To examine the role of CTLA-4 in Th cell differentiation, we used two newly generated CTLA-4-deficient (CTLA-4−/−) mouse strains: DO11.10 CTLA-4−/− mice carrying a class II restricted transgenic TCR specific for OVA, and mice lacking CTLA-4, B7.1 and B7.2 (CTLA-4−/− B7.1/B7.2−/− ). When purified naive CD4+ DO11.10 T cells from CTLA-4−/− and wild-type mice were primed and restimulated in vitro with peptide Ag, CTLA-4−/− DO11.10 T cells developed into Th2 cells, whereas wild-type DO11.10 T cells developed into Th1 cells. Similarly, when CTLA-4−/− CD4+ T cells from mice lacking CTLA-4, B7.1, and B7.2 were stimulated in vitro with anti-CD3 Ab and wild-type APC, these CTLA-4−/− CD4+ T cells produced IL-4 even during the primary stimulation, whereas CD4+ cells from B7.1/B7.2−/− mice did not produce IL-4. Upon secondary stimulation, CD4+ T cells from CTLA-4−/− B7.1/B7.2−/− mice secreted high levels of IL-4, whereas CD4+ T cells from B7.1/B7.2−/− mice produced IFN-γ. In contrast to the effects on CD4+ Th differentiation, the absence of CTLA-4 resulted in only a modest effect on T cell proliferation, and increased proliferation of CTLA-4−/− CD4+ T cells was seen only during secondary stimulation in vitro. Administration of a stimulatory anti-CD28 Ab in vivo induced IL-4 production in CTLA-4−/− B7.1/B7.2−/− but not wild-type mice. These studies demonstrate that CTLA-4 is a critical and potent inhibitor of Th2 differentiation. Thus, the B7-CD28/CTLA-4 pathway plays a critical role in regulating Th2 differentiation in two ways: CD28 promotes Th2 differentiation while CTLA-4 limits Th2 differentiation.
The proliferation and differentiation of T lymphocytes requires Ag, costimulation, and cytokines. The best-defined costimulators for T cells are B7.1 (CD80) and B7.2 (CD86), both of which are expressed on professional APCs, and the expression of which is increased by microbial pathogens and by various cytokines produced during innate immune reactions (1). B7 molecules bind CD28 on T cells and optimize T cell activation (2). A second receptor for B7 molecules, CTLA-4 (CD152), down-regulates T cell activation (3). CTLA-4 expression is up-regulated on activated T cells, and CTLA-4 binds B7.1 and B7.2 with higher affinity than does CD28 (4). The critical role of CTLA-4 in controlling T cell activation is dramatically illustrated by the fatal lymphoproliferative disease seen in CTLA-4-deficient (CTLA-4−/−) mice (5, 6). This disease requires CD4+ cells, suggesting an inhibitory role for CTLA-4 in the initiation and/or maintenance of CD4+ T cell responses (7).
The goal of the studies in this paper was to examine the role of CTLA-4 in T cell proliferation and differentiation. Much of our current knowledge of the function of CTLA-4 is based on experiments using agonistic and blocking Abs. Such studies have shown that CTLA-4 cross-linking inhibits cytokine secretion and cell cycle progression in T cells (8, 9). Many of these effects are the opposite of the demonstrated effects of CD28, which promotes early cytokine (e.g. IL-2) production, T cell survival by inducing expression of anti-apoptotic genes (e.g., Bcl-xL), and T cell differentiation into effector cells (10). CD28 plays an important role in differentiation of naive CD4+ T cells and influences Th2 differentiation more than Th1 development (11, 12). The role of CTLA-4 in Th cell differentiation is not yet clear.
