CD152 (CTLA-4) Determines the Unequal Resistance of Th1 and Th2 Cells against Activation-induced Cell Death by a Mechanism Requiring PI3 Kinase Function

The decision between survival and apoptosis of T cells is of particular importance for adaptive immune responses to ensure that a defined number of specialized T cells remain in the organism, thus maintaining homeostasis and memory. The primary form of apoptosis of clonally expanded T cells is activation-induced cell death (AICD), mainly controlled by the Fas (CD95) system (1, 2). Fas-mediated AICD is considered to be the primary mechanism in the deletion of mature CD4 T cells in the periphery.

In the absence of appropriate costimulation, TCR signaling induces Fas and Fas ligand (FasL) expression (3). Ligation of Fas initiates the recruitment of Fas-associated death domain and caspase-8 that triggers the proteolytic caspase cascade, resulting in the cleavage of various proteins and apoptosis (4). Several mechanisms exist to stop death processes either at the receptor, mitochondrial, or caspase level. Survival mechanisms are also mediated by phosphatidylinositol 3 kinase (PI3 K), which activates protein kinase B/Akt, an antiapoptotic kinase that inactivates proapoptotic molecules such as Bad and Forkhead transcription factor FKHRL1 (5, 6). CD28 costimulatory signaling and cytokines have been suggested to up-regulate the antiapoptotic molecules Bcl-x_L and c-FLIP during an immune response (7–9). Importantly, some effector T cells survive and contribute to long-term memory (10).

Cytotoxic T lymphocyte antigen 4 (CTLA-4, CD152) is a major down-regulator of immune responses (11–13). In most circumstances, the two homologues CD28 and CD152, which share common ligands, fulfill opposing functions during T cell activation. Triggering of CD28 strongly up-regulates IL-2 production and T cell proliferation, which is counterregulated by CD152-mediated inhibition of IL-2 transcription and cell cycle progression (14). CD28 also leads to the stabilization of IL-2 mRNA and up-regulation of Bcl-x_L, events that are not counterregulated by CD152 (15). The underlying mechanisms responsible for these effects are unclear. CD28 has been reported to bind to PI3 K, the adapters Grb-2/GADS, and the phosphatase PP2A, whereas CD152 binds to PI3 K and the phosphatases PP2A and SHP-2 (16–19). The coupling of CD28 and CD152 to different signaling mediators might indicate that CD152 does not simply counteract CD28-mediated costimulation.

During suboptimal activation of the T cell, CD152 engagement leads to impaired T cell activation and to G1 cell cycle arrest by inhibition of cyclin D3, CDK4, and CDK6 production (11–14, 20). Although CD28 and CD152 bind to the same ligands, B7.2 and B7.1, on APCs, CD152 binds with a 20–50-fold higher affinity. CD28 is constitutively expressed on CD4⁺ T cells, whereas CD152 is inducible reaching maximal surface levels 48 h after T cell stimulation (21). CD152 is transported from intracellular vesicles to the cell membrane, where it is expressed at the immunological synapse (22). This restricted regulation of CD152 trafficking and localization represents a major control point for its inhibitory function. It has been proposed that CD152 prevents some T cell clones with a high-affinity TCR by decreasing their competitive advantage over clones with a low-affinity TCR (22). Suggestive evidence and correlations have also shown that CD152 ligation in previously activated ConA blasts or αCD3-stimulated T cells might induce apoptosis to terminate the T cell response (23, 24).

After initial activation, naive CD4⁺ cells differentiate into Th1 and Th2 cells that secrete different cytokines. Interestingly, Th2 cells are substantially more resistant to AICD than Th1 cells (25, 26). Whether the expression of Fas or other apoptotic regulators is involved in this differential sensitivity after antigenic restimulation is controversial (27). Although Fas-mediated apoptosis is functional in CD152^−/− T cells in vivo (28), the role of CD152 in AICD of activated primary T cells that are prone to AICD remains unknown. Suggestive stainings for surface CD152, which were, however, only performed in long-term T cell clones, indicate that surface expression of CD152 might be present in Th2 rather than in Th1 cell clones (29). CD152 has also been proposed as an inhibitor of Th2 cell differentiation (30, 31). Owing to its low surface expression, detection of surface CD152 on primary T cells by conventional techniques has been difficult. Therefore, the role of CD152 in the response of differentiated primary Th1 and Th2 cells is almost entirely unknown.

