Cyclosporin A-Sensitive Transcription Factor Egr-3 Regulates Fas Ligand Expression

Abstract

Activation-induced transcriptional upregulation of the ligand for Fas (FasL) and the resulting apoptosis of Fas-bearing cells constitute essential steps in a host of normal and pathological processes. Here we describe an activation-inducible cis-acting regulatory element in the fasL promoter that is required for gene expression. Oligonucleotide competition and antibody supershift analyses identified two activation-induced DNA-binding species: Egr-1 (NGFI-A, krox-24, zif268, TIS-8), a transcription factor that has been implicated in growth, differentiation, and apoptosis; and Egr-3 (PILOT), a transcription factor of no previously known function. Activation-induced expression of Egr-3, like that of FasL, was inhibited by cyclosporin A, whereas expression of Egr-1 was unaffected. Transient expression of Egr-3 alone increased fasL promoter activity in a cyclosporin A-insensitive manner, whereas expression of Egr-1 had little effect. Moreover, endogenous fasL mRNA was induced in nonlymphoid cells by forced expression of Egr-3 in the absence of any other stimulus. These studies identify a critical Egr family-binding site in the fasL promoter and demonstrate that activation-induced Egr-3, but not Egr-1, directly upregulates fasL transcription in response to activating stimuli.

Fas (APO-1/CD95)-mediated apoptosis plays a key role in regulating the ability of the immune system to respond to an antigenic challenge, and in recent years it has become clear that the Fas-induced death pathway is important in normal and pathological physiology. For example, activation-induced upregulation of FasL and its interaction with Fas account for downregulation of immune responses and for elimination of T cells expressing self-reactive T-cell receptors (TCRs) (12). Direct evidence for a role of Fas and FasL in maintenance of peripheral lymphocyte homeostasis is provided by the lpr and gld mice, which have loss-of-function mutations in the genes encoding Fas (64) and FasL (39, 59), respectively. These mice develop fatal autoimmunity and lymphadenopathy as a result of the accumulation of a population of TCR⁺ CD4⁻ CD8⁻ peripheral T cells. These accumulated T cells display features indicative of prior activation and are refractory to activation-induced cell death in vitro (62). Mutations in Fas have been described in patients with a similar lymphoproliferative syndrome (15, 50). Inducible expression of FasL is also important in immune effector functions, such as killing of Fas⁺ targets by FasL⁺ CD8⁺ cytotoxic T cells and natural killer cells (2, 28, 29). FasL has been implicated in the maintenance of tissue-specific immune privilege, a situation in which a tissue is not rejected even when transplanted across a major histocompatibility complex barrier (5, 46). Two such tissues, the testis and the anterior chamber of the eye, contain cells that constitutively express FasL, and disruption of FasL function in gld mice abrogates their immune-privileged status (5, 18, 19).

Aberrant expression of Fas and FasL has been implicated in diseases other than the lymphoproliferative syndromes. For example, some tumors express FasL, which protects them from the immune response by inducing the apoptosis of responding T and natural killer cells (21, 48, 55). It has also been proposed that aberrant interleukin-1 (IL-1)-induced expression of Fas causes apoptosis of thyrocytes, which constitutively express FasL, leading to the development of Hashimoto’s thyroiditis (17). Fas-dependent killing of pancreatic islet β cells is required for development of autoimmune diabetes in the NOD mouse (10). Increased constitutive levels of Fas on lymphocytes in human immunodeficiency virus-infected persons and its interaction with upregulated FasL cause the ex vivo apoptosis of these cells, implicating this pathway as a mechanism for the immunodeficiency of AIDS (4, 30, 65, 66). Thus, the control of Fas-FasL interactions by regulation of each molecule’s expression is an essential feature of normal and pathological physiology.

In T cells, fasL mRNA expression is induced by TCR-mediated activation or by stimuli, such as a phorbol ester plus a Ca²⁺ ionophore, that bypass the TCR (1, 7). Induction of fasL mRNA is prevented by cyclosporin A (CsA), an immunosuppressive drug that inhibits calcineurin activity and activation of the NF-AT transcription factor, which is essential for production of IL-2 (1, 8, 26). In fact, NF-AT has been implicated as a direct regulator of fasL transcription by binding positive regulatory elements upstream of the fasL coding sequence (33, 34). In this study, we analyze the fasL enhancer-promoter region to determine what cis-acting elements are required for fasL mRNA upregulation. These studies identify an 8-bp sequence upstream of the initiator codon that binds members of the Egr family of transcription factors and confers most of the activation inducibility of fasL promoter reporter constructs. The data indicate that one family member in particular, Egr-3, mediates activation of fasL transcription through this single response element.

