A Role for Inducible Costimulator Protein in the CD28- Independent Mechanism of Resistance to Toxoplasma gondii¹

Infection with Toxoplasma gondii results in the activation of a strong cell-mediated Th1-type immune response characterized by the production of IFN-γ, the major mediator of resistance to infection (1). Given the critical role of T cells in the initiation and maintenance of immunity to T. gondii (2, 3, 4), it is important to understand the events that regulate their responses. Since the description of the two signal hypothesis of lymphocyte activation, it has been recognized that costimulation is an important process involved in the development and regulation of T cell responses. The B7/CD28 pathway is an important costimulatory interaction involved in T cell activation (5). However, previous studies from this laboratory have shown that CD28^−/− mice infected with T. gondii have a reduced CD4⁺ T cell response, but do mount a T cell response that is sufficient to provide resistance to this infection (6, 7). These results suggest that other costimulatory pathways are likely involved in the regulation of T cell responses during toxoplasmosis, or may reflect the ability of T. gondii infection to provide a TCR signal that is sufficient to overcome the requirement for CD28-mediated costimulation (8, 9).

Costimulatory functions have been ascribed to molecules such as ICAM-1, CD2, CD30, CD40 ligand, CD44, CD137, and 4-1BB (10, 11, 12, 13), but our studies have suggested that neither ICAM-1, CD40 ligand, or CD44 are critical for the development of T cell responses required for resistance to T. gondii (7, 14). Recently, inducible costimulator protein (ICOS),⁴ a molecule that is homologous to CD28 has been shown to be a potent costimulatory molecule in T cell activation (15, 16). Unlike CD28, which is constitutively expressed on T cells, ICOS is expressed only on activated CD4⁺ and CD8⁺ T cells and a subpopulation of memory T cells (15, 16). The ligand for ICOS is B7-related protein 1 (B7RP-1; B7 h, GL50, ICOSL), which is structurally related to members of the B7 family, and is expressed on macrophages, B cells, TNF-α-activated fibroblast cells, as well as on other nonlymphoid tissues (15, 17, 18, 19). Early studies demonstrated that ICOS was involved in the production of IL-4, IL-10, and IFN-γ, but not IL-2 by T cells, and subsequent studies reported that ICOS is expressed preferentially on Th2-type T cells and is important during Th2-type immune responses (20, 21, 22, 23, 24, 25). Additional studies have shown that ICOS is involved in the regulation of Th1 and Th2 responses to the parasites Nippostrongylus brasiliensis and Leishmania mexicana as well as the viral pathogens lymphocytic choriomeningitis virus and vesicular stomatitis virus but the importance of ICOS in the outcome of these infections is unclear (26, 27). Nevertheless, since ICOS is capable of providing a costimulatory signal to activate T cells, this was a candidate costimulatory molecule for the CD28-independent regulation of Th1-type T cell responses during toxoplasmosis. The studies presented here reveal that ICOS expression is up-regulated on activated T cells following infection with T. gondii and that this is independent of CD28-mediated signaling. Moreover, ICOS and CD28 are both required for optimal production of IFN-γ by parasite-specific T cells and the in vivo blockade of the B7RP-1-ICOS interaction results in increased susceptibility of CD28^−/− mice to infection. Together, these results identify a critical role for ICOS in the protective Th1-type response to T. gondii and suggest that ICOS and CD28 are parallel costimulatory pathways, either of which is sufficient to mediate resistance to T. gondii.

C57BL/6, Swiss Webster, and CBA/CaJ mice were obtained from The Jackson Laboratory (Bar Harbor, ME). C57BL/6 CD28-deficient animals (28) were bred and housed in the animal facilities at the University of Pennsylvania (Philadelphia, PA). All mice used were maintained within microisolator caging units at the University Laboratory Animal Resource facilities at the University of Pennsylvania. Six- to 12-wk-old C57BL/6 or CD28^−/− mice were used for all experiments with three to six mice per group.

