Structure/Function Analysis of the Murine CD95L Promoter Reveals the Identification of a Novel Transcriptional Repressor and Functional CD28 Response Element*

CD28 costimulation, an important second signal for antigen-mediated T cell activation, is known to enhance expression of several genes important for the regulation of CD4 (cid:1) T cell effector function including interleukin-2 and CD154. Previous studies demonstrate CD28-medi-ated enhancement of the transcription and expression of Fas ligand (CD95L) in T cell lines, suggesting a regulatory link between CD28 and CD95L expression. These results served as the basis for structure/function analysis of the CD95L promoter to elucidate the mechanism for CD28-mediated enhancement of CD95L. In this re-port, we describe a novel response element, located at (cid:2) 210 to (cid:2) 201 bp upstream of the transcription start site, that confers CD28 responsiveness to the CD95L gene. This response element is homologous to the CD28 response element (CD28RE) previously identified in the IL-2 promoter and bears structural similarities to a newly identified CD28RE in the CD154 promoter. We further demonstrate that CD28-mediated enhancement of promoter activity correlates with enhanced expression of CD95L mRNA, cell surface expression of CD95L protein, and increased apoptosis of CD95 (cid:1) target cells. These results demonstrate a direct transcriptional regulatory role for CD28 in CD95L-mediated functional activity in CD4 (cid:1) T cells. Mutational analysis of the CD95L promoter also reveals a novel transcriptional repressor Assays— The functional activity of CD95L expressed by antibody-stimulated A1.1 and primary CD4 (cid:2) T cells was determined by the ability of stimulated, costimulated, or nonstimulated cells to induce apoptosis in Fas (cid:2) L1210 target cells. Briefly, L1210 and L1210-Fas cells were labeled with 200 (cid:3) Ci/ml 51 Cr (1 mCi/ml; Amersham Bio- sciences) for 1.5 h at 37 °C and washed three times in complete medium. as follows: of triplicate and are representative of least four groups investigating CD95L regulation in human T cells have reported that specific promoter regions diminish activation-dependent transcription (13, 14). Previous studies have shown that deletions of the (cid:4) 2365 to (cid:4) 454 bp, the (cid:4) 900 to (cid:4) 486 bp, and the (cid:4) 1204 to (cid:4) 860 bp regions of the human CD95L gene enhance basal and inducible transcriptional activity (13, 14, 47). Taken together, these previous studies suggest the presence of multiple elements capable of attenuating both basal and inducible transcriptional activity of the human CD95L promoter. Our current data demonstrate that the (cid:4) 270 to (cid:4) 260 bp repressor element constitutively binds c-Fos, a member of the AP-1 family of proteins previously shown to be associated with transcriptional activation of CD95L, IL-2, and other cytokines in the activated T cell. The ability of c-Fos to repress activation-dependent transcriptional activity via the binding of AP-1 elements in the promoters of several genes has been previously described (48–51). The observation that c-Fos was constitutively bound to the AP-1 element and that the reporter constructs possessing a functional repressor AP-1 element were not hindered in the ability to be transactivated by TCR and CD28-mediated co-activation is novel. A motif search of the human CD95L promoter reveals several putative AP-1 elements located within each of the regions shown to repress basal and inducible transcriptional activity. Whether the multiple putative AP-1 elements located throughout the human CD95L promoter region mediate transcriptional repression will have to be determined empirically. These data suggest the possibility that AP-1 elements within the murine CD95L promoter and perhaps other gene promoters play roles as not only positive and negative transcriptional regulators but also as constitutive transcriptional “attenuators.” Further analysis of other repressor elements in AP-1-responsive genes such as IL-2

Ligation of receptors on the surface of T cells induces a series of signaling events culminating in specific gene expression via the activation and translocation of specific transcription factors to the nucleus. Ligation of the T cell receptor (TCR) 1 is known to activate several transacting factors that mediate cytokine induction and proliferation in activated T cells. NF-B and members of the activator protein-1 (AP-1) and nuclear factor of activated T cells play important roles in regulating genes important for T cell-specific proliferation and function (1,2). Op-timal activation of T cells, however, requires at least two signals: the primary signal transduced via the TCR upon ligation to the antigen-major histocompatibility complex complex and a second signal delivered via "costimulatory" molecules (3,4). Prototypical of these costimulatory molecules is CD28, a 44-kDa homodimer expressed constitutively by most T cells (3)(4)(5)(6). The ligands of CD28 include B7.1 (CD80) and B7.2 (CD86), whose expression is restricted primarily to antigen-presenting cells (4). Costimulation through CD28 by CD80/CD86 is not only required for optimal T cell activation but also promotes T cell survival (3,4). Whereas the scope of cellular alterations induced by CD28 signaling is not clearly defined, CD28-mediated costimulation is known to enhance expression of several genes important in CD4ϩ T cell effector function, including interleukin-2 (IL-2), granulocyte-macrophage colony-stimulating factor, interferon ␥ (IFN␥), and CD154 (7)(8)(9)(10).
