IL-12 Signaling Drives CD8⁺ T Cell IFN-γ Production and Differentiation of KLRG1⁺ Effector Subpopulations during Toxoplasma gondii Infection¹

Like many obligate intracellular microbial pathogens, Toxoplasma gondii infection induces a strong type 1 polarized immune response, engaging both innate and adaptive immune systems. It is clear that the major mechanism of host resistance to toxoplasmosis is mediated by the effector cytokine IFN-γ (1, 2, 3), which activates microbicidal defense mechanisms in both nonhematopoietic cells and hematopoietic cells, especially macrophages (4). The major sources of IFN-γ are CD4⁺ Th1 lymphocytes, CD8⁺ CTLs, and NK cells (5). Although it is likely that lymphocytes work together for optimal immune activation (3, 6, 7), seminal reports revealed that Th1 cells are dispensable for resistance to acute toxoplasmosis (3) and NK cells contribute minimally to protective immunity (8). Furthermore, the transfer of memory CTLs from vaccinated mice into Ag-inexperienced, naive mice confers resistance to parasite challenge (9, 10). Therefore despite the heterogeneous pool of effector cells responding to infection, CD8⁺ T cells play an essential role in vaccine-induced protective immunity against T. gondii.

The generation of CTL immune responses is likely to be influenced by the cytokine milieu produced by innate immune cells responding to microbial products. During T cell activation, these innate cytokines (also known as signal 3 cytokines) synergize with signals from the TCR and costimulatory receptors to induce functional maturation (11). Reductionist experiments have shown that T cells activated in vitro with Ag and costimulation alone undergo proliferation but ultimately become tolerant; functional maturation is only acquired with the addition of a signal 3 cytokine (12). The two main candidate signal 3 cytokines produced in response to intracellular pathogens are type 1 IFNs and IL-12 (11). However, the rules that dictate which signal 3 cytokine drives CD8⁺ T cell responses to microbial infection remain uncertain. In the case of lymphocytic choriomeningitis virus (LCMV)³ infection in mice, a deficiency in type 1 IFN signaling resulted in a blunted primary CD8⁺ T cell response and a constricted pool of memory cells (13, 14). Because microbial products typically induce the production of multiple signal 3 cytokines, it has been argued that type 1 IFNs and IL-12 may both contribute to CTL effector responses in either a compensatory or a completely redundant fashion (11, 15). As an example of the latter, type 1 IFNs and IL-12 were both found to be dispensable for the activation of CD8⁺ T cells during murine listeriosis (16), suggesting that additional inflammatory cytokines are induced by Listeria that exert signal 3 functions. Thus, the spectrum of innate mediators induced by different microbial agents might govern the requirements for individual signal 3 cytokines for the generation of pathogen-specific CTL responses (17). Because the production of type 1 IFNs in response to Toxoplasma infection has not been reported and a hallmark of toxoplasmosis is the production of IL-12 by dendritic cells (18) and macrophages (19), IL-12 emerges as the prime candidate signal 3 cytokine in this infection. From previous studies using in vivo IL-12 neutralizing Abs (19, 20) and mice genetically deficient in the p40 subunit of IL-12 (21, 22, 23), it remains unclear to what extent an IL-12 signaling deficiency impacts CTL effector differentiation during T. gondii infection. Therefore, in this study we comprehensively address the role of IL-12 in CD8⁺ T cell activation during T. gondii infection in vivo.

CD8⁺ T cells have two canonical effector functions: effector cytokine production and target cell cytolysis. Although the role of the CTL IFN-γ response for protective immunity during toxoplasmosis is well established, IFN-γ expression requires restimulation with Ag and may also be heavily influenced by noncognate cytokine signals, including those of IL-12. The current lack of identified T. gondii CTL epitopes precludes the tracking of Ag-specific effector populations using MHC-peptide tetramers. Moreover, in contrast to viruses and bacteria, it is unlikely that the totality of parasite-reactive CTLs can be defined by only a few “immunodominant” clonotypes. We therefore initially used the presence of intracellular granzyme B (GrB) as a marker to identify putative parasite-reactive CTLs, given that cytotoxic granules are prestored within armed effector cells (24). Subsequent correlative mapping of effector molecule (GrB and IFN-γ) expression with cell surface activation markers (CD62L and killer cell lectin-like receptor G1 (KLRG1)) was conducted on putative parasite-reactive CTLs harvested from a peripheral effector site (peritoneum) of Toxoplasma-infected mice. Using this multiparametric phenotyping approach, we were able to define four distinct CTL subpopulations exhibiting varying effector profiles and differential IL-12 dependencies. Our observation that effector site CTLs present such a highly heterogeneous activation profile supports current models of effector differentiation (25, 26) whereby CD8⁺ T cells initially develop along divergent pathways for the generation of acute effector or memory precursor cells.

