Report

Regulation of Antigen-Specific CD8+ T Cell Homeostasis by Perforin and Interferon-γ

See allHide authors and affiliations

Science  17 Nov 2000:
Vol. 290, Issue 5495, pp. 1354-1357
DOI: 10.1126/science.290.5495.1354

Abstract

T cell memory depends on factors that regulate expansion and death of these cells after antigenic stimulation. Mice deficient in perforin and interferon-γ (IFN-γ) exhibited increased expansion, altered immunodominance, and decreased death of antigen-specific CD8+ T cells after infection with an attenuated strain ofListeria monocytogenes, which was cleared from these mice. Expansion of CD8+ T cells was controlled by perforin, whereas IFN-γ regulated immunodominance and the death phase. Thus, perforin and IFN-γ regulate distinct elements of CD8+ T cell homeostasis independently of their role as antimicrobial effector molecules.

In response to infection, CD8+ T cells expand and differentiate into effector cells, which mediate pathogen clearance through perforin-dependent cytolysis or elaboration of IFN-γ or tumor necrosis factor (TNF) (1). The magnitude of expansion varies, depending on the infection, and reproducible hierarchies of immunodominance are observed in pathogen-specific CD8+ T cell responses. After pathogen clearance, antigen-specific (Ag-specific) CD8+ T cells undergo a reproducible death phase in which 80 to 90% of the expanded CD8+ T cells are eliminated, irrespective of the magnitude of the initial expansion. The remaining Ag-specific memory CD8+ T cells are maintained throughout the life of the host. The factors controlling the expansion, immunodominance hierarchies, and the death phase of CD8+ T cells remain undefined. Perforin and IFN-γ might regulate CD8+ T cell homeostasis through their role in resistance to infection (1) or through pathways that are independent of their antimicrobial effector functions (2). We addressed this issue by analyzing Ag-specific CD8+ T cell homeostasis in mice with perforin and/or IFN-γ deficiency.

CD8+ T cell responses were elicited in wild-type and gene knockout (KO) mice by infection with an attenuated ActA-deficientListeria monocytogenes (LM) strain (DP-L1942) (3) that elicits protective CD8+ T cell–mediated resistance in IFN-γ–deficient mice (4). At 7 days post infection (p.i.) of BALB/c (H-2d MHC) wild-type mice, CD8+ T cells specific for the dominant LLO91-99 and subdominant p60217-225 epitopes of LM represented 1/50 and 1/256, respectively, of splenic CD8+ T cells (Fig. 1, A and B) as detected by intracellular staining (ICS) for TNF. This technique detects the majority of Ag-specific CD8+ T cells identified by ICS for IFN-γ (5, 6) or by tetrameric major histocompatibility complex (MHC) class I–peptide staining (7). Similar results were reported after infection of wild-type mice with virulent LM (8–10). Expansion of Ag-specific CD8+ T cells in wild-type mice is followed by a death phase, which is essentially complete by 14 days p.i. (Fig. 1B). The resulting memory levels of Ag-specific CD8+ T cells were maintained for at least 60 days p.i.

Figure 1

Kinetics of Ag-specific CD8+ T cell responses after LM infection in wild-type (WT) and perforin- and/or IFN-γ–deficient mice. Wild-type and PKO/GKO mice were infected with 1.3 × 106 ± 0.4 LM strain DP-L1942 (37) and Ag-specific CD8+ T cells in the spleen were measured by ICS for TNF (38). (A) Data are from representative mice at day 7 p.i. Numbers represent the percent of TNF+ CD8+ T cells in the absence or presence of LLO91-99 (39) and/or p60217-225 (40) peptides. (B) Total number per spleen (mean ± SD) and frequencies of Ag-specific CD8+ T cells obtained from 4 to 15 wild-type or PKO/GKO mice per time point. (□), WT-LLO; (▪), WT-p60; (○), PKO/GKO-LLO; (•), PKO/GKO-p60. Dashed line represents limit of detection. (C) Total number per spleen (mean ± SD) and frequencies of Ag-specific CD8+ T cells obtained from three wild-type or PKO/GKO mice per group at day 7 p.i. with the indicated dose of DP-L1942. Symbols are as in (B). (D) Wild-type, PKO, and GKO mice were infected with 1.2 × 106 ± 0.2 DP-L1942, and the frequency and total number (mean ± SD) of LLO91-99- (top row) or p60217-225-specific (bottom row) CD8+ T cells in the spleen was determined by ICS for TNF at the indicated days. Data was obtained from 3 to 12 mice per time point.

