γ_c deficiency precludes CD8⁺ T cell memory despite formation of potent T cell effectors

Abstract

Several cytokines (including IL-2, IL-7, IL-15, and IL-21) that signal through receptors sharing the common γ chain (γ_c) are critical for the generation and peripheral homeostasis of naive and memory T cells. Recently, we demonstrated that effector functions fail to develop in CD4⁺ T cells that differentiate in the absence of γ_c. To assess the role of γ_c cytokines in cell-fate decisions that condition effector versus memory CD8⁺ T cell generation, we compared the response of CD8⁺ T cells from γ_c⁺ or γ_c⁻ P14 TCR transgenic mice after challenge with lymphocytic choriomeningitis virus. The intrinsic IL-7-dependent survival defect of γ_c⁻ naive CD8⁺ T cells was corrected by transgenic expression of human Bcl-2. We demonstrated that although γ_c-dependent signals are dispensable for the initial expansion and the acquisition of cytotoxic functions following antigenic stimulation, they condition the terminal proliferation and differentiation of CD8⁺ effector T cells (i.e., KLRG1^high CD127^low short-lived effector T cells) via the transcription factor, T-bet. Moreover, the γ_c-dependent signals that are critical for memory T cell formation are not rescued by Bcl2 overexpression. Together, these data reveal an unexpected divergence in the requirement for γ_c cytokines in the differentiation of CD4⁺ versus CD8⁺ cytotoxic T lymphocytes.

CD8⁺ T cells are an essential component of the adaptive immune response to pathogens including viruses, bacteria, and protozoa. Multiple parameters condition the generation of short-term effector CD8⁺ T cells and long-term memory CD8⁺ T cells (1). Upon recognition of pathogen-encoded peptides and appropriate costimulation, CD8⁺ T cells undergo massive clonal expansion and activation with profound modifications in their gene expression profile, leading to the differentiation of potent antiviral effector cells (2). These effector T cells patrol nonlymphoid tissues and, via rapid cytokine secretion and granule exocytosis, eliminate infected cells. The majority of CD8⁺ effector cells die from apoptosis, mediated by the B cell lymphoma 2 (Bcl-2) family proteins and the cell surface receptor CD95 (3). Immune protection is established by the T cells surviving this “contraction” phase (4) and maintained by slow basal homeostatic proliferation (5). Memory CD8⁺ T cells conserve key effector traits and high proliferative potential, thus ensuring rapid protection upon reinfection (6).

In recent years, the cell-fate programming of CD8⁺ T cells has been a major focus of interest. Indeed, a better understanding of the signals involved in the transition from effector to memory T cells could lead to the design of better vaccination strategies. The IL-7 receptor alpha-chain (CD127) has been proposed as a marker for CD8⁺ T cells destined to become memory cells (7) and the differential expression of CD127, and the killer cell lectin-like receptor G1 (KLRG1) by activated CD8⁺ T cells identified two subsets with distinctive cell fates: the KLRG1^high CD127^low short-lived effector cells (SLEC) and KLRG1^low CD127^high memory precursor effector cells (MPEC) (8, 9). The strength and duration of the antigenic signal, the CD4⁺ help received and the cytokine milieu have all been shown to influence the outcome of the CD8⁺ T cell response (1, 10, 11). However, specific versus redundant roles for cytokines signaling through the common gamma chain (γ_c) in this cell fate decision programming are still poorly defined.

Interleukin (IL)-2, -7, -15, and -21 share the γ_c receptor chain and its downstream signaling pathway and influence distinct steps in the CD8⁺ T cell immune response. The indispensable role for γ_c cytokines in central T cell development and peripheral T cell homeostasis is well established (5, 12). Although IL-7 is vital for the survival and homeostatic proliferation of naive and memory CD8⁺ T cells (13), IL-2 and IL-15 are essential for the acute proliferation, contraction and cell-renewal capacity of the CD8⁺ T cells (14, 15). Moreover, IL-21 production by CD4⁺ T cells promotes the cytotoxic function and maintenance of CD8⁺ effector T cells in the context of chronic viral infections (16, 17). Although these studies define roles for γ_c cytokines in CD8⁺ T cell memory generation, it is unclear at which step of the differentiation process these cytokines impact and what is their importance on the cell-fate decision toward terminal differentiation versus memory generation.

