The Cytoplasmic Domain of Fas Costimulates TCR Signals Mingyi Sun, Kristina T. Ames, Ivy Suzuki and Pamela J. Fink This information is current as of September 25, 2021. J Immunol 2006; 177:1481-1491; ; doi: 10.4049/jimmunol.177.3.1481 http://www.jimmunol.org/content/177/3/1481 Downloaded from

References This article cites 72 articles, 29 of which you can access for free at: http://www.jimmunol.org/content/177/3/1481.full#ref-list-1

Why The JI? Submit online. http://www.jimmunol.org/

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 25, 2021

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2006 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

The Cytoplasmic Domain of Fas Ligand Costimulates TCR Signals1

Mingyi Sun, Kristina T. Ames, Ivy Suzuki, and Pamela J. Fink2

Productive activation generally requires costimulation in addition to a signal delivered through the TCR. Although FasL is well-characterized for its capacity to deliver a death signal through Fas, this TNF family member can also transmit a reverse signal to enhance Ag-driven T cell proliferation. In this study, we define this reverse signal through FasL as costimulation by showing it requires TCR coengagement and is CD28 independent. We demonstrate that FasL-mediated costimulation drives FasL recruitment into lipid rafts and association with select Src homology 3 (SH3)-containing . We further show that the proline-rich intracellular domain of FasL is sufficient to costimulate by enhancing the phosphorylation of Akt, ERK1/2, JNK, and FasL itself, by activating the transcription factors NFAT and AP-1, and by enhancing IFN-␥ production. These results elucidate

the pathway of costimulation through the death inducer FasL, and comprise the first mechanistic analysis of a newly emerging Downloaded from group of costimulators, the TNF family. The Journal of Immunology, 2006, 177: 1481–1491.

ptimal T cell activation generally requires a costimula- (16, 17), CD30L (18), CD27L (19), the TNF-related activation- tory signal in addition to the signal delivered through the induced TRANCE (20), LIGHT (21–24), TNF-␣ (25, O TCR upon recognition of the MHC/peptide complex. 26), TRAIL (27), the B cell-activating factor BAFF (28, 29), and There are two main classes of molecules with known costimula- FasL (30–33). Because these molecules also induce signals deliv- http://www.jimmunol.org/ tory capacity for T cells, one of which includes Ig superfamily ered through their respective receptors, the costimulatory signals members such as ICOS and CD28 (1–3). Upon ligand binding, delivered in the reverse direction demonstrate the ability of these phosphorylation of the tyrosine residue within the YXXM motifs molecules to signal bidirectionally. How costimulatory TNF fam- in CD28 or ICOS cytoplasmic tails by Src family members pro- ily members balance these multiple tasks still remains unclear (34). motes association of these costimulators with adaptor proteins FasL is a 40-kDa type II transmembrane that has been such as PI3K and Grb2 (4). This association in turn leads to acti- well-characterized for its induction of cell death through the Fas vation of transcription factors, cytokine production, and cell pro- (35). A naturally arising point mutation in the FasL gene liferation (5). The second main class of T cell costimulators in- in gld mice results in the expression of nonfunctional FasL that can cludes TNFR family members such as OX40 (CD134), 4-1BB no longer bind Fas (36). Exploration of the reverse signaling ca- by guest on September 25, 2021 (CD137), CD30, CD27, and Fas (CD95) (6, 7). In contrast to pacity of FasL has provided both in vivo and in vitro evidence that CD28, which is constitutively expressed and costimulates naive T this molecule costimulates thymocyte maturation (33) and mature cells, and ICOS, which costimulates previously activated T cells, T cell proliferation (30–32). Despite the solid biological evidence most molecules in the TNFR family potentiate T cell activity at the that cross-linking both the TCR and FasL augments Ag-specific T peak of an (8, 9). Costimulation mediated by this cell proliferation by ϳ3-fold, little is known about the molecular second category of molecules is induced by association with members mechanism of this enhancement. Clues come from the structure of of the TNFR-associated factor (TRAF)3 family of adaptor molecules, the FasL cytoplasmic domain, which is the longest of the 19 leading to JNK activation and nuclear translocation of the transcrip- known TNF family members (37). Despite little cross-family se- tion factor NF-␬B (8). Recent studies have shown that the functional quence homology, the cytoplasmic domain of FasL is nearly as boundary between these classes of costimulators is not rigid, and that conserved across species (Fig. 1) as is the ligand-binding extra- cross-talk can occur between these two superfamilies (10). cellular domain (38). The FasL cytoplasmic domain does not con- In addition to the Ig and TNFR family members, a third group tain any ITAM, ITIM, YXXM, or TRAF interacting motifs, and of newly emerging costimulatory molecules consists of TNF fam- hence is likely to transduce signals distinct from those delivered by ily members such as CD40L (11–13), 4-1BBL (14, 15), OX40L the TCR and costimulators of the Ig or TNFR family. Two putative casein kinase I (CKI) binding motifs have been identified within the FasL cytoplasmic tail, including a conserved site in the TNF-␣ Department of Immunology, University of Washington, Seattle, WA 98195 tail that can be phosphorylated upon cell activation in vitro (39). Received for publication February 14, 2006. Accepted for publication May 10, 2006. Of interest in the search of a focal point for a signaling pathway, The costs of publication of this article were defrayed in part by the payment of page the FasL tail contains a striking proline-rich region not found in charges. This article must therefore be hereby marked advertisement in accordance any other TNF family member. This region includes several con- with 18 U.S.C. Section 1734 solely to indicate this fact. sensus Src homology 3 (SH3) and WW-binding domains, and 1 This work was supported by National Institutes of Health Grant AI 44130 (to P.J.F.) and by a Predoctoral Training grant from the Research Institute (to M.S.). FasL peptides spanning this region can associate in vitro with 2 Address correspondence and reprint requests to Dr. Pamela J. Fink, Department of many SH3- and WW-containing proteins, including Fyn, Grb2, Immunology, University of Washington, I-607H, Campus Box 357650, Seattle, WA FE65 and FBP11 (40, 41). Although it is unclear whether these 98195. E-mail address: pfi[email protected] proteins associate with FasL under physiological conditions in T 3 Abbreviations used in this paper: TRAF, TNFR-associated factor; CKI, casein ki- cells, they offer the promise of enabling cross-talk between the nase I; SH3, Src homology 3; HuIgG, human IgG1; KO, knockout; TM, transmem- brane domain; WT, wild type; CID, chemical inducers of dimerization; CIP, calf intestinal TCR and a potential costimulator. In T cells, FasL is synthesized phosphatase; HA, hemagglutinin; HPRT, hypoxanthine phosphoribosyltransferase. and retained within secretory vesicles to be transported to the cell

Copyright © 2006 by The American Association of Immunologists, Inc. 0022-1767/06/$02.00 1482 FasL REVERSE SIGNALING COSTIMULATES T CELLS

FIGURE 1. The cytoplasmic tail of FasL is highly conserved and contains potential signaling domains and docking sites. Schematic diagram of the FasL intracellular, TM, and extracellular domains, including amino acid sequences of murine, rat, and human FasL cytoplasmic regions. Trimerization, glyco- sylation, and receptor-binding sites are denoted in the extracellular domain. Small dots represent spaces left to maximize sequence alignment. Proline residues of the FasL cytoplasmic tail are marked in bold, the consensus SH3-binding sites are underlined, the SH3-binding site thought to interact with Fyn is overlaid with a thick line, boxes mark the two CKI binding motifs, and dots overlay serine, tyrosine, and threonine residues. There are no ITAM, ITIM, Downloaded from YXXM, or TRAF interacting motifs in the FasL tail. surface upon activation, and the proline-rich region in the FasL Cappel Pharmaceuticals supplied alkaline phosphatase-conjugated goat anti- cytoplasmic tail has been reported to be responsible for sorting rabbit IgG. Affinity purified goat anti-human IgG, anti-mouse IgG, and anti-hamster IgG were obtained from Rockland. FasIgG fusion protein,