MATERIALS AND METHODS

Cell lines and reagents.

2B4.11 is a murine T-cell hybridoma specific for peptide 81-104 of pigeon cytochrome c presented by I-E^k (22) and was maintained in RPMI 1640 (Biofluids Inc., Rockville, Md.) supplemented with 4 mM glutamine, 50 μM β-mercaptoethanol, 100 U of penicillin per ml, 150 μg of gentamicin per ml, and 10% heat-inactivated fetal calf serum (complete medium). HeLa (human cervical carcinoma) cells were cultured in Dulbecco’s modified Eagle’s medium with the above supplements. 145-2C11 (2C11), a hamster anti-mouse CD3-ɛ monoclonal antibody (36), and OKT3, an anti-human CD3 monoclonal antibody (32), were purified from culture supernatant by protein A chromatography. CsA was obtained from Sandoz. To generate human T-cell blasts, human peripheral blood mononuclear cells were isolated by density centrifugation with the use of a lymphocyte separation medium (Biofluids) cultured in complete medium for 2 days with 2 ng of phorbol myristate acetate (PMA) (Sigma) per ml and 1 μg of ionomycin (Sigma) per ml, washed, and cultured for an additional day with 10 U of recombinant human IL-2 (Cetus/Chiron, Emeryville, Calif.) per ml, provided by J. Wunderlich (National Institutes of Health).

Plasmids.

A 1.2-kb human fasL upstream genomic region fragment was isolated with the Promoterfinder PCR-based kit from Promega Corp. (Madison, Wis.) and cloned into the luciferase reporter construct pGL3 (Promega) to create the construct 1.2-FasL-GL3. The promoter constructs containing fasL sequences from −511, −370, −305, and −225 at the 5′ end were made with the use of the EcoRI, StuI, PstI, and PvuII sites. Other 5′-truncated fasL promoter constructs (−212, −204, −191, −161, and −137) and the plasmids −214:Δ−206/−138 (R-8-mer) and −218:Δ−210/−138 (L-8-mer) were made by PCR. The plasmids −220:Δ−204/−138 (16-mer), −225:Δ−200/−138:m−214/−211 (m16-mer) and −511:m−214/−211 were made by the overlap PCR technique (25). All PCR-based inserts were cloned into 1.2-FasL-GL3 via SmaI and HindIII. The expression plasmids encoding NGFI-A (Egr-1) and Egr-3 and the luciferase reporter construct A2ProLuc, containing two GCGGGGGCG (EBS-1) motifs upstream of the minimal prolactin promoter, are reported elsewhere (51).

Transient transfection assays.

In transient reporter assays of anti-CD3 responsiveness, human T-cell blasts were transiently transfected with 5 μg of reporter construct DNA by electroporation into 5 × 10⁶ cells as described elsewhere (45). Cells were diluted in medium containing 10 U of recombinant IL-2 per ml and distributed as duplicate or triplicate samples in 250-μl aliquots in 96-well plates. To cross-link CD3, wells were coated with 10 μg of the purified monoclonal antibodies OKT3 for human cells or 2C11 for mouse cells per ml. 2B4.11 cells were transiently transfected with 5 μg of reporter construct in the presence of 0.15 mg of DEAE-dextran per ml (3). In cotransfection experiments, duplicate aliquots of 10⁶ 2B4.11 cells were electroporated with 2.5 μg of luciferase reporter plasmid, 0.5 μg of cytomegalovirus-driven β-galactosidase reporter plasmid, and 5 μg of expression plasmid at 960 μF at 220 V in a volume of 200 μl. The plasmid pCB6⁺ (6) was used to control for the cytomegalovirus promoter-driven expression plasmids. Error bars represent the standard deviations of duplicate fold inductions. In reporter assays, triplicate 200-μl cultures of HeLa cells were transfected with 100 ng of luciferase reporter plasmid, 20 ng of β galactosidase reporter plasmid, and 150 ng of expression plasmid by the calcium phosphate technique (3). Error bars represent standard errors of the means. In reverse transcriptase PCR (RT-PCR) assays, 2 ml of cultures of HeLa cells were transfected with 3 μg of expression plasmid.