Swiss Webster and CBA/CaJ mice were used to maintain the ME49 strain of T. gondii and used as sources of tissue cysts for these experiments. For infection with ME49 tissue cysts, brains of chronically infected CBA/CaJ mice were harvested and prepared as previously described (29, 30). Mice were inoculated i.p. with 20 ME49 cysts in a volume of 0.2 ml. Serum samples were collected at different time points and survival was monitored. The RH strain of T. gondii was maintained in human foreskin fibroblast as previously described (7) and used to prepare soluble Toxoplasma Ag (STAg) as previously described (31). STAg activity was titrated to determine the optimal concentration for splenocyte proliferation and cytokine production (25 μg/ml).

At the time of sacrifice, peritoneal exudate cells were harvested with 5 ml of ice-cold 1× PBS without Ca²⁺/Mg²⁺ (BioWhittaker, Walkersville, MD) and 5 × 10⁵ cells/100 μl were used to prepare cytospins to quantitate parasite burden. Cells were fixed and stained using Protocol Hema3 (Biochemical Sciences, Swedesboro, NJ) as described in the manufacturer’s manual, then mounted and sealed using Cytoseal (Stephens Scientific, Kalamazoo, MI).

Complete RPMI 1640 (Life Technologies, Gaithersburg, MD) medium containing 10% heat-inactivated FCS (HyClone Laboratories, Logan, UT), sodium pyruvate, nonessential amino acids, penicillin (100 U/ml), streptomycin (1 mg/ml), and amphotericin B (25 ng/ml) (BioWhittaker). The 145-2C11 hybridoma for anti-CD3ε was provided by Dr. J. Bluestone (University of Chicago, Chicago, IL) and used to prepare purified anti-CD3. HuCTLA-4Ig, a fusion protein comprised of the human CTLA-4 extracellular domain and Fc portion of human IgG, and the chimeric control fusion protein L6 (Chi-L6) was supplied by Bristol-Myers Squibb Research Institute (Princeton, NJ) and used at a concentration of 20 μg/ml. B7RP-1Fc, a fusion protein consisting of the extracellular portion of the mouse B7RP-1 and the Fc portion of human IgG, was used to detect expression of ICOS and was supplied by Amgen (Thousand Oaks, CA). mAb specific to B7RP-1 (1B7; Amgen) was used at a concentration of 20 μg/ml. Infected mice were treated with 1 mg of anti-B7RP-1 or rat Ig (Sigma-Aldrich, St. Louis, MO) at days 0 and 5 postinfection.

Isolation of brain mononuclear cells (BMNCs) were performed as previously described (6). Briefly, mice were anesthetized with 200 mg/kg Kitaset (Fort Dodge Animal Health, Fort Dodge, IA) and 10 mg/kg Xyla-Ject (Phoenix Pharmaceuticals, St. Joseph, MO). Mice were perfused through the left ventricle with 50–60 ml of ice-cold PBS to remove peripheral blood. Brains were then collected, minced, resuspended in complete RPMI 1640, and digested with 100 μg of collagenase/dispase (Roche Diagnostics, Indianapolis, IN) and 300 μg of DNase (Roche Diagnostics) for 45 min at 37°C. Brain digests were then pelleted at 2000 × g for 10 min at 4°C and resuspended in 60% isotonic Percoll solution and overlaid with 30% Percoll solution. The Percoll gradient was centrifuged at 1000 × g for 25 min at 25°C and the top myelin layer was carefully removed before harvesting BMNCs at the 60 and 30% interphase layer. Cells were washed with complete RPMI 1640 before further analysis. Due to the low number of BMNCs obtained per animal, cells from at least three mice were pooled in each experiment.

Individual spleens from animals were harvested, dissociated into single-cell suspensions, and depleted of erythrocytes using 0.83% w/v ammonium chloride (Sigma-Aldrich). Cells were washed and resuspended in complete RPMI 1640 before further analysis. Splenocytes or BMNCs were plated at a cell density of 2 × 10⁵ cells/well in a final volume of 200 μl in 96-well plates (Costar, Costar, NY). Cells were stimulated with soluble anti-CD3 (145-2C11) at a final concentration of 1 μg/ml or STAg for 48 h at 37°C/5% CO₂ in the presence of CTLA-4Ig and/or anti-B7RP-1. Levels of IFN-γ from splenocyte cultures were measured using ELISA as previously described (32).

