The Journal of Immunology

CD70-Driven Costimulation Induces Survival or Fas-Mediated of T Cells Depending on Antigenic Load

Felix M. Wensveen,* Peter-Paul A. Unger,*,† Natasja A. M. Kragten,*,† Ingrid A. M. Derks,* Anja ten Brinke,‡ Ramon Arens,x Rene A. W. van Lier,*,† Eric Eldering,* and Klaas P. J. M. van Gisbergen*,†

Apoptosis plays an essential role in the removal of activated CD8 T cells that are no longer required during or postinfection. The Bim-dependent intrinsic pathway of apoptosis removes effector CD8 T cells upon clearance of viral infection, which is driven by withdrawal of growth factors. Binding of Fas to Fas mediates activation-induced death in vitro and cooperates with Bim to eliminate CD8 T cells during chronic infection in vivo, but it is less clear how this pathway of apoptosis is initiated. In this study, we show that the costimulatory TNFR CD27 provides a dual trigger that can enhance survival of CD8 T cells, but also removal of activated CD8 T cells through Fas-driven apoptosis. Using in vitro stimulation assays of murine T cells with cognate peptide, we show that CD27 increases T cell survival after stimulation with low doses of Ag, whereas CD27 induces Fas-driven T cell apoptosis after stimulation with high doses of Ag. In vivo, the impact of constitutive CD70-driven stimulation on the accumulation of memory and effector CD8 T cells is limited by Fas-driven apoptosis. Furthermore, introduction of CD70 signaling during acute infection with influenza virus induces Fas-dependent elimination of influenza-specific CD8 T cells. These findings suggest that CD27 suppresses its costimulatory effects on T cell survival through activation of Fas-driven T cell apoptosis to maintain T cell during infection. The Journal of Immunology, 2012, 188: 4256–4267.