Although the phenotype of the CTLA-4−/− mouse strain has been informative, the early activation of CTLA-4−/− T cells and death of these mice by 4 wk of age have limited the analyses that can be done using this strain. To overcome these problems, we have used two approaches to examine the obligatory roles of CTLA-4 in CD4+ T cell activation and differentiation. We have bred the DO11.10 class II-restricted TCR transgenic mice with CTLA-4−/− mice, and compared Ag-specific responses of wild-type and CTLA-4−/− CD4+ DO11.10 T cells. To exclude in vivo activation of CTLA-4−/− T cells by as yet unknown self or foreign Ags, we also have bred the CTLA-4−/− mice with mice lacking B7.1 and B7.2 (13, 14) and analyzed the responses of these CTLA-4−/− CD4+ T cells to anti-CD3 Ab plus wild-type APCs. Both approaches show that if CD4+ T cells are activated by TCR engagement and costimulation in the absence of CTLA-4, the cells exhibit a dramatic tendency to differentiate into Th2 populations. Surprisingly, the absence of CTLA-4 has only a modest effect on T cell proliferation, and only during secondary responses in vitro. These results suggest there is reciprocal regulation of CD4+ Th cell differentiation by CD28 and CTLA-4. CD28 promotes Th2 differentiation, whereas CTLA-4 is a critical, potent inhibitor of this differentiation pathway.
Materials and Methods
Mice expressing the DO11.10 TCR transgene, which responds to the peptide residues 323–339 of chicken OVA in the context of I-Ad (BALB/c), were obtained from Dr. Dennis Loh (Roche, Nutley, NJ) (15). CTLA-4 heterozygous (+/−) mice were backcrossed onto the BALB/c background for six generations before they were bred with the DO11.10 TCR transgenic mice and intercrossed. PCR typing for the DO11.10 transgene was conducted using primers of sequence 5′-CAG GAG GGA TCC AGT GCC AGC-3′ and 5′-TGG CTC TAC AGT GAG TTT GGT-3′. Genotyping of CTLA-4 was done by Southern blot analysis or PCR analysis of tail DNA for the wild-type locus and the neomycin resistance gene in the targeted locus (5). In addition, a combination of primers was used to amplify neomycin and adjacent CTLA-4 sequence, using 5′-TTG CTC AAA GAA ACA GCA GAG-3′ and 5′-GAC TAG GGG AGG AGT AGA AG-3′. CTLA-4−/− B7.1/B7.2−/− mice were generated by breeding 129S4Sv/Jae CTLA-4+/− mice with inbred 129S4/SvJae B7.1/B7.2−/− mice as described elsewhere (13). Mice were cared for according to institutional guidelines. Brigham and Women’s Hospital and Harvard Medical School are AAALAC (American Association for the Accreditation of Laboratory Animal Care) accredited institutions. Age-matched animals 6–18 wk old were used for experiments.
HPLC-purified OVA peptide 323–339 (OVA323–339) with the sequence I-S-Q-A-V-H-A-A-H-A-E-I-N-E-A-G-R-COOH was obtained from Analytical Biotechnology Services (Boston, MA). The GK1.5 (anti-CD4), ADH4 (anti-CD8), 145-2C11 (anti-CD3), and PV-1 (anti-CD28) hybridomas were obtained from the American Type Culture Collection (Manassas, VA). Anti-Thy1.2 Ab was obtained from Serotec (Oxford, U.K.). Anti-CD4-FITC, anti-L-selectin-PE (CD62L), streptavidin-CyC, anti-IL4-PE, and PE-conjugated isotype control Abs were obtained from PharMingen (San Diego, CA). Purified, nonconjugated anti-IL-4 Ab, 11B11, was also obtained from PharMingen. The hybridoma for the KJ1-26 Ab, which recognizes the DO11.10 TCR transgene, was provided by Dr. P. Marrack (National Jewish Center, Denver, CO) (16) and affinity-purified Ab was biotinylated by standard methods.
CD4+ T cells were obtained from pooled lymph nodes by positive magnetic selection using anti-CD4 (L3T4) Dynabeads and Detachabead (Dynal, Oslo, Norway) or magnetic columns (MACS) from Miltenyi Biotec (Auburn, CA). CD4+ cells were enriched for L-selectinhigh cells using MACS columns with anti-L-selectin beads from Miltenyi Biotec, or by staining cells with biotinylated KJ1-26 Ab and streptavidin-CyC, together with anti-CD4-FITC and anti-L-selectin-PE Abs and sterile cell sorting with a Coulter flow cytometer (Miami, FL). Comparable results were obtained with either method of cell isolation. Purities of populations were assessed by flow cytometry, with over 90% CD4+ purity and 80–90% purity of L-selectinhigh cells. APCs were prepared from splenocytes of BALB/c (for DO11.10) or 129/SvJae mice (for CTLA-4−/− B7.1/B7.2−/−) by treatment with anti-CD4 (GK1.5), anti-CD8 (ADH4), and anti-Thy1.2 Abs followed by rabbit Low-Tox complement (Accurate Chemical and Scientific, Westbury, NY). APCs were inactivated by treatment with 50 μg/ml Mitomycin C for 40 min at 37°C (Sigma, St. Louis, MO).