Here, we investigated the functional role of CD152 on individual primary Th1 and Th2 cells using a novel sensitive immunofluorescent labeling technique. We show for the first time that, after antigenic restimulation, CD152 surface expression is induced at high frequencies in Th2 cells compared with Th1 cells. Interestingly, CD152 rendered activated T cells resistant to AICD, whereas inhibition of CD152 sensitized cells for apoptosis. We identified that apoptosis protection by CD152 was dependent on the suppression of the Fas system and mediated by PI3 K, which induced the up-regulation of Bcl-2 and the inactivation of FKHRL1, a transcription factor regulating FasL expression. Thus, CD152 regulates AICD by targeting PI3 K and, due to higher frequencies of CD152 surface expression in Th2 than in Th1 populations, induces resistance to AICD mainly in Th2 cells. This novel activity of CD152 could explain the resistance of Th2 cells to apoptosis and might be important for T cell homeostasis.

Mice transgenic (tg) for the DO11.10 α/β-TCR (OVA-specific TCR^tg/tg) on a BALB/c background (a gift from D.Y. Loh, Washington University School of Medicine, St. Louis, MO), CD152 knockout mice on a C57/B6 background (a gift from J.P. Allison, University of California at Berkeley, Berkeley, CA), and C57/B6 and BALB/c mice were bred under pathogen-free conditions in the animal facility of the Bundesgesundheitsamt in Berlin and used at the age of 5–7 wk. Offspring of CD152^−/− mice were used at the age of 3 wk and genotyped by PCR using the following oligonucleotide primers: CD152, 5′-TGGAGTCCTTCATAGTTAGG-3′ and 5′-GCAAGATGGTGAGTGTGATGTT-3′; and Neo, 5′-CATAGTTAGGTCTGTGATGC-3′ and 5′-CGTCAAGAAGGCGATAGAAGGC-3′.

The following antibodies against murine antigens were used: αCD152 (UC10–4F10–11; Becton Dickinson), αCD25 (BD Biosciences), αTCR^tg (KJ1–26.1), αFasL (MFL3; BD Biosciences), and αFas (Jo2; BD Biosciences) in their respective forms of FITC, PE, and Cy5 conjugates. αCD28 (37.51), αCD3 (145–2C11), αCD4 (GK-1.5/4), αCD8 (196), αCD62L (MEL-14), αCD152 (UC10–4F10–11), control antibodies (560–31.1B9; provided by J.P. Allison), αIL-12 (C17–8.6), and αIFN-γ (AN18.17.24) were purified from hybridoma supernatants with protein G and controlled by HPLC and FACS^® analysis. αCD152 Fab fragments were prepared with the Immunopure Fab preparation kit (Pierce Chemical Co.). The fragments were controlled by HPLC and by Ag-specific T cell stimulations. The αFasL Ab (3C82; BioCheck) was used at a concentration of 20 μg/ml in neutralization experiments. Control antibodies such as IgG1 (R3–34), FITC-conjugated ham IgG2 (αKLH, HA4/8), PE-conjugated ham IgG, and annexin-V–PE conjugates as well as αIL-2 (JES-61A12) and αIL2Rα (3C7) that were used at 20 μg/ml were obtained from BD Biosciences. Recombinant IL-12 was purchased from R&D Systems and used at 10 ng/ml. IL-4 (supernatant of myeloid cell line P3X63 Ag.8.653 transfected with murine IL-4 cDNA) was added at an equivalent of 30 ng/ml to the cultures. IL-2 was purified from supernatants of X63.IL-2.6 cells. IL-2 and IFN-γ were detected in culture supernatants by ELISAs (R&D Systems) with a detection limit of 2 pg/ml. Dibutyryl cAMP (dB-cAMP), forskolin, wortmannin, and LY294002 were obtained from Sigma-Aldrich and used at 0.5 mM and 60, 1, and 30 μM, respectively. L-mimosine (ICN Biomedicals) was used at 60 μg/ml. Magnetic microbeads αCD4, αCD62L, αCD90, and αFITC multisort microbeads were purchased from Miltenyi Biotech. Sulfate polystyrene latex microspheres of 5 μm ± 0.1 μm diameter were obtained from Interfacial Dynamics.