pon recognition of pathogen-derived peptides on APCs, Xl antagonize proapoptotic molecules such as Bim and Puma, and Ag-specific CD8 T cells within the naive CD8 T cell pool the balance of these molecules determines whether mitochondrial U expand and differentiate into effector CD8 T cells to outer membrane integrity is maintained or lost and survival or combat infection. After pathogen clearance, the effector CD8 T apoptosis ensues, respectively (4, 5). The other major pathway of cell population contracts, and only a fraction survives to become apoptosis, known as the extrinsic pathway, is mediated by death memory CD8 T cells that provide enhanced protection upon re- domain-containing molecules such as Fas and TRAIL , encounter with the pathogen (1, 2). Two separate signaling path- which can directly activate -8 and induce apoptosis upon ways, the intrinsic and extrinsic pathways of apoptosis, are in- ligand binding (6). volved in mediating T cell death to control the population size of The importance of cell death in CD8 T cell responses has been the T cell pool during infection (3). The intrinsic pathway is shown using mice lacking or overexpressing of the in- mediated by members of the Bcl-2 family of proteins, which trinsic and extrinsic pathway. Molecules such as Bim and Puma control integrity of the mitochondrial outer membrane and regu- play an instrumental role in executing apoptosis under conditions of late release of from the mitochondrial intermem- limiting growth factors. Mice lacking these proapoptotic proteins brane space to activate caspase-9 and induce formation of the display strongly enlarged T cell responses during acute infection apoptosome (4, 5). Prosurvival molecules such as Bcl-2 and Bcl- and delayed contraction upon pathogenic clearance (7, 8). The extrinsic cell death pathway mediated by (FasL), and *Department of Experimental Immunology, Academic Medical Center, 1105 AZ Fas does not respond to withdrawal. In contrast, Amsterdam, The Netherlands; †Department of Hematopoiesis, Sanquin Research at in vitro and in vivo experiments using Fas-deficient lpr mice Central Laboratory of The Netherlands Blood Transfusion Service and Landsteiner Laboratory, 1066 CX Amsterdam, The Netherlands; ‡Department of Immunopathol- have shown that Fas induces apoptosis of activated T cells upon ogy, Sanquin Research at Central Laboratory of The Netherlands Blood Transfusion repetitive antigenic stimulation in a process called activation- Service and Landsteiner Laboratory, 1066 CX Amsterdam, The Netherlands; and x induced cell death (9, 10). More recently, using mice deficient Department of Immunohematology and Blood Transfusion, Leiden University Med- ical Center, 2333 ZA Leiden, The Netherlands for both Fas and Bim, several groups have shown that the intrinsic Received for publication October 5, 2011. Accepted for publication March 1, 2012. and extrinsic pathways cooperate to control the size of CD8 T This work was supported by an Academic Medical Center Ph.D. grant and a Vici grant cell responses during some but not all viral infection models from The Netherlands Organization for Scientific Research. (11–13). After HSV infection, contraction depends exclusively on Address correspondence and reprint requests to Dr. Klaas P.J.M. van Gisbergen, Bim, whereas contraction after murine g-herpes virus infection Department of Hematopoiesis, Sanquin Research at Central Laboratory of The Neth- requires both Fas- and Bim-dependent pathways. Therefore, it was erlands Blood Transfusion Service and Landsteiner Laboratory, Plesmanlaan 125, 1066 CX Amsterdam, The Netherlands. E-mail address: [email protected] proposed that Bim controls contraction upon acute infection in Abbreviations used in this article: DN, double-negative; FasL, Fas ligand; HPRT, contrast to Fas that is predominantly involved in eliminating cells hypoxanthine phosphoribosyltransferase; LCMV, lymphocytic choriomeningitis vi- during chronic responses (11). rus; mLN, mediastinal lymph node; PI, propidium iodide; TCID50, 50% tissue CD27 is a member of the TNFR superfamily that upon binding of culture-infective dose; Tg, transgenic; WT, wild-type. its unique ligand CD70 provides costimulation and favors T cell Copyright Ó 2012 by The American Association of Immunologists, Inc. 0022-1767/12/$16.00 survival (14, 15). The expression of CD70 is tightly regulated www.jimmunol.org/cgi/doi/10.4049/jimmunol.1102889 The Journal of Immunology 4257 and only transiently present on dendritic cells, B cells, and T cells Quantitative PCR upon activation, which ensures that signaling through CD27 is RNAwas extracted using TRIzol reagent (Invitrogen); cDNAwas synthesized limited during times of infection or inflammation (16, 17). CD27 using RevertAid H Minus Reverse Transcriptase (Fermentas) and random enhances T cell responses through multiple mechanisms that hexamers (Invitrogen) in combination with poly T oligonucleotides (Invi- include induction of autocrine IL-2 production and shifting the trogen). Quantitative real-time PCR was performed on a StepOnePlus RT- balance of pro- and antiapoptotic molecules toward a prosurvival PCR system (Applied Biosystems). Transcription levels were obtained using FAST SYBR Green Mix (Applied Biosystems). The following primer sets state (18–20). The crucial role of CD70-driven costimulation were used for determining hypoxanthine phosphoribosyltransferase (HPRT) becomes evident postinfection with acute viruses such as vaccinia and FasL levels: HPRT (forward: 59-TGAAGAGCTACTGTAATGATCA- virus, vesicular stomatitis virus, and the A/NT/60/68 strain of the GTCAAC-39 and reverse: 59-AGCAAGCTTGCAACCTTAACCA-39)and influenza virus, during which CD70 and CD27 essentially con- FasL (forward: 59-TACCACCGCCATCACAACC-39 and reverse: 59-GAT- TTGTGTTGTGGTCCTTCTTCTT-39). Values are represented relative to tribute to the size of the CD8 T cell response (14, 21). Recently, that of HPRT and calibrated relative to expression in naive CD4 T cells. we found that CD27 does not have an apparent role in the regu- lation of the size of primary CD8 T cell responses postinfection Western blot with other strains of influenza (22). Close inspection revealed that Cells were lysed in 1% Triton X-100–containing lysis buffer at the in- CD27 favors the outgrowth of low-affinity CD8 T cells at the cost dicated time points after stimulation, and cell lysates were boiled in of high-affinity CD8 T cells, suggesting that the costimulatory sample buffer for Western blot analysis. SDS-PAGE gel electrophoresis effects of CD27 depend on the affinity of the Ag interactions (22). was performed using the Bio-Rad mini-PROTEAN electrophoresis sys- tem (Bio-Rad). Lysates were analyzed using primary Abs against Bim In fact, CD27 enhances the cell death of CD8 T cells stimulated (Stressgen), Bcl-Xl (Transduction Laboratories), and b-actin (Santa Cruz with high-affinity Ags, whereas CD27 improves T cell survival Biotechnology) and secondary IRDye 680- or 800-labeled Abs (Li-Cor). after stimulation with similar amounts of low-affinity Ags (22). A Western blots were visualized on an Odyssey Imager (Li-Cor). Densito- reverse role for CD27 signaling in effector CD8 T cell differen- metric analysis was performed using Odyssey Application software v3.0 (Li-Cor). tiation is supported by the CD27-driven reduction rather than enhancement of CD8 T cell responses after chronic infection with Flow cytometry lymphocytic choriomeningitis virus (LCMV) (23, 24). In line with Spleen, lungs, and mediastinal lymph nodes (mLN) were isolated and these findings, we have observed that CD70 transgenic (Tg) mice grinded over nylon filters (BD Biosciences) to obtain single-cell suspen- that have constitutive expression of CD70 on T cells are unable sions in PBS containing 0.5% BSA. Blood, spleen, and lung cell suspen- to maintain memory populations of CD8 T cells after influenza sions were subjected to erylysis (155 mM NH4Cl, 10 mM KHCO3, and infection (25). Taken together, this suggests that under certain 1 mM EDTA) to remove RBCs. An automated cell counter (CaseyCounter, Innovatis, or Beckman Coulter Counter) was used to obtain absolute cell conditions, CD70 is able to downregulate CD8 T cell responses, numbers. Cell suspensions were stained with fluorochrome-conjugated or possibly at the level of apoptosis. biotinylated primary Abs for 30 min at 4˚C in PBS 0.5% BSA. For bio- In this study, we addressed whether signaling through CD70 and tinylated Abs, fluorochrome-conjugated streptavidin (eBioscience) was CD27 directly impairs CD8 T cell differentiation. We found that used in the secondary step. FcgRII/RIII blockers were added to prevent nonspecific binding of Abs (2.4G2; kind gift of Louis Boon, Bioceros BV). costimulation through CD70 and CD27 enhances T cell survival Expression levels were analyzed using FACSCalibur and Canto flow after low antigenic stimulation, but increases T cell apoptosis upon cytometers (BD Biosciences). high antigenic stimulation. Using in vitro and in vivo settings, we show that CD70-driven costimulation engages the Fas pathway of In vitro T cell stimulation apoptosis to induce T cell death. OTI cells were isolated from the spleen of OTI TCR Tg mice using anti- CD8–coated beads, and total polyclonal T cells were isolated from the spleen of wild-type (WT) mice using anti-CD4– and anti-CD8–coated Materials and Methods microbeads (Miltenyi Biotec). To measure proliferation, T cells were la- Mice beled with 1 mM CFSE (Molecular Probes) in PBS for 10 min at 37˚C. As 2/2 2/2 3 2/2 APCs, splenocytes of Cd27 mice (referred to as control) and Cd27 C57BL/6, OTI, lpr, gld (all from The Jackson Laboratory), Cd27 (14), CD70-B cell Tg mice (referred to as CD70 Tg) were used. To analyze 2/2 3 2/2 3 Cd27 OTI (22), CD70-B cell Tg (26), Cd27 CD70-B cell Tg the effect of endogenous CD70 in these culture assays, we have made use 3 (26), CD70-T cell Tg (25), and lpr CD70-T cell Tg mice (27) were of activated WT B cells as APCs. WT B cells were isolated using anti- maintained in specific pathogen-free conditions at the animal department CD19 microbeads (Miltenyi Biotec) and stimulated for 4 d in the presence of the Academic Medical Center (Amsterdam, The Netherlands). Unless of 5 mg/ml anti-IgM (HB88), 5 mg/ml LPS (Sigma-Aldrich), and 5 mg/ml explicitly stated otherwise, CD70-T cell Tg mice that express CD70 on anti-CD40 (FGK45; kind gift of Louis Boon, Bioceros BV). We have also T cells are used throughout and referred to as CD70 Tg for simplicity. used mock and CD70-transfected COS cells as APCs in other experiments, Mice were used at 8–16 wk of age unless stated otherwise, and within as indicated. For the generation of COS-CD70 cells, murine CD70 mRNA individual experiments, mice were strictly age- and gender-matched. All was isolated from purified B cells stimulated in vitro for 3 d with LPS animal experiments were performed according to institutional and national and anti-IgM and converted into cDNA using Superscript II reverse tran- guidelines. scriptase (Invitrogen). Murine CD70 cDNA was amplified using forward (59-TTGGATCCACCATGCCGGAGGAAGGTCGC-39) and reverse pri- Abs mers (59-TTGCGGCCGCTCAAGGGCATATCCACTG-39) and cloned The following mAbs from eBioscience were used: anti-CD3 (145-2C11), into pcDNA3.1 (Invitrogen). Linearized plasmid was transfected into COS anti-CD4 (RM4-5), anti-CD8 (53-6.7), anti-CD27 (LG.3A10), anti-CD44 cells with Fugene 6 (Roche), and cells were selected with G418 (Invi- (IM7), anti-CD45.1 (A20), anti-CD62L (MEL-14), anti-CD70 (FR70), trogen). COS-CD70 cells were subjected to multiple rounds of sorting by anti-B220 (RA3-6B2), anti–IL-7Ra (A7R34), and anti–IL-2 (JES6-5H4 flow cytometry to select for cells expressing high levels of CD70. APCs and JES6-1A12). Anti-FasL (MFL3) and anti-Fas (Jo2) were purchased were irradiated (30 Gray) and labeled with 1 mM DDAO (Molecular from BD Biosciences. Probes) in PBS for 10 min at 37˚C. T cells and APCs were cocultured in a 1:1 ratio in 96-well, round-bottom plates (Corning) containing 200 ml Reverse transcriptase multiplex ligation-dependent probe RPMI 1640 10% FCS. OTI cells were stimulated with the indicated con- amplification centration of the cognate peptide SIINFEKL, and polyclonal T cells were stimulated with anti-CD3 (145-2C11; kind gift of Louis Boon, Bioceros Total RNA was extracted using the TRIzol isolation method (Invitrogen). BV) that was coated onto the 96-well round-bottom plates for 1 h in PBS Levels of mRNA were analyzed with the Apoptosis Mouse mRNA RT- in a concentration of 5 mg/ml. Proliferation of DDAO2 T cells was MLPA (MRC-Holland) according to the manufacturer’s instructions. measured using CFSE dilution and cell death by staining for Annexin V Samples were run through a Genescan, analyzed with GeneMapper soft- (BD Biosciences) or propidium iodide (PI; Molecular Probes) after 3 d of ware (Applied Biosystems), and further processed using Excel software. stimulation. 4258 CD70 INDUCES Fas-DRIVEN T CELL DEATH

Influenza infection and adoptive transfer specific CD8 T cells were enumerated using allophycocyanin-conjugated tetramers of H-2Db containing the LCMV-derived peptide GP 3 33–41 Mice were intranasally infected with 10 50% tissue culture-infective KAVYNFATC (Sanquin Reagents). dose (TCID50) of the H1N1 influenza A virus A/PR8/34 for analysis of primary immune responses. Heterotypic infection with 1003 TCID50 of Statistical analysis the H3N2 influenza A virus HKx31 and 103 TCID50 A/PR8/34 was performed to examine secondary responses. Blood samples were drawn All experiments have been performed at least twice with similar results. from the tail vein to determine the levels of influenza-specific CD8 Figures represent means, and error bars denote SEM. Two-tailed Student t test was used to analyze for statistical significance. A p value ,0.05 was T cells at the specified days postinfection. Mice were sacrificed, and , , blood, spleen, mLN, and lungs were collected for analysis at the indi- considered statistically significant (*p 0.05, **p 0.005). cated days postinfection. Influenza-specific CD8 T cells were enumerated using anti-CD8 Abs and PE- or allophycocyanin-conjugated tetramers Results b of H-2D containing the influenza-derived peptide NP366–374 ASNEN- The outcome of CD70-driven costimulation depends on METM (Sanquin Reagents). For adoptive transfer, CD8 T cells were purified from the spleen by selection using anti-CD8–coated microbeads antigenic strength 6 (Miltenyi Biotec), and 10 3 10 CD8 T cells were i.v. injected per re- Upon binding of CD70, CD27 enhances expansion of T cells by cipient mouse. improving survival (15) and proliferation (28). Indeed, as previ- LCMV infection ously published (18, 28), we found that TCR Tg OVA-specific OTI cells activated with low amounts of peptide proliferate faster Mice were i.v. infected with 2 3 106 PFU LCMV clone 13 virus for analysis of chronic immune responses. Mice were sacrificed, and spleens and/or display enhanced survival in the presence of CD70 Tg 2/2 were collected for analysis at the indicated days postinfection. LCMV- APCs compared with control APCs (Fig. 1A). Cd27 OTI cells