Wild-type (WT) C57BL/6NCrl mice were purchased from Charles River Laboratories. IL-12p35^−/− (B6.129S1-Il12a^tm1Jm/J), IL-12Rβ2^−/− (B6.129S1-Il12rb2^tm1Jm/J), TCRβ^−/− (B6.129P2-tcrb^tm1Mom/J), and WT C57BL/6J mice were purchased from The Jackson Laboratory. Tyk2^−/−, STAT4^−/−, and T-bet^−/− mice (B6 background) were bred and housed under specific pathogen-free conditions at Brown University (Providence, RI). Sex- and age-matched mice were used in all experiments. All mice were handled according to Brown University Institutional Animal Care and Use Committee guidelines.

T. gondii tachyzoites were passed in and harvested from monolayers of human foreskin fibroblasts. Vaccinations consisted of a single i.p. dose of 1 × 10⁶ live, uracil auxotrophic tachyzoites. These parasites (cps1-1 or CPS) were genetically rendered avirulent via disruption of the carbamoyl phosphate synthetase (CPS) II gene from the parental type I RH strain (27). Growth medium for CPS cultures consisted of DMEM (Invitrogen) supplemented with 1% FBS, 1% penicillin/streptomycin, 1% l-glutamine, and 300 μM uracil. Before injection, CPS parasites were irradiated (15 kilorad) using a ¹³⁷Cs source to ensure that in vivo reversion to virulence will not occur.

Mice were sacrificed on the specified day postvaccination. To harvest peritoneal exudate cells (PECs), peritonea were lavaged with RPMI 1640 (Invitrogen) plus 5% FBS. Spleens were also harvested for select experiments. Single cell suspensions of tissues were washed and erythrocytes were lysed using a Tris-buffered NH₄Cl solution. Live cell numbers were counted on a hemocytometer using trypan blue exclusion. For in vitro culture, 1–2 × 10⁶ cells were plated on 48-well plates with 0.5 ml RPMI 1640 plus 10% FBS. Restimulated cultures were inoculated with live CPS tachyzoites (multiplicity of infection of 0.1) and incubated at 37°C for ∼12 h.

Both ex vivo cells and in vitro cultured (restimulated) cells were stained for flow cytometric analysis. Each sample was washed in FACS buffer (PBS plus 1% BSA) and then surface stained for 15 min on ice. Mouse-specific, surface-staining Abs included anti-CD62L-FITC, anti-TCRβ-PE, anti-CD44-PE, anti-CD70-PE, anti-CD8α-PerCP, and anti-TCRβ-allophycocyanin from BD Pharmingen and anti-CD27-FITC, anti-CD62L-PE, anti-CD127(IL-7Rα)-PE, anti-CD8α-PE-Cy7, and anti-KLRG1-allophycocyanin from eBioscience. Appropriate Ab isotypes were used when needed. Samples were then fixed in CytoFix/CytoPerm solution (BD Bioscience) and washed with Perm/Wash (BD Bioscience) according to the manufacturer’s instructions. To stain intracellular cytokines, samples were incubated in Perm/Wash plus Abs for 30 min on ice. Mouse-specific intracellular staining (ICS) Abs included anti-IFN-γ-FITC, anti-IFN-γ-PE, and anti-IFN-γ-allophycocyanin from BD Pharmingen. Mouse anti-human GrB-PE (Caltag Laboratories) is cross-reactive with murine GrB (28) and was also used for ICS. Appropriate Ab isotypes were used when needed. For T-bet ICS, fixed cells were incubated with a blocking buffer (Perm/Wash plus 2% normal goat serum plus mouse BD Fc Block) on ice for 20 min. Then, mouse-specific anti-T-bet Ab was added directly to the blocking buffer and stained for 30 min on ice. Stained samples were acquired on a BD FACSCalibur flow cytometer and analyzed using BD CellQuest Pro or FlowJo (Tree Star) software.

Phospho-STAT4 ICS was performed using published methods for intracellular staining of phosphorylated proteins (29). Briefly, cells were surface stained with anti-CD62L-FITC and anti-KLRG1-allophycocyanin then fixed with 1.6% formaldehyde. Cells were then treated with cold 100% methanol to allow for greater access to intracellular proteins. The final staining step included anti-pSTAT4(pY693)-PE (BD Biosciences) plus anti-CD8α-PE-Cy7 in Perm/Wash solution.

For in vitro neutralization of IL-12 during restimulation, anti-IL-12p40 (clone C17.8; BioExpress) was added to cultures at a final concentration of 20 μg/ml.

PECs were harvested from day 7 primed mice and spleens from unprimed mice were harvested for naive controls. The PECs were washed and surface stained with KLRG1-biotin, CD62L-PE, and CD8α-PE-Cy5. After the primary stain, the cells were washed and secondarily stained with SAv-FITC. Labeled cells were then sorted for FI, FII, FIII, and FIV CD8⁺ subpopulations (where “F” stands for “fraction”) using the BD FACSVantage SE cell sorter. Meanwhile, CD8α⁺ cells were column purified from the naive splenocytes through positive selection using magnetic microbeads (Miltenyi Biotec). Total RNA from the five groups of CD8s (Naive, FI, FII, FIII, and FIV) was subsequently isolated using the Qiagen RNeasy Micro kit. The RNA concentrations were calculated from A₂₆₀ values obtained by UV spectrophotometry. Input RNA from each group was normalized to 500 ng and reverse transcribed into cDNA using oligo(dT) primers (Qiagen Omniscript RT kit) and stored at −20°C.