The role for γ_c cytokines in programming effector CD4⁺ T cells has been previously reported. Using the Marilyn TCR transgenic model of skin graft rejection, we showed that γ_c cytokines condition the progressive differentiation of CD4⁺ T cells (18). In the absence of γ_c, “spurious” CD4⁺ T cells are generated that show an activated phenotype (CD44^hiCD62L^lo) and migrate to the skin but are unable to elicit graft rejection secondary to severe deficiencies in cytotoxic effector molecules and cytokine production capacities (18). The γ_c-dependent signals involved in programming cytotoxic CD4⁺ T cells engage both STAT5- and PI3K-dependent pathways (19). Based on these results, we hypothesized an analogous role for γ_c cytokines in the differentiation of CD8⁺ T cells. However, γ_c cytokines are pleiotropic factors that can play complementary or overlapping roles in the CD8 differentiation process. To assess the impact of γ_c cytokines on CD8 T cell programming, we derived TCR transgenic mice on the recombination-activating gene 2 (Rag2) deficient background with or without γ_c. Furthermore, we corrected the intrinsic survival defect of γ_c^−/− naïve CD8⁺ T cells by expression of human Bcl2. This approach allowed us to study the entire CD8⁺ T cell differentiation program in the absence of γ_c.

Results

Expression of Bcl2 Rescues Peripheral Naive P14 CD8⁺ T Cells in the Absence of γ_c.

γ_c-deficiency affects not only the survival of naive T cells but also leads to continuous accumulation of activated T cells in secondary lymphoid organs (20). In contrast, γ_c^−/− TCR transgenic (Tg) mice on the Rag2 deficient background harbor monoclonal populations of naive T cells, thus providing an approach to study the role of γ_c cytokines during immune responses (18, 21). P14 TCR Tg mice that develop CD8⁺ αβ T cells specific for the envelope glycoprotein 33–41 (GP_33–41) of the lymphocytic choriomeningitis virus (LCMV) were crossed onto the Rag2^−/−γ_c^+/+ or Rag2^−/−γ_c^−/− background. In the absence of γ_c, intrathymic development of P14 CD8⁺ SP T cells was strongly reduced and cells failed to accumulate in peripheral lymphoid organs (Fig. S1 A and B). The residual peripheral γ_c^−/− CD8⁺ P14 T cells display markedly reduced Bcl-2 levels (Fig. S1A) similar to that previously described for γ_c-deficient CD4⁺ T cells (22). Expression of the human Bcl-2 transgene could rescue the peripheral T cell survival defect in Rag2^−/−γ_c^−/− P14 mice, generating naïve splenic CD8⁺ T cells with a normal phenotype and number (Fig. S1). Bcl-2 transgenic Rag2^−/−γ_c^−/− P14 mice therefore provide an experimental model to assess the importance of γ_c signals in the differentiation of effector and memory CD8⁺ T cells that elicit anti-viral immunity.

γ_c-Dependent Cytokines Condition the Proliferation and Terminal Differentiation of KLRG1^high CD127^low SLEC.

We adoptively transferred P14 Bcl2 γ_c-competent or γ_c-deficient CD8⁺ T cells into naive C57BL/6 (Ly5.1) recipients and infected them with LCMV Armstrong. Controls included transfer of P14 CD8⁺ T cells that did not harbor the Bcl2 transgene. Following antigenic stimulation in vivo, both P14 Bcl2 γ_c^+/+ and γ_c^−/− cells, as well as P14 γ_c^+/+ cells, proliferated initially with the same kinetics, although the peak of expansion was significantly reduced in the absence of γ_c (50 × 10⁶ versus 6 × 10⁶ antigen-specific cells respectively, P < 0.0001) (Fig. 1A). Interestingly, the dynamics of activation (assessed by monitoring cell surface markers) was largely unchanged in the absence of γ_c (Fig. 1B and Fig. S2A). Because recent reports have described two CD8⁺ T cell subsets (based on the expression of KLRG1 and CD127) with distinct functional properties (8, 9), we analyzed splenocytes of infected mice and found a preferential accumulation of KLRG1^low CD127^high MPEC over KLRG1^high CD127^low SLEC in γ_c-deficient CD8⁺ T cell effectors (Fig. 1 C and D and Fig. S2B). As the transcription factors T-bet and eomesodermin are implicated in CD8 T cell differentiation (23, 24), and as SLEC formation requires T-bet expression (8), we analyzed their expression in activated WT and γ_c-deficient P14 Bcl2 CD8⁺ T cells. Interestingly, the reduction in the SLEC subset was correlated with decreased T-bet expression in γ_c^−/− CD8⁺ T cells (Fig. 1E). Furthermore, Tbx21 and Klrg1 transcripts were markedly reduced at a single-cell level in γ_c^−/− CD8⁺ T cells, whereas Eomes levels remained comparable to γ_c⁺ P14 Bcl2 CD8⁺ T cells (Fig. 1F and S2C). Together, our data indicate that γ_c cytokines regulate T-bet expression and thereby condition the generation of KLRG1^high CD127^low SLEC.