FIGURE 1. CD70-driven costimulation enhances T cell survival after low anti- genic triggering, but induces apoptosis after strong antigenic triggering. (A–C) WT or Cd272/2 OTI cells were stimulated for 3 d with the indicated concentrations of SIINFEKL peptide in the presence of splenocytes of Cd272/2 mice (control APCs) or splenocytes of Cd272/2 3 CD70 Tg mice (CD70 Tg APCs). (A) Represen- tative dot plots display proliferation using CFSE dilution and cell death using staining with PI of WT and Cd272/2 OTI cells stimulated in the presence of control or CD70 Tg APCs. (B) Cell death was ana- lyzed using PI staining of proliferating (CFSElo) WT OTI cells stimulated with control or CD70 Tg APCs (left panel) and, similarly, of proliferating WT and Cd272/2 OTI cells stimulated with CD70 Tg APCs (right panel). (C) Apoptosis was determined using staining for Annexin V of proliferat- ing WT (top panel) and Cd272/2 OTI cells (bottom panel) stimulated for 3 d with 10 pg/ml peptide in the presence of control or CD70 Tg APCs. (D) Histograms display expression of CD70 on B cells of Cd272/2 mice (control) and Cd272/2 3 CD70 Tg mice (CD70 Tg, top panel) and on non- stimulated (control) and activated B cells of WT mice (bottom panel). (E and F) Ac- tivated WT B cells were used as APCs to stimulate WT and Cd272/2 OTI cells for 3 d in the presence of the indicated concen- trations of peptide Ag. (E) Representative dot plots display proliferation using CFSE dilution and cell death using PI staining of WT and Cd272/2 OTI cells. (F) Cell death of proliferating (CFSElo) WT and Cd272/2 OTI cells was quantified using staining with PI. *p , 0.05, **p , 0.005. The Journal of Immunology 4259 did not have this advantage, showing that interactions between CD70 and CD27 are required for proliferation and/or survival of OTI cells to occur after low antigenic triggering (Fig. 1A). Using PI to examine T cell death, we observed that proliferating WT OTI cells remained alive in contrast to Cd272/2 OTI cells after stim- ulation with low doses of peptide in the presence of CD70 Tg APCs (Fig. 1A, 1B). Strikingly, under higher concentrations of peptide, the positive effects of CD70 and CD27-driven costimu- lation on T cell proliferation and/or survival were not apparent (Fig. 1A, 1B). In fact, activation of WT OTI cells, but not Cd272/2 OTI cells, with high amounts of peptide resulted in enhanced cell death in the presence of CD70 Tg APCs compared with control APCs (Fig. 1B). Staining for Annexin V demonstrated that CD70-driven T cell death under these conditions occurred through apoptosis (Fig. 1C). Next, we tested whether endogenous CD70 was able to induce similar effects in OTI cells as Tg CD70. Therefore, isolated WT B cells were stimulated with anti-IgM, LPS, and anti-CD40, which has been shown previously to induce endogenous CD70 expression (29). Although the expression level of CD70 on activated WT B cells was lower than on B cells of CD70 Tg mice (Fig. 1D), these levels of endogenous CD70 ex- pression were sufficient to enhance proliferation of OTI cells after stimulation with low doses of peptide and to induce cell death of OTI cells after stimulation with high doses of peptide FIGURE 2. CD70 engages the extrinsic pathway of apoptosis. (A and B) 2 2 (Fig. 1E, 1F). Taken together, this shows that CD70 induces Ly5.1+ WT and Ly5.2+Cd27 / OTI cells were separately cultured or T cell death rather than T cell proliferation after strong antigenic cocultured with 0.3, 3, or 30 pg/ml SIINFEKL peptide in the presence of triggering. CD70 Tg APCs. Staining for Ly5.1 was used to distinguish between WT and Cd272/2 OTI cells. (A) The percentage of proliferating WT and Cell-intrinsic and -extrinsic factors mediate CD70-driven Cd272/2 OTI cells in separate and cocultures was determined using CFSE costimulation dilution. (B) The percentage of proliferating OTI cells that underwent cell death was determined using PI staining. **p , 0.005. To study whether soluble or cell-bound factors mediated the co- stimulatory effects of CD70 and CD27, we setup a coculture system, in which Ly5.1+ WT and Ly5.2+Cd272/2 OTI cells were the contribution was not very prominent in the presence of the mixed in a 1:1 ratio and stimulated with peptide in the presence of higher doses of peptide Ag that were required to induce CD70- CD70 Tg APCs. After stimulation with low doses of peptide, driven T cell death (Fig. 3A). FasL was also upregulated on Cd272/2 OTI cells cocultured with WT OTI cells proliferated OTI cells in the presence of CD70-driven costimulation, but in more extensively than Cd272/2 OTI cells cultured separately on contrast to the induction of Fas, this effect occurred in particular CD70 Tg APCs (Fig. 2A). This shows that the Cd272/2 OTI cells after stimulation with intermediate to high peptide doses (Fig. benefited from CD70-driven costimulation of WT OTI cells. De- 3A). Consistent with the induction of FasL , CD70 induced spite this benefit on proliferation, Cd272/2 OTI cells displayed expression of FasL at the mRNA level in the presence of high but a higher degree of cell death compared with cocultured WT OTI not low doses of peptide (Fig. 3B). This suggested that CD70 cells on CD70 Tg APCs (Fig. 2B). Thus, under low antigenic induced T cell death predominantly through transcriptional up- triggering, CD70 induced costimulation through a cell-extrinsic regulation of FasL expression and engagement of the Fas pathway factor that increased T cell proliferation and through a cell- of apoptosis. Indeed, blockade of FasL–Fas signaling through that enhanced T cell survival. In the presence of blocking anti-FasL Abs impaired CD70-induced cell death in WT intermediate Ag doses, the beneficial effects of CD70-driven co- OTI cells (Fig. 3C), but did not have a dramatic effect on cell 2/2 stimulation disappeared, and proliferation and cell death of WT death of Cd27 OTI cells (Fig. 3D). Thus, strong antigenic and Cd272/2 OTI cells in the presence of CD70 Tg APCs was stimulation allows CD70 to induce T cell death via upregulation similar under separate and coculture conditions (Fig. 2A, 2B). of FasL and subsequent triggering of Fas. Strikingly, the enhanced cell death that was observed after stimu- To examine whether costimulation through CD70 and CD27 lation of WT OTI cells with high amounts of peptide in the pres- similarly affected polyclonal T cell populations after high anti- ence of CD70 Tg APCs was also evident in cocultured Cd272/2 genic triggering, total T cells of naive WT mice were stimulated OTIcells,butnotinseparatelyculturedCd272/2 OTI cells (Fig. with high levels of anti-CD3 Abs in the presence of control or 2B). This indicates that CD70-driven T cell death after strong an- CD70 Tg APCs. We did not observe that CD70 augmented the tigenic triggering is mediated by the extrinsic pathway rather than survival or the proliferative capacity of CD4 and CD8 T cells under the intrinsic pathway of apoptosis. these conditions (Fig. 4A and data not shown). In contrast, the presence of CD70 Tg APCs increased T cell death in both CD4 CD70 induces T cell death through FasL and Fas and CD8 T cells (Fig. 4A, 4B). T cells of Cd272/2 mice were not Fas (CD95) is an important death-inducing receptor that mediates affected by CD70-driven costimulation (Fig. 4A, 4B). Staining for activation-induced cell death of T cells upon binding to its ligand Annexin V demonstrated that CD70 enhanced cell death of WT 2 2 FasL (CD178). To examine whether CD27-driven costimulation but not Cd27 / T cells through apoptosis (Fig. 4C). To examine activated FasL and Fas-driven apoptosis, we analyzed expression of whether these findings were reproducible using different APCs, FasL and Fas on OTI cells in response to peptide stimulation and we made use of COS cells that were transfected with CD70. In CD27 triggering. CD70 induced expression of Fas on OTI cells, but contrast to mock-transfected COS cells, CD70-transfected COS 4260 CD70 INDUCES Fas-DRIVEN T CELL DEATH