The following cDNA oligonucleotide probes (Invitrogen) were used for PCR: KLRG1 no. 1, 5′-ACCTCCAGCCATCAATGTTC-3′; KLRG1 no. 2, 5′-CCTCTGGACGAGGAATGGTA-3′; LKLF no. 1, 5′-GTGGCACTGAAAGGGTCTGT-3′; LKLF no. 2, 5′-ACCAAGAGCTCGCACCTAAA-3′; IL-12Rβ1 no. 1, 5′-GAGGAGGCGGCTCTCCTCAG-3′; IL-12Rβ1 no. 2, 5′-ACATTCCTCCTGCTCCAGGG-3′; IL-12Rβ2 no. 1, 5′-CTGCACCCACTCACATTAAC-3′ IL-12Rβ2 no. 2, 5′-CAGTTGGCTTTGCCCTGTGG-3′ (LKLF is lung Krüppel-like factor). Input cDNA was diluted as described in Fig. 6B. Standard PCR were conducted and then resolved using agarose gel electrophoresis.

Splenocytes were harvested from naive WT and IL-12Rβ2^−/− mice. CD4⁺ and CD8α⁺ cells of both genotypes were column purified through positive selection using magnetic microbeads (Miltenyi Biotec). Each group was analyzed for nonselected T cells of the opposite lineage and was found to have <2% T cell contamination. Purified T cells were combined into four donor groups: 1) WT CD4 plus WT CD8 cells; 2) IL-12Rβ2^−/− CD4 plus WT CD8 cells; 3) WT CD4s plus IL-12Rβ2^−/− CD8 cells; and 4) IL-12Rβ2^−/− CD4s plus IL-12Rβ2^−/− CD8 cells. Donor T cells (2 × 10⁶ CD4 and 1 × 10⁶ CD8 cells in 0.2 ml) were adoptively transferred i.v. (retro-orbital) into recipient TCRβ^−/− mice at 3 mice/group. Immediately after transfer, each mouse was i.p. vaccinated with 1 × 10⁶ irradiated CPS. Primed, chimeric mice were sacrificed on day 7 postvaccination and assessed for CTL effector function and differentiation as described.

Effector CD8⁺ T cells were generated in vivo by i.p. vaccination of C57BL/6 mice with live, uracil auxotrophic T. gondii parasites termed CPS that were further rendered nonreplicative by irradiation. CPS vaccination has been previously shown to induce protective immunity in BALB/c (27) and C57BL/6 mice (30). Furthermore, this strategy is advantageous for assessing the role of innate cytokines during CD8⁺ T cell activation because there is no risk of host mortality, especially under cytokine-deficient conditions. Using this approach, the presence of effector CD8⁺ T cells was probed at the site of infection among PECs using both cell surface activation markers and intracellular effector molecule expression. Despite being CD44^high in the peritoneum, CD8⁺ T cells (96–98% TCRβ⁺ from naive mice) did not express GrB and did not produce IFN-γ upon in vitro restimulation (Fig. 1,A). On day 7 postvaccination, most PEC CD8⁺ T cells (98–99% TCRβ⁺) became CD44^high and 28% up-regulated GrB ex vivo. IFN-γ production was observed after 12 h of in vitro restimulation, with 11% of the CTLs expressing this effector molecule (Fig. 1 B). Generally, we have observed 10–25% IFN-γ positivity among PEC CD8⁺ T cells, with two-thirds of the IFN-γ⁺ cells also expressing GrB (data not shown).

We sought to identify putative parasite-responsive GrB⁺ CTLs using cell surface activation markers. A panel of such markers was tested for the ability to further enrich GrB⁺ effector cells when used in combination with CD62L down-regulation, a conventional T cell activation event. From this analysis KLRG1, a marker previously associated with CTL terminal differentiation (31, 32), emerged as the best marker (data not shown). Interestingly, its coexpression profile with CD62L revealed four subpopulations of CD8⁺ T cells that we have designated as the following four fractions (F): FI is CD62L^high and KLRG1⁻, FII is CD62L^low and KLRG1⁻, FIII is CD62L^low and KLRG1⁺, and FIV is CD62L^high and KLRG1⁺ (Fig. 1 C). FI cells exhibited the lowest frequencies of effector molecule expression, consistent with their naive-like cell surface phenotype. The subpopulations FII, FIII, and FIV were universally CD44^high and contained significantly higher frequencies of GrB⁺ as well as Ag-responsive IFN-γ⁺ cells, suggesting that they represent putative effector cells. Of these three subsets, FIII was the most enriched for effector cells (72% GrB⁺ and 38% IFN-γ⁺), whereas FII was the least enriched for effector cells (38% GrB⁺ and 23% IFN-γ⁺). FIV, however, displayed an interesting, and somewhat contradictory phenotype, exhibiting an intermediate frequency of effector molecule expression (53% GrB⁺ and 22% IFN-γ⁺) but retaining a CD62L^high status. Thus, among effector site CTLs, three effector subpopulations can be distinguished. Based on their cell surface activation and effector molecule profile (CD62L^low, KLRG1⁺, GrB-rich, IFN-γ -rich), FIII may represent the most highly differentiated effector CTL subpopulation.