Although CD28 costimulation is known to enhance target gene expression both by post-transcriptional stabilization of mRNA and direct transcriptional regulation, the promoter regions of the aforementioned CD28-responsive genes possess a functional CD28RE along with several TCR-responsive elements (3,6,8,9,11). CD28-induced transcriptional regulation of gene expression is mediated through the induction of transcription factor complex binding to specific elements in the 5Ј regulatory regions of CD28-responsive genes (12)(13)(14)(15)(16). Originally identified in the IL-2 gene, the CD28 response element (CD28RE) that is located Ϫ162 to Ϫ153 bp upstream of the transcriptional start site is required for CD28-mediated enhancement of IL-2 mRNA and protein expression in TCR-activated T cells (9). CD28 ligation induces the binding of a multiprotein complex to the response element that is capable of increasing transcriptional activity and gene expression in cooperation with TCR-mediated signaling but is not sufficient to transactivate by CD28 ligation alone (6,9,11). Similar dependence of TCR signaling also is seen in the CD28RE identified in the CD154 gene (17,18). The multiprotein complex that binds to the CD28RE is composed of several transcription factors ubiquitously activated by TCR signaling, including the NF-B/ Rel and AP1 family of proteins (19,20). However, the transcription complex only binds the CD28RE after CD28 costimulation (3,21). Whereas the identified CD28REs in CD28-responsive genes other than IL-2 have some sequence homology with the IL-2 CD28RE, no specific motif has yet been defined. Instead, the response elements in other CD28-responsive genes have been defined empirically, through structure/function analysis of the promoter regions.
The focus of this study was to identify the role and mechanism of CD28-mediated enhancement of CD95L expression in CD4ϩ T cells. Through the use of deletion/mutation mapping of the murine CD95L promoter, we identified a novel response element that conferred CD28 responsiveness to the CD95L gene. This response element is homologous to the CD28RE in the IL-2 promoter and bears structural similarities to a newly identified CD28RE in the CD154 promoter (9,18). Moreover, we demonstrate that CD28-mediated enhancement of promoter activity via this novel CD28RE correlates with enhanced expression of CD95L mRNA, cell surface expression of CD95L protein, and increased CD95L-mediated apoptosis. These results demonstrate a direct transcriptional regulatory role for CD28 in CD95L-mediated functional activity of CD4ϩ T cells. In addition, during the course of our studies, a response element with sequence homology to an AP-1 binding element was demonstrated to attenuate transcriptional activity of the CD95L promoter, suggesting a dual role for the AP-1 family of transcription factors in the modulation of CD95L.
Plasmid Constructions-Genomic DNA was isolated from A1.1 cells using DNAeasy kit (Qiagen, Valencia, CA). A 720-bp PCR fragment corresponding to the promoter region of CD95L (Ϫ689 to ϩ65) was generated using the upstream primer 5Ј-GTACCTCAGTTTTCATCTG-GTGACCAGAAG-3Ј and the downstream primer 5Ј-GCACCCAGC-CCCAGGAAAGG-3Ј in a PCR using platinum Pfx high fidelity polymerase (Invitrogen) and 50 ng of A1.1 genomic DNA template. The resultant 720-bp fragment was gel-purified, an A overhang was generated by treatment with Taq polymerase (Invitrogen) and dATP for 15 min, and cloned into the Topo-PCR2.1 T/A cloning vector (Invitrogen) and sequenced for orientation and fidelity. This plasmid (PCR2.1-689pCD95L) served as the source of the insert for further plasmid constructions. To make the Ϫ689 to ϩ65 luciferase reporter plasmid, the 720-bp insert was excised with SacI and XhoI and ligated into the SstI/XhoI site of pGL-3 Basic luciferase vector (Promega, Madison, WI). To make the 5Ј truncation mutants, shorter fragments were generated using new upstream primers, and the common downstream primer was used to make the Ϫ689 to ϩ65 insert. For the Ϫ253 to ϩ65 fragment, the upstream primer 5Ј-CAGGCAAGCCTGGTTTACCAGCC-3Ј and the common downstream primer were used. To generate the Ϫ201 to ϩ65 fragment, the upstream primer 5Ј-CGAAGACTTGTCGTCAGAAATT-TCTGGGC-3Ј was used. The Ϫ168 to ϩ65 fragment was produced using 5Ј-CTTCCTGGGGTTGCTGTGAGCTTTTTG-3Ј. The Ϫ138 to ϩ65 fragment was produced by using the upstream primer 5Ј-GCTTCTCAGCT-TCAGATGCAAGTGAGTG-3Ј. The resultant fragments were subcloned into the polylinker site of Topo PCR2.1 T/A vector and sequenced for orientation and fidelity. Luciferase reporter constructs were produced by excising the various fragments with SacI and XhoI and then subcloning the fragments into the polylinker site of pGL-3 Basic.
Site-directed mutations to putative elements in the CD95L promoter region for the reporter constructs were produced using the PCR-ligation-PCR method previously described (36). Briefly, two fragments flanking the region to be deleted or mutated were produced using specific primers and Pfx polymerase (Invitrogen). The fragments generated were gel-purified, phosphorylated with T4 polynucleotide kinase (New England Biolabs, Beverly, MA), mixed, and ligated with T4 ligase (Invitrogen). Finally, full-length fragments containing the internal deletions or mutations were amplified using the 5Ј and 3Ј distal primers. The fragment was subcloned into Topo 2.1 PCR, sequenced for fidelity and orientation, and then subcloned into pGL-3 Basic.