T cell cultures
For proliferation assays, purified T cells and APC were cultured with various concentrations of OVA peptide or anti-CD3 Ab in 200 μl of RPMI 1640 medium, as previously described (5). Supernatants for cytokine analysis, and T cells for restimulation assays, were harvested from 1 ml cultures initiated with 2 × 105 T cells, 2 × 106 APC, and OVA peptide concentrations as indicated. After 4 days, dead cells were removed with Lymphocyte Separation Medium (Organon Teknika, Durham, NC) and T cells were incubated overnight before restimulation with Ag and fresh APC. In some experiments, anti-IL-4 Ab (11B11) was added to the primary cultures at a concentration of 2 μg/ml.
Supernatants were analyzed for IL-2, IFN-γ, IL-4, IL-5, and IL-10 by sandwich ELISA using Ab pairs and standards purchased from PharMingen, according to the manufacturer’s instructions.
Intracellular cytokine staining
CD4+ T cells (5 × 105) were restimulated with 5 × 106 APC and OVA peptide for 3 days. Intracellular staining for IL-4 was performed as recommended by PharMingen. In brief, cells were activated with 10 ng/ml PMA and 1 μM ionophore for 4–6 h. Brefeldin A (10 μg/ml) was added during the last 2 h of the activation. Cells were fixed in 4% paraformaldehyde, then lysed in 0.5% saponin/1% BSA/0.1% NaN3 and incubated with a mixture of anti-CD4-FITC and anti-IL-4-PE or a PE-conjugated isotype control Ab.
In vivo administration of CD28
CTLA-4−/− B7.1/B7.2−/− or wild-type mice were injected i.v. with 100 μg of purified anti-CD28 Ab (PV-1) on days 0, 7, and 14 and sacrificed on day 19. Lymph node cells (2 × 105) were stimulated with various concentrations of anti-CD3 Ab (145-2C11), and cell cultures supernatants were harvested to measure cytokines at the indicated times.
Proliferation and differentiation of CTLA-4−/− DO11.10 T cells
Wild-type DO11.10 and CTLA-4−/− DO11.10 mice show comparable thymic maturation of T cells and lymphocyte numbers early in life (data not shown). At 6–8 wk of age, CTLA-4−/− DO11.10 mice show mild lymphadenopathy and splenomegaly. This may be due to activation of T cells expressing TCRs other than the DO11.10 TCR, or activation of DO11.10 T cells by cross-reactive ligands. In contrast to nontransgenic CTLA-4−/− mice, which die by 3–4 wk of age, the CTLA-4−/− DO11.10 mice survive for >16 wk, presumably because the transgenic TCR reduces the frequency of autoreactive T cells.
To compare the responses of DO11.10 wild-type and CTLA-4−/− T cells, we purified CD4+/KJ1–26+/L-selectinhigh T cells from both strains. These T cells were activated in vitro with OVA323–339 and APCs, and proliferation and cytokine production were measured during primary and secondary responses. DO11.10 wild-type and CTLA-4−/− T cells showed comparable proliferation (Fig. 1,A) and, in most experiments, similar IL-2 production during primary stimulation, and essentially no IFN-γ was produced, as expected from naïve T cells (Fig. 1 B). The only consistent difference between these cell populations was that after 3–4 days of priming in vitro, the DO11.10 CTLA-4−/− T cells produced detectable amounts of IL-4. This result suggests that naive T cells are capable of producing IL-4 during primary responses to Ag, but are normally prevented from doing so by CTLA-4.