Next, the kinetics of CTL expansion and contraction in the peritoneal cavity was analyzed over 4 wk following CPS vaccination. The expansion phase for total CD8⁺ lymphocytes was observed between days 3 and 10 (Fig. 2,A). The climax in numbers of effector cells on day 10 coincided nicely with the peak frequency of GrB positivity in total CD8⁺ T cells (Fig. 2,B). Interestingly, the peak frequency of IFN-γ production in total CD8⁺ T cells was prolonged until day 14 (Fig. 1,C), during which GrB expression and total cell numbers had already started to decrease. CTL contraction was observed from day 10 to day 21 (Fig. 2 A), and their numbers remained low thereafter.

We further analyzed the kinetics of the CTL response with respect to the four phenotypic subpopulations described above. From this analysis, FIII CTLs emerged as the dominant subpopulation, reaching 5-fold greater absolute cell numbers relative to FII and FIV at the peak of the CTL response on day 10 (Fig. 2,D). The massive accumulation and subsequent rapid contraction of the FIII compartment suggests that cells in FIII may be end-stage effector cells, consistent with their terminally differentiated phenotype (see Fig. 1,C). In contrast to FIII, the relative frequencies and absolute numbers of CTLs in both FII and FIV remained steadily low throughout the entire expansion and contraction phases (Fig. 2 D and E), implying that either or both of these fractions may represent transitionary precursor(s) for FIII.

KLRG1 expression is thought to mark terminal effector CD8⁺ T cells. To test whether the KLRG1⁺ fractions FIII and FIV display other markers of terminal differentiation, we probed each subpopulation for down-regulation of the costimulatory receptor CD27. Stimulation of CD27 on T cells by CD70⁺ APCs induces rapid T cell activation (33), ultimately leading to activation-induced cell death (34). Conversely, it has been shown that a minority of Ag-specific CD8⁺ T cells expressing IL-7R during the peak of the primary CTL response persist long term and subsequently progress into memory cells (35). Considering this, CD27 down-regulation and IL-7Rα expression were used to determine which fraction likely contains acute effector cells or memory precursor effector cells, respectively. In the peritoneum on day 7 postvaccination, CD27 was down-regulated en masse by only the KLRG1⁺ CTL subpopulations, FIII and FIV (Fig. 3,A). During the contraction phase on day 15 there was no change in the frequency of CD27⁻ CTLs among FI and FII; however, FIII remained enriched for these cells while FIV showed reduced but still elevated frequencies of CD27⁻ cells. When the same PEC CTL fractions were queried for the presumptive memory cell marker, a contrasting profile was observed. The KLRG1⁻ fractions FI and FII had the greatest concentration of IL-7Rα⁺ cells on day 7, whereas FIII and FIV had the lowest frequencies of IL-7Rα⁺ cells (Fig. 3 B). This disparity was even more pronounced during the time of CTL contraction, where IL-7R expression was retained in FI and FII and nearly absent in FIII and FIV. Taken together with the dichotomy in KLRG1 positivity between FII vs FIII and FIV, the patterns of CD27 and IL-7R expression suggest that FII represents memory precursor-like effector cells while FIV and FIII could be acutely generated, highly activated effector cells.

Although both FIII and FIV exhibit characteristics of terminal differentiation, FIV is consistently the subdominant population both numerically and in terms of effector potential. Therefore, FIII cells are the most likely the terminal effector population while FIV may represent a less mature effector or effector-precursor subpopulation. Consistent with this notion, FIII cells in the spleen, similar to their peritoneal counterparts, were uniformly IL-7R negative (Fig. 3 C). In contrast, splenic FIV cells expressed similar levels of IL-7R expression as those of the memory precursor-like FII.