Transfections/Luciferase Assays-A1.1 cells (10 7 ) were transfected using DEAE-dextran with 10 g of the indicated reporter plasmid and 1 g of the pCMV-␤Gal control vector. Twenty-four hours following transfection, the cells were replated in duplicate on uncoated 6-well plates or plates coated with either anti-CD3 (5 g/ml) alone or with anti-CD28 (10 g/ml) for 4 h at 37°C. Following incubation, cells were harvested by repeated washing with PBS and centrifuged, and extracts were prepared by lysing cells in 70 l of reporter lysis buffer (Promega). Luciferase activity was determined by mixing 40 l of extract with 100 l of luciferase substrate (Promega) and immediately reading the sample in a Monolight 2010 luminometer (BD Biosciences, Franklin Lakes, NJ). ␤-Galactosidase activity was determined by mixing 20 l of extract with galactolyte ␤-galactosidase substrate (Tropix, Bedford, MA), incubating for 1 h at 25°C, adding 300 l of galactolyte accelerator (Tropix), and reading the sample in the luminometer. Luciferase activity was normalized by dividing the mean luciferase RLU by the mean ␤-galactosidase RLU. The normalized luciferase RLU from the stimulated samples were divided by the normalized RLU of the untreated sample, and values were expressed as "normalized -fold induction." Preparation of Nuclear Extracts-After stimulation on the antibodycoated plates, cells were harvested by washing with cold PBS and centrifuged at 1100 rpm for 5 min to pellet. All subsequent steps were done on ice. The cell pellet was resuspended in 200 l of buffer A (10 mM Hepes, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, and 0.5 mM dithiothreitol) and lysed by passing through a 28-gauge needle four times. The nuclei were then pelleted by centrifugation for 10 s, and the supernatant was aspirated. The crude nuclei preparation was then extracted by adding 120 l of buffer C (20 mM HEPES, pH 7.9, 25% (v/v) glycerol, 420 mM KCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM dithiothreitol, and 0.5 mM phenylmethylsulfonyl fluoride (added fresh just before use) and incubating for 15 min on ice. 120 l of buffer D was then added (20 mM HEPES, pH 7.9, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, and 0.5 mM dithiothreitol) and centrifuged for 10 min, and the supernatant was harvested, snap frozen in LN 2 , and stored at Ϫ80°C. Total protein determinations were done using the BCA protein determination kit from Pierce, using albumin as a protein standard.
Electrophoretic Mobility Shift Assay and Supershift Analysis-Sense and antisense oligonucleotides corresponding to elements in the murine CD95L promoter were synthesized by Integrated DNA Technologies (Coralville, IA). To make double-stranded oligonucleotides, equimolar quantities of the sense and complementary antisense oligonucleotides were mixed in 1ϫ S1 nuclease buffer (0.05 M sodium acetate, 1 mM zinc acetate, 0.25 M NaCl, and 0.05 mg/ml bovine serum albumin), heated to 100°C for 3 min, and then allowed to cool to room temperature in a water bath over several hours. The double-stranded oligonucleotides were 5Ј-end-labeled by mixing 1 pmol of double-stranded oligonucleotide with 10 Ci of [␥-32 P]ATP (Amersham Biosciences), 8 units of T4 polynucleotide kinase (New England Biolabs) in 1ϫ PNK buffer and incubating for 1 h at 37°C. After incubation, the end-labeled doublestranded oligonucleotides were purified from free ATP by passing over a NICK column (Amersham Biosciences).
A total of 7 l of nuclear extract corresponding to 5-10 g of protein was mixed with 1 l of radiolabeled oligonucleotide (ϳ50,000 cpm) in a reaction mix containing 1ϫ binding buffer (10 mM Tris, 1 mM EDTA, 1 mM dithiothreitol, 100 mM KCl, 10% (v/v) glycerol) and 1 g of poly(dI-dC) (Amersham Biosciences) as a nonspecific inhibitor in a final volume of 25 l for 30 min at 25°C. The samples were resolved on a 6% polyacrylamide gel in Tris borate EDTA that has been prerun for 30 min. The gel was visualized on a STORM PhosphorImager (Amersham Biosciences). For the specific and nonspecific competition analyses, equimolar amounts of the cold double-stranded oligonucleotide competitors were added to the binding reaction before the addition of the labeled oligonucleotide probe. For the supershift analyses, various amounts of the appropriate polyclonal or nonimmune control sera, purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), were added to the reaction mixtures during the 20-min incubation prior to electrophoresis.
Reverse Transcriptase-PCR-Total RNA was isolated from A1.1 cells following activation on antibody-coated plates using the RNAeasy kit (Qiagen). First strand cDNA synthesis was performed using Superscript II reverse transcriptase (Invitrogen) in a reaction using 2 g of total RNA primed with random hexamers in a total reaction volume of 20 l. Following first strand synthesis, 10% of the reaction volume was used as a DNA template for amplification by PCR. For detection of murine CD95L, the forward and reverse primers were 5Ј-CTT-GGGCTCCTCCAGGGTCAGT-3Ј and 5Ј-TCTCCTCCATTAGCACCA-GATCC-3Ј, respectively. The PCR products generated from the CD95L primers were normalized against PCR products generated from murine ␤-actin forward (5Ј-TCATGAAGTGTGACGTTGACATCCGTAAAG-3Ј) and reverse (5Ј-CCTAGAAGCATTTGCGGTGCACGATGGAGG-3Ј), after electrophoresis on a 1.5% agarose gel and visualization of the PCR product with UV light. The band intensity of PCR products was determined using an EpiChemi digital image analysis system (UVP, Upland, CA).