Isolation of naive CD62L^highCD4⁺ T cells from OVA-specific TCR^tg/tg mice was performed using two-parameter high-gradient magnetic cell separation (MACS) according to the manufacturer's instructions. CD4⁺ cells were isolated by positive selection on an AUTOMACS system to a purity of >99.3%, and CD62L^high CD4⁺ cells were positively selected on AUTOMACS to a purity of >99% as determined by flow cytometry. Isolated cells were stimulated with 1 μg/ml OVA_323–339 peptide (Neosystem) and APCs (2 × 10⁶ cells/ml) in RPMI 1640 medium containing 10% fetal calf serum, 0.3 mg/ml glutamine, and 10 μM 2-mercaptoethanol. CD90⁺ cell–depleted splenocytes from syngeneic BALB/c mice were used as APCs (1.5 × 10⁶ cells/ml).

Naive (CD4⁺CD62L⁺) CD4 cells were differentiated in a primary stimulation into Th1 or Th2 cells in the presence of OVA_323–339 and T cell–depleted splenocytes as APCs. For Th1 polarization, IL-12 and αIL-4 were added to the cultures and for Th2 polarization, IL-4, αIL-12, and αIFN-γ were used during a primary stimulation period of 6 d. Cultures were split on day 3 after the onset of the primary stimulation. Dead cells were removed by Ficoll gradient centrifugation on day 6. Restimulation (secondary stimulation) was performed with OVA_323–339 and APCs. Polarities of Th1 and Th2 cells were routinely checked 3 d after the secondary Ag stimulation by mitogen stimulation with PMA/ionomycin for 6 h and incubation with brefeldin A for the last 2 h. To this end, intracellular IFN-γ and IL-4 were stained in fixed (2% formaldehyde) and permeabilized (0.5% saponin) cells and analyzed by FACS^®.

CD4 T cells (10⁶ cells/ml) were stimulated at a ratio of 1:1 with Ab-coupled microspheres. Cross-linking of CD152 on CD4 cells was performed using latex microspheres coated with antibodies as described previously (14). In brief, 10⁷ microspheres/ml were suspended in PBS with 0.1 μg/ml αCD3, 2 μg/ml αCD28, αCD152, or a hamster control Ab (NF18, 2.9 μg/ml) and incubated for 1.5 h at 37°C, followed by washing with PBS and blocking with 10% fetal calf serum. In some experiments, 10⁷ microspheres/ml were suspended in PBS with 0.5 μg/ml αCD3, 1 μg/ml αCD28, αCD152, or a hamster control Ab (NF18, 4.5 μg/ml). In case of CD152 cross-linking in the presence of inhibitors, secondary Th2 cells restimulated with Ag and APCs for 48 h were used. After removal of dead cells, Th2 cells were sorted for CD4 expression and replated with Ab-coated microspheres with or without CD152 cross-linking in the presence of 50 U/ml IL-2 and the cell cycle inhibitors L-mimosine or dB-cAMP. IL-2 was added during replating to avoid apoptosis caused by growth factor deprivation or the addition of L-mimosine alone. For inhibition of PI3 K, Th2 cells were restimulated with plate-bound antibodies (2 μg/ml of each αCD3 and αCD28) for 48 h. Cells were pretreated with wortmannin, LY294002, or vehicle controls for 30 min before they were replated with microspheres coated with 0.5 μg/ml αCD3, 1 μg/ml αCD28, 4.5 μg/ml αCD152, or control antibodies in the presence of IL-2. For analyses of cells from CD152^−/− mice, splenocytes were stimulated with 1.5 μg/ml Con A. After removal of dead cells, CD152^−/− cells were restimulated with 2 μg/ml αCD3 and T cell–depleted APCs from syngeneic CD152^+/+ mice.

Cell cycle progression was measured by labeling 10⁷ cells/ml T cells with 10 μM 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes) in PBS/0.1% BSA for 6 min at room temperature. The reaction was stopped by resuspending the cells in RPMI 1640 medium. To measure cell death, cells were stained with annexin-V–PE and propidium iodide (PI) according to the manufacturer's instructions to exclude early apoptotic and late apoptotic/dead cells, respectively. The frequency of nonapoptotic cells was taken into consideration for all experiments. Routinely, staining was controlled by calcium chelation with 2 mM EGTA. Apoptotic DNA fragmentation was measured by TdT-mediated dUTP nick-end labeling (TUNEL) staining using the FLOWTACS kit (R&D Systems). For measurement of caspase activation, cells were incubated for 20 min with the permeable substrate FITC-VAD-fmk (CaspACE; Promega) at a final concentration of 5 μM, followed by subsequent flow cytometric analysis.