IL-12 is a proinflammatory cytokine known to regulate type 1 effector responses of Th cells in numerous intracellular infection models. Bioactive IL-12 (p70) exists as a heterodimer of the p35 and p40 subunits that binds with high affinity to the cell surface IL-12R complex, another heterodimer consisting of the IL-12Rβ1 and IL-12Rβ2 subunits (36). Upon ligation of IL-12p70 with the IL-12R, the intracellular IL-12R-bound Janus kinases Tyk2 and Jak2 become activated and subsequently phosphorylate specific tyrosine residues on the IL-12R (36). These residues serve to create a docking site for STAT molecules, especially STAT4, the major transcription factor activated downstream of IL-12R signaling (37, 38). An earlier study had reported that IL-12p40 knockout mice exhibited low CTL precursor frequencies when immunized with a temperature-sensitive mutant of T. gondii, suggesting a regulatory role for IL-12 in CTL type 1 effector development (22). We therefore tested the requirements of several key constituents of the IL-12 signaling cascade for CD8⁺ T cell subpopulation activation and effector differentiation during murine toxoplasmosis. Knockout mice deficient for IL-12p35, Tyk2, or STAT4 were vaccinated with CPS parasites and assessed for CD8⁺ T cell IFN-γ production and subpopulation development. As previously shown, vaccinated WT mice generate high frequencies of IFN-γ-producing CTLs on day 7. Priming of Tyk2^−/− mice, whose IL-12 signaling is only partially impaired (39, 40), resulted in a reduced frequency of IFN-γ⁺CD8⁺ T cells (Fig. 4,A). Interestingly, the deficit of IFN-γ-producing CTLs in Tyk2^−/− mice was mostly contained within the end-stage effector subpopulation FIII (Fig. 4,B), suggesting that end-stage effector development requires full signal strength from the IL-12 receptor. Unlike Tyk2^−/− mice, STAT4^−/− and IL-12p35^−/− mice are completely deficient in IL-12 signaling (41, 42) and, under these conditions, the generation of IFN-γ producing CTLs was completely abrogated (Fig. 4, A and B).

When the subpopulation frequency distributions were compared between WT and IL-12 signaling-deficient mice, a pattern emerged. Among the effector subsets, reduced or absent IL-12 signaling during vaccination similarly resulted in an increase in frequency of FII and decreases in the frequencies of FIII and FIV on day 7 (Fig. 4,C). Of the CD8⁺ T cells that were generated under IL-12 signaling-deficient conditions, effector subpopulations were still activated as seen by normal or increased frequencies of GrB expression (Fig. 4 D). Thus, whereas IL-12 globally regulates IFN-γ expression in CTL subpopulations during Toxoplasma infection, it appears to be selectively required for the generation of the KLRG1⁺ subsets FIII and FIV.

To test whether the lack of FIII and FIV differentiation persists throughout the course of the CD8⁺ T cell response in IL-12p35^−/− mice, subpopulation frequency distributions were analyzed on days 5, 7, and 16 postvaccination. On day 5, an early time point in effector T cell development, subpopulation frequencies were comparable between WT and IL-12p35^−/− mice, except for a 2-fold increase in the frequency of FII cells in the latter (Fig. 5). Two days later, the effect of IL-12 deficiency was evident among IL-12p35^−/− PEC CTLs, where FIII and FIV failed to develop and FII became the dominant effector subpopulation. On day 16, FIII and FIV remained low in IL-12p35^−/− mice. The strong developmental blockade of CTLs in the IL-12p35^−/− mice at the FII stage indicates that IL-12 acts to drive the differentiation of the KLRG1⁺ effector subpopulations.

Because the effector CD8⁺ subpopulations have differential developmental requirements for IL-12, we explored whether they also differ in their inherent responsiveness to IL-12 stimulation. To assay IL-12 sensitivity, we measured the frequency of STAT4 phosphorylation in each CD8⁺ fraction following a short pulse of rIL-12 in vitro. STAT4 phosphorylation was subsequently determined by intracellular staining and quantified among CTL subpopulations by flow cytometry. After in vitro stimulation with rIL-12, STAT4 phosphorylation was observed in 22% of day 8 PEC CTLs, whereas naive CD8⁺ T cells did not activate STAT4 under the same conditions (Fig. 6,A). Interestingly, the effector CTL subpopulations (FII, FIII, and FIV) were highly enriched for IL-12-responsive cells, with the KLRG1⁺ fractions containing somewhat higher frequencies of pSTAT4⁺ cells. IL-12 responsiveness appears to be acquired progressively as effector cells differentiate, which is reflected in the increased expression of transcripts for both IL-12R chains in FII, FIII, and FIV (Fig. 6,B), as well the uniform sensitivity of their IFN-γ responses to IL-12p40 neutralization in vitro (Fig. 6 C). Given that the emergence of FI and FII cells are clearly IL-12 independent, the presence of IL-12-sensitive cells in these fractions suggests that IL-12 responsiveness is acquired before the IL-12-dependent signaling events leading to the differentiation of the KLRG1⁺ subpopulations.