Gel shift assays.

Whole-cell extracts were prepared by resuspending phosphate-buffered saline-washed cells in 10 mM HEPES, pH 7.9, 400 mM KCl, 10% glycerol, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride and subjecting them to three rounds of freezing on dry ice followed by rapid thawing in a 37°C water bath. Extracts were obtained from the supernatants after 5 min of centrifugation at 14,000 × g at 4°C. Binding reactions were carried out at 22°C for 30 min by combining 3-μl extracts with 12 μl of binding buffer (10 mM HEPES, pH 7.9, 10% glycerol, 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride) containing 1 μg of poly-dI-dC and 0.02 pmol of ³²P-end-labeled double-stranded oligonucleotide probe. Three picomoles (a 150-fold excess) of unlabeled double-stranded competitor oligonucleotide was added where indicated. In antibody supershift assays, 1 μl (1 μg) of antiserum specific for Egr family member proteins (Santa Cruz Technologies Inc., Santa Cruz, Calif.) was included in the binding reaction. The antisera have been determined by the manufacturer to be specific within the Egr family of proteins. The complexes and the unbound probe were separated on 5% polyacrylamide gels.

RT-PCR.

Total RNA was isolated by using the Trizol denaturant (Life Technologies, Grand Island, N.Y.). cDNA was synthesized by using random hexamers and avian myeloblastosis virus reverse transcriptase (Invitrogen, San Diego, Calif.). The PCR primers specific for human fasL were as follows: 5′-AAGAAGAGAGGGAACCACAGC AC-3′ (sense strand) and 5′-TCACTCCAGAAAGCACAATTC-3′ (antisense strand). The PCR primers specific for human gapdh were as follows: 5′-AGGTCGGAGTCAACGGATTT-3′ (sense strand) and 5′-CAGCAGAGGGGGCAGAGATG-3′ (antisense strand). One microliter of cDNA was amplified in a 20-μl PCR mixture in buffer containing 2 mM MgCl₂ and 0.5 U of Amplitaq Gold (Perkin-Elmer, Branchburg, N.J.). PCR products derived from an incubation of 12 min at 95°C , followed by 25 cycles of 1 min at 95°C, 1 min at 62°C, and 1 min at 72°C (fasL) and 15 cycles of 1 min at 95°C, 1 min at 56°C, and 1 min at 72°C (gapdh) were separated on 2% agarose gels and transferred to Hybond membranes (Amersham Corp., Arlington Heights, Ill.). The membranes were probed with ³²P-labeled cDNAs encoding the extracellular part of human FasL or a PstI fragment of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (67).

Northern blot analysis.

Total RNA (3 μg) was separated by electrophoresis through a 1.5% agarose gel containing 6% formaldehyde and buffered with 3-morpholinopropanesulfonic acid (MOPS) (Quality Biological, Inc., Gaithersburg, Md.). After transfer to a Genescreen membrane (NEN, Boston, Mass.), RNA was covalently bound by UV cross-linking, and hybridization with ³²P-labeled cDNA probes was carried out at 65°C in 0.5 M sodium, 7% sodium dodecyl sulfate (SDS), and 1 mM EDTA and buffered to pH 7.2 with phosphate (11). The cDNA encoding the extracellular part of mouse FasL (67) and gapdh were used as probes. Final washes were performed at 60°C in 80 mM sodium and phosphate buffered as before, with 1% SDS and 1 mM EDTA. After exposure to detect fasL, the membrane was stripped by boiling in 0.1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) and 1% SDS and probed for gapdh.

RESULTS

Identification of a TCR-responsive element in the fasL promoter.

To identify DNA elements that control the regulation of fasL transcription, regions of the human fasL gene 5′ of the translation start site were used to drive the expression of a luciferase reporter in the T-cell hybridoma 2B4.11, which has been shown to undergo activation-induced apoptosis owing to fasL upregulation (7, 27, 67). In these transient transfections, stimulation with anti-CD3 antibodies induced transcription of a reporter construct driven by the 511-bp sequence of the fasL promoter (Fig. 1A). The level of inducible activity achieved was comparable to that obtained with constructs containing as much as 3.5 kb of the region 5′ of fasL (data not shown), suggesting that this 511-bp sequence contains the major 5′ regulatory element(s) for this gene. Whereas truncation of the fasL promoter to nucleotide −225, which deletes the putative NF-AT binding site at −275 (33), had little inhibitory effect, truncation beyond this point abrogated activation-induced transcriptional activity. Step-by-step deletion in this region revealed that the sequence between −225 and −212 is indispensable for activation-induced fasL promoter activity (Fig. 1B). Transcriptional regulation by the fasL promoter was also assessed in normal preactivated human peripheral blood cells (Fig. 1C). As with the murine T-cell hybridoma, truncation of the 511-bp promoter to nucleotide −225 had no effect, but truncation to −212 prevented activation-induced luciferase expression.