Flow Cytometry-Following stimulation, 10 6 A1.1 cells were harvested by repeated washing with cold PBS and centrifuged, and the pellet was resuspended in 50 l of fluorescence-activated cell sorting wash (PBS, 1% fetal calf serum, 0.05% sodium azide), containing 1 l of biotinylated anti-murine CD95L antibody MFL3 (Pharmingen) or 1 l of biotinylated isotype control antibody, and incubated on ice for 20 min. Cells were washed twice with fluorescence-activated cell sorting wash and resupended in 50 l of fluorescence-activated cell sorting wash and streptavidin-phycoerythrin (Sigma) for 10 min on ice. Cells were washed three times and resuspended in 300 l of PBS containing 3% paraformaldehyde. Samples were read on a Becton-Dickinson FACScan (San Jose, CA), and data were analyzed using CellQuest software (Becton-Dickinson).
Cytotoxicity Assays-The functional activity of CD95L expressed by antibody-stimulated A1.1 and primary CD4ϩ T cells was determined by the ability of stimulated, costimulated, or nonstimulated cells to induce apoptosis in Fasϩ L1210 target cells. Briefly, L1210 and L1210-Fas cells were labeled with 200 Ci/ml 51 Cr (1 mCi/ml; Amersham Biosciences) for 1.5 h at 37°C and washed three times in complete medium. A1.1 cells or recently activated primary T cells were incubated for 4 h on flat bottom 96-well microtiter plates (BD Biosciences) coated with anti-CD3 antibody, anti-CD28 antibody, both, or neither before the addition of the radiolabeled target cells. The plates were incubated at 37°C for 16 h before the supernatants were harvested using Skatron supernatant harvesting frames (Sterling, VA) and counted on a Cobra II ␥-counter (Packard Instrument Co.). The percentage of specific release was calculated as follows: 100 ϫ (cpm of experimental Ϫ cpm of spontaneous release/cpm of maximum release Ϫ cpm of spontaneous release). The results are expressed as the mean Ϯ S.E. of triplicate wells per assay and are representative of at least four similar analyses.
Preparation of Primary T Cells-Single cell suspensions of splenocytes were obtained from D011.10 TCR transgenic mice following red blood cell lysis. CD4ϩ T cells were enriched to greater than 96% CD4ϩ as determined by flow cytometry using a negative selection/magnetic bead CD4ϩ T cell isolation kit from Miltenyi Biotech (Sunnyvale, CA). The CD4ϩ T cells were stimulated and expanded for 2 weeks in coculture with a 10-fold excess of irradiated splenocytes (30 grays) obtained from BALB/c mice that were pulsed with OVA 323-339 peptide (Research Genetics, Huntsville, AL). The medium was supplemented with 100 units/ml recombinant murine IL-2 (R&D Systems, Minneapolis, MN). Following co-culture, dead cells were removed by Ficoll density centrifugation, and the remaining viable cells were rested by placing them back in culture for 24 h in complete medium containing 100 units/ml IL-2. For the cytotoxicity assays, the recently activated, rested T cells were harvested and replated on antibody-coated plates as described above.

Activity of the CD95L Promoter Is Induced by TCR Ligation and Enhanced by CD28
Costimulation-To define the CD28responsive sequence elements within the CD95L promoter, we utilized deletion/mutation mapping of the 5Ј regulatory region of murine CD95L using reporter constructs and transient transfection. The readily transfectable A1.1 T cell hybridoma served as an in vitro cell model for this objective. This line has been used extensively in studying the regulation of CD95L expression and activation-induced cell death, and these cells are sensitive to nonphysiological stimuli that mimic TCR-mediated stimulation, including agonistic antibodies to both CD3 and CD28, phorbol esters, and calcium ionophore (14,22,37). Fig. 1A shows schematically the Ϫ689 to ϩ65 bp 5Ј regulatory region of the murine CD95L gene and the Ϫ689 to ϩ65 bp CD95L luciferase construct (p689). Response elements previously shown to be important for TCR-mediated regulation of CD95L are depicted (13-15, 19 -21, 38). To dissect the molec-ular mechanisms underlying co-stimulation-mediated CD95L expression, mutational analyses were initiated to determine whether a response element sensitive to CD28 was contained within the Ϫ689 bp region of the CD95L promoter. Initially, A1.1 cells were transfected with a luciferase construct encoding the minimal inducible promoter region (Ϫ256 to ϩ45 bp) of the murine IL-2 gene to serve as both a control for CD28-mediated enhancement of promoter activity and a comparison of relative promoter activities following activation. As shown in Fig. 1B, TCR-mediated signaling induced a 4 -5-fold increase in luciferase activity over unstimulated cells. Costimulation through CD28 further enhanced CD95L transcriptional activity an additional 4 -5-fold over the activity observed with TCR stimulation alone. As expected, CD28 costimulation greatly increased transcriptional activity of the minimal IL-2 reporter construct (35-fold) versus stimulation only through the TCR (Fig. 1B). These data show that the Ϫ689 region is sufficient for TCR and CD28-mediated activation and served as a basis for mutational analysis.