IL-12 could be acting directly on the CD8⁺ T cells themselves to drive effector differentiation, or indirect mechanisms could be involved. For instance, IL-12 could be required for CD4⁺ lymphocyte help to CTLs through the up-regulation of CD40L costimulation (43). To determine whether effector CTL differentiation is due to CD8⁺ T cell-intrinsic IL-12 signaling and whether IL-12-regulated CD4 help is involved, chimeric mice lacking IL-12Rβ2 in either CD4⁺ or CD8⁺ T cells or both lineages were constructed. Genetic deletion of the β2 chain of the IL-12R specifically abolishes IL-12 signaling, including downstream STAT4 phosphorylation, despite the sustained ability of IL-12 to bind IL-12Rβ1 (44). On day 0 donor T cells (2:1; CD4:CD8) were adoptively transferred into TCRβ^−/− recipient mice (which lack T lymphocytes, but all other cells are IL-12 signaling proficient) and then vaccinated. Seven days later, PECs were harvested and analyzed for CTL subpopulation differentiation and IFN-γ production. When chimeric mice with both WT CD4⁺ and CD8⁺ T cells were vaccinated, differentiation of KLRG1⁺ effector subpopulations was observed (Fig. 7,A), as previously noted in vaccinated, nonchimeric mice. Nevertheless, the FI fraction was underrepresented, potentially due to the effects of homeostatic proliferation of transferred T cells in these lymphopenic recipients. Effector subpopulation distribution appeared normal in mice lacking IL-12Rβ2 only on CD4⁺ T cells. However, when either CD8⁺ T cells alone or when both CD8⁺ and CD4⁺ T cells could not signal through the IL-12R, altered subpopulation differentiation was observed (Fig. 7,A). The defect in these mice was similar to that observed in IL-12p35^−/− mice, such that the frequency of FII increased while the frequencies of FIII and FIV decreased compared with WT CD8⁺ subpopulations. Similar to the effects of IL-12Rβ2 deficiency on subset distribution, deficits in CD8⁺ T cell IFN-γ production were observed only in the groups of chimeric mice that received IL-12Rβ2^−/− CD8⁺ T cells (Fig. 7 B). The results from the chimeric mouse experiments indicate that the acquisition of effector function in CD8⁺ T cells is not dependent on the effects of IL-12 signaling in the CD4⁺ T cell compartment. Rather, CD8⁺ T cell-intrinsic IL-12 signaling drives both CTL effector subpopulation differentiation and the induction of IFN-γ production.

Finally, we wanted to explore the involvement of T-bet, a “master regulator” of the type 1 T cell lineage (45), in IL-12-dependent KLRG1⁺ effector cell development. It was recently shown that T-bet controls IFN-γ effector competence in peripheral CTLs of mice infected with Toxoplasma cysts (46). Furthermore, in LCMV-specific CD8⁺ T cells, IL-12 appears to be an important factor for maximal expression of T-bet (47). Joshi et al. have demonstrated that quantitative induction of T-bet provides a linkage between the pathogen-driven inflammatory cytokines and KLRG1 up-regulation in LCMV-specific CD8⁺ T cells (48). Indeed, following CPS vaccination we find that T-bet-deficient mice exhibit impaired generation of the KLRG1⁺ CTL subpopulations (FIII and most notably FIV) similar to but less severe than that exhibited by the IL-12 signaling-deficient mice in both PECs (Fig. 8,A) and the spleen (Fig. 8,B). These data suggest that IL-12 controls KLRG1⁺ effector development through a T-bet-dependent mechanism. In culture, IL-12 stimulation alone was shown to induce high-level T-bet expression and the formation of KLRG1⁺ effectors among LCMV-specific CD8⁺ T cells (48). To test whether IL-12 controlled T-bet expression in CD8⁺ T cells induced by T. gondii, we stained for T-bet in PEC effector cells derived from day 7 CPS-primed WT and IL-12Rβ2^−/− mice. As shown in Fig. 8,C, a small but discernible population (∼10%) of T-bet⁺CD8⁺ T cells can be detected directly ex vivo from primed WT mice, whereas T-bet positivity in IL-12Rβ2^−/− CD8⁺ T cells was only slightly above the background staining (using T-bet^−/− PECs). As expected, the three effector subpopulations (FII, FIII, and FIV) contained greater frequencies of T-bet⁺ cells compared with the naive-like FI fraction. However, the frequency of T-bet positivity was unexpectedly much lower than the corresponding incidence of GrB or IFN-γ positivity within each effector fraction (see Fig. 1,C). The observation that only a minority of effector cells express T-bet implies that T-bet expression maybe transient and highly regulated. To amplify T-bet induction, mice were given a booster injection of CPS, which would trigger in vivo IL-12 production, 1 day before harvesting PECs. CPS boosting increased T-bet positivity among all CD8⁺ fractions, but no discernible T-bet induction occurred in IL-12Rβ2^−/− CD8⁺ T cells (Fig. 8 D). In conclusion, in vivo IL-12 signaling was found to control T-bet induction in effector CD8⁺ T cells, which is an important factor for the generation of KLRG1⁺ subpopulations during T. gondii infection.