To assess surface expression of ICOS on T cells, purified splenocytes were resuspended in FACS buffer (1× PBS, 0.2% BSA fraction V, and 4 mM sodium azide) at a concentration of 1 × 10⁷ cells/ml. A total of 10⁶ cells was preincubated with saturating concentrations of Fc Block for 20 min on ice and stained with PE or PerCP-conjugated anti-CD4, FITC-conjugated anti-CD8 (BD PharMingen, San Diego, CA), and B7RP-1Fc (1 mg/ml) for 30 min on ice. Cells were washed with FACS buffer and then incubated with biotinylated goat anti-human IgG (Vector Laboratories, Burlingame, CA) for 30 min on ice. Cells were washed with FACS buffer and incubated with allophycocyanin-conjugated streptavidin (BD PharMingen) for an additional 20 min on ice. Cells were then washed, resuspended in FACS buffer, and analyzed using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA). For intracellular detection of IFN-γ, purified splenocytes were plated in a 96-well plate (Costar) at a density of 4 × 10⁵ cells/well in a final volume of 200 μl. Cells were stimulated with STAg (25 μg/ml) for 72 h, then PMA (50 ng/ml), ionomycin (500 ng/ml), and brefeldin A (10 μg/ml; Sigma-Aldrich) were added to cultures during the last 5 h of stimulation. Cells were harvested, washed, resuspended in FACS buffer, and then stained with various fluorochrome-conjugated Abs as described above. Cells were then washed with FACS buffer, fixed with 4% w/v paraformaldehyde, washed again, and permeabilized with 0.1% w/v saponin in FACS buffer. After permeabilization, cells were stained with allophycocyanin-conjugated anti-IFN-γ (BD PharMingen) for 30 min on ice. Cells were washed once with 0.1% saponin buffer and then with FACS buffer. Analysis of the cells was performed using a FACSCalibur flow cytometer (BD Biosciences). Results were analyzed using CellQuest software (BD Biosciences). Ab concentrations were empirically determined to give optimal staining for flow cytometric analyses.

INSTAT software (GraphPad, San Diego, CA) was used for unpaired or paired two-tailed Student’s t test. A value of p < 0.05 was considered to be significant.

The recent identification of the B7RP-1-ICOS interaction and its role in providing a costimulatory signal for T cell activation suggested a possible alternative costimulatory signal involved in the CD28-independent mechanism of T cell-mediated resistance to T. gondii (6, 7). To determine whether ICOS was involved in the immune response to T. gondii, experiments were performed to compare the expression of ICOS in uninfected and infected mice. C57BL/6 mice were infected i.p. with 20 cysts of the ME49 strain of T. gondii and at 7 days after infection, splenocytes were harvested, then analyzed for expression of ICOS. In uninfected C57BL/6 animals, only a small percentage of T cells expressed ICOS and all of these T cells were CD44^high (Fig. 1,A). By day 7 after infection, there was small increase in the percentage of CD8⁺ T cells which expressed ICOS and a marked expansion in the percentage of CD4⁺ T cells expressing ICOS (Fig. 1,B). Although a slight variation in the magnitude of ICOS up-regulation on CD4⁺ and CD8⁺ T cells was observed in four independent experiments, the levels of ICOS expressed on T cells from infected animals was consistently and reproducibly higher than uninfected C57BL/6 mice. In contrast, at 6 wk postinfection, ICOS expression on splenic CD4⁺ T cells had decreased compared with splenic T cells from 7-day infected animals, albeit ICOS expression on splenic T cells from 6-wk infected animals remained higher than those of uninfected mice (Fig. 1 C).