Surface expression of CD152 was detected using magnetofluorescent liposomes (32). T cells were incubated with unconjugated hamster αCD152 Ab at 1 μg/ml for 15 min at 4°C. Cells were washed twice, incubated with Cy5 dye-filled liposomes, and conjugated to αham Fab fragments for 30 min at 4°C. Unbound liposomes were removed by washing the cells twice with PBS-BSA. The specificity of CD152 staining was controlled by isotype control Ab conjugated with Cy5-filled liposomes as well as by incubation of cells with Cy5-filled liposomes only. The expression of Bcl-2 was detected using a Cytofix/Cytoperm staining kit obtained from BD Biosciences (3F11-PE). Cytometric analyses were performed using a FACScalibur™ (Becton Dickinson) and CellQuest™ software. Dead cells were excluded by forward and side scatter gating and PI staining in surface staining analyses.

Phosphorylation-induced inactivation of Forkhead transcription factor FKHRL1 was detected by immunoblot analysis as described previously (33). In brief, Th2 cells were cultured under the indicated conditions in the presence and absence of CD152 ligation and lysed after 24 h in sample buffer. Proteins (25 μg/lane) were separated on a SDS–polyacrylamide gel and electroblotted onto a polyvinylidene difluoride membrane. Membranes were blocked for 1 h with 5% milk powder in TBS containing 50 mM NaF and immunoblotted for 1 h with a phosphorylation-specific polyclonal Ab recognizing phosphorylated Thr-32 of FKHRL1 (Santa Cruz Biotechnology, Inc.). Specificity of the Ab was controlled by incubation of the Ab with a FKHRL1 peptide before immunoblotting. Membranes were washed four times with TBS/0.02% Triton X-100 and incubated with a peroxidase-conjugated goat anti–rabbit Ab for 1 h. After extensive washing, the reaction was developed by enhanced chemiluminescent staining.

CD152 is not expressed on the surface of resting T cells and difficult to detect even after stimulation of T cells (22, 24, 34, 35). To analyze CD152 expression in different Th populations, we stimulated naive (CD4⁺CD62L⁺) OVA-specific CD4⁺ T cells in a primary stimulation with the peptide OVA_323–339 and congenic APCs using Th1 and Th2 culture conditions, respectively. On day 6, cells were restimulated with congenic APCs and Ag. Polarization of primary Th1 and Th2 cells resulted in a purity of >99.5% in both effector populations, as assessed by determining IFN-γ producers in the Th2 and IL-4 producers in the Th1 population after stimulation with PMA and ionomycin on day 3 of a secondary Ag-specific stimulation (Fig. 1 A). To unambiguously identify T cells expressing surface CD152, we used a staining technique based on CD152-specific immunofluorescent liposomes, which increased the detection sensitivity at least 1,000-fold (32). Expression of surface CD152 was evaluated at various time points after primary activation of naive T cells in Th1 or Th2 conditions and a subsequent secondary Ag stimulation (Fig. 1, B and C). Specific surface expression of CD152 was detected with similar frequencies in primary Th1 and Th2 cultures after primary stimulation and peaked after 48 h. Thereafter, the frequency of surface CD152-expressing cells declined in Th2 cultures until day 5, whereas surface CD152-expressing Th1 cells were hardly detectable (1%) 3 d after stimulation. After secondary stimulation, >40% of the Th2 cells expressed CD152 and only ∼10% of the Th1 cells displayed CD152 on their surface 48 h after restimulation (Fig. 1, B and C). Thus, both primary Th1 and Th2 cells are able to express surface CD152, but CD152 is expressed at a four- to fivefold higher frequency in Th2 cells, indicating that CD152 might play a more prominent role in Th2 than in Th1 cells.

To examine the role of CD152 during AICD, serological inactivation of CD152 was performed by adding neutralizing αCD152 Fab fragments to Th1 and Th2 cells during primary and secondary stimulations. Nonapoptotic cells were evaluated after different times after restimulation using double staining with PI and annexin-V. PI was used to identify dead and late apoptotic cells. Annexin-V stains phosphatidylserine in a Ca²⁺-dependent way, which is exposed at early stages of apoptosis. The specificity of the staining was controlled by chelating Ca²⁺ with EGTA demonstrating no unspecific annexin-V staining (unpublished data). After restimulation of Ag-experienced cells for 6 d, only 36% of the Th1 cells remained viable, whereas 57% of the Th2 cells survived after the onset of secondary Ag stimulation (Fig. 2 A). Serological blockade of CD152 by αCD152 Fab fragments led to reduced frequencies of surviving cells and, interestingly, only low numbers of Th1 (22%) and Th2 cells (29%) remained viable (Fig. 2 A). The small difference in cell survival of 5–10% between activated, CD152-blocked Th1 and Th2 cells was highly reproducible. In principle, similar effects of CD152 in Th1 and Th2 cells were found after 3 d of restimulation, although cell death was clearly less pronounced at this earlier time point (Fig. 2 A). A survival-promoting effect of CD152 was confirmed when cell death was analyzed by measurement of DNA fragmentation using TUNEL staining or by the flow cytometric detection of caspase activation (Fig. 2, B and C). Thus, CD152 blockade enhanced the susceptibility to apoptosis in differentiated Th cells, which was more pronounced in Th2 cells compared with Th1 cells.