If the absence of CTLA-4 allows IL-4 production during priming, then this should lead to Th2 responses upon restimulation. Analysis of secondary responses showed that this is indeed the case. Upon priming and restimulation with Ag and APCs, wild-type DO11.10 cells produced the Th1 cytokine IFN-γ, whereas CTLA-4−/− DO11.10 T cells produced Th2 cytokines, IL-4, IL-5, and IL-10 (Fig. 2,B). This bias toward Th2 development in cells that lack CTLA-4 was observed over a range of peptide concentrations (0.1, 1, 10 μg/ml) during primary stimulation (data not shown). Interestingly, in these recall responses the CTLA-4−/− DO11.10 T cells proliferated more than wild-type DO11.10 cells (Fig. 2 A), suggesting that the role of CTLA-4 in limiting T cell expansion is seen late during immune responses.
Cytokine production by T cells from CTLA-4−/−B7.1/B7.2−/− mice.
Although the experiments with DO11.10 T cells suggest that CTLA-4 regulates Th2 differentiation, an alternative possibility is that some of these cells are activated in vivo by either environmental or endogenous Ag, and this may affect subsequent differentiation of the T cells. It is difficult to exclude such in vivo activation by unknown Ags. We have shown that stimulation of CD28 is critical for T cell activation in CTLA-4−/− mice, and that B7 antagonists prevent “spontaneous” lymphocyte activation and disease in CTLA-4−/− mice (17). More recently, we have demonstrated that breeding the CTLA-4−/− mice with mice lacking both B7.1 and B7.2 also completely prevents endogenous activation and disease (13). T cells from CTLA-4−/− B7.1/B7.2−/− mice have the phenotypic and functional characteristics of naive cells. We reasoned that studying the responses of CTLA-4−/− B7.1/B7.2−/− T cells to TCR cross-linking and wild-type APCs in vitro would provide an additional means of evaluating the functional capabilities of CTLA-4−/− T cells, without the confounding influence of in vivo activation.
CD4+ T cells were isolated from CTLA-4−/− B7.1/B7.2−/− mice and activated in vitro with anti-CD3 Ab and syngeneic wild-type APCs. The primary and secondary responses of these cells were compared with those of CD4+ T cells from mice lacking only B7.1 and B7.2. As was seen with DO11.10 T cells, the initial proliferation of these CTLA-4−/− B7.1/B7.2−/− T cells was similar to that of CTLA-4+/+ B7.1/B7.2−/− cells. However, these CTLA-4−/− cells proliferated more than the CTLA-4+/+ cells during secondary stimulation (data not shown). Strikingly, the CTLA-4−/− B7.1/B7.2−/− T cells produced IL-4, even during primary stimulation in the presence of wild-type APCs (Fig. 3,A), whereas the CTLA-4+/+ B7.1/B7.2−/− cells did not. Moreover, upon secondary stimulation, these CTLA-4−/− cells secreted much more IL-4 and much less IFN-γ than did CTLA-4+/+ B7.1/B7.2−/− cells (Fig. 3 B). This differentiation into Th2 cells did not appear to be solely determined by the strength of signal during the priming, because IL-4 was produced by CTLA-4−/− cells that had been primed over a range of anti-CD3 Ab concentrations (from 1:500 to 1:10,000) during priming (data not shown). Thus, when naive T cells lacking CTLA-4 are exposed to an Ag analogue (anti-CD3 Ab) and costimulation, they differentiate into Th2 populations.
Neutralizing IL-4 during priming only partially inhibits Th2 differentiation of CTLA-4−/− T cells.