Infection of H-2d MHC perforin/IFN-γ–deficient (PKO/GKO) mice (11) with DP-L1942 resulted in altered homeostasis of Ag-specific CD8+ T cells (Figs. 1 and2). Expansion of LLO91-99- and p60217-225-specific CD8+ T cells at day 7 p.i was significantly increased to 1/11 and 1/16, respectively, of splenic CD8+ T cells (Fig. 1, A and B). PKO/GKO mice contain 9-fold (LLO91-99) and 33-fold (p60217-225) greater total Ag-specific CD8+ T cells as a result of increased spleen size (12). Thus, CD8+ T cell expansion is aberrant in PKO/GKO mice. In addition, the nearly equivalent expansion of LLO91-99- and p60217-225-specific CD8+ T cells in PKO/GKO mice resulted in a dramatic alteration of the immunodominance hierarchy normally observed in wild-type mice, where LLO91-99 responses were four- to sixfold higher than p60217-225 responses (Fig. 1, A and B).

Figure 2

Immunodominance and Ag-specific CD8+ T cell death in PKO/GKO, PKO, GKO, and wild-type mice. (A through C) Increase in LLO91-99- and p60217- 225- specific CD8+ T cells in PKO/GKO, PKO, and GKO mice compared with wild-type mice at the indicated days after infection. The total number of LLO91-99- and p60217- 225-specific CD8+ T cells per spleen of the indicated KO mouse was divided by the total number of epitope-specific CD8+ T cells per spleen of wild-type mice. (▵), LLO-KO/WT; (▾), p60-KO/WT. (D through F) Normalized death of LLO91-99- and p60217-225-specific CD8+ T cells in PKO/GKO, PKO, and GKO mice compared with wild-type mice at the indicated time points after LM infection. At the peak of primary response (day 7 p.i.), the frequency of LLO91-99- or p60217-225-specific CD8+ T cells are presented as 100%. (□), WT-LLO; (▪), WT-p60; (○), KO-LLO; (•), KO-p60.

The death phase of CD8+ T cells is also abnormal in PKO/GKO mice (Figs. 1B and 2D). No significant decrease in Ag-specific CD8+ T cells is observed at 14 days p.i. of PKO/GKO mice, a time at which the death phase (∼75% loss) is essentially complete in wild-type mice. At 28 days p.i., the frequency and total numbers of Ag-specific CD8+ T cells in PKO/GKO mice are less than twofold reduced compared with the levels at day 14 p.i. Thus, the rate of Ag-specific CD8+ T cell death is dramatically reduced in PKO/GKO mice (Fig. 2D). Together, these data show that Ag-specific CD8+ T cell expansion, immunodominance, and death are disrupted in PKO/GKO mice.

PKO/GKO mice harbor more organisms than wild-type mice at 2 days p.i. (12), which could contribute to the increased expansion of Ag-specific CD8+ T cells. However, the ∼10-fold increase in bacterial numbers does not appear sufficient to cause the ∼1000-fold shift in the dose response of CD8+ T cell expansion in PKO/GKO compared with wild-type mice (Fig. 1C). Furthermore, the altered immunodominance hierarchy in PKO/GKO mice does not result from increased antigen load because it is observed at all doses of infection, whereas the normal immunodominance hierarchy in wild-type mice is preserved at all doses (Fig. 1C). Lastly, PKO/GKO mice clear infection by 10 days p.i. (12). Thus, the aberrant death phase of Ag-specific CD8+ T cells does not result from persistent infection. These data suggest that the impact of perforin and IFN-γ on CD8+ T cell homeostasis is distinct from their roles in clearance of infection.

Perforin is a critical component of the cytolytic effector pathways of CD8+ T cells in resistance to infection (13). In addition, perforin plays a role in exhaustion of Ag-specific CD8+ T cells after lymphocytic choriomeningitis virus (LCMV) infection (2, 14). To determine the contribution of perforin deficiency to altered CD8+ T cell homeostasis, H-2d MHC perforin-deficient (PKO) (15) and wild-type mice were infected with DP-L1942 and Ag-specific CD8+ T cell responses were analyzed. Expansion of both LLO91-99- and p60217-225-specific CD8+ T cells was increased three- to fourfold in PKO mice compared with wild-type mice (Figs. 1D and 2B). However, the total number and frequency of LLO91-99-specific CD8+ T cells remained four- to fivefold greater than the p60217-225 response (Fig. 1D), and the normal immunodominance hierarchy was preserved in PKO mice. In addition, CD8+ T cells in PKO mice exhibited a relatively normal death phase (Figs. 1D and 2E). Clearance of DP-L1942 infection was indistinguishable between PKO and wild-type mice (12). Furthermore, immunization of PKO mice with peptide-coated dendritic cells (DC) resulted in a threefold higher expansion of LLO91-99-specific CD8+ T cells compared with wild-type mice (Fig. 3). Thus, perforin regulates CD8+ T cell expansion independently of its role as an effector molecule in resistance to infection.