We further used cells from mice harboring a genetic inactivation of CD152. After polyclonal stimulation of splenocytes with ConA, substantially more CD152^−/− splenocytes were killed on day 5 compared with wild-type cells (Fig. 3 A). Excluding early apoptotic (annexin-V⁺, PI⁻) and dead (PI⁺) T cells, ∼30% more of the wild-type cells than CD152^−/− cells remained viable. A comparable difference between CD152^−/− and CD152-expressing cells was found, when dead cells were removed and CD4⁺ T cells were restimulated with αCD3 for a further 48 h (Fig. 3 B). Enhanced apoptosis of CD152^−/− cells was also confirmed by TUNEL staining (Fig. 3 C). Thus, although it must be considered that CD152^−/− cells differ even after activation from wild-type cells, in the presence of functional CD152, less apoptosis occurred after primary and secondary cell stimulation, which supports the idea that CD152 could be involved in apoptosis resistance.

Because CD152 blockade may lead to enhanced IFN-γ production (35), which has been implicated in AICD (36), IFN-γ levels were analyzed in the cultures using a sensitive ELISA. After serological inactivation of CD152, enhanced IFN-γ production was detectable in supernatants of Th1 but not Th2 populations (Fig. 4 A). Thus, at least in Th2 cells, enhanced IFN-γ production was unlikely to be responsible for the reduced survival in the absence of CD152 engagement.

Blockade of CD152 during antigenic stimulation of primary T cells also enhances IL-2 synthesis, which could affect apoptosis induction (26, 37, 38). After neutralization of CD152 with Fab fragments, we found enhanced IL-2 levels in supernatants of Th1 cells, but also Th2 cells, increasing from 0.01 to 2 ng/ml (unpublished data). The enhanced apoptosis in CD152-blocked Th2 cells was not reversed by inhibition of IL-2 signaling (Fig. 4 B). Furthermore, addition of exogenous IL-2 even slightly enhanced survival of Th1 and Th2 cells (Fig. 4 C), but the CD152 effect was still clearly detectable, which indicated that the susceptibility to apoptosis conferred by CD152 blockade was not due to effects of IL-2.

In the next experiments, we asked whether CD152-mediated protection against AICD was a consequence of altered cell cycle progression or a direct effect of CD152. Because CD152 exerts its main function at its peak expression, we monitored its effect on an equally activated CD4 T cell population 48 h after the onset of cell stimulation. To avoid potential effects of IFN-γ on AICD (36), we focused on Th2 cells and induced AICD in the presence of IL-2 as described previously (8). Th2 cells were restimulated for 48 h with congenic APCs and OVA, and Ab-coupled microspheres were used to cross-link CD152 in combination with CD3 and CD28 coligation. Individual nonapoptotic cells were determined 3 d (Fig. 5, top) and 4 d later (not depicted) with similar results. 21% more Th2 cells remained resistant to AICD after stimulation with αCD3, αCD28, and αCD152 compared with CD3/CD28-stimulated T cells. Thus, CD152 engagement of Th2 cells triggered enhanced resistance to AICD, which was even observed in the presence of αCD28.

Another hallmark of CD152 signaling is the induction of G1 cell cycle arrest in lymphocytes (14). Because cell cycle arrest might induce resistance to apoptosis (39), we performed similar experiments in the presence of L-mimosine or dB-cAMP, two drugs inducing G1 arrest (Fig. 5). As reported previously (40), measurement of CFSE staining revealed that both drugs arrested cell cycle progression (unpublished data), and further analysis was only performed in nonproliferating cells. In the presence of IL-2, L-mimosine and dB-cAMP caused a moderate increase in the viability of cells that had been restimulated solely with αCD3 and αCD28. Nevertheless, CD152 ligation induced a comparable increase in resistance against AICD (∼20%), regardless of whether mimosine or dB-cAMP was used. Replating of the L-mimosine–treated cells without IL-2 caused dramatic cell death that could not be analyzed. CD152 ligation of dB-cAMP–treated cells was protective also without IL-2, although to a lesser extent than in the presence of IL-2 (unpublished data). Thus, CD152-induced resistance of Th2 cells against apoptosis was not caused indirectly by reduced IL-2 production or G1 arrest, but rather by a direct CD152 effect.