FIGURE 3. CD70 upregulates FasL and in- duces Fas-driven T cell death. (A) Histograms depict the expression of FasL (top panel) and Fas (bottom panel) on WT and Cd272/2 OTI cells that were stimulated for 3 d with the indicated concentration of SIINFEKL peptide in the pres- ence of CD70 Tg APCs. (B) The levels of FasL mRNA were determined in WT and Cd272/2 OTI cells using quantitative PCR after 3 d of stimulation with the indicated concentration of peptide in the presence of CD70 Tg APCs. WT OTI cells (C) and Cd272/2 OTI cells (D)were cultured for 3 d with or without blocking anti- FasL Abs in the presence of the indicated con- centrations of cognate peptide and CD70 Tg APCs. Cell death was analyzed in proliferating (CFSElo) cells using PI staining. *p , 0.05, **p , 0.005.

cells expressed intermediate to high levels of CD70 (Fig. 4D). death, costimulation through CD70 and CD27 induced upregu- Similar results were obtained after activation of T cells in the lation of Bcl-Xl and Bcl-2 expression. Interestingly, CD70 also presence of COS cells transected with CD70, demonstrating that induced downregulation of Bim expression, but did not interfere the CD70-driven induction of T cell death was independent of the with expression of other molecules involved in apoptosis or sur- APC (Fig. 4E). Thus, signaling through CD70 and CD27 also vival of T cells (Fig. 5C). To confirm these findings at the protein triggered T cell apoptosis in polyclonal T cell populations after level, Bcl-Xl and Bim expression were analyzed by Western blot strong stimulation with Ag. After triggering with anti-CD3, CD70 in T cells after stimulation with mock or CD70-transfected COS and CD27 induced upregulation of FasL in contrast to Fas on cells. CD70 upregulated the expression of Bcl-Xl (Fig. 5D) and polyclonal T cells (Fig. 4F), suggesting that CD70-driven T cell downregulated the expression of Bim in T cells after 24–48 h of death occurred through upregulation of FasL and engagement of stimulation (Fig. 5E). The effect of CD70 was in particular strong the Fas pathway of apoptosis. Indeed, T cells from lpr mice that on the largest isoform of Bim (Bim EL) and occurred to a lesser lack expression of Fas or gld mice that are deficient for FasL were extent on the other isoforms of Bim (Fig. 5E). Thus, under these protected against CD70-induced apoptosis (Fig. 4G). Moreover, conditions, CD70 and CD27 shifted the balance of Bcl-2 mole- addition of blocking Abs against FasL prevented T cell death in cules in T cells toward an apoptosis-resistant state. This supports the presence of CD70 Tg APCs, but had no effect on T cell the notion that CD70 elicits costimulatory activity after stimula- death in the presence of control APCs (Fig. 4H). Thus, signaling tion of T cells with high antigenic loads despite the induction of through CD70 and CD27 engages the Fas pathway of apoptosis T cell apoptosis. in polyclonal populations of CD4 and CD8 T cells after strong CD70 accelerates accumulation of Fas-sensitive B220+ antigenic triggering. double-negative T cells in vivo CD70 and CD27 induce T cell death despite costimulatory To study the Fas-dependent effects of CD70 and CD27 signaling activity in vivo, mice that constitutively express CD70 on T cells (25) were Costimulation through CD70 and CD27 has also been shown to crossed onto Fas-deficient lpr mice. We observed that these mice induce upregulation of IL-2 and the antiapoptotic Bcl-2 family had dramatically enlarged LNs compared with WT, CD70 Tg, and member Bcl-Xl, which promotes T cell survival (18–20). We lpr mice (Fig. 6A). In particular, cervical and axillary LNs were questioned whether CD70 differentially regulated expression of enlarged (Fig. 6A). The enlarged LNs of lpr 3 CD70 Tg mice IL-2 and Bcl-2 family members in T cells after stimulation with correlated with the unusual presence of a large proportion of high antigenic doses. The production of IL-2 was analyzed using double-negative (DN) T cells that, in contrast to conventional CD4 ELISA in OTI cells stimulated with increasing amounts of pep- and CD8 T cells, expressed the B cell marker B220 (Fig. 6B). tide in the absence or presence of CD70-driven costimulation. B220+ DN T cells are normally eliminated within lymphoid tissue After triggering of CD70, WT but not Cd272/2 OTI cells pro- through Fas-dependent apoptosis and therefore are hardly detect- duced much higher amounts of IL-2, particularly in the pres- able within WT mice, but accumulate in lpr mice (30). Similarly, ence of high doses of Ag (Fig. 5A). Similarly, CD70 induced patients with mutations in FasL or Fas develop autoimmune production of IL-2 in polyclonal T cells stimulated with high lymphoproliferative syndrome that is manifested by a dramatic amounts of anti-CD3 (Fig. 5B). The expression of Bcl-2 family increase in DN T cells and consequent enlargement of, in par- members was determined using reverse-transcriptase multiplex ticular, the cervical LNs (31). We observed that the accumulation ligation-dependent probe amplification in strongly activated of B220+ DN T cells was much larger within LNs and spleen of T cells of WT mice in the presence or absence of CD70-driven lpr 3 CD70 Tg mice than in those of lpr mice (Fig. 6C). Inter- costimulation. Under conditions that result in enhanced T cell estingly, lpr 3 Bim2/2 mice displayed a similar phenotype as lpr 3 The Journal of Immunology 4261

FIGURE 4. Fas ligation mediates CD70-driven ap- optosis of polyclonal T cells. (A and B) T cells were isolated from spleen, labeled with CFSE, and stimu- lated with anti-CD3 in the presence of irradiated and DDAO-labeled control or CD70 Tg APCs. After 3 d, cultures were labeled with anti-CD4 and anti-CD8 Abs and analyzed for proliferation using CFSE dilution and cell death using PI staining. (A) Dot plots show CFSE dilution and PI staining of CD4 T cells of WT and Cd272/2 mice stimulated with anti-CD3 in the pres- ence of control or CD70 Tg APCs. (B) Apoptosis in CD4 T cells (left panel) and CD8 T cells (right panel) of WT and Cd272/2 mice was enumerated after stim- ulation with anti-CD3 in the presence of control or CD70 Tg APCs. (C) Histograms display the expression of Annexin V on WT (left panel) and Cd272/2 T cells (right panel) stimulated for 3 d with anti-CD3 in the presence of control or CD70 Tg APCs. (D) The rep- resentative histogram displays expression of CD70 on mock and CD70-transfected COS cells. (E) WT and Cd272/2 T cells were stimulated for 3 d with anti-CD3 in the presence of mock or CD70-transfected COS cells. Cell death was determined in proliferating (CFSElo) T cells using staining for PI and Annexin V (AV). (F) Histograms show the expression of FasL (top panel) and Fas (bottom panel) on T cells from WT and Cd272/2 mice after anti-CD3 stimulation in the pres- ence or absence of CD70 triggering using control or CD70 Tg APCs. (G) The percentage of dead cells of total proliferating T cells from WT, lpr, and gld mice was determined using PI staining after incubation with anti-CD3 with or without CD70-driven costimulation. (H) Similarly, the percentage of dead cells was deter- mined of anti-CD3 stimulated WT T cells that were incubated with or without blocking anti-FasL Abs in the presence of control or CD70 Tg APCs. *p , 0.05, **p , 0.005.