Figure 3

Expansion of Ag-specific CD8+ T cells in wild-type and PKO mice in the absence of infection. LLO91-99-coated (2 × 105) DC (41) were injected intravenously into wild-type or PKO mice. Four wild-type and three PKO mice were analyzed on day 7 for LLO91-99-specific CD8+ T cells by ICS for IFN-γ. Two mice from each group were analyzed at day 14. LOD, limit of detection, 5 × 103 Ag-specific CD8+ T cells per spleen. The percent of CD8+ T cells and spleen sizes were similar between groups.

IFN-γ has multiple roles in response to infection, ranging from up-regulation of molecules involved in antigen presentation (16) to activation of microbicidal macrophages (17). Infection of H-2d MHC IFN-γ–deficient (GKO) mice (4) with DP-L1942 resulted in similar frequencies and total numbers of LLO91-99-specific CD8+ T cells at day 7 compared with wild-type mice (Fig. 1D). In contrast, the frequency and total number of CD8+ T cells specific for p60217-225 was higher in GKO mice (fourfold and threefold, respectively) compared with wild-type mice, indicating that normal immunodominance hierarchies are not maintained in GKO mice (Figs. 1D and 2C). In contrast to wild-type mice, the frequencies and total number (Figs. 1D and 2F) of CD8+ T cells specific for either epitope did not decrease at 14 days p.i. for GKO mice and remained elevated at 28 days p.i. This result occurred despite clearance of the infection by day 10 (12). Thus, IFN-γ deficiency had little impact on CD8+ T cell expansion against the dominant LLO91-99 epitope but resulted in altered immunodominance hierarchies and decreased death rate of Ag-specific CD8+ T cells.

We next examined CD8+ T cell homeostasis after infection with LCMV, which elicits vigorous CD8+ T cell priming against the dominant NP118-126epitope, representing >50% of splenic CD8+ T cells at day 8 p.i. for wild-type mice (Fig. 4A) (18). In wild-type mice, virus is cleared and a rapid death phase ensues in which the majority of Ag-specific CD8+ T cells are eliminated by day 15 p.i. In contrast, PKO mice do not clear the infection and eventually die (2,14); however, LCMV is cleared from GKO mice by 12 days p.i., and these GKO mice mount CD8+ T cell responses against the NP118-126 epitope (19). Thus, we examined the death phase of NP118-126-specific CD8+ T cells in wild-type and GKO mice after LCMV infection (Fig. 4). Both wild-type and GKO mice made a vigorous NP118-126-specific CD8+ T cell response as detected by MHC class I tetramer staining at 8 days p.i., although the response of wild-type mice was higher than that of GKO mice. Similar results were obtained with ICS for TNF (7). These data suggest that IFN-γ contributes to the massive expansion of CD8+ T cells after LCMV infection of wild-type mice, perhaps through its ability to promote optimal antigen presentation (20). Strikingly, the substantial death phase observed in wild-type mice did not occur in GKO mice (Fig. 4, A through C). The total number of Ag-specific cells per spleen in GKO mice was only slightly reduced at day 29, a time at which wild-type mice contain <10% of the levels obtained at the peak of expansion. Consistent with previous reports, we could detect no LCMV in blood samples obtained at 12 days p.i. of wild-type or GKO mice. Thus, IFN-γ also contributes to the death phase of Ag-specific CD8+ T cells after LCMV infection.

Figure 4

Kinetic analysis of NP118-126-specific CD8+ T cells during LCMV infection. Wild-type and GKO mice were infected with 5 × 105 plaque-forming units (PFU) of LCMV strain Armstrong (10), and the frequency of NP118-126-specific CD8+ T cells in the spleen was determined by MHC class I tetramer Ld(NP118) (18) staining. (A) Frequency of NP118-126-specific CD8+ T cells from representative mice at day 8, 15, and 29 p.i. Numbers represent the percent of CD8+ T cells that stain with Ld(NP118). Frequency of CD8+/Ld(NP118)+ in naı̈ve mice was <1%. (B) Total number (mean ± SD) and frequencies of NP118-126-specific CD8+ T cells per spleen from three mice per time point. (C) Normalized death of NP118-126-specific CD8+ T cells in GKO mice and wild-type mice. At day 8 p.i., the total number of NP118-126-specific CD8+ T cells per spleen is presented as 100%.