FIG. 1.

Deletional mapping of the region in the fasL enhancer-promoter required for TCR-induced transcriptional activity. (A and B) 2B4.11 cells were transiently transfected with reporter plasmids containing the indicated fasL region and stimulated with immobilized anti-CD3 antibodies. Luciferase activity in whole-cell lysates was determined 13 h later. The results are based on the averages of the luciferase light units obtained from duplicate samples from each transfection analyzed. Numbering is based on the translational initiation codon of the fasL gene. Similar results were obtained in six (A) and five (B) independent experiments. (C) Human T-cell blasts were electroporated with the indicated reporter constructs and immediately stimulated with immobilized anti-CD3 antibodies. Similar results were obtained in three independent experiments. The mean luciferase light units of unstimulated cultures, in order of appearance in the figure, were as follows: 3035, 1118, 2922, 2306, and 2630 (A); 3915, 2208, 2841, and 3672 (B); and 699, 949, 148, and 166 (C).

To determine the extent of the region sufficient to confer TCR inducibility of fasL transcription, short sequences from the critical 5′ region were placed adjacent to nucleotide −137 (which includes the TATA box starting at −128 [60] and is here referred to as the fasL minimal promoter). One such construct, containing nucleotides −220 to −205 (16-mer) (Fig. 2A, construct 1), had the same activity as the uninterrupted 0.5-kb promoter (construct 4), whereas the −137 construct alone (construct 5) was inactive. Thus, the regions upstream of −220 and downstream of −205 contain little if any regulatory activity. Mutation of the four nucleotides (underlined) from −214 to −211 completely abolished the activity of the 16-mer (construct 2). Importantly, mutation of this region resulted in complete loss of activity of the full 511-bp fasL promoter as well (construct 3). Further truncation of the 16-mer element in the context of the fasL minimal promoter revealed that the eight remaining nucleotides (from −214 to −207) contained activity similar to the wild-type (−511) fasL promoter (Fig. 2B). Therefore, the 8-bp segment from −214 to −207, termed the Fas ligand regulatory element (FLRE), is both necessary and sufficient for most of the TCR-mediated induction of the fasL promoter.

FIG. 2.

Determination of the minimal region required for TCR responsiveness of the fasL promoter-enhancer. (A) Four nucleotides (GTGG, underlined) in the center of the 16-bp sequence from −220 to −205 were mutated to CACC in construct 2 and to TGTT in construct 3. Transfected 2B4.11 cells were stimulated for 15 h with immobilized anti-CD3 antibodies, and luciferase activity was measured. (B) The required four nucleotides (see panel A) and the adjoining four 3′ (−214 to −207: 8mer A) or 5′ (−218 to −211: 8mer B) nucleotides were appended to the fasL minimal promoter at position −137 and compared with the wild-type fasL promoter extending to −511 for inducibility in 2B4.11 cells after 13 h of stimulation with immobilized anti-CD3 antibodies. Similar results were obtained in five independent experiments. The mean luciferase light units of unstimulated cultures, in order of appearance in the figure, were as follows: 12894, 11374, 6821, 9894, 8740, and 1862 (A); and 2611, 3113, 2095, and 1344 (B).

Activation of transcription by the FLRE is CsA sensitive.

Activation-induced upregulation of FasL expression is prevented by the immunosuppressive drug CsA (1, 8, 43). The ability of CsA to inhibit FLRE-dependent activation was tested (Fig. 3). CsA inhibited TCR-mediated activation of the 511-bp fasL promoter construct, confirming that the region studied contains the entire TCR-responsive region of the fasL gene. Moreover, CsA was equally effective at inhibiting activation of the 8-bp FLRE, consistent with a requisite role for this regulatory element in activation-induced fasL upregulation.

FIG. 3.

FLRE-mediated transcription is sensitive to CsA. 2B4.11 cells that had been transiently transfected with the indicated fasL-dependent reporter plasmids were distributed into wells that were untreated or coated with anti-CD3 antibodies. CsA (100 ng/ml) was added to half of these cultures.