CD70 Tg mice, as they also had strongly increased B220+ DN phenotype T cells compared with WT mice, underlining the role T cell numbers (11–13). The dramatic increase of B220+ DN of CD70 in costimulation of T cells (25, 26, 32). Indeed, staining T cells in lpr 3 CD70 Tg mice is consistent with a role for CD70 for CD44 and CD62L in the CD8 T cell compartment of the in the downregulation of Bim (see also Fig. 5C, 5E) that, in the spleen showed enhanced percentages and absolute numbers of absence of Fas, enables enhanced survival of B220+ DN T cells. CD44hiCD62Llo effector or effector memory phenotype CD8 Moreover, this may imply that CD70-driven inflammation may T cells in T-cell specific CD70 Tg mice compared with WT mice accelerate the onset of autoimmune lymphoproliferative syndrome (Fig. 7A, 7B). The CD44hiCD62Llo CD8 T cell population in in patients with mutations in Fas or FasL. CD70 Tg mice was predominantly of the effector phenotype, as the majority lacked expression of the memory-associated mole- CD70-driven costimulation triggers Fas-dependent T cell cule CD127 (Fig. 7A, 7B). Strikingly, in the absence of Fas, CD70 death in vivo induced the accumulation of not only effector CD8 T cells, but We took advantage of the lpr 3 CD70 Tg mice to study whether also of effector memory CD8 T cells (Fig. 7A, 7C). This dem- CD70 impacted differentiation of conventional T cells in a Fas- onstrates that CD70-driven expansion of effector phenotype and, dependent manner. CD70 Tg mice that express CD70 on T cells, in particular, of effector memory phenotype T cells is limited by B cells, or dendritic cells have increased numbers of effector CD70-induced and Fas-dependent T cell death. 4262 CD70 INDUCES Fas-DRIVEN T CELL DEATH

FIGURE 5. CD70 upregulates costimulatory path- ways despite the induction of T cell death. The pro- duction of IL-2 was measured using ELISA in the supernatants of WT and Cd272/2 OTI cells (A) stim- ulated with the indicated amount of peptide and poly- clonal WT and Cd272/2 T cells (B) stimulated with anti-CD3 in the presence of control or CD70 Tg APCs. (C) WT T cells were stimulated for the indicated days with anti-CD3 Abs in the presence of control or CD70 Tg APCs, and expression of pro- and antiapoptotic molecules was analyzed using reverse-transcriptase multiplex ligation-dependent probe amplification. Gene induction of pro- and antiapoptotic molecules is represented in a heat map after log transformation of expression levels (scale is from 23 to 3), relative to averaged values of unstimulated cells at day 0 (n = 3). Using Western blot, expression of Bcl-Xl (D) and Bim (E) was analyzed in WT T cells after stimulation with anti-CD3 and mock or CD70-transfected COS cells for the indicated hours. Actin was used as a loading con- trol. Blots were quantified for Bcl-Xl (D) and Bim EL (E) relative to actin levels using densitometric analysis. **p , 0.005.

CD70 signaling during the primary response downregulates 45 d after transfer of donor CD8 T cells, and the number of CD8 T cell recall responses influenza-specific CD8 T cells of donor origin was analyzed To study the timing of CD70-induced elimination of T cells during 11 d later (Fig. 8A). Donor CD8 T cells of WT mice, but not of in vivo immune responses, we used the influenza infection model. CD70 Tg mice, contributed significantly to the influenza-specific Previously, we have described that the constitutive presence of CD8 T cell response in recipient WT mice (Fig. 8B). Quantifi- CD70-dependent costimulation enhances primary CD8 T cell cation of donor-derived CD8 T cells showed that those of WT responses against influenza (25, 33). In contrast, secondary CD8 origin, in contrast to those of CD70 Tg origin, were able to induce T cell responses are impaired under chronic CD70-driven co- strong recall responses in blood, mLN, spleen, or lungs upon in- stimulation through the loss of memory CD8 T cells (25). We were fection (Fig. 8C, 8D). This implied that expression during the interested whether CD70 signaling during the primary response or, primary infection is sufficient for CD70 to induce impairment of thereafter, during the memory phase was important for the loss of secondary CD8 T cell responses. To establish whether expression memory CD8 T cells and impaired recall CD8 T cell responses. of CD70 during the memory phase also contributed to impaired Therefore, WT and CD70 Tg mice were infected with influenza, recall CD8 T cell responses, the reciprocal experiment was per- and total CD8 T cells were isolated at day 10 postinfection for formed, in which CD8 T cells of influenza-infected WT mice were transfer into noninfected WT mice. To prevent that adoptively transferred into WT or CD70 Tg recipients (Fig. 8E). Donor CD8 transferred T cells expressed CD70, we made use of CD70 Tg T cells of WT mice contributed significantly to the influenza- mice that express CD70 on B cells. These CD70 Tg mice have specific CD8 T cell response of WT and CD70 Tg recipients a similar phenotype as CD70 Tg mice that express CD70 on (Fig. 8F). Strong recall responses of WT donor CD8 T cells were T cells (25, 26). Recipient mice were infected with influenza observed in mLN, spleen, and lungs upon infection of both WT The Journal of Immunology 4263

FIGURE 7. Effector and memory phenotype CD8 T cells are susceptible to Fas-driven apoptosis under conditions of chronic CD70 costimulation. (A) Splenic CD8 T cells of WT, lpr, CD70 Tg, and lpr 3 CD70 Tg mice were analyzed for expression of CD44 and CD62L. The percentage of hi lo FIGURE 6. Chronic CD70 costimulation induces accumulation of CD44 CD62L CD8 T cells that represent effector or effector memory B220+ DN T cells in the absence of Fas. (A) Cervical LNs of WT, CD70 phenotype CD8 T cells was determined (top panel). The expression of hi lo Tg, lpr, and lpr 3 CD70 Tg mice are displayed (left panel). Lpr 3 CD70 CD127 was analyzed within the CD44 CD62L CD8 T cell population to identify the CD127+ effector memory subset and the CD1272 effector Tg mouse was dissected to display enlarged cervical and axillary LNs 2 indicated by arrows (right panel). (B) Dot plot displays expression of CD4 subset (bottom panel). The absolute number of total, CD127 , and + hi lo and CD8 on CD3+ T cells of lpr 3 CD70 Tg spleen (left panel), and CD127 CD44 CD62L CD8 T cells was determined within the spleen of B 3 histogram displays expression of B220 on CD4, CD8, and DN T cells WT and CD70 Tg mice ( ) and within the spleen of lpr and lpr CD70 C , , (right panel). (C) Absolute numbers of B220+ DN T cells were determined Tg mice ( ). *p 0.05, **p 0.005. in peripheral LNs and spleen of lpr and lpr 3 CD70 Tg mice. *p , 0.05, **p , 0.005. not only higher in the blood, but also in the spleen of lpr 3 CD70 Tg mice compared with CD70 Tg mice (Fig. 9D). This shows that and CD70 Tg recipient mice (Fig. 8G). Thus, CD70 instructs CD70 is able to trigger Fas-dependent elimination of CD8 T cells. 3 downregulation of memory CD8 T cell responses early during The impaired removal of CD8 T cell responses in lpr CD70 Tg primary infection when Ag levels are high and not during the mice was not caused by reduced viral clearance, as the influenza memory phase when the virus has been cleared. virus was undetectable in the lungs of all mice (data not shown). This is corroborated by earlier observations that show that Fas is CD70 enables Fas-dependent elimination of CD8 T cells not essential for elimination of virally infected cells after influenza Recently, it has been reported that CD70-driven costimulation infection (34). Reduced removal of effector CD8 T cells resulted reduces rather than enhances CD8 T cell responses during chronic in larger memory CD8 T cell responses in lpr 3 CD70 Tg mice infection with LCMV clone 13 (24). Consistent with this study, we compared with CD70 Tg mice (Fig. 9C, 9E). Levels of memory observed that Cd272/2 mice displayed enhanced CD8 T cell CD8 T cells in blood and spleen of lpr 3 CD70 Tg mice were responses against LCMV clone 13 compared with WT mice almost restored to WT and lpr levels (Fig. 9E). Thus, CD70-driven (Fig. 9A). Interestingly, Cd272/2 mice already had enlarged CD8 costimulation enables Fas-dependent elimination of influenza- T cell responses at day 8 postinfection, and the numbers of specific CD8 T cells in the contraction phase, which results in LCMV-specific CD8 T cells remained elevated until at least day the maintenance of lower numbers of memory CD8 T cells. 20 postinfection (Fig. 9A). Also, lpr mice contained larger num- bers of LCMV-specific CD8 T cells than WT mice within the Discussion spleen at day 8 postinfection (Fig. 9B), which is in line with In this study, we have demonstrated that Fas-driven apoptosis CD70-driven activation of the Fas pathway of apoptosis to ac- forms an integral aspect of CD70-driven costimulation that sup- complish elimination of LCMV-specific CD8 T cells. To deter- presses the prosurvival effects of CD27 on T cells. This striking role mine whether CD70 was able to engage the Fas-dependent of costimulatory CD70 and CD27 in downregulation of CD8 T cell pathway of T cell death during infection, CD70 Tg and lpr 3 responses may clarify the negative impact of CD27 signaling CD70 Tg mice were infected with influenza, and CD8 T cell during previously reported in in vitro and in vivo settings. responses were followed in time within the blood. CD8 T cell Members of the TNFR family, including CD27, mediate an responses were not different between CD70 Tg mice and lpr 3 important pathway of costimulation that essentially contributes to CD70 Tg mice at the peak of infection at day 10, but thereafter, expansion of CD8 T cells during acute immune responses (35). lpr 3 CD70 Tg maintained higher numbers of influenza-specific CD27 induces expansion of CD8 T cells through enhancement of CD8 T cells than CD70 Tg mice (Fig. 9C). At the time of con- proliferation and survival (15, 28). We also observed that CD70 traction at day 14, numbers of influenza-specific CD8 T cells were enhances the proliferation and survival of CD8 T cells in our 4264 CD70 INDUCES Fas-DRIVEN T CELL DEATH