FasL from the Golgi to secretory vesicles (42, 43). Moreover, re- http://www.jimmunol.org/ consisting of the extracellular domain of murine Fas joined to the hinge and cent studies suggest that the region between residues 2–33 in the constant regions of HuIgG, was purchased from R&D Systems, or used as cytoplasmic tail of FasL is both a negative regulator of FasL dialyzed sera from B6.SCID mice injected 5 wk previously with 1–2 ϫ 107 surface expression and a positive regulator of FasL-mediated cy- NIH 3T3 transfectants. The transfectants were derived by Dr. B. Stanger and totoxicity (44, 45). provided by Dr. A. Marshak-Rothstein (Boston University, Boston, MA). In this report, we investigate the molecular basis of FasL-me- Cell lines diated costimulation through reverse signaling in T cells. We show Jurkat cells were cultured in RP10 (RPMI 1640 with 10% FBS, 10 mM that costimulation mediated by FasL requires TCR coengagement HEPES, 4 mM L-glutamine, 100 U/ml penicillin, 100 ␮g/ml streptomycin, and is independent of costimulation through CD28. We also show 5.5 ϫ 10Ϫ5 M 2-ME, and 50 ␮g/ml gentamicin sulfate). H-2k-specific that FasL costimulation correlates with the association of FasL B6.WT, B6.lpr, and B6.gld CD8ϩ T cells were generated by coculturing by guest on September 25, 2021 with lipid rafts and select SH3-containing downstream adaptors, responder spleen and lymph node cells with irradiated (3000 rad) C3H splenocytes every 8–10 days in RP10, with the addition of 50 U/ml rIL-2 the phosphorylation of FasL on serine residues, the activation of ϩ after the third stimulation. H-2Ld-reactive CD8 L3 CTLs were stimulated Akt, ERK1/2, and JNK, the activation and nuclear translocation of every 8–9 days with irradiated BALB/c splenocytes and 50 U/ml rIL-2. transcription factors, and the induction of IFN-␥ transcription and Stable transfectants of L3 CTL were generated by electroporation at 250 secretion. Thus, while FasL is a death-inducing ligand for Fas, it mV, 975 ␮F (Bio-Rad), 5 days after restimulation. Electroporated cells also transduces a costimulatory signal through its intracellular do- were selected for 2–3 wk in medium containing 0.75 mg/ml G418 (In- vitrogen Life Technologies) and 25 U/ml rIL-2. main. This is the first comprehensive study of the mechanism of costimulation through a newly identified family of costimulators. Cell stimulation Responder T cells were seeded in plates coated overnight with 15 ␮g/ml Materials and Methods goat anti-hamster IgG alone, goat anti-HuIgG, or goat anti-mouse IgG Mice alone, or both, and after washing with PBS, for 4 h with 0.2–1.0 ␮g/ml suboptimal anti-CD3, and where indicated, with 5.0–10 ␮g/ml FasIgG or Wild-type C57BL/6 (B6.WT), B6.MRL-Faslpr (B6.lpr), B6 Smn.C3H- titered doses of anti-Ly49a. This system, in which the only cells were the Faslgld (B6.gld), C57BL/6-scid/SzJ (B6.SCID), and BALB/cJ mice were responders, eliminated the potential for nonresponder cell contamination in purchased from The Jackson Laboratory. C3HeB/FeJ and C57BL/6- subsequent experiments. Cd28tm1Mak mice (B6.CD28KO) were maintained in our animal facility using breeders obtained from The Jackson Laboratory. All mice were used Proliferation assays at 6–9 wk of age, in accordance with the Institutional Animal Care Use lpr gld Committee guidelines of University of Washington. For proliferation of B6. and B6. CTLs to mitogens in conjunction with FasL costimulation, 2 ϫ 104 responders were seeded in triplicate Reagents wells of 96-well plates to which appropriate mitogens were added at sub- optimal concentrations. Con A was added at 0.75 ␮g/ml, PMA and iono- Con A, cholera toxin, subunit B, and PMA were purchased from Sigma- mycin were used at 0.5 and 10 ng/ml, respectively, and azide-free anti- Aldrich, and ERK1/2 inhibitor PD98059, PI3K inhibitor LY294002, iono- Thy-1 was added in soluble form at 5 ␮g/ml. Con A stimulation required mycin, and human IgG1 (HuIgG) were obtained from Calbiochem. Anti- the presence of filler cells that were provided by irradiated B6 splenocytes. CD3 (145-2C11), anti-Thy-1 (G7), anti-FasL (Kay-10), and purified mouse For anti-CD3 activation with FasL costimulation, 2 ϫ 104 responders were anti-Ly49a were purchased from BD Pharmingen. Santa Cruz Biotechnol- seeded in triplicate wells of 96-well plates with 9.4 ␮g/ml plate-bound ogy supplied anti-FasL (Q-20), rabbit anti-Grb2 (C-23), anti-Fyn (FYN3), FasIgG plus 0.2 ␮g/ml plate-bound anti-CD3. For Thy-1 activation, 2 ϫ anti-Vav (C-14), anti-p85 PI3K (Z-8), anti-Lck (2102), anti-p-c-Jun (Ser63/ 105 spleen and lymph node cells pretreated with anti-CD4 plus guinea pig 73), anti-NFATc1 (H-110), and anti-PLC␥1 (1249). Rabbit anti-active p38 complement (Invitrogen Life Technologies) were used as responders. For was provided by Promega. Rabbit anti-phosphoserine and anti- FasL-blocking assays, soluble FasIgG was added to cultures at 2.8–10 hemagglutinin (HA) were purchased from Zymed Laboratories. Rabbit ␮g/ml on the day of Ag stimulation. HuIgG and/or B6.SCID preimmune anti-LAT, anti-ERK1/2, anti-phospho-ERK1/2, anti-phospho-JNK and anti- serum were used as negative controls. For proliferation of transfected L3 phospho-Akt were obtained from Technology, and Chemi- CTL, cells were added to triplicate wells with 0.2 ␮g/ml plate-bound anti- con International supplied mouse anti-IKK␤, anti-Brg1, and anti-actin. CD3 alone or with titered doses of plate-bound anti-Ly49a. Proliferation The Journal of Immunology 1483 was measured by [3H]TdR incorporation 18 h after pulsing with 1␮Ci/well pellets were resuspended in 1ϫ NEB buffer 3 with the addition of 20 U of on the indicated day. CIP, and the reaction mixtures were incubated at 37°C for 2 h. Recombinant DNA constructs Isolation of lipid rafts For constructs in chemical inducers of dimerization (CID) assay, the ␨ Equal numbers of L3 CTLs were lysed in 1 ml of ice-cold MBS (25 mM construct was provided by Dr. D. Spencer (Baylor College of Medicine, MES, 150 mM NaCl, 0.5% Triton X-100, protease inhibitor mixture (pH Houston, TX). In this construct, the myristoylation-target domain (M) is 6.5)). Lysates were homogenized 10 times through 22-gauge needles be- linked to the murine ␨ cytoplasmic domain, three FK506-binding protein fore being mixed with an equal volume of 85% sucrose (w/v) in lysis domains (FKBP3), and the HA epitope (HA), and is cloned into the eu- buffer, overlaid with 6 ml of 35% sucrose and 3 ml of 5% sucrose in lysis ϫ karyotic expression vector pBJ5 (46). To generate the murine FasL CID buffer with 1 mM Na3VO4, and centrifuged at 200,000 g for 16 h at 4°C. construct (FasL), ␨ and FasL cytoplasmic domains were exchanged by Eleven 1-ml fractions were then collected, starting at the top of the gradi- subcloning. The FasL cytoplasmic domain was generated by PCR from ent. Immunoblotting confirmed that lipid rafts were recovered from the murine FasL cDNA provided by Dr. D. Lynch (Amnis, Seattle, WA). low-density gradient fractions 3–6, while cytosolic material and soluble NFAT, AP-1, and NF-␬B firefly reporter constructs were provided by Dr. membranes were recovered predominantly in fractions 9–11. ␥ E. Clark (University of Washington, Seattle, WA), and IL-2 and IFN- ELISA reporters were provided by Dr. C. Dong (M. D. Anderson Cancer Center, Houston, TX). For the FasL/Ly49a chimeric construct, the extracellular Cells were left unstimulated or stimulated as indicated for 24 or 48 h. and transmembrane (TM) domains of murine Ly49a were fused to the Cell-free supernatants were collected and assessed for IFN-␥ protein levels cytoplasmic domain of murine FasL by PCR with overlapping primers, and by ELISA according to the manufacturer’s instructions (BD Pharmingen). cloned into the eukaryotic expression vector EGFP C1 (Invitrogen Life Technologies). Murine Ly49a cDNA was provided by Dr. L. Lanier (Uni- Results versity of California, San Francisco, CA). FasL costimulation requires concomitant TCR engagement Downloaded from Detection of IFN-␥ transcription by quantitative PCR Previous studies have demonstrated that FasL-mediated costimu- RNA was isolated free of contaminating genomic DNA using the Qiagen lation is required for CTLs to achieve maximal proliferation and RNeasy . Random hexamer-primed cDNA was generated using super- expansion in vitro and in vivo (30, 32), and that the effects of script II RNase reverse transcriptase kit (Invitrogen Life Technologies). To costimulation are most apparent under conditions of suboptimal ␥ detect IFN- transcription, real-time PCRs were performed and analyzed stimulation through the TCR (Ref. 30 and our unpublished data). on a Stratagene MX4000 using SYBR Green I solution (Molecular Probes) and Hot Start Taq polymerase (Qiagen). IFN-␥ primers for real-time PCR Costimulation can be measured by the ability of soluble FasIgG http://www.jimmunol.org/ were provided by Dr. M. Bix (University of Washington, Seattle, WA). For a fusion protein to block proliferation, and by the ability of highly given cDNA, relative abundance of each target was normalized to hypoxan- cross-linked FasIgG to enhance proliferation of Ag-activated T Ϫ⌬CT thine phosphoribosyltransferase (HPRT) according to the formula 2 , cells. To further examine FasL costimulation in CD8ϩ T cells, we where ⌬C ϭ C TARGET Ϫ C HPRT, where C is the cycle threshold. T T T T compared the influence of FasL cross-linking in the context of CID assay TCR-dependent stimulation by suboptimal levels of Con A and Equal amounts totaling 3 ␮g of combinations of CID constructs were co- anti-CD3 (Fig. 2A), and TCR-independent stimulation by subop- transfected into 3.2 ϫ 106 Jurkat cells with 0.9 ␮g of firefly luciferase timal levels of anti-Thy-1 and PMA plus ionomycin (Fig. 2B). reporter and 0.1 ␮goftheRenilla luciferase reporter pRL-TK (provided by Although both Con A and anti-Thy-1 stimulations require the ex- Dr. S. Levin, Zymogenetics, Seattle, WA) using 10 ␮l of Lipofectamine pression of the TCR/CD3 complex by the responding T cells, Con by guest on September 25, 2021 (Invitrogen Life Technologies). After 24 h, cells were stimulated with 50 A activates T cells in a TCR-dependent manner (47), whereas ng/ml PMA and 1.5 ␮M ionomycin (data not shown) or different dilutions of FK1012 (provided by Dr. D. Spencer) or AP1510 dimerizers (provided Thy-1 cross-linking operates via a distinct pathway (48). Both ϩ by Ariad Pharmaceuticals). Cells were lysed 10–14 h later, and firefly types of stimuli up-regulated FasL surface expression on CD8 T reporter activation was quantitated using the Promega Dual-luciferase re- cells (data not shown), demonstrating that TCR engagement is not porter assay kit. Values are presented as the ratio of emission from the an essential component of this process. However, the increase in firefly reporter relative to the Renilla transfection control. Each signal is normalized to the signal generated in the presence of medium alone. proliferation of WT and lpr (Fas mutant) responders upon FasL cross-linking by plate-bound FasIgG with a suboptimal dose of Preparation of nuclear extracts anti-CD3 was observed only with TCR-dependent stimulation (top Cells were counted and equal numbers of cells were washed with pre- panels, Fig. 2A), and not with TCR-independent stimulation (top chilled PBS and resuspended with five times the volume of the pellet using panels, Fig. 2B). The heightened sensitivity to FasL cross-linking buffer H (20 mM HEPES (pH 7.9), 1.5 mM MgCl2, 10 mM KCL, 1.0 mM of lpr responders may be due to their enhanced FasL expression DTT, and protease inhibitors). The lysates were passaged 10 times through 26 gauge needles and centrifuged at 20,000 ϫ g for 2 min. The pellets were (49, 50). FasL cross-linking did not enhance proliferation of gld resuspended in prechilled buffer D (20 mM HEPES (pH 7.0), 1.5 mM (FasL mutant) responders with either type of stimulus.