Although C57BL/6 mice are able to control the acute phase of infection, they do develop toxoplasmic encephalitis (TE) in which large numbers of CD4⁺ and CD8⁺ T cells infiltrate into the brain, a process required to control parasite replication at this site (3, 4). Moreover, the majority of T cells present in the brain during TE display an effector/memory phenotype (i.e., CD25^low, CD44^high, and CD62L^low) (6). Whereas the brains of uninfected mice contain no lymphocytes, analysis of brain mononuclear cells from chronically infected wild-type (WT) mice revealed that >90% of the CD44^high, CD4⁺, and CD8⁺ T cells expressed ICOS (Fig. 1 D). Together, these data reveal that following infection, the activation of T cells in the spleen is accompanied by increased expression of ICOS during the acute phase of infection. As the infection is controlled in the periphery, the numbers of ICOS⁺ T cells in the spleen return to near-normal levels. However, T cells isolated from the brains of mice with TE continued to express high levels of ICOS. Together, these results suggest that the infection-induced expression of ICOS is restricted to parasite-specific effector T cells.

Since the ability of T cells to produce IFN-γ is essential for resistance to T. gondii (1), the role of ICOS and CD28 in the ability of splenocytes from naive and acutely infected mice to produce IFN-γ was compared. Stimulation of naive splenocytes with anti-CD3 resulted in the production of IFN-γ and, as previously reported (33), this response was almost completely blocked by CTLA-4Ig (p < 0.0004). In contrast, anti-B7RP-1 had no effect on the production of IFN-γ by these resting T cells (Fig. 2,A, p = 0.7). These results indicate that although CD28 is critical in activation and production of IFN-γ by resting T cells, ICOS plays a minimal role in this initial activation process. When splenocytes from acutely infected mice were stimulated with STAg, the production of IFN-γ by parasite-specific T cells was partially inhibited by the addition of CTLA-4Ig (Fig. 2,B, p < 0.003). Similarly, the addition of anti-B7RP-1 also had an inhibitory effect on the production of IFN-γ in these cultures (Fig. 2,B, p < 0.001). Moreover, the addition of CTLA-4Ig plus anti-B7RP-1 resulted in an additive effect and reduced the levels of IFN-γ production to background (Fig. 2,B, p < 0.0001). Similar results were observed when infected splenocytes from C57BL/6 mice were stimulated with anti-CD3 (data not shown). Further analysis of Ag-stimulated splenocytes from acutely infected animals revealed that few CD8⁺ T cells were IFN-γ⁺ (data not shown) but that >75% of the CD4⁺ T cells that expressed IFN-γ⁺ also expressed ICOS (Fig. 2 C).

When BMNCs from chronically infected mice were isolated and stimulated with STAg, there was increased production of IFN-γ compared with unstimulated cells. The addition of CTLA-4Ig or anti-B7RP-1 only resulted in a small, but significant, reduction in the levels of IFN-γ produced in response to Ag (Fig. 2,D, p < 0.005 and 0.003, respectively). However, when both costimulatory pathways were blocked, there was a 50% reduction in levels of IFN-γ produced (Fig. 2 D, p < 0.001). Together, these results indicate that following infection with T. gondii, costimulation through ICOS, in combination with CD28, is required for optimal production of IFN-γ. Thus, in WT mice there are distinct CD28- and ICOS-dependent components required for optimal production of IFN-γ.