FIGURE 8. CD70 costimulation in the primary response reduces recall CD8 T cell responses. (A–D) WT and CD70-B cell Tg mice were infected with the influenza strain HKx31, and after 10 d, total CD8 T cells were isolated, and 10 3 106 cells were transferred into Ly5.1+ WT recipients. After 45 d, recipient mice were infected with the influenza strain A/PR8/34, and 11 d later, they were sacrificed for analysis. (A) Schematic representation is shown of the experimental setup to determine whether CD70 eliminates memory CD8 T cells during the priming phase of influenza infection. (B) Representative dot plots display influenza-specific CD8 T cells by CD8 and tetramer staining of splenocytes at day 11 of rechallenge (top panel). Tetramer+ CD8 T cells were gated to display host- and donor-derived cells by Ly5.1 staining (bottom panel). (C) The percentage of tetramer+ donor CD8 T cells (Ly5.12) postinfection was followed in time within the blood. (D) The percentage of tetramer+ donor CD8 T cells (Ly5.12) was determined within the CD8 T cell population of mLN, spleen, and lungs at day 11 after rechallenge. (E–G) Ly5.1+ WT mice were infected with HKx31, and after 17 d, total CD8 T cells were isolated and transferred into WT or CD70 Tg recipients. After 52 d, recipient mice were infected with A/PR8/34 45, and 10 d later, they were sacrificed for analysis. (E) Schematic representation is shown of the experimental setup to determine whether CD70 eliminates memory CD8 T cells during the memory phase of influenza infection. (F) Representative dot plots of total splenocytes display staining for CD8 and tetramer to enumerate influenza-specific CD8 T cells at day 10 after rechallenge (top panel). Tetramer+ CD8 T cells were gated to display host- and donor-derived cells using Ly5.1 staining (bottom panel). (G) The percentage of tetramer+ donor CD8 T cells (Ly5.1+) was determined within the CD8 T cell population of mLN, spleen, and lungs at day 10 after rechallenge. **p , 0.005. in vitro cultures after stimulation with low doses of antigenic Bcl-Xl. Moreover, we observed that CD70 induced downregula- peptide. The effect on T cell proliferation is separable from the tion of proapoptotic Bim, indicating that Bim-dependent T cell prosurvival effect of CD27, underlining the dual ability of CD27 apoptosis is impaired during CD70-driven costimulation. to enhance expansion of CD8 T cells through proliferation and In contrast to these prosurvival effects of CD27, we found that survival. A multitude of complementary mechanisms accounts for CD27 is also able to induce T cell death through induction of FasL- the costimulatory activity of CD27, including induction of IL-2 and Fas-dependent apoptosis after stimulation with high amounts production and upregulation of antiapoptotic members of the Bcl- of Ag. Previously, we have described that CD70 sensitizes T cells 2 family (18–20). Confirming these studies, we have shown that for Fas-driven apoptosis, as agonistic anti-Fas Abs were more CD70 induced production of IL-2 and upregulated antiapoptotic effective in triggering T cell apoptosis in the presence of CD70 The Journal of Immunology 4265