MgCl2, 0.2 mM EDTA, 20% glycerol, 0.42 M KCl, 1.0 mM DTT, and In line with these data, the addition of soluble FasIgG to block protease inhibitors), and rotated at 4°C for 20 min before being centrifuged FasL cross-linking inhibited FasL costimulation in cells activated ϫ for 2 min at 20,000 g, and the supernatants saved as the nuclear extracts. by TCR-dependent (bottom panels, Fig. 2A), but not by TCR-in- Coimmunoprecipitation and immunoblotting dependent stimuli (bottom panels, Fig. 2B). In addition, soluble FasIgG blocked FasL costimulation when allogeneic H-2k spleno- Cells were counted and equal numbers of cells were lysed at a concentra- ϩ tion of 1 ϫ 107 cells/ml in lysis buffer (1% Triton X-100, 20 mM Tris, 150 cytes were used to stimulate naive CD8 B6 T cells (Fig. 2C). mM NaCl, and protease inhibitor mixture (pH 7.4)) for 20 min with rock- Thus, assays of both plate-bound FasIgG-mediated induction and ing. The lysates were passaged 10 times through 22-gauge needles and soluble FasIgG-mediated blocking of FasL costimulation demon- ϫ centrifuged at 20,000 g for 10 min. The supernatants were precleared strate that TCR coengagement is required for the FasL-mediated 2- with protein A/G beads (Amersham Biosciences) for 3–4 h before incu- bation with 1–5 ␮g mouse anti-FasL for1hat4°C. Protein A/G beads were to 3-fold enhancement of T cell proliferation. Therefore, FasL added to supernatants, which were then incubated overnight at 4°C. The cross-linking provides a signal that costimulates the TCR-induced beads were washed three times with lysis buffer and boiled in the presence proliferation of both naive and previously activated CD8ϩ T cells. of SDS-loading buffer with 10 mM DTT. Protein complexes were resolved by 10%–15 SDS-PAGE, transferred to polyvinylidene difluoride mem- Costimulation through FasL is independent of CD28 branes, and immunoblotted with specific Abs. Calf intestinal phosphatase costimulation (CIP; New England Biolabs) treatment was included to address the phos- phate group specificity of the anti-phosphoserine Ab. For CIP treatment of The relationship between costimulation through the FasL and immunoprecipitates, samples were washed with three times lysis buffer, the CD28 molecules was explored by testing whether cross-linking 1484 FasL REVERSE SIGNALING COSTIMULATES T CELLS

FIGURE 2. FasL-mediated costimulation requires TCR coengagement and is independent of CD28 co- stimulation. A and B, B6.WT, B6.lpr, and B6.gld long- term CTL lines or anti-CD4 plus complement pretreated spleen and lymph node cells (for anti-Thy-1 stimulation) were cultured with (A) suboptimal doses of Con A (in the presence of irradiated syngeneic splenocytes) or plate-bound anti-CD3; B, suboptimal doses of PMA and ionomycin or anti-Thy-1. Proliferation was measured by 3 [ H]TdR uptake in cultures pulsed on day 3, with the Downloaded from exception of PMA and ionomycin stimulation, in which wells were pulsed on day 1. All data are averages of triplicate wells and represent the mean Ϯ SD of two experiments. FasL surface expression was up-regulated by each stimulus, as measured by flow cytometry of cells stained with anti-murine FasL. Top panels, FasL- mediated costimulation is provided by cross-linking http://www.jimmunol.org/ FasL with plate-bound FasIgG only in cultures activated with TCR-dependent stimuli. Percentages indicate pro- liferation normalized to that measured in the presence of HuIgG. Bottom panels, The addition of soluble FasIgG blocks FasL-mediated costimulation only in conjunction with TCR-dependent stimuli. Percentages indicate pro- liferation normalized to that measured in the absence of IgG reagents. C,[3H]TdR uptake by naive CD8ϩ T cells

from B6.WT and B6.CD28KO mice pulsed on day 3 of by guest on September 25, 2021 culture with allogeneic H-2k stimulators alone, or in the presence of 10 ␮g/ml soluble HuIgG or FasIgG during the entire culture period. Right panel includes 50 U/ml rIL-2. Percentages indicate normalized comparison to proliferation levels of B6.WT responders in the absence of soluble IgG. Data are representative of two indepen- dent experiments.

FasL on CD28-deficient cells results in costimulation of Ag-driven The FasL cytoplasmic domain alone can mediate costimulation T cell proliferation. Although CD8ϩ T cells from CD28 knockout To explore the mechanism of costimulatory (KO) mice proliferated poorly in response to allogeneic spleno- cytes compared with WT CD8ϩ T cells, this proliferative capacity through FasL, pharmacological regulators of oligomerization were was further reduced by blocking FasL costimulation with the ad- used to measure costimulation through the FasL cytoplasmic tail. ␨ dition of soluble FasIgG (left panel, Fig. 2C). Furthermore, the In these experiments, the and FasL cytoplasmic tails, lacking deficiency in CD28 costimulation, but not FasL costimulation, both TM and ligand-binding extracellular domains, were targeted could be overcome by addition of exogenous IL-2 (right panel, to the inner leaflet of the plasma membrane of Jurkat cells through Fig. 2C), consistent with previous results (51). These data dem- a myristoylation motif (Fig. 3A). These membrane-bound domains onstrate that FasL costimulation is independent of costimulation were then cross-linked by FK1012 or AP1510, chemical inducers through CD28 and cannot be substituted for by the addition of that oligomerize fusion molecules containing FK506-binding exogenous IL-2. domains (46, 52). The Journal of Immunology 1485 Downloaded from