To investigate whether the Th2 differentiation of CTLA-4−/− T cells was due to IL-4 produced by CTLA-4−/− T cells or an intrinsic property of CTLA-4−/− T cells, we examined the consequences of priming CTLA-4−/− T cells in the presence of neutralizing anti-IL-4 Ab. CD4+/KJ1–26+/L-selectinhigh wild-type DO11.10 and CTLA-4−/− DO11.10 T cells were primed with peptide Ag and APC in the presence of a saturating concentration of anti-IL-4 Ab. During priming of CTLA-4−/− DO11.10 T cells with 10 μg/ml peptide for 5 days, the concentration of IL-4 in the supernatants was 0.07 ± 0.03 ng/ml, whereas in the presence of anti-IL-4 Ab no IL-4 was detectable (detection limit for IL-4 ELISA was <0.005 ng/ml). Therefore, IL-4 was effectively neutralized by the Ab. In both wild-type and CTLA-4−/− DO11.10 T cells, the presence of anti-IL-4 Ab during priming increased IFN-γ production upon secondary stimulation (Fig. 4). However, the presence of anti-IL-4 Ab during priming resulted in only a modest decrease in IL-4 upon restimulation of CTLA-4−/− cells, compared with CTLA-4−/− cells primed in the absence of anti-IL-4. In secondary cultures, IL-4 was detected by ELISA in supernatants of CTLA-4−/− DO11.10 T cells that had been primed in the presence of anti-IL-4 (Fig. 4,A). Furthermore, intracellular cytokine staining showed IL-4 production by CTLA-4−/− DO11.10 cells that had been primed in the presence of anti-IL-4 (Fig. 4 B). In addition, IL-4 mRNA was detected by RNase protection assays (RPAs)4 of CTLA-4−/− DO11.10 cells that had been primed in the presence of anti-IL-4 (data not shown).
Similarly, the addition of anti-IL-4 Ab to the primary cultures of CTLA-4−/− B7.1/B7.2−/− T cells failed to completely block IL-4 production during secondary stimulation of these CTLA-4−/− T cells(data not shown). Analyses of cytokine transcripts by RPAs gave similar results. When RPAs were performed with CTLA-4−/− B7.1/B7.2−/− and B7.1/B7.2−/− T cells that had been primed with anti-CD3 and wild-type APC in the presence of anti-IL-4 Ab and subsequently restimulated with plate-bound anti-CD3 Ab for 6 h, significant amounts of mRNA for the Th2 cytokines, IL-4, IL-5, IL-10, and IL-13 were only observed in the CTLA-4−/− B7.1/B7.2−/− T cells (data not shown). These data suggest that the propensity of CTLA-4−/− T cells to differentiate toward the Th2 pathway might be an intrinsic property of these T cells.
Anti-CD28 Ab stimulates Th2 differentiation in vivo in the absence of CTLA-4
The experiments described so far have investigated in vitro responses of CTLA-4−/− T cells to peptide Ag or TCR cross-linking and costimulation. We postulated that if CTLA-4 has an essential role in regulating Th2 differentiation, then providing mice lacking CTLA-4 with a CD28-mediated signal in vivo also might induce a Th2 response. To test this, a stimulatory anti-CD28 Ab was administered i.v. to CTLA-4−/− B7.1/B7.2−/− mice or wild-type mice, lymph node cells were harvested 19 days later, and IL-4 production was assayed following restimulation of lymph node cells with anti-CD3 Ab. We had previously shown that this anti-CD28 Ab (PV-1) produces a lymphoproliferative phenotype in CTLA-4−/− B7.1/B7.2−/− mice, but has no in vivo proliferative effect in wild-type mice (13). As shown in Fig. 5, exposure to this agonistic anti-CD28 Ab in vivo induced IL-4 production in CTLA-4−/− B7.1/B7.2−/− cells, but not in wild-type cells. The CTLA-4−/− cells made less IFN-γ than wild-type mice (data not shown). The lack of an effect of anti-CD28 Ab in wild-type mice suggests that the B7-CTLA-4 interaction can provide signals that limit CD28-mediated Th2 differentiation, even when CD28 is triggered with a cross-linking Ab.
This study was designed to explore the role of CTLA-4 in regulating the functional responses of CD4+ T cells. To do this, we first compared T cells expressing a transgenic TCR, either on a wild-type or CTLA-4−/− background. This experimental approach allowed us to quantitatively analyze the responses of T cells to physiologic stimulation with a defined Ag and APCs. However, we could not exclude potential in vivo activation of the TCR transgenic T cells, because the transgenic TCR does not exclude all endogenous TCRs. This possibility can be eliminated by breeding the mice into RAG-2−/− mice, which is in progress. However, even in RAG-2−/− TCR mice, it is possible that T cells expressing the transgenic TCR might be activated by positively selecting ligands in vivo. Therefore, we used a second experimental system, in which the CTLA-4−/− mice were bred with mice lacking B7.1 and B7.2. In the absence of B7 costimulation, T cells from CTLA-4−/− B7.1/B7/2−/− are not activated in vivo (13), so the responses of naïve T cells can be examined.