Because AICD is mainly mediated by the Fas system, we investigated whether CD152 interfered with Fas/FasL interaction. Th1 and Th2 effectors were generated under our standard conditions by stimulation of naive T cells with Ag plus APCs in the presence or absence of neutralizing αCD152 Fab fragments. Upon Ag-specific restimulation, Fas/FasL interaction was blocked with the FasL-specific Ab 3C82. Ag-stimulated Th1 cells showed only a 14% increase in the number of cells undergoing AICD after CD152 inhibition, whereas the blockage of CD152 in Th2 cells resulted in ∼30% more cell death 6 d after restimulation (Fig. 6 A). Analysis on day 5 gave similar results (unpublished data). Prevention of Fas signaling by αFasL reversed the sensitizing effect induced by CD152 blockade completely, and also substantially inhibited AICD. Despite different frequencies of Th1 and Th2 cells undergoing AICD, in both populations the CD152-mediated effect was dependent on the Fas/FasL pathway.

Previous data on the expression of Fas and FasL in Th1 and Th2 cells are controversial, due to the use of unspecific antibodies, insufficient T cell polarization, or the use of different Th effector cell clones and cell lines. Therefore, we analyzed Fas and FasL expression using polarized, pure Th1 and Th2 cells upon restimulation with APCs and Ag, which allowed physiological TCR signaling in the presence of dynamic B7 expression. In unstimulated cells, FasL was not detectable and also expression of Fas was low. Both Th cell populations showed increased surface expression of Fas and FasL after the onset of secondary stimulation, but at different densities (Fig. 6 B). Fas was strongly expressed 8 h after restimulation and steadily declined thereafter. In general, Th1 cells showed a higher density of Fas molecules per cell compared with Th2 cells over the entire time course. Serological inactivation of CD152 enhanced the expression of Fas in both populations (Fig. 6 B). Also FasL was expressed at lower levels in Th2 cells. Blockade of CD152 during stimulation substantially increased FasL expression in Th2 cells, whereas it had no enhancing effect in Th1 cells. Thus, CD152 modulated expression of Fas in Th1 and Th2 cells, whereas FasL levels were regulated by CD152 mainly in Th2 cells. Consistently, FasL was also expressed to substantially higher levels in CD152^−/− T cells compared with wild-type cells after restimulation with APCs and αCD3 (Fig. 6 C). Therefore, these results suggest that down-regulation of FasL expression in Th2 cells might contribute to CD152-mediated resistance against AICD.

CD152 binds to and activates PI3 K and, therefore, we examined whether PI3 K was necessary for apoptosis resistance induced by CD152. Th2 cells were stimulated for 48 h with plate-bound αCD3 and αCD28, before CD152 was cross-linked together with αCD3 and αCD28 coupled to microspheres. Because inhibition of PI3 K might prevent IL-2 production (41) and replating could cause growth factor deprivation leading to cell death, cells were supplied with 50 U/ml of exogenous IL-2. When cells were treated with the PI3 K inhibitors LY294002 or wortmannin, inhibition of PI3 K activity completely prevented the survival effect of CD152 ligation. The survival effect of CD152 ligation was attenuated by Ly294002 also in the absence of IL-2 addition (unpublished data). Thus, T cells restimulated with αCD3/αCD28 plus αCD152 in the presence of the PI3 K inhibitors revealed similar frequencies of nonapoptotic cells as T cells stimulated with only αCD3/αCD28 (Fig. 7 A). The inhibitors themselves were not cytotoxic as determined by incubation of nonstimulated T cells for 2 d (Fig. 7 A, inset). Also, LY294002 did not affect the increased cell death induced by neutralizing αCD152, but only reduced the number of viable Th2 cells in control cultures, where CD152 signaling was still functional (Fig. 7 B).