In this study, we have characterized the local anti-Toxoplasma CD8⁺ T cell response following vaccination with CPS-deficient T. gondii, an attenuated strain of the parasite that induces long-term protective immunity. Phenotypic profiling of cells harvested from the peritoneal site of vaccination revealed four distinct subpopulations of CD8⁺ T cells. Although these four subpopulations were also discernible in the spleen, their frequency was considerably lower, owing to dilution by the resident naive CD8⁺ T cell population. We therefore chose to study the peritoneal site, where a depot of parasite Ags would predictably enrich for putative effector CTLs. Remarkably, these GrB⁺ and IFN-γ-inducible cells were encompassed within the three subpopulations: FII, FIII, and FIV. Based on their kinetic behavior and on the differential expression of functionally important cell surface receptors, we propose that these three effector CTL subpopulations might represent distinct maturational states of CD8⁺ effector differentiation. FIII (CD62L^lowKLRG1⁺), we believe, is the dominant, end-stage effector subpopulation. This CD8⁺ fraction had the highest concentration of cells expressing the effector molecules (GrB and IFN-γ) and was the most abundant in frequency and absolute cell number during the peak of the CD8⁺ T cell response in the peritoneum. Furthermore, these cells are likely to be terminally differentiated as they express KLRG1, a marker associated with replicative senescence in T cells exposed to repetitive antigenic challenge or chronic viral infection (32, 49). In contrast to the terminal FIII population, we believe the effector-like FII population may contain precursors for long-term memory cells. Although they are already CD62L^low, this fraction has not yet acquired KLRG1 expression and is the least enriched for GrB⁺ cells. Of note, FII selectively retained the expression of IL-7R during the CD8⁺ T cell contraction phase in the periphery. Because the retention of IL-7R expression in a subset of effector cells generated during primary infection has been previously associated with long-term survival (35), their relative immaturity and the high frequency of IL-7R expression in FII suggest that some of these cells might progress to become effector memory T cells.

Unlike the more clear cut profiles associated with FIII and FII, FIV CD8⁺ T cells present a somewhat puzzling and contradictory set of features because this subset expresses markers associated with both precursor and terminal cells. It can be argued that by virtue of its CD62L^high status, FIV contains precursors to the central memory T cell lineage. We believe that this notion is unlikely for several reasons. First, these cells are KLRG1 positive, a feature shared with the terminal FIII subset. Furthermore, the FIV subpopulation in the peritoneum remains IL-7R⁻ and down-regulates CD27 expression, a feature indicative of high effector potential (50, 51, 52) and maintenance (53) of peripheral CD8⁺ T cells. Alternatively, FIV cells could arise from the FIII subpopulation. Currently, it is not yet settled whether re-expression of CD62L in activated T cells occurs in vivo. In an adoptive transfer system using physiologically appropriate numbers of TCR transgenic T cells, conversion of effector memory T cells (T_EM) to central memory T cells (T_CM) cells by re-expression of CD62L was not observed (54), whereas in a recent paper CD62L reacquisition by a polyclonal population of antiviral effector CTLs was reported (55). Given this uncertainty, we favor the scenario that most FIV CD8⁺ T cells probably give rise to the dominant, terminal FIII subpopulation. Nevertheless, the retention of IL-7R expression in a proportion of splenic FIV cells leaves open the possibility that, within secondary lymphoid tissues, memory precursors exist in this subpopulation.

Our analysis of the CD8⁺ T cell response to Toxoplasma immunization revealed a highly heterogeneous population of acute effector cells. Although we have yet to directly compare their protective functions, the three effector subpopulations seem to express comparable levels of effector molecules. The observed heterogeneity is consistent with current models of CD8⁺ T cell effector/memory differentiation postulating that naive CTL precursors generate a heterogeneous pool of primary effector cells with divergent potentials for terminal differentiation and memory cell formation (25, 26). Specifically, our observations are highly compatible with the “fate commitment with progressive differentiation” model for effector and memory CTL differentiation recently proposed by Kaech and Wherry (26). It is proposed that during the primary CTL response, exposure to high signal strength (signals 1 plus 2 plus 3) favors the differentiation of short-lived effector cells (SLECs) that are highly susceptible to activation-induced cell death. In a parallel pathway, low activation signal strength is thought to induce memory precursor effector cells (MPECs) that later progress to long-term memory cells. Additionally, MPECs are thought to retain the potential to continually regenerate the SLEC pool. A distinguishing feature of SLECs is the early acquisition of KLRG1, which appears to mark effector cells committed to down-regulating IL-7R expression (48). MPECs, in contrast, do not acquire KLRG1 and retain responsiveness to IL-7. Based on the expression of KLRG1, our FII subpopulation must therefore represent MPECs whereas the SLEC lineage comprises of both the FIII and FIV subpopulations. Our resolution of a minor CD62L^high fraction among KLRG1⁺ effector cells indicates that an alternative developmental route to generating FIII cells can be initiated through the acquisition of KLRG1 before CD62L down-regulation.