Both of these experimental systems show that CTLA-4 has a critical role in regulating Th2 differentiation. CTLA-4−/− T cells produced IL-4 even during primary stimulation and rapidly develop into almost pure populations of Th2 cells. This was seen not only in the DO11.10 strain, which is on a Th2-prone BALB/c background, but also in the 129Sv strain, in which T cells have a tendency to develop into Th1 cells when primed under “neutral” conditions. Furthermore, in vivo anti-CD28 administration led to striking Th2 differentiation in CTLA-4−/− B7.1/B7/2−/−, but not wild-type mice. Blocking IL-4 during priming of CTLA-4−/− T cells did not prevent Th2 development completely, whereas anti-IL-4 Ab inhibited Th2 responses of wild-type T cells (18). This may be because CTLA-4−/− T cells produce IL-4 whenever they encounter costimulation, i.e., either during primary or secondary stimulation. A number of studies have shown that the strength of the signal that a T cell receives contributes to the outcome of the T helper phenotype (19, 20). For this reason, we examined CTLA-4−/− T cells over a range of Ag doses. In all studies, we observed differentiation toward the Th2 phenotype.
The results described in this manuscript are consistent with a study using anti-CTLA-4 Ab, in which blockade of CTLA-4 during an staphylococcal enterotoxin B (SEB)-mediated immune response induced IL-4 production, especially by CD4+ cells (21). These results complement studies showing an important role for B7-CD28-mediated costimulation in Th2 development, because CD28−/− DO11.10 T cells and naive T cells stimulated with B7.1/B7.2−/− APC have impaired Th2 differentiation (11, 12).
The absence of CTLA-4 had no influence on initial proliferative responses and a relatively modest effect on proliferation upon restimulation. The modest effects of CTLA-4 on T cell proliferation are surprising in light of studies with Abs showing that CTLA-4 ligation limits T cell expansion very early after triggering (8, 9). In fact, our results suggest that CTLA-4 has an obligatory role in inhibiting T cell expansion late in immune responses. The kinetics of this effect are consistent with the kinetics of CTLA-4 expression, which typically increases 24–48 h after T cell activation in vitro (22).
We conclude from these studies that one of the major functions of CTLA-4 is to limit the magnitude of Th2 differentiation. When T cells recognize Ag on APCs, engagement of CD28 by B7 molecules promotes Th2 differentiation, whereas B7 binding to CTLA-4 blocks Th2 differentiation. The mechanism of these opposing effects of CD28 and CTLA-4 on cytokine profiles remains to be established. One potential mechanism could be that CTLA-4 competes with CD28 for B7 binding and therefore counteracts the Th2 differentiation induced by CD28 triggering. An alternative possibility is that CD28 transduces signals that directly activate the transcription of Th2 cytokine genes, and CTLA-4 inhibits these signals. We are currently comparing the effects of CD28 and CTLA-4 ligation on transcription factors that are known to bind to cytokine gene promoters (23) to identify the critical signals involved. Finally, these results raise the possibility that the skewed Th2 response of CTLA-4−/− T cells plays a role in the lymphoproliferative and infiltrative disease of CTLA-4−/− mice. Studies to define the roles of various cytokines in the disease of CTLA-4−/− mice, using cytokine antagonists and knockouts, are in progress.
We thank S. Scott for technical assistance, B. Chang and J. Burgess for their help with genotyping mice, and Drs. Alex J. McAdam and Nicky A. Schweitzer for critically reading the manuscript.
This work was supported by National Institutes of Health Grants RO1 AI40614 (A.H.S.), RO1 AI25022 (A.K.A.), and PO1 AI35297 (A.K.A. and A.H.S.); a fellowship from the Human Frontier Science Program (M.A.O.); Grant K11 AI01212 (D.A.M.); and National Institutes of Health Training Grant T32 HL07627 (R.B.L.).
Abbreviation used in this paper: RPA, RNase protection assay.