FIGURE 4. The FasL intracellular domain costimulates TCR-induced CD8ϩ T cell proliferation. A, Schematic diagram of the FasL/Ly49a chi- meric construct encoding murine FasL cytoplasmic and Ly49a TM and extracellular domains. B, L3 CTLs stably transfected with a construct en- coding the chimeric FasL/Ly49a molecule or the empty EGFP C1 vector ␮ were cultured with 0.2 g/ml plate-bound anti-CD3 alone or with the in- http://www.jimmunol.org/ dicated doses of plate-bound anti-Ly49a. Proliferation was measured by FIGURE 3. The cytoplasmic domain of FasL is sufficient to mediate [3H]TdR uptake in cultures pulsed on day 2. costimulation. A, Murine FasL and ␨ fusion constructs encode an N-ter- minal myristylation motif (M) that tethers the molecules to the membrane, the relevant cytoplasmic domains, FKBP3, and a C-terminal HA epitope. The FasL cytoplasmic domain mediates costimulation in The addition of FK1012 or AP1510 induces hetero-oligomers of these nontransformed murine CD8ϩ T cells membrane-bound proteins. B, Jurkat cells cotransfected with combinations of CID constructs, firefly luciferase reporter (NFAT or AP-1) and Renilla To test further the capacity of the FasL cytoplasmic tail to co- ϩ luciferase reporter (pRL-TK) were subjected to CID analysis. The combi- stimulate TCR-derived signals in nontransformed murine CD8 T nations were: empty vector alone, ␨ϩempty vector, FasLϩempty vector, cells in a less artificial system, an additional construct was gener- by guest on September 25, 2021 and ␨ϩFasL. Protein analysis by immunoblotting with anti-HA showed ated to encode a fusion protein composed of the murine FasL cy- similar expression levels of constructs among all the transfectants (our toplasmic tail and the Ly49a TM and extracellular domains (Fig. unpublished data). Values are presented as the ratio of emission from the 4A). Using an Ly49a-specific Ab, FasL cross-linking was induced firefly reporter relative to the Renilla transfection control. The ratio of in stably transfected L3 cells, an established H-2d-specific CTL reporter to transfection control for cells incubated with medium alone was line of B6 origin (53). Cross-linking the fusion protein boosted the 1.2–3.9 ϫ 102. Each signal was normalized to the medium control and Significant differences anti-CD3-induced proliferative capacity of the transfectants in an ,ء .represents the mean Ϯ SD of three experiments relative to the ␨ only control. Using the two-tailed Student’s t test on the Ab dose-dependent manner (Fig. 4B). Further increasing the dose data from the optimal dimerizer concentration, p Յ 0.0058 for NFAT and of anti-Ly49a gradually decreased the proliferation of L3 transfec- p Յ 0.002 for AP-1. Similar results (p Յ 0.00035) were obtained with an tants, suggesting that the degree of cross-linking of the FasL chi- NF-␬B reporter. C, Jurkat cells cotransfected with combinations of CID meric molecules is crucial to its costimulatory capacity. Ly49a- constructs, Renilla and IL-2 or IFN-␥ firefly luciferase reporters were sub- specific Abs did not influence the anti-CD3-induced proliferation jected to CID analysis. Data represent the mean Ϯ SD of two experiments. of L3 CTLs transfected with empty vector alone (Fig. 4B). These data clearly illustrate that anti-CD3 stimulated nontransformed CD8ϩ T cells are susceptible to costimulatory signals transduced To assess the delivery of signals dependent on CID, Jurkat cells through the FasL cytoplasmic tail of a chimeric molecule cross- were cotransfected with the relevant ␨ and FasL CID constructs linked by an extracellular ligand. Thus, we are now capable of and firefly luciferase reporters of NFAT, AP-1, IFN-␥, IL-2, or measuring FasL-mediated costimulation through reverse signaling NF-␬B activity. Cotransfection with a Renilla luciferase reporter at both the molecular and cellular levels. served as a transfection control. cross-linking ␨ fusion proteins generated a minor up-regulation of NFAT reporter activity (Fig. FasL-mediated costimulation recruits FasL into lipid rafts 3B), recapitulating earlier experiments establishing the CID system To examine whether FasL localizes to lipid rafts upon FasL-me- (46). Although cross-linking FasL on its own had no effect on diated costimulation, sucrose gradient centrifugation was per- reporter activity, co-cross-linking ␨ and FasL cytoplasmic tails formed to separate lipid rafts and soluble membranes from L3 generated a dimerizer-dependent synergistic up-regulation of CTLs cultured with a suboptimal dose of plate-bound anti-CD3 NFAT, AP-1 (Fig. 3B), IFN-␥, IL-2 (Fig. 3C), and NF-␬B (our alone or with FasIgG. Although endogenous FasL in L3 CTLs was unpublished data) activity. These results offer the first direct evi- recruited into lipid rafts upon TCR cross-linking, this recruitment dence at the molecular level that oligomerization of the FasL cy- from soluble to raft-associated fractions was more complete with toplasmic domain is sufficient for FasL-mediated costimulation, TCR and FasL co-cross-linking (Fig. 5). Interestingly, the Src fam- and further emphasize the dependence of this signal on TCR ily tyrosine kinase Fyn has a similar redistribution pattern in rafts. cross-linking. In contrast, another Src family member, Lck, redistributed to rafts 1486 FasL REVERSE SIGNALING COSTIMULATES T CELLS

FIGURE 5. FasL-mediated costimulation recruits FasL into lipid rafts. A total of 3 ϫ 107 L3 CTLs were cultured for 1 h with either medium alone, or 0.72 ␮g/ml plate-bound anti-CD3 alone or with 5 ␮g/ml plate-bound FasIgG before being subjected to lipid raft isolation by sucrose gradient ultracen- trifugation. Equal volumes of gradient fractions were separated by 12% SDS-PAGE and blotted with anti-FasL (Q-20), anti-Fyn, anti-LAT, anti-Lck, and cholera toxin B subunit to detect respectively FasL, Fyn, LAT, Lck, and GM1 in lipid rafts. Blots with anti-LAT, anti-Lck, and cholera toxin B subunit confirmed that lipid rafts were recovered from low density gradient fractions 3–6, and cytosolic material and soluble membranes were recovered in fractions Downloaded from 9–11. The tabulated values represent the percentage of each protein in rafts, calculated as band densities in fractions 3–6/band densities in fractions 1–11. Data are representative of three experiments. upon TCR cross-linking, but translocated away from rafts with alone or with FasIgG in the presence or absence of the PI3K in- FasL co-cross-linking. Lipid rafts may thus provide a platform to hibitor LY294002. Compared with culture with a suboptimal dose http://www.jimmunol.org/ which FasL is translocated to facilitate recruitment of downstream of anti-CD3 alone, co-cross-linking TCR and FasL on L3 but not adaptors and induction of costimulatory signaling. gld T cells resulted in the enhancement of Akt phosphorylation by 9-fold, after normalization against FasL (Fig. 7A). This enhance- FasL is phosphorylated upon FasL-mediated costimulation ment was inhibited by LY294002. These data suggest that FasL ϩ Although there is only a single tyrosine residue and no YXXM transduces a costimulatory signal in CD8 T cells by associating motif in the murine FasL cytoplasmic tail, this domain contains with the p85 subunit of PI3K and activating Akt. Thus, FasL- two potential CKI-binding motifs and three threonine and eight mediated costimulation may operate through the PI3K pathway, serine residues (Fig. 1). To determine whether these serine resi- and would be expected to activate transcription factors such as dues are phosphorylated upon FasL-mediated costimulation, the AP-1 via the Ras/MAPK pathway. by guest on September 25, 2021 phosphoserine content was measured of FasL molecules immuno- To assess directly whether the Ras/MAPK pathway is involved precipitated from L3 and gld CTLs stimulated through the TCR in in FasL-mediated costimulation, the phosphorylation of ERK1/2 the presence or absence of plate-bound FasIgG. Although FasL upon TCR and FasL co-cross-linking was examined in cells cul- expression levels in the two T cell populations were very similar tured with a suboptimal dose of plate-bound anti-CD3 alone, plate- under each condition, FasL costimulation of L3 but not gld T cells bound FasIgG alone, or both. The level of phosphorylated ERK1/2 resulted in an almost 4-fold increase in phosphatase sensitive was augmented 6-fold in L3 but not gld T cells stimulated with serine residues on FasL (Fig. 6A). anti-CD3 plus FasIgG over that in cells stimulated with suboptimal anti-CD3 alone, after normalization against FasL (Fig. 7B and our Costimulation drives association of FasL with select unpublished data). Treatment of L3 cells with the ERK1/2 inhib- SH3-containing proteins itor PD98059 blocked ERK1/2 activation in the presence or ab- Given the SH3-binding domains found in the FasL cytoplasmic tail sence of FasL costimulation. Furthermore, the PI3K inhibitor (Fig. 1), SH3-containing molecules were investigated to identify LY294002 also blocked ERK1/2 activation, confirming our pre- ϩ proteins that interact with the FasL intracellular domain to induce diction that the costimulatory signal transduced by FasL in CD8 downstream signaling. Although FasL expression in both L3 and T cells depends on MAPK activation via the PI3K pathway. gld T cells was similar under each condition tested, TCR and FasL Similarly, FasL and TCR co-cross-linking enhanced JNK phos- co-cross-linking led to enhanced association between WT but not phorylation compared with that in cells stimulated with suboptimal gld FasL molecules and Fyn, Grb2, and the p85 subunit of PI3K, anti-CD3 alone (Fig. 7C), indicating that FasL costimulation in- compared with that with suboptimal anti-CD3 stimulation alone duces AP-1 activation indirectly through phosphorylating JNK. No (Fig. 6B). FasL cross-linking alone resulted in no enhancement of p38 activation was observed (our unpublished data). association (Fig. 6B). No association between FasL and Lck, Vav, or phospholipase C␥1 in L3 CTLs was detected, although these molecules were expressed (Fig. 6B and our unpublished data). The FasL cytoplasmic domain is sufficient to activate the MAPK Thus, FasL associates with some, but not all, SH3-containing mol- pathway ϩ ecules upon FasL-mediated costimulation in CD8 T cells. To determine whether the FasL intracellular domain alone is suf- ficient to enhance ERK1/2 phosphorylation, we measured the FasL-mediated costimulation enhances Akt, ERK1/2, and JNK effect of dimerizers on Jurkat transfectants of CID constructs. Al- phosphorylation though the level of total ERK1/2 expression in each transfectant To investigate whether Akt is involved in FasL-mediated costimu- was similar, cells transfected with constructs encoding the ␨ and lation, Akt phosphorylation was measured after L3 and gld CTLs FasL tails had 3- to 4-fold more ERK1/2 phosphorylation than were cultured with a suboptimal dose of plate-bound anti-CD3 cells transfected only with the construct encoding the ␨ tail (Fig. The Journal of Immunology 1487