The Ϫ210 to Ϫ201 Region of the CD95L Promoter Is a Functional CD28RE-To identify the CD28-responsive element within the Ϫ689 bp region, a series of luciferase constructs containing serial 5Ј truncations to the Ϫ689 bp sequence were generated ( Fig. 2A). The truncation mutant reporter constructs p324 (Ϫ324 to ϩ65), p253 (Ϫ253 to ϩ65), p201 (Ϫ201 to ϩ65 bp), p168 (Ϫ168 to ϩ65), and p135 (Ϫ135 to ϩ65), along with the p689 construct, were assessed for transcriptional activity following activation. Whereas differences in the relative activity of the various constructs were noted, a comparison of nor-malized -fold induction of the ␣CD3-stimulated and the ␣CD3and ␣CD28-co-stimulated constructs demonstrated that truncation of the CD95L promoter from Ϫ253 to Ϫ201 bp abrogated CD28-mediated transcriptional enhancement of the CD95L promoter (Fig. 2B). These results suggest that sequence upstream of Ϫ201 bp confers CD28 responsiveness.
A motif search of the Ϫ253 to Ϫ201 bp region of the CD95L promoter revealed a 10-bp sequence located from Ϫ210 to Ϫ201 with homology to the CD28RE of the IL-2 promoter (murine CD95L (GGAACTTCGA) versus murine IL-2 CD28RE (AA-GAAATTCCA); the conserved sequence is underlined). Based on this homology, the functionality of this region as a CD28RE in the CD95L promoter was determined. Site-directed mutagenesis was used to generate constructs in p253 with mutations generated in the Ϫ210 to Ϫ201 region. In one variant (p253mut1), the Ϫ210 to Ϫ201 region was deleted. The second variant was constructed with the sequence of the same region randomized (p253mut2). The p253, p253mut1, and p253mut2 constructs were transfected, and cells were stimulated as previously described. As predicted, both mutations to the Ϫ210 to Ϫ201 region of p253 abrogated the CD28-mediated enhancement of transcriptional activity (Fig. 3A), demonstrating that the Ϫ210 to Ϫ201 bp region confers CD28 responsiveness to the p253 reporter construct.
Because of the differences in the relative transcriptional activity between the full-length p689 and the p253 constructs, we wanted to confirm the importance of the Ϫ210 to Ϫ201 region in the context of the longer upstream sequence. To this end, the Ϫ210 to Ϫ201 bp region in p689 was deleted (p698mut1) and tested for CD28 responsiveness. Similar reduction in CD28 transcriptional enhancement was observed in p689mut1 as observed in the shorter p253mut1 and p253mut2 constructs (Fig. 3B). Thus, these data identify the Ϫ210 to Ϫ201 region as critical for CD28-mediated enhancement of transcription of the CD95L gene.
CD28 Costimulation Enhances mRNA, Protein, and CD95Lmediated Lytic Activity in Both Primary CD4ϩ T Cells and A1.1 T Cell Hybridoma Cells-To determine whether the increased transcriptional activity observed with CD28 stimulation correlated with enhanced CD95L-mediated function, we first determined whether CD28 costimulation increased steady state levels of CD95L mRNA and cell surface protein expression. CD95L mRNA expression by A1.1 T cell hybridoma cells was assessed using semiquantitative reverse transcriptase-PCR after TCR stimulation with or without CD28 costimulation using agonistic anti-CD3 and anti-CD28 antibodies, respectively. The strength of TCR-mediated stimulation was titrated using various amounts of soluble anti-CD3 antibody. Fig. 4A demonstrates that CD28 costimulation enhances CD95L mRNA steady state levels over TCR stimulation alone. Greater CD28-mediated enhancement was observed with suboptimal TCR signaling via lower levels of anti-CD3 (3.6-fold enhancement with 0.1 g/ml anti-CD3 versus 2.3-2.7-fold enhancement at higher levels of anti-CD3), suggesting that costimulation is required for optimal activation. To determine whether enhanced steady state levels of mRNA correlates with enhanced cell surface expression of CD95L protein, A1.1 cells were stimulated on antibody-coated plates as before, harvested, stained with an antibody specific for murine CD95L, and subjected to flow cytometry. As indicated in Fig. 4B, an increase in the number of cells staining positive for surface CD95L expression is seen in A1.1 cells costimulated with anti-CD28 over TCR stimulation alone, demonstrating CD28-mediated enhancement of cell surface CD95L protein expression correlates with enhanced levels of mRNA.