The data presented in the previous section support the hypothesis that ICOS has a role in the regulation of IFN-γ responses during toxoplasmosis and that its effects may be independent of CD28. However, previous in vitro studies have reported that the ability of activated T cells to express ICOS was largely dependent on costimulation through CD28 (20, 34). Therefore, to help determine whether ICOS was involved in CD28-independent pathways of resistance to T. gondii, studies were performed to assess whether the infection-induced increase in expression of ICOS occurred in CD28^−/− mice. Analysis of ICOS expression on T cells revealed that a small percentage of CD4⁺ (3%) and CD8⁺ T cells (1%) express ICOS and expression was restricted to CD44^high T cells (Fig. 3,A). However, as described previously, infection of CD28^−/− mice results in normal numbers of activated CD8⁺ T cells, but there are decreased numbers of activated CD4⁺ T cells (6, 7). This defect, as well as the infection-induced increase in the number of splenocytes, makes it difficult to directly compare the percentage of activated CD4⁺ T cells that express ICOS in CD28^−/− and WT mice. Therefore, the data presented in Fig. 3,B show the total numbers of CD4⁺ and CD8⁺ T cells which express ICOS in the spleens of uninfected and infected mice. At day 7 postinfection, analysis of ICOS expression revealed that there was no significant difference (p = 0.26) in the total numbers of ICOS⁺CD8⁺ T cells in WT and CD28^−/− mice, and although uninfected and infected CD28^−/− mice had fewer ICOS⁺CD4⁺ T cells than WT controls (uninfected, p < 0.02; infected, p < 0.01), there was a 2-fold increase in the numbers of ICOS⁺CD4⁺ T cells in the infected CD28^−/− mice (Fig. 3,B). In contrast to the spleen, analysis of CD4⁺ and CD8⁺ T cells isolated from the brains of chronically infected CD28^−/− mice revealed that the majority of these T cells were CD44^high and expressed high levels of ICOS (Fig. 3 C). These results demonstrate that although there are fewer ICOS⁺ T cells in the spleens of acutely infected CD28^−/− mice compared with WT mice, there is still a significant increase in the number of CD28^−/− T cells which have up-regulated ICOS during the acute phase of infection. Moreover, by the chronic phase of the infection, expression of ICOS by T cells from the brains of infected CD28^−/− mice was comparable to that observed in chronically infected WT mice. Thus, the ability of T cells to increase expression of ICOS in response to infection can occur independently of CD28.

To determine the possible role of ICOS as an alternative costimulatory molecule in the CD28-independent resistance to T. gondii (6, 7), splenocytes from infected CD28^−/− mice were stimulated with STAg and production of IFN-γ by T cells in the presence of CTLA-4Ig, anti-B7RP-1, or in combination were measured by ELISA. Splenocytes from CD28^−/− mice infected for 7 days and stimulated with STAg produced low levels of IFN-γ and the addition of CTLA-4Ig did not alter IFN-γ levels (7) (data not shown). However, the addition of anti-B7RP-1 resulted in a marked decrease (p < 0.002) in the production of IFN-γ (Fig. 4,A). To further explore the in vitro role of ICOS in regulating effector T cell function in the absence of CD28, BMNCs from the brains of chronically infected CD28^−/− animals were stimulated with toxoplasma Ag, and IFN-γ production in the presence or absence of anti-B7RP-1 was measured. BMNCs from chronically infected CD28^−/− mice stimulated with STAg produced significantly higher levels of IFN-γ compared with unstimulated cells (p < 0.001), and anti-B7RP-1 significantly inhibited (p < 0.001) the response to STAg (Fig. 4 B). These results strengthen the hypothesis that the B7RP-1/ICOS pathway is required for the CD28-independent mechanisms involved in the production of IFN-γ by parasite-specific T cells during the acute and chronic phases of infection.

To determine the in vivo role of ICOS, WT mice were infected with T. gondii and treated with anti-B7RP-1 at days 0 and 5 after infection. Initial studies revealed no significant difference in mortality, serum levels of IFN-γ, or parasite burden in treated WT mice compared with control treated mice (data not shown). However, CD28^−/− mice treated with anti-B7RP-1 on days 0, 3, and 5 of infection showed increased susceptibility to infection compared with infected CD28^−/− mice that received the sham treatment (Fig. 5,A). The ability of anti-B7RP-1 treatment to increase susceptibility of CD28^−/− mice to infection correlated with a significant reduction in serum levels of IFN-γ (Fig. 5,B, p < 0.03) and an increase in the parasite burden compared with sham-treated CD28^−/− mice (Fig. 5,C, p < 0.005). Analysis of splenocytes from infected CD28^−/− mice treated with anti-B7RP-1 revealed that they produced reduced levels of IFN-γ when stimulated with STAg compared with control CD28^−/− mice. Although these results failed to reach statistical significance (p = 0.17) a similar trend (7.9 vs 4.1 ng; n = 3 mice/group) was also observed in a repeat experiment. However, for both experimental groups, the addition of anti-B7RP-1 in vitro resulted in a significant reduction in the levels of IFN-γ (Fig. 5 D, p < 0.03 for both experimental groups). Taken together, these results demonstrate that in the absence of CD28, ICOS has an important role in the production of IFN-γ and resistance to T. gondii.