Analysis of FLRE-binding proteins.

Electrophoretic mobility shift assays (gel shifts) with extracts of 2B4.11 T cells were performed to detect nuclear proteins that bind to the FLRE (Fig. 4A). When extracts from unactivated cells were incubated with a labeled 16-bp FLRE-containing oligonucleotide, a single retarded complex appeared. In contrast, extracts from cells activated with PMA plus ionomycin or anti-CD3 antibodies contained two new species: a major band with low mobility (band I) and a minor band with higher mobility (band II). The induced binding activities were present in equivalent amounts in freeze-thaw extracts of whole cells and isolated nuclei but were not detected in purified cytoplasmic extracts, indicating that they are predominantly nuclear in location (data not shown). CsA prevented the activation-induced appearance of band II but had little if any effect on band I or on the constitutive lower band (band III, Fig. 4A).

FIG. 4.

Analysis of FLRE-binding proteins in 2B4.11 extracts. (A) Extracts from 2B4.11 cells either untreated, treated for 3 h with 10 ng of PMA per ml and 1 μg of ionomycin per ml or with immobilized anti-CD3 antibodies in the absence or presence of 100 ng of CsA per ml were incubated with the end-labeled oligonucleotide corresponding to the region from −220 to −205 (16-mer) of the fasL promoter. Complexes were then resolved by electrophoresis on 5% nondenaturing polyacrylamide gels. Shown are the resulting retarded complexes. (B) Extracts from 2B4.11 cells stimulated for the indicated times with immobilized anti-CD3 antibodies were incubated with the end-labeled 16-mer oligonucleotide, and complexes were resolved as described for panel A. (C) Extracts from 2B4.11 cells stimulated for 3 h with either immobilized anti-CD3 antibodies or PMA and ionomycin and either in the absence or in the presence of 35 μM cycloheximide and were analyzed as described for panel A.

Kinetic analyses revealed that band I began to appear approximately 60 min and band II approximately 2 h after activation; both bands were maximally expressed 3 to 4 h after activation (Fig. 4B). To determine if the FLRE-binding proteins in these complexes preexisted or were synthesized de novo, 2B4.11 cells were activated in the presence of the protein synthesis inhibitor or the RNA synthesis inhibitors, and the appearance of the shifted complexes was ascertained. Both cycloheximide (Fig. 4C) and actinomycin D (data not shown) prevented the appearance of the FLRE-binding complexes. If either of the FLRE-binding factors is required to induce the transcription of fasL mRNA, then inhibition of protein synthesis may prevent activation-induced upregulation of fasL. To test this possibility, 2B4.11 cells were activated in the absence or presence of cycloheximide, and fasL mRNA levels were determined. As shown by Northern blot analysis (Fig. 5), induction of fasL mRNA by both PMA plus ionomycin and anti-CD3 antibodies was prevented by inhibition of de novo protein synthesis. Expression of gapdh mRNA was unaffected by cycloheximide. Thus, development of the FLRE-binding activity is relatively late, and, like expression of fasL itself, depends on de novo protein synthesis, consistent with the possibility that the factors are synthesized in response to activation signals and are required for induction of fasL mRNA.

FIG. 5.

Induction of fasL mRNA requires protein synthesis. 2B4 cells were stimulated for 5 h either with PMA plus ionomycin or with immobilized anti-CD3 antibodies in either the absence or the presence of 100 μM cycloheximide and subjected to Northern blot analysis. After having been probed with ³²P-labeled fasL cDNA, the blot was stripped and reprobed with ³²P-labeled gapdh cDNA.

The FLRE-containing segment closely resembles the Egr-1-binding site in the thymidine kinase promoter (16). Therefore, competition assays with the use of the labeled FLRE were performed in the presence of unlabeled oligonucleotides bearing Egr-family recognition sequences. Although an excess of the mutated FLRE had no effect, an unlabeled consensus-binding site for Egr-1 (EBS-1) (9) or a variant Egr-1-binding site found within the IL-2Rβ promoter (37) blocked the appearance of activation-induced bands I and II (Fig. 6A). The binding complexes were further characterized with antisera specific for different members of the Egr family. Two anti-Egr-1 antibodies specifically prevented formation of the upper shifted complex (band I) but had no effect on band II (Fig. 6B). In contrast, antisera against Egr-3 caused the disappearance of band II without affecting band I. Antisera specific for Egr family members Egr-2 and WT, the latter being the product of the Wilms’ tumor suppressor gene and a variant Egr family member with Egr-1-like DNA sequence specificity, had no effect on binding of the FLRE-binding complexes. Therefore, the induced FLRE-binding proteins detected in lysates of activated 2B4.11 cells are Egr-1 and Egr-3.