PI3 K–mediated generation of the second messenger PIP₃ leads to activation of the survival kinase Akt (42). Akt phosphorylates and thereby inhibits several proapoptotic factors, including the Forkhead transcription factor FKHRL1, which is involved in transactivation of the FasL promoter (43) and, therefore, might also contribute to CD152-mediated down-regulation of FasL. Therefore, phosphorylated and inactivated FKHRL1 was analyzed in Ag-primed Th2 cells using a phosphorylation-specific Ab. Unstimulated cells revealed only low amounts of phosphorylated FKHRL1, whereas restimulation of Th2 cells resulted in a strong increase of FKHRL1 phosphorylation (Fig. 7 C, left). Most importantly, when cells were restimulated and CD152 was blocked by αCD152 Fab, a significant decrease of FKHRL1 phosphorylation was observed. The specificity of the analysis was verified by competition with FKHRL1 peptide as well as by experiments in the presence of LY294002, both of which resulted in disappearance of the phosphospecific protein band (Fig. 7 C, right).

Another consequence of PI3 K–triggered Akt activation is induction of Bcl-2 expression, which was recently observed in αCD3-stimulated T cells (44). Inhibition of CD152 strongly reduced the amounts of Bcl-2 in activated Th2 cells upon restimulation with Ag and APCs, which was evident in particular on day 1 after restimulation (Fig. 8 A). In addition, experiments with CD152 cross-linking were performed. In primary Th2 cells as well as in Th2 cells preactivated for 48 h and restimulated with αCD3/αCD28-coupled microspheres, CD152 activation resulted in increased Bcl-2 expression (Fig. 8 B). The induction of Bcl-2 expression by CD152 was completely reversed by LY294002. Thus, Bcl-2 expression was regulated by CD152-mediated PI3 K activity during CD3/CD28 stimulation on primary activated Th2 cells. Altogether, our data show that CD152 is predominantly expressed on activated Th2 cells and contributes to enhanced resistance of activated Th2 cells against AICD by down-regulating Fas/FasL expression, an event closely associated with CD152/PI3 K–mediated up-regulation of Bcl-2.

Our results identify enhanced CD152 surface expression and signaling in Th2 cells as an important functional difference between Th1 and Th2 cells. CD152 cross-linking in the Th2 population revealed that a fraction of cells was rescued from apoptosis corresponding to the number of Th2 cells expressing surface CD152. The heterogeneity of CD152 expression in individual primary T cells has not been analyzed thus far. Our paper shows clearly that Th1 and Th2 cells differ in the frequency of CD152 expression and, therefore, are remarkably different in their susceptibility to AICD. Although more Th1 effectors than Th2 cells underwent Fas/FasL-mediated AICD, mainly Th2 effectors could escape this fate through CD152 engagement. We demonstrate that antiapoptotic consequences of CD152 ligation are associated with up-regulation of Bcl-2, phosphorylation of FKHRL1, and reduced Fas/FasL expression.

Expression of CD152 on the surface of primary T cells is difficult to detect using conventional cytometric and microscopic techniques (14). Studies on CD152 expression have been generally performed intracellularly without discriminating CD152 surface expression, and mostly in Th1 and Th2 cell clones or cell lines. The sensitive liposome-based staining technique used in this work allowed the detection of as few as 100–200 surface molecules per cell. We show through CD152-specific liposome staining that only a fraction of primary cells, activated in an appropriate Th1 or Th2 milieu, express surface CD152. The induction of CD152 expression was rather similar during primary responses of naive T cells activated under Th1 or Th2 conditions. However, upon secondary stimulation a pronounced increase of surface CD152-expressing cells was consistently found in particular in the Th2 population. Hence, the differentiation history of an activated naive T cell apparently correlates with CD152 surface expression upon reencountering an Ag. The fact that activated Th1 and Th2 cells differentially express CD152 during an immune response was reflected by a functional analysis as follows. When CD152 was ligated at the time point of maximal expression after the onset of T cell activation, a substantial fraction of the Th2 cells remained viable.

Our paper is the first demonstrating that CD152 ligation and genetic or serological ablation of CD152 signaling affects AICD in different primary T cell subsets. At first view, our data would concur with the observation that transfection of CD152 could inhibit apoptosis in a T cell hybridoma cell line (45). However, in this work, the hybridoma cells were not preactivated before CD152 ligation. As CD152 cross-linking at the beginning of stimulation leads to a general shutdown of T cell activation associated with impaired up-regulation of activation molecules, inducible FasL expression will be also abolished. Thus, it is very likely that in the reported experiments, T cell activation and not AICD was inhibited by CD152.