tive-feedback mechanism of CD27 signaling that enables CD27 to counteract its effects on CD8 T cell expansion. We show that CD70-driven Fas-dependent apoptosis occurs after stimulation with higher levels of Ag than the prosurvival effects of CD70. Therefore, the opposite effects of CD70 may optimize the size of CD8 T cell responses. Under conditions of limiting Ag, CD70 may promote expansion of effector CD8 T cells, whereas during Ag abundance when survival factors for CD8 T cells are optimal, CD70 may enable removal of CD8 T cells. Consistent with this, during acute infection such as with influenza, when viral and antigenic loads are likely limiting, CD70 enhances the survival of effector and memory CD8 T cells (14, 15). Interestingly, it has been reported that the prosurvival effects of CD70 on CD8 T cell responses against influenza depend at least in part on the capacity of CD70 to prevent FasL upregulation on CD4 T cells and con- sequent Fas-dependent elimination of CD8 T cells (40). A role for CD70 in downregulation of CD8 T cell responses may have rel- evance for chronic infection, during which Ag and Ag-related survival signals persist, and continuous turnover of effector CD8 T cells occurs to ensure the replacement of senescent or ineffec- tive T cells. This is consistent with the increased numbers of ef- fector CD8 T cells under blockade of CD70-driven costimulation that we and others observed during infection with chronic strains of the LCMV virus (23, 24). The role of Fas-driven apoptosis is also more apparent during chronic infection than during acute infection. Fas and Bim cooperate to limit CD8 T cell numbers in the chronic model of murine g-herpes virus infection in contrast to the acute model of HSV infection, during which contraction is FIGURE 9. CD70 and Fas eliminate CD8 T cells during infection. (A) exclusively Bim dependent (11). Extending these studies, we have WT and Cd272/2 mice were infected with LCMV clone 13, and the found that Fas also eliminates effector CD8 T cells during chronic number of LCMV-specific CD8 T cells was enumerated in spleen after 8 infection with LCMV clone 13. Taken together, these studies b and 20 d of infection using H2-D tetramers loaded with GP33–41 peptide. highlight the importance of CD70 and Fas in restricting CD8 (B) Similarly, the number of LCMV-specific CD8 T cells was quantified in T cell responses during infection, and in this study, we clarify that C spleen of WT and lpr mice at day 8 postinfection. ( ) The percentage of they in fact operate in the same pathway. tetramer+ CD8 T cells was followed in time within the blood of CD70 Tg We have previously shown that phenotypically, CD70 Tg mice and lpr 3 CD70 Tg animals upon infection with influenza. (D) The ab- solute number of tetramer+ CD8 T cells was analyzed in mLN, spleen, and that display constitutive expression of CD70 on B or T cells lungs of CD70 Tg and lpr 3 CD70 Tg mice at day 14 after influenza display remarkable similarities to situations of chronic infection infection. (E) The absolute number of tetramer+ CD8 T cells was analyzed (25, 27, 41). Constitutive expression of CD70 on B cells induces in spleen, mLN, and lungs of WT and lpr mice at day 59 after influenza effector CD8 T cell differentiation that eventually culminates in infection (left panel) and CD70 Tg and lpr 3 CD70 Tg mice at day 64 depletion of the naive T cell pool and lethal immunodeficiency after influenza infection (right panel). *p , 0.05, **p , 0.005. (41). The accumulation of effector CD8 T cells and lethal disease was dramatically accelerated in B cell-specific CD70 Tg mice on a Fas-deficient background (36). This has shown the importance of (36). Our current findings show that CD70 instructs rather than Fas in restricting the pathological impact of constitutive CD70 permits Fas-driven apoptosis, as CD70 upregulated the expression expression, but the rapid onset of disease and death has prevented of FasL on activated T cells and engaged Fas-dependent cell careful analysis of the events during CD8 T cell differentiation. In death without requirement for external stimulation of the Fas this study, using T cell-specific lpr 3 CD70 Tg mice that display pathway. CD70 and CD27 triggered T cell death despite the in- a similar but milder phenotype than B-cell specific lpr 3 CD70 Tg duction of IL-2 production and the stimulatory and suppressive mice, we have demonstrated that memory rather than effector effects on expression of Bcl-Xl and Bim, respectively. Previously, phenotype CD8 T cells are targeted by CD70-induced and Fas- it has been documented that activation-induced cell death, which dependent apoptosis. The development of memory CD8 T cells involves repetitive exposure of T cells to Ag occurs in an IL-2– that are able to provide enhanced protection upon reinfection is dependent manner (37). Therefore, it is possible that the induction a cardinal feature of acute CD8 T cell responses. However, dif- of IL-2 does not conflict with T cell apoptosis during CD70-driven ferentiation into memory CD8 T cells is compromised after costimulation. Consistent with a supportive role for IL-2 in CD70- chronic LCMV infection. LCMV-specific CD8 T cells that are induced T cell apoptosis, IL-2 triggers activation-induced cell transferred from chronically infected mice into noninfected mice death through downregulation of the expression of cFLIP that do not survive in the absence of Ag (42). Supporting a role of antagonizes Fas-driven apoptosis and upregulation of FasL ex- CD70 in this process, we have previously observed the disap- pression (38). Moreover, it has been reported previously that pearance of bona fide memory CD8 T cells after acute infection CD70-driven expansion of T cells is more pronounced in the with influenza in the presence of CD70 (25). In this study, we have absence of IL-2 (28). In T cells, Fas triggers apoptosis indepen- demonstrated that constitutive expression of CD70 during the dent of Bcl-Xl and Bim, which enables CD27 signaling to induce memory phase did not drive the loss of these memory CD8 T cells. T cell death during suppression of the intrinsic pathway of apo- Rather, expression of CD70 during the primary response com- ptosis (39). Thus, Fas-dependent apoptosis may represent a nega- promised CD8 T cell memory in a Fas-dependent fashion. This 4266 CD70 INDUCES Fas-DRIVEN T CELL DEATH eliminates long-term signaling through CD70 and CD27 as the 14. Hendriks, J., L. A. Gravestein, K. Tesselaar, R. A. van Lier, T. N. Schumacher, and J. Borst. 2000. CD27 is required for generation and long-term maintenance main driving force of downregulation of CD8 T cell responses of T cell immunity. Nat. Immunol. 1: 433–440. in this setting. The early effect of CD70 on elimination of CD8 15. Hendriks, J., Y. Xiao, and J. Borst. 2003. CD27 promotes survival of activated T cell memory was consistent with the timeframe, in which T cells and complements CD28 in generation and establishment of the effector T cell pool. J. Exp. Med. 198: 1369–1380. CD70 induced Fas-driven removal of CD8 T cells during influ- 16. Hintzen, R. Q., S. M. Lens, M. P. Beckmann, R. G. Goodwin, D. Lynch, and enza infection. This indicates that CD70 and CD27 mediate a R. A. van Lier. 1994. Characterization of the human CD27 ligand, a novel Fas-dependent pathway that compromises the differentiation of member of the TNF gene family. J. Immunol. 152: 1762–1773. 17. Nolte, M. A., R. W. van Olffen, K. P. van Gisbergen, and R. A. van Lier. 2009. memory CD8 T cells during chronic infection. Timing and tuning of CD27-CD70 interactions: the impact of signal strength in The dual capacity of CD27 to up- and downregulate T cell setting the balance between adaptive responses and immunopathology. Immunol. responses is shared by the related TNFR-like molecule 4-1BB. Rev. 229: 216–231. 18. Peperzak, V., Y. Xiao, E. A. Veraar, and J. Borst. 2010. CD27 sustains survival of During murine CMV infection, 4-1BB suppresses CD8 T cell CTLs in virus-infected nonlymphoid tissue in mice by inducing autocrine IL-2 responses against acute epitopes, but enhances CD8 T cell production. J. Clin. Invest. 120: 168–178. 19. Peperzak, V., E. A. Veraar, A. M. Keller, Y. Xiao, and J. Borst. 2010. The Pim responses during the latent phase (43). Supporting this study, kinase pathway contributes to survival signaling in primed CD8+ T cells upon agonistic anti-4-1BB Abs improve or impair T cell responses CD27 costimulation. J. Immunol. 185: 6670–6678. against acute infection with LCMV and influenza depending on 20. van Oosterwijk, M. F., H. Juwana, R. Arens, K. Tesselaar, M. H. van Oers, E. Eldering, and R. A. van Lier. 2007. CD27-CD70 interactions sensitise naive the time of application (44). Similar to CD27, 4-1BB required Fas- CD4+ T cells for IL-12-induced Th1 cell development. Int. Immunol. 19: 713– driven apoptosis to induce contraction of CD8 T cell responses 718. (44). The costimulatory pathway through CD28 has not been re- 21. Schildknecht, A., I. Miescher, H. Yagita, and M. van den Broek. 2007. Priming of CD8+ T cell responses by pathogens typically depends on CD70-mediated ported to downregulate T cell responses by itself but is counter- interactions with dendritic cells. Eur. J. Immunol. 37: 716–728. acted by the inhibitory family member CTLA4 that also binds B7 22. van Gisbergen, K. P., P. L. Klarenbeek, N. A. Kragten, P. P. Unger, molecules (45). This shows that the ability to adequately control M. B. Nieuwenhuis, F. M. Wensveen, A. ten Brinke, P. P. Tak, E. Eldering, M. A. Nolte, and R. A. van Lier. 2011. The costimulatory molecule CD27 T cell responses through negative feedback is an important and maintains clonally diverse CD8(+) T cell responses of low antigen affinity to inherent aspect of costimulation. Therefore, it is important to take protect against viral variants. Immunity 35: 97–108. the potential to downsize immune responses into account in the 23. Matter, M., B. Odermatt, H. Yagita, J. M. Nuoffer, and A. F. Ochsenbein. 2006. Elimination of chronic viral infection by blocking CD27 signaling. J. Exp. Med. design of vaccination strategies that target costimulation either 203: 2145–2155. through CD70 and CD27 or other pathways. 24. Penaloza-MacMaster, P., A. Ur Rasheed, S. S. Iyer, H. Yagita, B. R. Blazar, and R. Ahmed. 2011. Opposing effects of CD70 costimulation during acute and chronic lymphocytic choriomeningitis virus infection of mice. J. Virol. 85: 6168– Acknowledgments 6174. 25. van Gisbergen, K. P. J. M., R. W. van Olffen, J. van Beek, K. F. van der Sluijs, We thank the staff of the animal facility of the Academic Medical Center for R. Arens, M. A. Nolte, and R. A. van Lier. 2009. Protective CD8 T cell memory excellent animal care, Gijs van Schijndel for making the tetramers, Claudia is impaired during chronic CD70-driven costimulation. J. Immunol. 182: 5352– Brandao for technical assistance, and Dr. Martijn Nolte for critical reading 5362. of the manuscript and helpful discussion. 26. Arens, R., K. Tesselaar, P. A. Baars, G. M. van Schijndel, J. Hendriks, S. T. Pals, P. Krimpenfort, J. Borst, M. H. van Oers, and R. A. van Lier. 2001. Constitutive CD27/CD70 interaction induces expansion of effector-type T cells and results in Disclosures IFNgamma-mediated B cell depletion. Immunity 15: 801–812. 27. Beishuizen, C. R., N. A. Kragten, L. Boon, M. A. Nolte, R. A. van Lier, and The authors have no financial conflicts of interest. K. P. van Gisbergen. 2009. Chronic CD70-driven costimulation impairs IgG responses by instructing T cells to inhibit germinal center B cell formation through FasL-Fas interactions. J. Immunol. 183: 6442–6451. References 28. Carr, J. M., M. J. Carrasco, J. E. Thaventhiran, P. J. Bambrough, M. Kraman, 1. Harty, J. T., and V. P. Badovinac. 2008. Shaping and reshaping CD8+ T-cell A. D. Edwards, A. Al-Shamkhani, and D. T. Fearon. 2006. CD27 mediates memory. Nat. Rev. Immunol. 8: 107–119. -2-independent clonal expansion of the CD8+ T cell without effector 2. Prlic, M., M. A. Williams, and M. J. Bevan. 2007. Requirements for CD8 T-cell differentiation. Proc. Natl. Acad. Sci. USA 103: 19454–19459. priming, memory generation and maintenance. Curr. Opin. Immunol. 19: 315–319. 29. Tesselaar, K., Y. Xiao, R. Arens, G. M. van Schijndel, D. H. Schuurhuis, 3. Bouillet, P., and L. A. O’Reilly. 2009. CD95, BIM and T cell homeostasis. Nat. R. E. Mebius, J. Borst, and R. A. van Lier. 2003. Expression of the murine CD27 Rev. Immunol. 9: 514–519. ligand CD70 in vitro and in vivo. J. Immunol. 170: 33–40. 4. Opferman, J. T., and S. J. Korsmeyer. 2003. Apoptosis in the development and 30. Davidson, W. F., F. J. Dumont, H. G. Bedigian, B. J. Fowlkes, and H. C. Morse, maintenance of the . Nat. Immunol. 4: 410–415. III. 1986. Phenotypic, functional, and molecular genetic comparisons of the 5. Youle, R. J., and A. Strasser. 2008. The BCL-2 protein family: opposing activ- abnormal lymphoid cells of C3H-lpr/lpr and C3H-gld/gld mice. J. Immunol. 136: ities that mediate cell death. Nat. Rev. Mol. Cell Biol. 9: 47–59. 4075–4084. 6. Strasser, A., P. J. Jost, and S. Nagata. 2009. The many roles of 31. Straus, S. E., M. Sneller, M. J. Lenardo, J. M. Puck, and W. Strober. 1999. An signaling in the immune system. Immunity 30: 180–192. inherited disorder of apoptosis: the autoimmune lymphoproliferative 7. Fischer, S. F., G. T. Belz, and A. Strasser. 2008. BH3-only protein Puma con- syndrome. Ann. Intern. Med. 130: 591–601. tributes to death of antigen-specific T cells during shutdown of an immune re- 32. Keller, A. M., A. Schildknecht, Y. Xiao, M. van den Broek, and J. Borst. 2008. sponse to acute viral infection. Proc. Natl. Acad. Sci. USA 105: 3035–3040. Expression of costimulatory ligand CD70 on steady-state dendritic cells breaks + 8. Hildeman, D. A., Y. Zhu, T. C. Mitchell, P. Bouillet, A. Strasser, J. Kappler, and CD8 T cell tolerance and permits effective immunity. Immunity 29: 934–946. P. Marrack. 2002. Activated T cell death in vivo mediated by proapoptotic bcl-2 33. Arens, R., K. Schepers, M. A. Nolte, M. F. van Oosterwijk, R. A. van Lier, family member bim. Immunity 16: 759–767. T. N. Schumacher, and M. H. van Oers. 2004. Tumor rejection induced by CD70- + 9. Bonfoco, E., P. M. Stuart, T. Brunner, T. Lin, T. S. Griffith, Y. Gao, H. Nakajima, mediated quantitative and qualitative effects on effector CD8 T cell formation. P. A. Henkart, T. A. Ferguson, and D. R. Green. 1998. Inducible nonlymphoid J. Exp. Med. 199: 1595–1605. expression of Fas ligand is responsible for superantigen-induced peripheral de- 34. Topham, D. J., R. A. Tripp, and P. C. Doherty. 1997. CD8+ T cells clear in- letion of T cells. Immunity 9: 711–720. fluenza virus by or Fas-dependent processes. J. Immunol. 159: 5197– 10. Russell, J. H., B. Rush, C. Weaver, and R. Wang. 1993. Mature T cells of au- 5200. toimmune lpr/lpr mice have a defect in antigen-stimulated suicide. Proc. Natl. 35. Croft, M. 2009. The role of TNF superfamily members in T-cell function and Acad. Sci. USA 90: 4409–4413. diseases. Nat. Rev. Immunol. 9: 271–285. 11. Hughes, P. D., G. T. Belz, K. A. Fortner, R. C. Budd, A. Strasser, and P. Bouillet. 36. Arens, R., P. A. Baars, M. Jak, K. Tesselaar, M. van der Valk, M. H. van Oers, 2008. Apoptosis regulators Fas and Bim cooperate in shutdown of chronic im- and R. A. van Lier. 2005. Cutting edge: CD95 maintains effector T cell ho- mune responses and prevention of autoimmunity. Immunity 28: 197–205. meostasis in chronic immune activation. J. Immunol. 174: 5915–5920. 12. Hutcheson, J., J. C. Scatizzi, A. M. Siddiqui, G. K. Haines, III, T. Wu, Q. Z. Li, 37. Lenardo, M. J. 1991. Interleukin-2 programs mouse alpha beta T L. S. Davis, C. Mohan, and H. Perlman. 2008. Combined deficiency of proap- for apoptosis. Nature 353: 858–861. optotic regulators Bim and Fas results in the early onset of systemic autoim- 38. Refaeli, Y., L. Van Parijs, C. A. London, J. Tschopp, and A. K. Abbas. 1998. munity. Immunity 28: 206–217. Biochemical mechanisms of IL-2-regulated Fas-mediated T cell apoptosis. Im- 13. Weant, A. E., R. D. Michalek, I. U. Khan, B. C. Holbrook, M. C. Willingham, munity 8: 615–623. and J. M. Grayson. 2008. Apoptosis regulators Bim and Fas function concur- 39. Strasser, A., A. W. Harris, D. C. Huang, P. H. Krammer, and S. Cory. 1995. Bcl-2 rently to control autoimmunity and CD8+ T cell contraction. Immunity 28: 218– and Fas/APO-1 regulate distinct pathways to lymphocyte apoptosis. EMBO J. 230. 14: 6136–6147. The Journal of Immunology 4267