FIG. 6.

Identification of FLRE-binding proteins as Egr family members. (A) Extracts from 2B4.11 cells activated for 3 h with PMA plus ionomycin were incubated with the labeled 16-mer oligonucleotide in the presence of an excess of the indicated unlabeled competitor oligonucleotide and were analyzed by gel shift. (B) The extracts used in panel A were incubated with the labeled 16-mer in the presence of the indicated antisera. WT, wild type.

Egr-3 expression is sufficient to induce FLRE-dependent transcription.

These results demonstrate that the FLRE is required for fasL promoter activity and that this region bind Egr family members Egr-1 and Egr-3 in activated T cells. To directly determine whether Egr proteins are sufficient to activate FLRE-dependent gene transcription, cDNAs encoding either Egr-1 or Egr-3 were transiently transfected into 2B4.11 cells together with the 16-mer fasL luciferase reporter construct. Egr-1 had little effect on FLRE-mediated transcription (Fig. 7A). In contrast, Egr-3 caused a marked increase in luciferase activity. Notably, the 16-mer construct with a mutated FLRE was not activated by Egr-3 overexpression. Activation-induced expression of Egr-3, like that of FasL, is inhibited by CsA (40, 53), as is transactivation by the FLRE (Fig. 3). To determine whether the effect of CsA is proximal or distal to Egr-3 upregulation, 2B4.11 T cells were transfected with Egr-3-encoding cDNA, and FLRE-dependent gene transcription was assessed in the absence or presence of CsA. This drug did not inhibit the Egr-3-mediated increase in luciferase activity (Fig. 7B), indicating that, when Egr-3 has been expressed, CsA does not affect activation-induced transcription of the fasL promoter.

FIG. 7.

Expression of exogenous Egr-3 protein is sufficient to induce FLRE-dependent transcription in 2B4.11 cells in a CsA-independent manner. 2B4.11 cells were electroporated with either the wild-type (16-mer) or the mutated (m16-mer) reporter constructs plus an expression plasmid encoding the indicated Egr family protein or without insert. Half of each transfected sample was treated with CsA (100 ng/ml). Luciferase activity was determined after 15 h. Similar results were obtained in six independent experiments. The mean luciferase light units of the control plasmid transfected cultures were 230 (16-mer), 206 (16-mer, CsA-treated), 324 (m16-mer), and 326 (m16-mer, CsA-treated).

Heterologous expression of Egr-3 induces fasL mRNA in nonlymphoid cells.

The ability of Egr-3 to activate the fasL promoter was tested in the nonlymphoid cell HeLa, which does not express Egr-3 even when stimulated with PMA and ionomycin (40). As in 2B4.11 cells, overexpression of Egr-3 induced FLRE-dependent luciferase activity, whereas Egr-1 had little effect (Fig. 8A). The function of the Egr-1-expressing construct was tested by coexpression with a luciferase reporter dependent on a pair of consensus Egr-1-binding sites (EBS-1) upstream of the minimal prolactin promoter (13). In contrast with the FLRE-dependent reporter, the EBS-1-dependent reporter was induced more strongly by Egr-1 than Egr-3 (Fig. 8A). Thus, the FLRE is an Egr family-binding site that, despite being able to bind to both Egr-1 and Egr-3 in vitro (Fig. 6), is preferentially activated by Egr-3 in vivo.

FIG. 8.

(A) Expression of exogenous Egr-3 protein is sufficient to induce FLRE-dependent transcription in nonlymphoid cells. HeLa cells were transfected with the indicated expression plasmids and either the Egr-1 consensus binding site-dependent reporter (EBS-1) or the 16-mer reporter (FLRE). Luciferase activity was determined after 35 h. The average values of the control plasmid-transfected cultures were 382 (16-mer) and 355 (EBS-1). (B) Expression of exogenous Egr-3 protein induces production of fasL mRNA in a heterologous cell. HeLa cells were transfected with the indicated expression plasmids, and the levels of endogenous fasL mRNA were determined by RT-PCR after 35 h. The PCR primers used amplify a segment of cDNA derived from four separate exons; the lanes marked (−RT) contain the PCR product of a mock reverse transcription reaction in which RNA from the transfected samples was used, but RT was omitted.