Our results demonstrate that perforin and IFN-γ regulate distinct elements of Ag-specific CD8+ T cell homeostasis. Perforin deficiency could enhance CD8+ T cell expansion through decreased killing of antigen presenting cells (APC) (21,22), resulting in prolonged Ag-display and the stimulation of additional naı̈ve precursor cells. Alternatively, perforin may act directly on Ag-specific CD8+ T cells by controlling the balance between division and death during the expansion phase. The possibility that CD8+ T cells become sensitive to their own perforin was originally suggested after demonstration that perforin-deficient CD8+ T cells fail to undergo exhaustive deletion after LCMV infection (2).

Immunodominance in CD8+ T cell responses is a well-recognized but poorly understood phenomenon that may be controlled by the available T cell receptor (TCR) repertoire, strength of peptide-MHC binding, antigen processing, and/or antigen levels (23). Extensive studies of these parameters in the LM infection model suggested that immunodominance may be a consequence of the available TCR repertoire (9,24–26). However, it is not clear how the absence of IFN-γ would impact the development of the naı̈ve T cell repertoire. One possibility is that the proteosome changes associated with IFN-γ signaling (16) actually decrease the number of p60217-225 epitopes generated during in vivo LM infection of wild-type mice, causing the reduced CD8+ T cell response to this epitope. In this case, in vitro peptide quantitation experiments showing more p60217-225 than LLO91-99 epitope after LM infection (24) may reflect the situation that occurs in vivo in GKO mice. This issue can be addressed by quantitation of peptides after infection of wild-type and IFN-γ–deficient mice.

Currently, little is known regarding the regulation of the death phase of Ag-specific CD8+ T cells (27). Recent studies with LCMV infection demonstrate normal death of Ag-specific CD8+ T cells in the absence of TNF receptor 1 (TNFRI) and CD95 (28, 29), receptors that are critical for the phenomenon of activation-induced cell death. Two recent studies suggest that IFN-γ regulates apoptosis of activated CD4+ T cells during M. tuberculosis infection or in the development of experimental autoimmune encephalitis (EAE) (30, 31). IFN-γ could regulate the death phase of Ag-specific CD8+ T cells directly; by controlling the expression of death factors, death receptors, or survival receptors such as IL-15R (32); or by sensitizing cells to the signals delivered by these receptors. The existence of IFN-γ receptor knockout mice and the capacity to identify and purify Ag-specific CD8+ T cells via MHC class I tetramer technology should facilitate the testing of these hypotheses.

Alternatively, IFN-γ could indirectly regulate the death phase of CD8+ T cell homeostasis. Recent studies reveal that naı̈ve but not memory CD8+ T cells require self-peptide–MHC complexes for homeostatic proliferation in lymphopenic hosts (27). Thus, it was suggested that the memory pool of CD8+ T cells might be selected for the absence of low-affinity self-peptide–MHC interaction, so as to minimize the potential for inappropriate activation and autoimmune disease (27). In seeming contrast to this notion, analysis of TCR Vβ usage suggests that the repertoire of effector and memory CD8+ T cells is similar (33,34). Studies with higher resolution may be required to address this issue. On the basis of our results, we speculate that IFN-γ could participate in selection of the memory pool by increasing the expression of self-peptide–MHC complexes, thus determining a threshold for survival that would prevent the inappropriate activation of memory cells during infection with unrelated pathogens. Consistent with this idea, preliminary studies reveal decreased MHC class I expression on spleen cells of LM infected GKO compared with wild-type mice (7). More detailed analyses are required to determine the relation between self-peptide–MHC class I expression and the death phase.

Our results demonstrate that perforin and IFN-γ not only contribute to pathogen clearance but also play major roles in CD8+ T cell homeostasis that are distinct from their roles in control of infection. Because T cell memory is determined by the magnitude of expansion minus the death phase (35), our results may suggest strategies to enhance T cell memory in response to vaccination.

  • * To whom correspondence should be addressed. E-mail: john-harty{at}uiowa.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article