c-Jun phosphorylation and translocation into the nucleus, a mea- surement of AP-1 activation (Fig. 8).

MAPK-dependent costimulation through FasL enhances IFN-␥ production To probe the immunologic consequences of FasL costimulation and the effect of MAPK inhibition on this costimulation, IFN-␥ transcription and secretion from L3 cells were measured. Co-cross- linking TCR and FasL on L3 cells augmented IFN-␥ transcription 2- to 3-fold compared with that in cells stimulated with suboptimal anti-CD3 alone (Fig. 9A). Similarly, total IFN-␥ secreted by L3 cells upon TCR and FasL co-cross-linking was enhanced ϳ4-fold in the presence of FasIgG (Fig. 9B). FasL cross-linking had no effect on IFN-␥ transcription or secretion in gld T cells (data not shown). Suppression by the ERK inhibitor PD98059 indicates that MAPK activation is essential for the induction of IFN-␥ transcription and secretion upon TCR cross-linking alone or with FasL co-cross- linking (Fig. 9). Thus, costimulation through FasL augments ϩ IFN-␥ production in CD8 T cells in a MAPK-dependent way. Downloaded from

Discussion Our results indicate that coengagement of the TCR is necessary for FasL cross-linking to enhance proliferation, defining this reverse signaling as costimulation (Fig. 2, A and B). This requirement en-

sures that FasL costimulation enhances neither the proliferation of http://www.jimmunol.org/ bystander T cells nor of Ag-specific T cells after Ag clearance. Because both TCR-dependent and TCR-independent means of ac- tivation can induce the expression of surface FasL (our unpub- lished data), this requirement for TCR coengagement by FasL co- stimulation is unrelated to FasL expression. Instead, after TCR and FasL co-cross-linking, downstream molecules such as Fyn, Grb2, and PI3K may be recruited to lipid rafts, along with the TCR and FasL, and thereby gain access to the FasL intracellular domain to form a scaffold for TCR and FasL cross-talk. The degree of FasL by guest on September 25, 2021 cross-linking appears to be important for delivery of a costimula- tory signal, as suggested by the influence on costimulation of FIGURE 6. FasL is phosphorylated and associates with select SH3-con- cross-linking Ab (Fig. 4) and dimerizer (Fig. 3) concentration. taining proteins upon FasL-mediated costimulation. A,L3andgld T cells Also suggestive is the absence of positive FasL reverse signaling ␮ were cultured for 1 h with either medium alone, or 0.72 g/ml plate-bound in the presence of soluble FasIgG (Fig. 2) or FasIgG bound directly ␮ anti-CD3 alone or with 5 g/ml plate-bound FasIgG. Lysates from 5 to to plates in the absence of a first stage Ab (our unpublished data 10 ϫ 106 cells were immunoprecipitated overnight with anti-FasL (Kay- and Ref. 54). 10). Cell lysates with or without CIP pretreatment were blotted with Ab specific for phosphoserine, and the blot was reprobed with anti-FasL (Q- Although CD28 is widely considered the primary costimulatory 20). Data are representative of two experiments. B, L3 cells were cultured receptor for initial expansion and survival of T cells, a number of for 1 h with medium alone, 0.72 ␮g/ml plate-bound anti-CD3 alone, 5 costimulators can modulate the quantity, quality, or duration of ␮g/ml plate-bound FasIgG alone, or both. Lysates from 5 to 10 ϫ 106 cells immune responses independently of CD28 (55). Our data indicate were immunoprecipitated overnight with anti-FasL (Kay-10). L3 cell ly- that FasL is one of these molecules that induces CD28-indepen- sates were blotted with Abs specific for Fyn, the p85 subunit of PI3K, dent costimulation (Fig. 2C). In addition, exogenous IL-2 com- Grb2, Lck, and FasL, as indicated. Cellular extracts corresponding to one- pensates for the deficiency of CD28 but not of FasL costimulation fifth to one-tenth of the cell lysates used for the immunoprecipitations were (Fig. 2C). Similar to CD28 and ICOS, FasL can costimulate both blotted at the same time to confirm the expression of each protein in the naive and preactivated T cells, but enhances proliferation by a cells. Data are representative of three experiments. comparatively subtle, but consistent, 2- to 3-fold. Using two separate approaches, we provide here the first direct evidence at the molecular level for FasL-mediated costimulation 7D). These data illustrate that FasL intracellular domain-mediated through reverse signaling. The CID system offers several advan- costimulation plays an essential role in the activation of ERK1/2. tages for assessing costimulation through FasL. It can generate higher order oligomers of the type we suspect generate the positive FasL costimulation drives NFAT nuclear translocation and AP-1 reverse signal through FasL. Furthermore, the absence of extra- activation cellular domains eliminates the death-inducing function of FasL To further investigate the molecular basis of FasL costimulation, and obviates the ensuing complications, thereby facilitating the NFAT translocation and AP-1 activation in the nucleus were ex- direct demonstration of the synergy between TCR- and FasL-de- amined. Co-cross-linking TCR and FasL in L3 but not gld T cells pendent signaling. Balancing these advantages is the artificiality of enhanced NFATc1 nuclear translocation compared with that in the system, both from cellular and molecular standpoints. In an cells stimulated with suboptimal anti-CD3 alone (Fig. 8). Again, alternative approach, the use of nontransformed murine transfec- L3 but not gld T cells costimulated through FasL showed enhanced tants of a construct encoding a chimeric FasL/Ly49a molecule 1488 FasL REVERSE SIGNALING COSTIMULATES T CELLS

solves this problem (Fig. 4). Here, the readout is cellular prolifer- ation, presumably the sum of the numerous molecular readouts our CID assay measures. Both systems complement each other. The CID data indicate that the FasL cytoplasmic domain is sufficient to activate the transcription factors NFAT, AP-1, and NF-␬B, and to up-regulate IL-2 and IFN-␥ promoter activity (Fig. 3). NFATc1 nuclear translocation and c-Jun phosphorylation upon TCR co- cross-linking with endogenous FasL molecules in nontransformed T cells (Fig. 8) confirm these findings and link transcription factor activation to IFN-␥ transcription and secretion in WT but not gld CD8ϩ T cells (Fig. 9 and our unpublished data). These data are consistent with a recent report that FasL cross-linking can activate NF-␬B and AP-1, although in this case, induction of IL-8 expres- sion was the final readout (56). Recent studies have shown that plasma membrane compartmen- talization plays an essential role in T cell activation and costimu- lation (57–59). After cross-linking, the TCR translocates into lipid rafts that function as platforms to recruit downstream signaling

molecules such as Src family kinases (60). It has been demon- Downloaded from strated that the related proteins TNFR1, Fas, and LIGHT localize into lipid rafts (23, 61, 62). Our data show that FasL is similarly recruited into lipid rafts after translocation to the cell surface upon TCR and FasL co-cross-linking (Fig. 5). Thus, rafts may help or- ganize the downstream mediators of FasL costimulation. Recent