The balance between effector vs memory precursor cell development is thought to be influenced by the inflammatory cytokine microenvironment during the primary CTL response (11). Our data show that IL-12, the candidate signal 3 cytokine induced during T. gondii infection, controls not only the induction of IFN-γ production but also effector CTL subpopulation differentiation. Our work extends an earlier report that CD8⁺ T cell IFN-γ responses to Toxoplasma are reduced in IL-12p40-deficient mice (22). Specifically, we found that direct IL-12 signaling in CD8⁺ T cells is essential for the generation of the KLRG1⁺ effector subpopulations FIII and FIV, but not for the memory precursor-containing FII subset. A similar role for IL-12 was recently established in the murine model of listeriosis, where an IL-12 signaling deficiency resulted in the decreased generation of primary effector CD8⁺ T cells and, interestingly, an increased pool of memory cells (56). However, in contrast to the Toxoplasma system, IFN-γ production by Listeria-specific CD8⁺ T cells was predominantly IL-12 independent during infection (16, 56), dissociating the inductive effect of IL-12 on IFN-γ expression from its developmental role. Therefore, despite obvious differences in the nature of signals 1 and 2 influencing CTL activation in these two infections, IL-12 signaling acts as a critical determinant driving the formation of the terminal effectors in vivo.

An outline of the mechanism by which IL-12 promotes KLRG1⁺ effector cell development is just emerging. As previously discussed, in LCMV-specific CD8⁺ T cells an IL-12 deficiency resulted in diminished T-bet expression and a paucity of IL-7R⁻ effector cells (47). Indeed, T-bet overexpression alone appears to be sufficient for the repression of IL-7R expression in activated CTLs (48, 57). Furthermore, CD8⁺ T cells conditioned by IL-12 in vitro expressed high levels (increased mean fluorescence intensity) of T-bet, which led to the generation of KLRG1⁺ and IL-7R⁻ CD8⁺ effectors (48). Additionally, KLRG1⁺ and IL-7R⁻ CTLs were shown to naturally express more T-bet transcripts and protein than their KLRG1⁻ and IL-7R⁺ CTL counterparts (48). Hence, by modulating T-bet, inflammatory cytokines like IL-12 may influence the balance between acutely generated effector cells and long-lived memory cell precursors. In the present study we obtained clear in vivo evidence that IL-12 plays a decisive role in the generation of KLRG1⁺CD8⁺ T cells during T. gondii infection. Furthermore, we provide ex vivo data showing that T-bet expression is indeed dependent on IL-12 receptor signaling. Finally, the phenocopying of the IL-12 signaling-deficient CD8⁺ effector profiles by T-bet deficiency generally lend credence to the IL-12 → T-bet → KLRG1 inductive pathway outlined by Joshi et al. (48).

The model recently proposed by Kaech and colleagues (48) postulates that graded levels of T-bet induce KLRG1⁺ effector cell differentiation. A prediction of this model is that in our system the FIII subpopulation should have significantly higher levels of T-bet compared with the FII subset. We observed, however, that the WT FII and FIII subpopulations not only contained similar concentrations of T-bet⁺ cells, but also expressed comparable levels of T-bet as measured by mean fluorescence intensity. Even after in vivo restimulation, FIII cells did not express significantly higher levels of T-bet ex vivo. Moreover, our KLRG1⁻ and KLRG1⁺ effector CTL subpopulations were found to be similarly responsive to IL-12 signaling in vitro, which does not support the presence of graded T-bet levels among the subpopulations given that T-bet positively autoregulates IL-12Rβ2 expression (58). Our results suggest that in vivo T-bet expression is not stably maintained among most differentiated effector cells. Rather, T-bet could be expressed only transiently following exposure to parasite Ags and innate cytokines. Although T-bet seems to be an important transcription factor for the induction and maintenance of KLRG1 positivity, other factors must be involved because some KLRG1⁺CD8⁺ T cells develop in the CPS-primed T-bet^−/− mouse and only a minority of KLRG1-expressing CTLs contemporaneously express T-bet. Therefore, IL-12 likely induces multiple regulatory factors for the generation of the KLRG1⁺ effector subpopulations in vivo. Given the broad expression of IL-12 receptors, uniform IL-12 responsiveness, and equivalent T-bet levels in all three effector cell subpopulations, we envision that the regulatory influence of IL-12 signaling extends throughout the developmental progression of CD8⁺ T cells.

In summary, we have characterized the CD8⁺ T cell response to T. gondii vaccination and identified three distinct subpopulations of effector cells that coexist in the peritoneal effector site. These results support the general notion that primary CTL responses are heterogeneous by nature (25) and that CTL subsets likely develop along distinct lineages for the generation of effector or memory precursor cells (26). Importantly, our data underscore the key role of IL-12 for the induction of antiparasite CD8⁺ effector responses and for the generation of end-stage effector CTLs. The propensity of IL-12 in favoring CTL effector differentiation into short-lived effector cells, perhaps at the expense of memory cell precursors, has important implications for the preservation of CTL function during chronic infection and for the use of adjuvants to improve vaccines against intracellular pathogens.

We thank Steve Reiner, Laurent Brossay, Loren Fast, and Pamela Gaddi for constructive comments and also Laurie Glimcher for providing us with T-bet-deficient mice. We are grateful to former and current members of the Yap Lab: Mike Shaw, Mike Ling, and YanLin Zhao.

The authors have no financial conflict of interest.