40. Dolfi, D. V., A. C. Boesteanu, C. Petrovas, D. Xia, E. A. Butz, and P. D. Katsikis. 43. Humphreys, I. R., S. W. Lee, M. Jones, A. Loewendorf, E. Gostick, D. A. Price, 2008. Late signals from CD27 prevent Fas-dependent apoptosis of primary CD8+ C. A. Benedict, C. F. Ware, and M. Croft. 2010. Biphasic role of 4-1BB in the T cells. J. Immunol. 180: 2912–2921. regulation of mouse cytomegalovirus-specific CD8(+) T cells. Eur. J. Immunol. 41. Tesselaar, K., R. Arens, G. M. van Schijndel, P. A. Baars, M. A. van der Valk, 40: 2762–2768. J. Borst, M. H. van Oers, and R. A. van Lier. 2003. Lethal T cell immunodefi- 44. Zhang, B., C. H. Maris, J. Foell, J. Whitmire, L. Niu, J. Song, B. S. Kwon, ciency induced by chronic costimulation via CD27-CD70 interactions. Nat. A. T. Vella, R. Ahmed, J. Jacob, and R. S. Mittler. 2007. Immune suppression or Immunol. 4: 49–54. enhancement by CD137 T cell costimulation during acute viral infection is time 42. Wherry, E. J., D. L. Barber, S. M. Kaech, J. N. Blattman, and R. Ahmed. 2004. dependent. J. Clin. Invest. 117: 3029–3041. Antigen-independent memory CD8 T cells do not develop during chronic viral 45. Chen, L. 2004. Co-inhibitory molecules of the B7-CD28 family in the control of infection. Proc. Natl. Acad. Sci. USA 101: 16004–16009. T-cell immunity. Nat. Rev. Immunol. 4: 336–347.