimulated cells transfected with the empty expression vector. All transfections were repeated four or five times, and results are shown as means ± standard errors of the means.

DISCUSSION

Genes whose expression is fully sensitive to CsA can be broadly subdivided into two groups. The first group encompasses genes which are only expressed in the lymphoid system (e.g., those for IL-2, gamma interferon, and TRAP/CD40 ligand) (6, 11, 34, 37, 43). Genes of the second group can be induced within and also outside of the lymphoid system: within the lymphoid system, they are suppressed by CsA; outside of the lymphoid system their expression is CsA resistant (e.g., genes for IL-8, TNF-α, Fas ligand, and PILOT/EGR3) (10, 17, 18, 23, 26, 28, 30). This differential sensitivity of the second group of genes allows a systematic study of the molecular mechanism underlying the tissue specificity of CsA. Such an analysis is attractive, since an understanding of the mechanism of action of this immunosuppressive drug will intrinsically provide information on the lymphoid-cell-specific aspects of gene regulation. The present report is based on the differential expression pattern of the EGR3 gene in T cells versus that in fibroblasts, which we have reported earlier (23).

In order to define the differences at the gene regulation level, comparative EGR3 promoter studies of CsA-sensitive Jurkat T cells and CsA-resistant Hs913T fibroblasts were performed, since these two cell lines truly represent the regulation of EGR3 in the respective primary cells. The first series of experiments determined that the EGR3 promoter region from −226 to −99 plays a key role in the regulation of EGR3 in T cells, because deletions of this region strongly reduced inducible promoter activity in these cells. Within this region we identified a critical 27-bp element (−134 to −108) which could confer two-signal inducibility and also CsA sensitivity not only to the EGR3 promoter but also to a heterologous promoter, as shown when it was functionally tested in the Jurkat line. These experiments clearly determined that the isolated 27-bp element contains all necessary information for the characteristic two-signal expression pattern of EGR3 in T cells. In clear contrast to the findings with T cells, we observed that the promoter region from −226 to −99 is not required for the expression of EGR3 in fibroblasts, and the isolated 27-bp element was found to be inactive when functionally tested in this cell type. Instead, a different promoter region (from −2952 to −895), which is not utilized in T cells, turned out to be essential for PMA-induced expression of EGR3 in fibroblasts.

Several lines of evidence clearly indicate that members of the NF-AT family of TF account for the T-cell-specific usage of the 27-bp element and that this regulatory complex is the main target for CsA. In T cells, (i) the 27-bp element forms major complexes with NF-ATc and NF-ATp proteins; (ii) the binding of NF-ATc and NF-ATp to the 27 bp-element is fully abrogated in the presence of CsA; and (iii) the introduction of a mutation into the 27-bp element prevents the binding of the NF-AT proteins and concomitantly abolishes the two-signal-dependent, CsA-sensitive regulation of the entire EGR3 promoter in T cells. When the same experiments were performed with fibroblasts, (i) binding studies with the 27-bp element revealed only a faint complex reactive with an NF-ATc-specific antibody and the binding of NF-ATp could not be detected in any of the experiments performed and (ii) the mutation of the 27-bp element was functionally silent in these cells.

Our functional and DNA-binding studies suggested that the inactivity of the T-cell-specific 27-bp element in fibroblasts was caused by insufficient expression levels of NF-AT proteins. This assumption was confirmed by cotransfection studies performed to determine the relative roles of NF-ATp and NF-ATc in this system. The overexpression of NF-ATp restored the two-signal phenotype and CsA sensitivity of EGR3 promoter induction in fibroblasts, but only if the NF-AT binding site remained intact. On the other hand, enhancement of EGR3 promoter activity in fibroblasts achieved by ectopic expression of NF-ATc could not be inhibited by CsA. This observation is in agreement with our finding that Hs913T fibroblasts already contain an NF-ATc binding complex, and yet EGR3 gene expression is not influenced by CsA. Why overexpression of NF-ATc failed to compensate for the missing NF-ATp in Hs913T cells remains unclear at present. Our findings thus indicate that, at least in certain cell types, NF-ATp expression is a prerequisite for the CsA-sensitive, two-signal induction of EGR3.

Whether our observation of the differential usage of a specific NF-AT binding region can be extrapolated to all genes whose expression is CsA sensitive in T cells but CsA insensitive outside of the lymphoid system remains to be determined. Little information in this regard is available, since studies similar to ours have not been systematically undertaken to date. Interestingly, the few available data suggest that our results concerning the regulation and CsA sensitivity of EGR3 are representative of a number of CsA-sensitive genes. The data from two reports, one dealing with the PMA-inducible expression of IL-8 in a fibrosarcoma cell line and the other analyzing the PMA-plus-Ca²⁺ ionophore-inducible, FK506-sensitive expression of IL-8 in Jurkat T cells, allow the conclusion that the promoter elements necessary for the activation of the IL-8 gene in nonlymphoid versus lymphoid cells are partly different (27, 31). However, the role of NF-AT proteins in the differential regulation of the IL-8 gene was not investigated in these studies. In another publication, NF-AT binding sites were identified as critical for the CsA-sensitive expression of the Fas ligand in Jurkat T cells but were found to be irrelevant for the constitutive expression of the Fas ligand in Sertoli cells (18). Finally, when the differential regulation of the TNF-α gene in dendritic cells (FK506 resistant) and mast cells (FK506 sensitive) was investigated, it was found that an AP-1 site adjacent to an indispensable κ3 promoter element is required for the binding of the NF-AT protein(s) and for the induction of the TNF-α promoter in mast cells but not in dendritic cells, where no binding of NF-AT was observed (35). Although not accompanied by functional analyses, this study thus also supports the central role of NF-AT proteins in the differential regulation of certain genes.

Taken together, our data and the results of other groups indicate that certain genes partially utilize different promoter elements when expressed within or outside of the lymphoid system. In particular, these genes strictly require specialized promoter elements functionally interacting with certain members of the NF-AT protein family for expression in lymphoid cells. CsA interferes with this critical regulatory unit and thus suppresses the induction of these genes in lymphoid tissues. Other tissues apparently do not express NF-AT proteins capable of functionally interacting with such lymphoid cell-specific promoter regions, and the same genes utilize instead other CsA-insensitive regulatory elements for transcription. The observation of a differential usage of specialized NF-AT binding promoter elements thus provides one molecular mechanism for the tissue specificity of the immunosuppressive drug CsA at the gene transcription level.