data also demonstrate that the localization of FasL to lipid rafts http://www.jimmunol.org/ regulates FasL-induced killing activity (63). One molecular signpost of reverse signaling through FasL is the phosphorylation state of FasL itself. Co-cross-linking FasL and the TCR on WT but not gld T cells leads within an hour to the ap- pearance of phosphoserine residues in FasL, with no concomitant increase in FasL surface expression (Fig. 6A and our unpublished data). Because the cytoplasmic domain of gld FasL is identical with that of WT, these data also emphasize that Fas-mediated cross-linking of FasL is the essential trigger for the phosphoryla- by guest on September 25, 2021 tion of FasL serine residues and the events that follow. It is un- known which specific serine residues in the FasL cytoplasmic tail are phosphorylated, or which serine kinases phosphorylate them, although the presence of two conserved CKI sites in the cytoplas- mic domain of FasL is notable (Fig. 1). In contrast, costimulation through members of the Ig superfamily results in phosphorylation of tyrosine residues (4), suggesting that these two costimulatory pathways are distinct. Our work identifies several potential downstream mediators of FIGURE 7. FasL-mediated costimulation activates Akt, ERK1/2, and the FasL costimulation pathway. Unlike CD28 and the TCR, FasL JNK. A and C, T cells were cultured with either medium alone, or 0.72 does not associate with Lck, the Src family protein tyrosine kinase ␮g/ml plate-bound anti-CD3 Ϯ 5 ␮g/ml plate-bound FasIgG in the absence whose binding to the SH3 domain of CD28 leads to kinase acti- or presence of 20 ␮M PI3K inhibitor LY294002 (ϩLY). A, The levels of vation and sustained signaling at the immunological synapse (64). phosphorylated Akt, total actin, and total FasL expression were assessed An association between Lck and an SH3 domain-containing pep- after a 1-h incubation (judged to be the peak of Akt phosphorylation during tide of FasL has been reported in PMA- and ionomycin-stimulated a 0.5–3 h time course). The tabulated values represent the normalized ratios Jurkat cells (65), but our data are not consistent with this finding. of L3 band densities generated from phospho-Akt, FasL, or actin under ϩ each condition to that generated with medium alone. Data are representa- In our hands, endogenous FasL in normal murine CD8 T cells tive of three experiments. B, L3 cells were cultured with either medium interacts in an activation-dependent manner with another Src fam- alone, 5 ␮g/ml plate-bound FasIgG alone, or 0.72 ␮g/ml plate-bound anti- ily protein tyrosine kinase, Fyn (Fig. 6B). Interestingly, Fyn but CD3 Ϯ 5 ␮g/ml plate-bound FasIgG in the absence or presence of either 20 ␮M PI3K inhibitor LY294002 (ϩLY) or 30 ␮M ERK inhibitor PD98059 (ϩPD). The levels of phosphorylated ERK1/2, FasL, total alone. Data are representative of three experiments. D, A total of 4 ϫ 106 ERK1/2, and actin expression were assessed after a 3-h incubation (judged Jurkat cells were cotransfected with combinations of CID constructs. Sixty to be the peak of ERK phosphorylation during a 0.5–6 h time course). The to 70 h later, equal numbers of transfectants were incubated with 500 nM tabulated values represent the normalized ratios of L3 band densities gen- AP1510 for 30 min, and the levels of phosphorylated ERK1/2 and total erated from phospho-ERK1/2, FasL, ERK1/2 or actin under each condition ERK1/2 expression were assessed by immunoblotting. The tabulated val- to that generated with medium alone. Data are representative of two ex- ues represent the normalized ratios of band densities generated from phos- periments. C, The levels of phosphorylated JNK, FasL and total actin pho-ERK1/2 or total ERK1/2 from each transfectant to that generated from expression were assessed after a 3-h incubation. The tabulated values rep- cells transfected with empty vector. Protein analysis by immunoblotting resent the normalized ratios of L3 band densities generated from phospho- with anti-HA showed similar expression levels of constructs among all the JNK, FasL or actin under each condition to that generated with medium transfectants. Data are representative of three experiments. The Journal of Immunology 1489

FIGURE 8. FasL costimulation drives NFAT nuclear translocation and AP-1 activation. Cells were cultured for 6 h with either medium alone, or 0.72 ␮g/ml plate-bound anti-CD3 alone or with 5 ␮g/ml plate-bound FasIgG. The presence of NFATc1 and phospho-c-Jun in each cellular component was assessed by immunoblotting. Probing for the nuclear marker Brg1 and the cytoplasmic marker IKK␤ showed no cross-contamination of nuclear and Downloaded from cytoplasmic extracts. The tabulated values represent the normalized ratios of band densities generated from nuclear NFATc1, phospho-c-Jun or Brg1 under each condition to that generated with medium alone. Data are representative of three experiments. not Lck can also be recruited into rafts upon FasL costimulation Our data identify Grb2 as an additional activation-dependent (Fig. 5), suggesting that the association between FasL and Fyn FasL-interacting protein (Fig. 6B). Grb2 is an adaptor that consti- may be responsible for their raft localization. Although it is unclear tutively associates with Sos, thereby triggering the activation of http://www.jimmunol.org/ what role Fyn plays in FasL costimulation, we predict that it works Ras/MAPK pathway and the transcription of a number of proteins at the interface of signals through the TCR and FasL, and likely required for T cell activation and differentiation (68). We observe functions to phosphorylate downstream proteins. FasL shares with ERK1/2 activation upon FasL costimulation (Fig. 7B), similar to CD28 an association with PI3K and Grb2 (Fig. 6B), although FasL that observed through LIGHT costimulation in mouse T cells (23). interacts via the SH3 domain of this molecule and not, as does The FasL cytoplasmic tail is sufficient for this activation (Fig. 7D), CD28, with the SH2 domain (4). suggesting that signaling proteins such as Fyn, Grb2, and PI3K Akt is a known target of PI3K, and Akt activation in turn drives may interact with the FasL tail and indirectly activate ERK1/2. the activation of GSK3, further enhancing the production of IL-2 PI3K inhibition also blocks ERK1/2 activation induced by FasL by guest on September 25, 2021 by prolonging the nuclear residency of NFAT (66). Recent studies costimulation (Fig. 7B), further confirming that the PI3K pathway show Akt phosphorylation is induced by cross-linking another is involved in ERK1/2 activation upon FasL costimulation. We do TNF family member, CD27L (67). Our data indicate that FasL costimulation promotes the association of FasL with the p85 sub- not observe p38 activation through FasL costimulation, although unit of PI3K (Fig. 6B), and that Akt activity is enhanced by TCR this protein apparently can be activated by CD40L, TRANCE, and and FasL co-cross-linking (Fig. 7A). How this is achieved is not TRAIL cross-linking (20, 27, 69, 70). JNK is activated through clear, but we predict that FasL associates with and activates PI3K, FasL costimulation (Fig. 7C), which phosphorylates c-Jun, a com- which in turn activates Akt. ponent of the transcriptionally active AP-1 complex. Another com- ponent of the AP-1 complex is Fos, which is activated by the MAPK pathway (71). Suppression of IFN-␥ production through FasL costimulation by the ERK inhibitor suggests that the MAPK pathway plays an important role in FasL costimulation (Fig. 9). Taken together, these data suggest that FasL induces costimulation through the PI3K pathway, which leads to MAPK activation, tran- scription factor activation, cytokine production, and cell prolifer- ation. However, it is still not clear what links Akt and ERK1/2 activation, and how the MAPKs mediate transcription factor activation. The biological significance of TNF family member-mediated costimulation through reverse signaling is underexplored. Al- though FasL delivers a strong death signal to Fas-bearing cells, FIGURE 9. FasL-mediated costimulation induces IFN-␥ production. A reverse signaling through FasL costimulates FasL-expressing T total of 1.5 ϫ 106 L3 cells were cultured with either medium alone, or 0.5 cells at a very modest level. Why does one single molecule play ␮g/ml plate-bound anti-CD3 alone or with 5 ␮g/ml plate-bound FasIgG in two opposite roles by sending and receiving bidirectional signals? the presence or absence of 10 ␮M ERK inhibitor PD98059 (ϩPD). A, The answer is unclear, but what we know is that a very subtle ␥ Twenty-four hours later, RNA was isolated and IFN- transcription was variation of immune behavior can make all the difference between analyzed by real-time PCR and data were normalized to HPRT. Normal- health and illness (72). It is likely that bidirectional signaling ized IFN-␥ transcription from cells cultured in PBS was 0.1. B, Forty-eight hours later, cell-free supernatants were assessed for IFN-␥ production by through FasL-Fas interaction helps modulate the balance between ELISA. IFN-␥ secretion from cells cultured in PBS was below the level of life and death of the target cell and contributes to the fine tuning of detection. Data represent the mean Ϯ SD under each condition. the cellular and functional interactions in the . The 1490 FasL REVERSE SIGNALING COSTIMULATES T CELLS differential control of FasL expression may regulate these distinct 25. Eissner, G., S. Kirchner, H. Lindner, W. Kolch, P. Janosch, M. Grell, functions played by this complex molecule (31). P. Scheurich, R. Andreesen, and E. Holler. 2000. Reverse signaling through transmembrane TNF confers resistance to lipopolysaccharide in human mono- cytes and macrophages. J. Immunol. 164: 6193–6198. Acknowledgments 26. Harashima, S., T. Horiuchi, N. Hatta, C. Morita, M. Higuchi, T. Sawabe, H. Tsukamoto, T. Tahira, K. Hayashi, S. Fujita, and Y. Niho. 2001. Outside-to- We thank Ariad Pharmaceuticals, and Drs. D. Spencer, inside signal through the membrane TNF␣ induces E- (CD62E) expres- A. Marshak-Rothstein, E. Clark, C. Dong, A. Weinmann, M. Bix, S. Levin, sion on activated human CD4ϩ T cells. J. Immunol. 166: 130–136. L. Lanier, and D. Lynch for reagents, protocols, and constructs. We also 27. Chou, A. H., H. F. Tsai, L. L. Lin, S. L. Hsieh, P. I. Hsu, and P. N. Hsu. 2001. ␥ thank G. Turk for excellent technical assistance, K. Otto for contributions Enhanced proliferation and increased IFN production in T cells by signal trans- duced through TNF-related -inducing ligand. J. Immunol. 167: in the early stages of this work, and our colleagues for reading the 1347–1352. manuscript. 28. Huard, B., P. Schneider, D. Mauri, J. Tschopp, and L. E. French. 2001. T cell costimulation by the TNF ligand BAFF. J. Immunol. 167: 6225–6231. 29. Huard, B., L. Arlettaz, C. Ambrose, V. Kindler, D. Mauri, E. Roosnek, Disclosures J. Tschopp, P. Schneider, and L. E. French. 2004. BAFF production by antigen- The authors have no financial conflict of interest. presenting cells provides T cell co-stimulation. Int. Immunol. 16: 467–475. 30. Suzuki, I., and P. J. Fink. 1998. Maximal proliferation of cytotoxic T lympho- cytes requires reverse signaling through Fas ligand. J. Exp. Med. 187: 123–128. References 31. Suzuki, I., and P. J. Fink. 2000. The dual functions of Fas ligand in the regulation 1. Weiss, A., B. Manger, and J. Imboden. 1986. Synergy between the T3/antigen of peripheral CD8ϩ and CD4ϩ T cells. Proc. Natl. Acad. Sci. USA 97: receptor complex and Tp44 in the activation of human T cells. J. Immunol. 137: 1707–1712. 819–825. 32. Suzuki, I., S. Martin, T. E. Boursalian, C. Beers, and P. J. Fink. 2000. Fas ligand 2. Mueller, D. L. 2000. T cells: a proliferation of costimulatory molecules. Curr. costimulates the in vivo proliferation of CD8ϩ T cells. J. Immunol. 165:

Biol. 10: R227–R230. 5537–5543. Downloaded from 3. Frauwirth, K. A., and C. B. Thompson. 2002. Activation and inhibition on lym- 33. Boursalian, T. E., and P. J. Fink. 2003. Mutation in Fas ligand impairs maturation phocytes by costimulation. J. Clin. Invest. 109: 296–299. of thymocytes bearing moderate affinity T cell receptors. J. Exp. Med. 198: 4. Rudd, C. E., and H. Schneider. 2003. Unifying concepts in CD28, ICOS and 349–360. CTLA4 co-receptor signalling. Nat. Rev. Immunol. 3: 544–556. 34. Eissner, G., W. Kolch, and P. Scheurich. 2004. Ligands working as receptors: 5. Kane, L. P., J. Lin, and A. Weiss. 2002. It’s all Rel-ative: NF␬B and CD28 reverse signaling by members of the TNF superfamily enhance the plasticity of costimulation of T cell activation. Trends Immunol. 23: 413–420. the immune system. Cytokine Rev. 15: 353–366. 6. Siegel, R. M., F. K. M. Chan, H. J. Chun, and M. J. Lenardo. 2000. The multi- 35. Nagata, S., and P. Golstein. 1995. The Fas death factor. Science 267: 1449–1456. faceted role of Fas signaling in immune cell and autoimmunity. Nat. 36. Nagata, S., and T. Suda. 1995. Fas and Fas ligand: lpr and gld mutations. Im- Immunol. 1: 469–474. munol. Today 16: 39–43. http://www.jimmunol.org/ 7. Croft, M. 2003. Costimulation of T cells by OX40, 4-1BB, and CD27. Cytokine 37. Bodmer, J. L., P. Schneider, and J. Tschopp. 2002. The molecular architecture of Growth Factor Rev. 14: 265–273. the TNF superfamily. Trends Biochem. Sci. 27: 19–26. 8. Croft, M. 2003. Co-stimulatory members of the TNFR family: keys to effective 38. Orlinick, J. R., K. B. Elkon, and M. V. Chao. 1997. Separate domains of the T-cell immunity? Nat. Rev. Immunol. 3: 609–620. human Fas ligand dictate self-association and receptor binding. J. Biol. Chem. 9. Watts, T. H. 2005. TNF/TNFR family members in costimulation of T cell re- 272: 32221–32229. sponses. Annu. Rev. Immunol. 23: 23–68. 39. Watts, A. D., N. H. Hunt, Y. Wanigasekara, G. Bloomfield, D. Wallach, 10. Croft, M. 2005. The evolving crosstalk between co-stimulatory and co-inhibitory B. D. Roufogalis, and G. Chaudhri. 1999. A casein kinase I motif present in the receptors: HVEM-BTLA. Trends Immunol. 26: 292–294. cytoplasmic domain of members of the tumour necrosis factor ligand family is 11. Cayabyab, M., J. H. Phillips, and L. L. Lanier. 1994. CD40 preferentially co- ϩ implicated in “reverse signalling”. EMBO J. 18: 2119–2126. stimulates activation of CD4 . J. Immunol. 152: 1523–1531. 40. Hane, M., B. Lowin, M. Peitsch, K. Becker, and J. Tschopp. 1995. Interaction of 12. van Essen, D., H. Kikutaki, and D. Gray. 1995. CD40 ligand-transduced co- peptides derived from the Fas ligand with the Fyn-SH3 domain. FEBS Lett. 373: by guest on September 25, 2021 stimulation of T cells in the development of helper function. Nature 378: 265–268. 620–623. 41. Wenzel, J., R. Sanzenbacher, M. Ghadimi, M. Lewitzky, Q. Zhou, D. R. Kaplan, 13. Blair, P. J., J. L. Riley, D. M. Harlan, R. Abe, D. K. Tadaki, S. C. Hoffmann, D. Kabelitz, S. M. Feller, and O. Janssen. 2001. Multiple interactions of the T. While, T. Francomano, S. J. Perfetto, A. D. Kirk, and C. H. June. 2000. CD40 ϩ cytosolic polyproline region of the CD95 ligand: hints for the reverse signal ligand (CD154) triggers a short-term CD4 T cell activation response that results transduction capacity of a death factor. FEBS Lett. 509: 255–262. in secretion of immunomodulatory and apoptosis. J. Exp. Med. 191: 42. Bossi, G., and G. M. Griffiths. 1999. Degranulation plays an essential part in 651–660. regulating cell surface expression of Fas ligand in T cells and natural killer cells. 14. Pollok, K. E., Y. Kim, J. Hurtado, Z. Zhou, K. K. Kim, and B. S. Kwon. 1994. Nat. Med. 5: 90–96. 4-1BB T-cell antigen binds to mature B cells and macrophages, and costimulates 43. Blott, E. J., G. Bossi, R. Clark, M. Zvelebil, and G. M. Griffiths. 2001. Fas ligand anti-␮-primed splenic B cells. Eur. J. Immunol. 24: 367–374. is targeted to secretory lysosomes via a proline-rich domain in its cytoplasmic 15. Salih, H. R., P. A. Kiener, and V. Nussler. 2002. 4-1BB ligand–just another tail. J. Cell Sci. 114: 2405–2416. costimulating molecule. Int. J. Clin. Pharmacol. Ther. 40: 348–353. 16. Stuber, E., M. Neurath, D. Calderhead, H. P. Fell, and W. Strober. 1995. Cross- 44. Xiao, S., U. S. Deshmukh, S. Jodo, T. Koike, R. Sharma, A. Furusaki, S. S. Sung, linking of OX40 ligand, a member of the TNF/NGF cytokine family, induces and S. T. Ju. 2004. Novel negative regulator of expression in Fas ligand (CD178) proliferation and differentiation in murine splenic B cells. Immunity 2: 507–521. cytoplasmic tail: evidence for translational regulation and against Fas ligand re- tention in secretory lysosomes. J. Immunol. 173: 5095–5102. 17. Matsumura, Y., T. Hori, S. Kawamata, A. Imura, and T. Uchiyama. 1999. Intra- cellular signaling of gp34, the OX40 ligand: induction of c-jun and c-fos mRNA 45. Jodo, S., V. J. Pidiyar, S. Xiao, A. Furusaki, R. Sharma, T. Koike, and S. T. Ju. expression through gp34 upon binding of its receptor, OX40. J. Immunol. 163: 2005. Fas ligand (CD178) cytoplasmic tail is a positive regulator of Fas ligand- 3007–3011. mediated cytotoxicity. J. Immunol. 174: 4470–4474. 18. Wiley, S. R., R. G. Goodwin, and C. A. Smith. 1996. Reverse signaling via CD30 46. Spencer, D. M., T. J. Wandless, S. L. Schreiber, and G. R. Crabtree. 1993. ligand. J. Immunol. 157: 3635–3639. Controlling signal transduction with synthetic ligands. Science 262: 1019–1024. 19. Lens, S. M., P. Drillenburg, B. F. den Drijver, G. van Schijndel, S. T. Pals, 47. Weiss, A., and J. B. Imboden. 1987. Cell surface molecules and early events R. A. van Lier, and M. H. van Oers. 1999. Aberrant expression and reverse involved in human T activation. Adv. Immunol. 41: 1–38. signalling of CD70 on malignant B cells. Br. J. Haematol. 106: 491–503. 48. Gunter, K. C., R. N. Germain, R. A. Kroczek, T. Saito, W. M. Yokoyama, 20. Chen, N. J., M. W. Huang, and S. L. Hsieh. 2001. Enhanced secretion of IFN␥ C. Chan, A. Weiss, and E. M. Shevach. 1987. Thy-1-mediated T-cell activation by activated TH1 cells occurs via reverse signaling through TNF-related activa- requires co-expression of CD3/Ti complex. Nature 326: 505–507. tion-induced cytokine. J. Immunol. 166: 270–276. 49. Watanabe, D., T. Suda, H. Hashimoto, and S. Nagata. 1995. Constitutive acti- 21. Tamada, K., K. Shimozaki, A. I. Chapoval, G. Zhu, G. Sica, D. Flies, T. Boone, vation of the Fas ligand gene in mouse lymphoproliferative disorders. EMBO J. H. Hsu, Y. X. Fu, S. Nagata, et al. 2000. Modulation of T-cell-mediated immu- 14: 12–18. nity in tumor and graft-versus-host disease models through the LIGHT co-stim- 50. Chu, J. L., P. Ramos, A. Rosendorff, J. Nikolic-Zugic, E. Lacy, A. Matsuzawa, ulatory pathway. Nat. Med. 6: 283–289. and K. B. Elkon. 1995. Massive upregulation of the Fas ligand in lpr and gld 22. Scheu, S., J. Alferink, T. Potzel, W. Barchet, U. Kalinke, and K. Pfeffer. 2002. mice: implications for Fas regulation and the graft-versus-host disease-like wast- Targeted disruption of LIGHT causes defects in costimulatory T cell activation ing syndrome. J. Exp. Med. 181: 393–398. and reveals cooperation with ␤ in mesenteric lymph node genesis. 51. Shahinian, A., K. Pfeffer, K. P. Lee, T. M. Kundig, K. Kishihara, A. Wakeham, J. Exp. Med. 195: 1613–1624. K. Kawai, P. S. Ohashi, C. B. Thompson, and T. W. Mak. 1993. Differential T 23. Shi, G., H. Luo, X. Wan, T. W. Salcedo, J. Zhang, and J. Wu. 2002. Mouse T cell costimulatory requirements in CD28-deficient mice. Science 261: 609–612. cells receive costimulatory signals from LIGHT, a TNF family member. Blood 52. Amara, J. F., T. Clackson, V. M. Rivera, T. Guo, T. Keenan, S. Natesan, 100: 3279–3286. R. Pollock, W. Yang, N. L. Courage, D. A. Holt, and M. Gilman. 1997. A 24. Ye, Q., C. C. Fraser, W. Gao, L. Wang, S. J. Busfield, C. Wang, Y. Qiu, versatile synthetic dimerizer for the regulation of protein-protein interactions. A. J. Coyle, J. C. Gutierrez-Ramos, and W. W. Hancock. 2002. Modulation of Proc. Natl. Acad. Sci. USA 94: 10618–10623. LIGHT-HVEM costimulation prolongs cardiac allograft survival. J. Exp. Med. 53. Glasebrook, A. L., and F. W. Fitch. 1979. T-cell lines which cooperate in gen- 195: 795–800. eration of specific cytolytic activity. Nature 278: 171–173. The Journal of Immunology 1491