Based on these results, we next determined whether en- hancement of mRNA and protein expression by CD28-mediated costimulation was sufficient to enhance CD95L-mediated apoptosis. Thus, a lytic assay was employed, where unstimulated A1.1 cells or cells stimulated with anti-CD3 alone, anti-CD28 alone, or both anti-CD3 and anti-CD28 were used as effector cells against CD95ϩ and CD95Ϫ target cells (L1210-Fas and L1210 cells, respectively). The data show that A1.1 cells stimulated with a combination of anti-CD3 and anti-CD28 mediated enhanced CD95L-dependent lysis of CD95ϩ target cells when compared with A1.1 cells stimulated with anti-CD3 alone (Fig. 5A), demonstrating a functional consequence for CD28 modulation of CD95L expression. Less than 5% specific lysis was observed in control CD95-negative L1210 cells, demonstrating that lysis was CD95L-dependent (data not shown). Finally, to confirm that the enhancement of CD95L-dependent lytic activity by CD28 costimulation observed in the A1.1 T cell hybridoma model was consistent with enhancement of lytic activity of CD28 costimulated primary T cells, in vitro-activated CD4ϩ splenic T cells were activated as described for A1.1 and tested as effector cells in a lytic assay against L1210-Fas and L1210 target cells. As observed with A1.1 cells, CD28mediated costimulation of primary T cells enhanced CD95Ldependent apoptosis (Fig. 5B), demonstrating that CD28-mediated enhancement of CD95L transcriptional activity of the CD95L promoter results in the qualitative enhancement of CD95L function.
The AP-1 Element Located in the Ϫ270 to Ϫ260 Region of the CD95L Promoter Is a Transcriptional Repressor Element-Having identified a functional CD28RE in the CD95L promoter, we were intrigued by the observation that the different 5Ј truncation constructs demonstrated highly variable relative transcriptional activity. Several groups have reported similar variation in transcriptional activity following truncation of the 5Ј regulatory regions of the human CD95L and IL-2 genes, FIG. 3. The ؊210 to ؊201 region of the CD95L promoter is required for CD28-mediated enhancement of transcriptional activity. A1.1 cells were transiently transfected with the p253 luciferase construct containing a deletion to the Ϫ210 to Ϫ201 region (p253mut1) or random Ϫ210 to Ϫ201 sequence (p253mut2) (A) or the p689 construct containing a deletion of the Ϫ210 to Ϫ201 region (p689mut1) (B). The transfected cells were left unstimulated (NS) or stimulated on immobilized anti-CD3 antibody only (␣CD3) or both anti-CD3 and anti-CD28 antibodies (␣CD3/28) for 4 h. The luciferase activity, normalized to ␤-galactosidase activity of the stimulated cells, was compared with the normalized activity of the transfected but not stimulated cells and expressed as normalized -fold induction. The bars represent the S.E. of duplicate transfections. These data are representative of at least three independent experiments.

FIG. 4. CD28 costimulation enhances CD95L mRNA and protein.
A, expression of CD95L mRNA was determined by reverse transcriptase-PCR of total RNA isolated from A1.1 cells stimulated with the indicated amounts of immobilized anti-CD3 (␣CD3) and/or anti-CD28 (␣CD28) for 4 h with murine CD95L-specific primers. Band intensity was normalized to the ␤-actin product generated using specific primers to murine ␤-actin. The PCR products were resolved on a 1.5% agarose gel. Bands were visualized by staining with ethidium bromide, and optical density was determined using densitometry. where truncations to the promoter region increased inducible transcriptional activity, seemingly through the removal of upstream undefined inhibitory elements (9,14). Our data show that stimulated cells transfected with the p253 construct had 20 -30-fold higher inducible transcriptional activity than the p324 construct, although the relative level of CD28-mediated enhancement of activity is similar. These data suggest the presence of a transcriptional repressor element within the Ϫ324 to Ϫ253 bp region. To further localize this element, additional 5Ј truncations were made to the Ϫ324 bp region. The truncation mutant reporter constructs p309 (Ϫ309 to ϩ65 bp), p298 (Ϫ298 to ϩ65 bp), and p276 (Ϫ201 to ϩ65 bp), along with the p324 and p253 constructs, were assessed for transcriptional activity following activation (Fig. 6A). Fig. 6B shows that a 25-bp region between Ϫ276 and Ϫ253 strongly attenuated transcriptional activity of the CD95L promoter. A motif search reveals the Ϫ270 to Ϫ260 bp region bears sequence homology to an AP-1 binding element. To assess whether repressor function is dependent on the putative AP-1 element, the activation-dependent transcriptional activities of the p276, p276mut (GGACTCAGG to GAGCTAATG; mutated bases underlined), and p253 constructs were compared. Fig. 6C shows that muta-tion to the putative AP-1 element abrogated repressor activity in the p276 construct. Further, mutation to the AP-1 element also relieved transcriptional repression in the context of the Ϫ689 region (Fig. 6C), suggesting that transcriptional repression of CD95L is dependent on an intact AP-1 element in the Ϫ270 to Ϫ260 bp region.
To verify that the Ϫ270 to Ϫ260 bp region is a binding site for AP-1 family proteins, electrophoretic mobility shift assay and supershifting were performed using nuclear extracts prepared from unstimulated and ␣CD3and ␣CD3/28-activated A1.1 FIG. 5. CD28 costimulation enhances CD95L-dependent lytic activity. The CD95L-dependent lytic activity of A1.1 cells (A) or primary CD4ϩ cells left unstimulated (NS) or stimulated with anti-CD3 antibodies alone (␣CD3), anti-CD28 alone (␣CD28), or both anti-CD3 and anti-CD28 (␣CD3/28) (B) was determined using a 51 Cr release assay against CD95-positive and CD95-negative target cells (L1210-Fas and L1210, respectively). Lytic activity of all effectors against the CD95-negative targets was less than 5% specific lysis (data not shown). The bars represent S.E. of triplicate wells. Data shown are representative of eight independent experiments.