Studies from this laboratory have shown that although CD28 was required for optimal CD4⁺ T cell responses during toxoplasmosis, there was a significant CD28-independent component which was sufficient for resistance to this infection (6, 7). The discovery of ICOS and its ability to provide costimulation to T cells suggested that this may be an alternative pathway to mediate CD28-indpendent activation of T cells required for resistance to T. gondii. Indeed, studies presented here indicate that the B7RP-1-ICOS interaction largely accounts for the CD28-independent component of T cell production of IFN-γ during the acute phase of toxoplasmosis. In addition, although blockade of the B7RP-1-ICOS interaction did not affect the overall susceptibility of WT mice to T. gondii, CD28^−/− mice treated with anti-B7RP-1 became susceptible to infection. In agreement with these findings, previous studies have shown that ICOS is involved in the regulation of CD28-dependent and CD28-independent Th1 and Th2 responses to the nematode N. brasiliensis and the viral pathogens lymphocytic choriomeningitis virus and vesicular stomatitis virus (26). However, the importance of ICOS in the outcome of these infections was unclear and blockade of ICOS in CD28^−/− mice did not alter the time of expulsion of N. brasiliensis. Similarly, ICOS^−/− mice are slightly more resistant to infection with L. mexicana than WT mice, although the Th1 and Th2 responses specific for L. mexicana in these mice are compromised (27). Thus, the data presented here provide the first demonstration that ICOS has a critical role in resistance to a pathogen.

Following the identification of ICOS, a consensus quickly emerged that the B7RP-1-ICOS interaction was more important in Th2- than Th1-type responses. Thus, blockade of the B7RP-1-ICOS interaction inhibited Th2- but not Th1-mediated lung inflammation (21), and several studies have shown that Th2 cells express higher levels of ICOS than do Th1 cells (20, 26). Moreover, ICOS-deficient mice have severe defects in Th2 responses, CD40-mediated Ab class switching, and germinal center formation but produced normal levels of IFN-γ (22, 23, 24). However, this paradigm has been challenged by studies which demonstrated that blockade of the B7RP-1-ICOS interaction ameliorated experimental allergic encephalomyelitis and graft rejection, which are both mediated by Th1 cells (25, 35). The studies presented here extend those findings and identify an important role for ICOS in Th1-type responses required for resistance to an intracellular pathogen.

Analysis of ICOS expression following infection with T. gondii revealed distinct patterns associated with different phases of infection. During the acute phase of infection, high levels of ICOS were expressed on activated CD4⁺ T cells from the spleen, consistent with the model that CD4⁺ T cells are the major source of T cell-derived IFN-γ during acute infection (36). In contrast, during toxoplasmic encephalitis, when the brain is the site most affected, expression of ICOS by splenic T cells had decreased, while CD4⁺ and CD8⁺ T cells isolated from the brains of infected mice expressed high levels of ICOS. Since ICOS can provide costimulation for both CD4⁺ (15, 23, 25, 35, 37) and CD8⁺ (38) T cells to make IFN-γ, this finding is consistent with their role in producing IFN-γ required for resistance to toxoplasmic encephalitis (3, 39). However, in the studies with T cells from the brains of mice with TE, blockade of ICOS alone had a small but reproducible inhibitory effect on their ability to produce IFN-γ. When the blockade was combined with CTLA4-Ig, there was an additive effect, although significant levels of IFN-γ were still produced. Nonetheless, the simultaneous blockade of CD28 and ICOS led to a 50% reduction in levels of IFN-γ produced by these T cells and suggest a role for these costimulatory pathways in the pathogenesis of toxoplasmic encephalitis. Similarly, although the administration of anti-B7RP-1 to infected CD28^−/− mice resulted in reduced serum levels of IFN-γ and reduced production of IFN-γ in recall responses, there were still significant levels of IFN-γ and some of these mice did survive the acute phase of infection. There are several possible explanations for these findings and perhaps the blockade of the B7RP-1-ICOS interaction in these studies was incomplete and studies with ICOS^−/− mice would be needed to address this issue. Alternatively, the studies presented here have focused on the role of ICOS in the regulation of T cell responses but have not addressed whether ICOS has a role in the regulation of NK cell production of IFN-γ which provides an innate mechanism of resistance to T. gondii (40, 41). Our previous studies have shown that infection of severe combined immunodeficient mice with T. gondii stimulates NK cells to express CD28 and that the CD28-B7 interaction is involved in the regulation of NK cell production of IFN-γ (42). Related to the studies presented here, although the NK cell responses to T. gondii are blunted in CD28^−/− mice (7), these NK cells may still contribute to the low systemic levels of IFN-γ seen in infected CD28^−/− mice treated with anti-B7RP-1. Alternatively, it is also possible that the administration of anti-B7RP-1 may inhibit NK cell responses since preliminary studies indicate that although resting NK cell do not express ICOS, IL-2-activated NK cells can express low levels of ICOS (our unpublished observations). Whether this has any functional significance for NK cell responses during toxoplasmosis is unclear and is the subject of ongoing studies.