AICD of already activated T cells is mainly mediated by the expression of Fas/FasL. We provide several lines of evidence that CD152 affects AICD mainly by impinging on the Fas pathway. First, sensitization of activated T cells to AICD induced by blockade of CD152 was completely abolished by FasL antibodies. Second, CD152 ligation led to a strong inhibition of FasL in Th2 cells, whereas decreased Fas expression was seen in both effector cell populations. The inhibition of Fas/FasL expression and a consequently reduced AICD were observed at various time points after antigenic restimulation, suggesting that CD152 did not simply delay the induction of Fas and FasL expression. In accordance with this concept is our finding that FasL was also strongly up-regulated during the onset of activation in CD152^−/− T cells. It might be argued that up-regulation of Fas/FasL in CD152^−/− T cells is discrepant to the lymphoproliferative disease observed in CD152^−/− mice. However, CD152 deficiency also strongly affects cell cycle control and proliferative events that might overcome the up-regulation of proapoptotic mediators. Thus, experiments with CD152^−/− cells, which might be already in activated state ex vivo, must certainly be interpreted with caution. Data with Rag2^−/− CD152^−/− T cells suggested that CD152, even though being involved in tolerance induction, does not affect AICD (46). Whether these differences to our results are caused by differences in the activation history of cells, effects of CD152 on cell cycle progression or different experimental systems remains to be investigated. Nevertheless, and more in line with the present results, it has been shown that Rag2^−/− mice reconstituted with a mixture of CD152^+/+ and CD152^−/− T cells do not show enhanced, but even reduced total numbers of lymphocytes after infection with LCMV and Leishmania major (47). This surprising observation does not only indicate that CD152 affects T cell survival in a nonautonomous fashion but eventually also via modulating the expression of a proapoptotic factor (47).

Our results identified PI3 K as the presumably crucial mediator of CD152–induced resistance against AICD. It has been demonstrated that Akt, a target of PI3 K, is an essential signaling intermediate in this antiapoptotic pathway. We show that cross-linking of CD152 during T cell activation induces Bcl-2, which might contribute to the early protection of T cells. PI3 K, either directly or through the Akt pathway, is involved in the up-regulation of Bcl-2 after CD152 ligation. The PI3 K/Akt survival pathway has been recently reported to be required for the up-regulation of Bcl-2 by IL-2 (45). As CD152 inhibits IL-2 transcription during T cell activation, and Bcl-2 is up-regulated by IL-2, this effect is probably independent of IL-2 and directly due to CD152 signaling. PI3 K/Akt also inhibits the Forkhead factor FKHRL1 involved in FasL transcriptional regulation. Akt phosphorylates FKHRL1 leading to its retention in the cytoplasm (42). Our results demonstrate that antigenic restimulation of Th2 cells results in phosphorylation of FKHRL1, which was significantly impaired when CD152 signaling was blocked by αCD152 Fab. The dephosphorylation of FKHRL1 induced by CD152 inhibition should promote nuclear translocation and activation of target genes such as FasL. It is interesting to note that suppression of Akt has been recently found to induce FasL expression in smooth muscle cells (48).

In most analyses, only suppressive functions and down-regulation of gene expression have been described for CD152. Apart from TGFβ, whose role in CD152 function is obscure, our data show for the first time that CD152 engagement is able to directly induce protein expression (e.g., of Bcl-2). The reason why most studies missed such an effect lies presumably in the fact that CD152 function has been generally examined after CD152 cross-linking at the onset of T cell responses, which will fully prevent T cell activation. Our results suggest that when CD152 is maximally expressed at the cell surface, CD152 might have more complex functions and actively determines a T effector cell's fate.

According to our data, T cells would be activated and primed to die unless they up-regulate surface CD152, which blocks AICD. As only a fraction of cells up-regulate surface CD152, and a similar fraction of cells responds to CD152 cross-linking, we assume that only those cells that express sufficient surface CD152 and receive appropriate additional signals will become resistant of apoptosis. That this restricted population being maintained after the T cell response fades away might be important to create the memory for the adaptive immune response. Although Bcl-2, a T cell memory marker (49, 50), is up-regulated by CD152 in a fraction of CD4 cells, future studies are needed to clarify the fate of them and their role in contribution to the memory cell compartment.

The authors thank A. Scheffold, J.P. Allison, D.Y. Loh, and G.-R. Burmester for kindly providing us with reagents, mice, and support.

The authors work was supported by Deutsche Forschungsgemeinschaft Br 1860/3-2, by Gemeinnützige Hertie-Stiftung GSH 191/00/15, and a grant of the Charité, Universitätsmedizin Berlin to M.C. Brunner-Weinzierl.