If the Egr-3-binding site is the primary control region for fasL, overexpression of Egr-3 protein alone might be expected to activate the endogenous fasL promoter. To test this possibility, HeLa cells were transfected with Egr-encoding plasmids and the levels of endogenous fasL mRNA was assessed by RT-PCR. One experiment representative of five similar experiments is shown in Fig. 8B. Whereas cotransfection of the control (empty vector) plasmid had no effect, the plasmid encoding Egr-3 induced fasL mRNA, an effect that was not inhibited by CsA (data not shown). Consistent with its limited effect on the FLRE-dependent reporter, Egr-1 induced a barely detectable increase in fasL mRNA. PCR of the same RT samples to detect gapdh mRNA demonstrated that all RNA samples were intact. Thus, expression of Egr-3 alone is sufficient to activate the endogenous fasL promoter.

DISCUSSION

Although expression of fasL mRNA in T cells is known to require activation signals, relatively little is known about the transcriptional regulation of the fasL gene. Iterative deletional analysis of the upstream region of fasL yielded one predominant region of activity, an 8-bp segment from −214 to −207, that interacts with induced protein factors. Antibody supershift analysis indicated that the FLRE-binding activity is predominantly composed of Egr-1 and Egr-3, members of a family of Zn²⁺-finger transcription factors. The FLRE is a G-rich region that loosely resembles the consensus Egr-1-binding sequence EBS-1 (GCG[G/T]GGGCG) (9). The Egr-1-binding site (CCGTGGGTG) that was identified in the thymidine kinase promoter (44) is quite similar to the FLRE, with seven sequential identical residues. Noncanonical Egr-1-binding sites that also contain a high proportion of G residues were identified in promoters of several other genes, including IL-2Rβ (GCGTAGGAGGCA) (37), platelet-derived growth factor A chain (GAGGAGGAGGAGGA) (63), and rat cardiac α-myosin heavy chain (GTGGGGGTG) (20).

Egr family mRNAs can be induced by a wide variety of stimuli, including mitogenic serum-derived growth factor, nerve growth factor, membrane depolarization, and B-cell antigen receptor- and TCR-mediated lymphocyte activation (for a review, see reference 16). Activation of target gene transcription by Egr family members requires their de novo synthesis, and therefore Egr-dependent transcription of their target genes is prevented by cycloheximide (56). The DNA-binding domains of Egr-1 and Egr-3, as well as those of other family members (i.e., Egr-2 and NGFI-C), have critical residues in common and similar binding sequence specificities (58). The differences between family members reside largely in non-DNA-binding domains, suggesting that the members may play different biological roles. In this regard, the herpes simplex virus type 1 latency-associated transcript promoter appears to be negatively regulated by Egr-2 in neurons, whereas Egr-1 and Egr-3 are inactive (61). Egr-1 has been implicated in the regulation a large number of genes (16), including some with important roles in immune function, such as those encoding IL-2 (54), IL-2Rβ (37), ICAM-1 (42), CD44 (41), TGF-β (38), and tumor necrosis factor alpha (31). To our knowledge, no specific function has been previously ascribed to Egr-3.

Antibody supershift analysis identified the major FLRE-binding factor in activated 2B4.11 cells as Egr-1. Nevertheless, Egr-1 does not appear to have a major role in activation-induced upregulation of fasL transcription. The relative ineffectiveness of Egr-1 at regulating fasL transcription in cell lines is manifested in the phenotype of mice made deficient in Egr-1, which display no obvious lymphoid defects (35). Why Egr-1 so poorly transactivates FasL promoter-dependent gene transcription is unclear. One possibility is that, for allosteric reasons, binding of Egr-1 to the FLRE does not efficiently recruit cofactors necessary for transcription. Another intriguing possibility is that the activity of Egr-1 is itself regulated by other proteins. In this regard, the constitutively expressed protein NAB1 (52) and the activation-inducible protein NAB2 (57) were shown to inhibit the activity of Egr proteins by interacting with repression domains, termed R1, that are found in Egr family members 1, 2, and 3. Moreover, Egr-1 was found to be more sensitive than Egr-3 to the inhibitory action of both NAB proteins. Studies are currently underway to evaluate these possibilities as well as to determine what effect Egr family members that were not detected in activated 2B4.11 T cells, such as Egr-2 and NGFI-C, have on fasL transcription.