Initial in vitro studies demonstrated that the stimulation of T cells with anti-CD3 resulted in ICOS expression which was largely dependent on the B7-CD28 interaction (20) and suggested that some of the effects attributed to CD28 could be due to signaling through ICOS. Since CD28 is expressed constitutively, whereas ICOS expression is inducible, a model in which stimulation through CD28 was required for the activation of T cells and increased expression of ICOS was proposed. However, when human CD4⁺ T cells are stimulated with suboptimal levels of anti-CD3, ICOS can provide costimulation (34) and has also been shown to have a role in the costimulation of CD28^−/− T cells (26). The in vivo data presented here show that in the absence of CD28 there is a marked reduction in the total number of CD4⁺ T cells that express ICOS. Although these data confirm the important role of CD28 for optimal CD4⁺ T cell responses during toxoplasmosis, we interpret the reduced levels of ICOS as a result of reduced numbers of activated CD4⁺ T cells in these mice. However, the high levels of ICOS observed on CD4⁺ and CD8⁺ T cells isolated from brains of CD28^−/− mice with TE demonstrate that stimulation through CD28 is not required for expression of ICOS during toxoplasmosis.

The close structural and functional relationship between ICOS and CD28 raises the question of whether these molecules are functionally redundant or whether they have distinct roles in the regulation of T cell responses required for resistance to infection. For example, in mice chronically infected with T. gondii, the CD28-mediated production of IL-2 is important for the ability of CD8⁺ T cells to make IFN-γ in recall responses and more recently studies have indicated an important role for IL-2 during ICOS-mediated costimulation (34). Thus, although these may be parallel pathways, both may be required for optimal T cell responses in this particular experimental system. Nevertheless, during toxoplasmosis, WT mice treated with anti-B7RP-1 are resistant to T. gondii, but in the absence of CD28, ICOS is critical for resistance. These findings correlate with the presence of CD28-dependent and ICOS-dependent pathways for the production of IFN-γ during toxoplasmosis and suggest that ICOS and CD28 are parallel costimulatory pathways, either of which is sufficient to mediate the production of IFN-γ and resistance to T. gondii. Several questions about the role of ICOS in the regulation of immunity to T. gondii remain unresolved. Our studies have not addressed whether the B7RP-1-ICOS interaction is restricted to the regulation of IFN-γ production or whether these molecules may also regulate other T cell functions necessary for resistance to T. gondii. Thus, further studies are required to delineate the functional relationship between CD28 and ICOS and how each pathway integrates to provide an optimal T cell response necessary for resistance to T. gondii.

We thank Claudia Baikalov for expression of the B7RP1-Ig and Phil Scott and Jay Farrell for discussions during the course of these studies.