54. Desbarats, J., R. C. Duke, and M. K. Newell. 1998. Newly discovered role for Fas 64. Holdorf, A. D., K. H. Lee, W. R. Burack, P. M. Allen, and A. S. Shaw. 2002. ligand in the cell-cycle arrest of CD4ϩ T cells. Nat. Med. 4: 1377–1382. Regulation of Lck activity by CD4 and CD28 in the immunological synapse. Nat. 55. Bachmann, M. F., R. M. Zinkernagel, and A. Oxenius. 1998. Immune responses Immunol. 3: 259–264. in the absence of costimulation: viruses know the trick. J. Immunol. 161: 65. Ghadimi, M. P., R. Sanzenbacher, B. Thiede, J. Wenzel, Q. Jing, M. Plomann, 5791–5794. A. Borkhardt, D. Kabelitz, and O. Janssen. 2002. Identification of interaction 56. Imamura, R., K. Konaka, N. Matsumoto, M. Hasegawa, M. Fukui, N. Mukaida, partners of the cytosolic polyproline region of CD95 ligand (CD178). FEBS Lett. ␬ T. Kinoshita, and T. Suda. 2004. Fas ligand induces cell-autonomous NF- B 519: 50–58. activation and -8 production by a mechanism distinct from that of 66. Kane, L. P., and A. Weiss. 2003. The PI-3 kinase/Akt pathway and T cell acti- -␣. J. Biol. Chem. 279: 46415–46423. vation: pleiotropic pathways downstream of PIP3. Immunol. Rev. 192: 7–20. 57. Ramnik, X., T. Brennan, Q. Li, C. McCormack, and B. Seed. 1998. Membrane compartmentation is required for efficient T cell activation. Immunity 8: 723–732. 67. Garcia, P., A. B. De Heredia, T. Bellon, E. Carpio, M. Llano, E. Caparros, 58. Bromley, S. K., W. R. Burack, K. G. Johnson, K. Somersalo, T. N. Sims, P. Aparicio, and M. Lopez-Botet. 2004. Signalling via CD70, a member of the C. Sumen, M. M. Davis, A. S. Shaw, P. M. Allen, and M. L. Dustin. 2001. The TNF family, regulates T cell functions. J. Leukocyte Biol. 76: 263–270. immunological synapse. Annu. Rev. Immunol. 19: 375–396. 68. Koretzky, G. A. 1997. The role of Grb2-associated proteins in T-cell activation. 59. Schwartz, J. C. D., X. Zhang, S. G. Nathenson, and S. C. Almo. 2002. Structural Immunol. Today 18: 401–406. mechanisms of costimulation. Nat. Immunol. 3: 427–434. 69. Brenner, B., U. Koppenhoefer, A. Lepple-Wienhues, H. Grassme, C. Muller, 60. Xavier, R., and B. Seed. 1999. Membrane compartmentation and the response to C. P. Speer, F. Lang, and E. Gulbins. 1997. The CD40 ligand directly activates antigen. Curr. Opin. Immunol. 11: 265–269. T-lymphocytes via tyrosine phosphorylation dependent PKC activation. Biochem. 61. Legler, D. R., O. Micheau, M. A. Doucey, J. Tschopp, and C. Bron. 2003. Re- Biophys. Res. Commun. 239: 11–17. cruitment of TNR receptor 1 to lipid rafts is essential for TNF␣-mediated NF␬B 70. Brenner, B., U. Koppenhoefer, H. Grassme, J. Kun, F. Lang, and E. Gulbins. activation. Immunity 18: 655–664. 1997. Evidence for a novel function of the CD40 ligand as a signalling molecule 62. Muppidi, J. R., and R. M. Siegel. 2004. Ligand-independent redistribution of Fas in T-lymphocytes. FEBS Lett. 417: 301–306. (CD95) into lipid rafts mediates clonotypic T cell death. Nat. Immunol. 5: 182–189. 71. Krawczyk, C., and J. M. Penninger. 2001. Molecular motors involved in T cell 63. Nachbur, U., D. Kassahn, S. Yousefi, D. F. Legler, and T. Brunner. 2006. Post- receptor clusterings. J. Leukocyte Biol. 69: 317–330. transcriptional regulation of Fas (CD95) ligand killing activity by lipid rafts. 72. Germain, R. N. 2004. An innately interesting decade of research in immunology. Downloaded from Blood 107: 2790–2796. Nat. Med. 10: 1307–1320. http://www.jimmunol.org/ by guest on September 25, 2021