FIG. 6. The ؊270 to ؊260 region of the CD95L promoter is a functional transcriptional repressor element. A, the p324 luciferase constructs and 5Ј truncation mutants; B, transcriptional activity of the p324, p309, p298, p276, and p253 luciferase constructs. C, transcriptional activity of the p276 luciferase construct containing mutations to the Ϫ270 to Ϫ260 region (p276mut) or the p689 construct containing a deletion of the Ϫ270 to Ϫ260 region (p689mut2). The bars represent the S.E. of duplicate transfections. These data are representative of at least three independent experiments. cells and the Ϫ276 to Ϫ239 bp region (276wt) as a probe. Fig.  7A shows non-activation-dependent, constitutive binding of a protein complex to the 276wt probe that was susceptible to competition by the cold 276wt oligonucleotide but not the 276mut oligonucleotide, suggesting AP-1-specific binding. To verify AP-1 family member protein binding to the repressor element, supershifting was employed using antibodies to human c-Fos and c-Jun. Fig. 7B demonstrates that the addition of c-Fos but not c-Jun polyclonal antisera blocked complex binding to the 276wt probe. The specificity of anti-Fos-mediated inhibition was further demonstrated by showing that binding inhibition of the lower band (b) by c-Fos antiserum is dose-dependent (Fig. 7C). Taken together, these data demonstrate that the distal AP-1 element in the CD95L promoter is a functional repressor element and suggest that AP-1 family proteins play both positive and negative regulatory roles in the modulation of CD95L expression. DISCUSSION Costimulatory signals play a critical role in T cell fate including proliferation, effector cell function, and cell death, specifically activation-induced cell death (14, 22-24, 26, 28, 29, 32, 33, 39). The association of cytokines, such as IL-2, to each of these possible aspects of T cell physiology is clearly established, making IL-2 prototypical of costimulation-regulated genes. CD28 is the most widely studied costimulatory molecule expressed by the T cell. For this reason, numerous studies have investigated the mechanism of CD28-mediated regulation of IL-2 (3,6,9). The observation that CD28 costimulation enhanced the expression of IL-2 mRNA in the presence of translational inhibitors such as cyclohexamide suggested that the biochemical events associated with CD28 ligation were acting directly at the level of transcriptional regulation and by an "immediate early" pathway (9). Specifically, preformed factors capable of modulating transcription were being activated, translocating to the nucleus and binding to specific elements in the 5Ј regulatory region of the IL-2 gene (3,6,8,9). By establishing this direct regulatory pathway, a model of costimulation-mediated regulation of gene programs associated with T cell fate was established. Subsequently, the expression of other genes has been linked to the CD28 response element defined for IL-2, including CD154 (17,18). We present evidence in the current study that CD95L, a molecule that plays a critical role T cell fate, follows a similar model of regulation.
Previous studies established a critical role for CD95L in T cell homeostasis via its role in activation-induced cell death, as a mediator of cytotoxic activity in murine models of human disease such as experimental autoimmune encephalomyelitis, and in peripheral tolerance (30,31,34,35,40). The majority of the previous studies investigating CD95L regulation focused on TCR-mediated induction of transcriptional activity, because TCR signaling alone is sufficient to induce mRNA, protein, and CD95L-dependent cytolytic activity including activation-induced cell death. These studies established direct regulatory pathways linking TCR-mediated signaling and activation of nuclear factors that mediate transcription of CD95L such as nuclear factor of activated T cells, the Fos/Jun dimers, and NF-B (12,15,19,20,38,(41)(42)(43)(44). The results from the current study are in agreement with these data, demonstrating that signaling via anti-CD3 induces transcriptional activity and expression of mRNA, protein, and CD95-dependent apoptosis over unstimulated A1.1 cells. Although earlier studies showed that CD28 costimulation enhanced CD95L-dependent cytotoxicity in T cell lines, a regulatory link between CD28 costimulation and CD95L expression was not established (30,31,34,35,40). Norian et al. (16) were the first to show CD28-mediated enhancement of CD95L transcriptional activity in CD4ϩ T cells. Subsequently, studies investigating inducible gene programs associated with T cell costimulation utilized gene expression arrays to identify CD95L among hundreds of other CD28 costimulation-responsive genes (45,46). Whereas these studies have identified CD95L as CD28-responsive, the precise mechanism for CD28-mediated coregulation was not defined.