Unlike egr-1, which is constitutively expressed in many tissues, egr-3 mRNA is not found in unactivated tissues (49). However, egr-3 is readily detected in T cells activated by the combination of PMA and Ca²⁺ ionophore but not by either agent alone (40). egr-3 was also detected in NIH 3T3 cells activated by the addition of serum (49) and in fibrosarcoma and lung fibroblast cell lines stimulated with PMA (40). Unlike T cells, the fibroblast lines did not require costimulation with a Ca²⁺ ionophore to induce egr-3 mRNA, and this induction was not sensitive to CsA (40). Therefore, egr-3 appears to be a mitogen-inducible gene whose transcriptional regulation may be different in lymphoid than in nonlymphoid cell types.

It is worth noting that potential binding sites for several known transcription factors are in the immediate vicinity of the FLRE. For example, a partial (TGAGT, a five of seven nucleotide match) activator protein-1 (AP-1) consensus site is located at nucleotides −217 to −213. We do not believe, however, that AP-1 contributes to the transactivating activity of this region, because a consensus AP-1-binding oligonucleotide did not compete for binding of any of the activation-induced factors in a gel shift assay, and the unlabeled Egr-binding site did not compete for binding of AP-1 to the consensus AP-1 site (data not shown). The region containing the Egr-binding activity also contains consensus binding sites for several interferon γ-inducible gene products. Adjacent to the minimum required 8-bp sequence is a consensus IRF-1- and IRF-2-binding site, AAGTGA (−220 to −215). In T cells, gamma interferon-regulated protein family members IRF-1, IRF-2, and ISGF3γ are expressed constitutively, whereas ICSBP is induced upon activation (47). A consensus binding site for ICSBP, NNTTTC, is located from nucleotides −209 to −204 (14). It is unlikely that any of these defined transcription factors contribute to the binding activity, however, because they do not coincide with the minimal 8-bp site recognized by Egr-1 and Egr-3 (from −214 to −207). Moreover, gamma interferon treatment of 2B4.11 cells did not induce binding activity for this region, although it did induce binding activity specific for the interferon-regulated transcription factor STAT1 (data not shown). Thus, although this critical region of the fasL promoter contains potential binding sites for other transcription factors, the data suggest that they do not greatly affect activation-induced transcriptional activity.

The lack of FasL expression observed in NF-ATp-deficient mice has indicated a role for NF-AT in FasL regulation (23). Two groups have provided data suggesting that an NF-AT site in the fasL promoter is critical for upregulation of this gene. Latinis et al. reported that an NF-AT binding site exists at nucleotides −275 to −271 of the fasL gene and that mutation of this site inhibits transcription of a luciferase reporter driven by the fasL promoter expressed in Jurkat T cells (33, 34). Holtz-Heppelmann et al. (24) reported that in Jurkat T cells a fasL reporter extending to nucleotide −318 responded modestly to TCR signals and that mutation of the NF-AT sequence at −275 inhibited this activity. In contrast, we found that deletion of the putative NF-AT at the −275 element had little effect on luciferase induction (Fig. 1). We have also introduced a mutation at the NF-AT site similar to that reported by Latinis et al. and found a reproducible but small decrease (mean reduction of 35%) in anti-TCR-inducible luciferase activity. Both studies that have suggested a direct role for NF-AT in transcriptional regulation of fasL were performed with Jurkat T cells. In contrast, the experiments presented here were performed with a T cell hybridoma, HeLa cells, and peripheral blood lymphocytes, and in all cases the putative NF-AT site at −275 was dispensable, while the FLRE was necessary and sufficient for activation of the fasL promoter. Furthermore, forced expression of Egr-3 induced fasL promoter-dependent luciferase activity and, more importantly, expression of endogenous fasL mRNA that was not prevented by CsA, indicating that NF-AT does not play a major role in fasL expression once Egr-3 is expressed. Because Egr-3 expression is itself CsA sensitive, we conclude that the NF-AT site at −275 has at most a modest role in regulating FasL expression and propose that the CsA sensitivity of FasL expression is, in fact, secondary to inhibition of Egr-3 expression, presumably because Egr-3 transcription itself is regulated by NF-AT. In any case, further characterization of how Egr-3 expression and function are regulated should greatly enhance our understanding of how FasL expression is controlled in health and disease.