The current study demonstrates that CD28-mediated transcriptional control of the murine CD95L promoter is mediated FIG. 7. The ؊270 to ؊260 repressor region of the CD95L promoter constitutively binds c-Fos. Nuclear extracts made from A1.1 cells that were either unstimulated or stimulated for 30 min with ␣CD3 or ␣CD3 and ␣CD28 were employed in an electrophoretic mobility shift assay with an oligonucleotide corresponding to the Ϫ276 to Ϫ239 bp region of the CD95L promoter (276wt). A, binding reactions contained no extract (P) or nuclear extract with either a specific competitor of unlabeled 276wt oligonucleotide (S) or the nonspecific competitor 276mut oligonucleotide (NS). B, supershifting analysis of the nuclear extract prepared from ␣CD3and ␣CD28-stimulated A1.1 cells, using nonimmune polyclonal sera (Ig Con) or anti-human c-Fos or c-Jun. C, supershifting analysis showing dose-dependent inhibition of complex binding using increasing amounts of the c-Fos antiserum. The control Ig reaction mixture (Ig Con) contained 8 g of the control antiserum. via a short element located ϳ200 bp upstream of the transcription start site and is homologous with the CD28RE of the IL-2 promoter. Although functionally similar to the CD28RE of the IL-2 gene with respect to the ability to enhance gene expression, the current study demonstrates that deletion or mutation of the CD28RE of the CD95L promoter diminishes both CD28mediated enhancement and TCR-inducible transcriptional activity. These data are in contrast with previous studies for IL-2 gene expression showing that mutation of the CD28RE in the IL-2 promoter abrogated CD28-mediated enhancement but had no effect on TCR-mediated transcriptional activity (9). Similar to CD95L, however, loss in TCR-inducible activity following mutation of the functional CD28RE also was observed for the CD154 gene (17,18). These data suggest that although CD28mediated transcriptional regulation via homologous CD28REs of different genes is similar, the role of these elements in modulating overall transcriptional activity is gene-specific and contextual within the promoter regions of the different CD28reponsive genes. Although the present study did not directly assess whether functional CD28REs are present in the human CD95L promoter region, a motif search of the human CD95L promoter reveals several sequences bearing homology to the core sequence of the CD28RE in the murine CD95L promoter (GAAnTTC) located within 350 bp upstream from the transcription start site. Previous studies investigating costimulation-mediated enhancement of transcription of other human genes, including IL-2, IL-3, and granulocyte-macrophage colony-stimulating factor, have demonstrated that the functional CD28REs of the aforementioned genes are localized to a few hundred base pairs proximal to the transcription start sites (6,9). Deletional/mutational analysis, however, is required to confirm CD28RE function in the human CD95L gene.
The present study also identified and partially characterized a novel repressor element within the murine CD95L promoter. Several groups investigating CD95L regulation in human T cells have reported that specific promoter regions diminish activation-dependent transcription (13,14). Previous studies have shown that deletions of the Ϫ2365 to Ϫ454 bp, the Ϫ900 to Ϫ486 bp, and the Ϫ1204 to Ϫ860 bp regions of the human CD95L gene enhance basal and inducible transcriptional activity (13,14,47). Taken together, these previous studies suggest the presence of multiple elements capable of attenuating both basal and inducible transcriptional activity of the human CD95L promoter. Our current data demonstrate that the Ϫ270 to Ϫ260 bp repressor element constitutively binds c-Fos, a member of the AP-1 family of proteins previously shown to be associated with transcriptional activation of CD95L, IL-2, and other cytokines in the activated T cell. The ability of c-Fos to repress activation-dependent transcriptional activity via the binding of AP-1 elements in the promoters of several genes has been previously described (48 -51). The observation that c-Fos was constitutively bound to the AP-1 element and that the reporter constructs possessing a functional repressor AP-1 element were not hindered in the ability to be transactivated by TCR and CD28-mediated co-activation is novel. A motif search of the human CD95L promoter reveals several putative AP-1 elements located within each of the regions shown to repress basal and inducible transcriptional activity. Whether the multiple putative AP-1 elements located throughout the human CD95L promoter region mediate transcriptional repression will have to be determined empirically. These data suggest the possibility that AP-1 elements within the murine CD95L promoter and perhaps other gene promoters play roles as not only positive and negative transcriptional regulators but also as constitutive transcriptional "attenuators." Further analysis of other repressor elements in AP-1-responsive genes such as IL-2 is required to verify this hypothesis.
In summary, the present study identified two mechanisms for regulating CD95L, a gene that plays a key role in CD4ϩ T cell effector function and cell fate. First, we identified a functional CD28RE in the murine CD95L promoter that suggests a direct regulatory mechanism for co-stimulation-mediated induction of CD95L. Similar to studies investigating CD28-mediated regulation of IL-2 expression, the current study demonstrated for the first time CD28-mediated transcriptional regulation of CD95L correlated with increased steady state levels of mRNA, protein, and CD95L-mediated lytic activity. The CD28RE identified in the promoter of the CD95L shares both sequence homology and CD28 functional activity with the CD28REs of IL-2, CD154, and other cytokine genes by playing a major role in TCR-mediated transcriptional regulation. By identifying a direct mechanism of CD28 coregulation of CD95L in this study, CD95L is now added to a growing list of costimulation-regulated or enhanced genes that play major roles in T cell function. The present study also identified and characterized a transcriptional repressor element in the CD95L promoter that binds the AP-1 family protein c-Fos. Whereas several groups have reported transcriptional repressors in the CD95L gene, the distal AP-1 element appears to attenuate transcriptional activity rather than block TCR-or CD28-mediated transcriptional activation, suggesting a novel role for AP-1 mediated regulation. Further studies are required to determine whether this novel function of AP-1-mediated regulation plays a role in the transcriptional regulation of IL-2 or other TCRmediated cytokine genes.