Fig. 1.

γ_c-dependent cytokines condition the proliferation and terminal differentiation of KLRG1^high CD127^low SLEC. 10⁵ P14, P14 Bcl2, or P14 Bcl2 γ_c^−/− CD45.2 CD8⁺ T cells were adoptively transferred into naive CD45.1 mice that were subsequently infected with 2 × 10⁵ PFU of lymphocytic choriomeningitis virus Armstrong. (A) Mean ± SEM number of antigen-specific CD8 T cells calculated based on GP_33–41 tetramer and CD45.2 congenic marker staining (n = 6–12). (B) Cell surface expression of the indicated molecules by P14 Bcl2 (shaded) and P14 Bcl2 γ_c^−/− (line) CD8⁺ T cells at baseline, day 5 and day 7 postinfection. Numbers indicate mean fluorescent intensity (P14 Bcl2 top, P14 Bcl2 γ_c^−/− bottom). (C and D) Memory precursor effector cells and short-lived effector cells subsets were analyzed at day 7 postinfection based on KLRG1 and CD127 expression. Bar graph represents the mean ± SEM percentage of cells from four separate experiments (n = 6–12). Numbers in dot plots indicate the percentage of each correspondent population. (E) T-bet expression was determined in P14 Bcl2 (shaded) and P14 Bcl2 γ_c^−/− (line) CD8⁺ T cells at day 7. Numbers indicate mean fluorescent intensity. (F) Individual P14 Bcl2 (black bars) and P14 Bcl2 γ_c^−/− (white bars) cells were recovered at day 7 postinfection and Klrg1, IL7r, Tbx21 and Eomes expression assessed. The percentage of tetramer⁺ CD45.2⁺ CD3ε⁺ cells expressing the indicated gene is given (mean ± SEM). (** P < 0.005, *** P < 0.0005, NS P ≥ 0.05)

CD8 T Cell Killing Function Is Unaffected by the Absence of the γ_c Chain, Despite Reduced Granzyme B Levels.

We next characterized the impact of these alterations on the functional capacities of γ_c^−/− P14 Bcl2 CD8⁺ T cells. As shown in Fig. 2A, granzyme B protein levels were strongly decreased in activated CD8⁺ T cells in the absence of γ_c, whereas perforin protein levels were unaffected. This granzyme B defect was restricted to KLRG1^lo cells (Fig. 2A and Fig. S3A). Morphological analysis of cytotoxic granules demonstrated that granule size and shape were unchanged in γ_c^−/− CD8⁺ T cells although average number of granules per cell was reduced (Fig. 2B). Regarding cytokine production, we found that a similar proportion of γ_c^−/− CD8⁺ effector T cells produced IFNγ and TNFα following in vitro restimulation as their γ_c⁺ counterparts, although a significant increase (twofold) in triple producers (IFNγ⁺ TNFα⁺ IL-2⁺ cells; P = 0.002) were found in the absence of γ_c (Fig. 2C and Fig. S3 B and C).

Fig. 2.

CD8 T cell killing function is unaffected by the absence of the γ_c chain, despite reduced granzyme B (GrzB) levels. (A) KLRG1 expression and intracellular granzyme B and perforin expression of day 7 infected P14 Bcl2 (shaded) and P14 Bcl2 γ_c^−/− (line) CD8⁺ T cells. Numbers on histograms indicate the mean fluorescent intensity; numbers in dot plots indicate the percentage of cells (n = 8). (B) Morphology and enumeration of granzyme B (GrzB) spots in CD8⁺ CD45.2⁺ cells at day 7 postinfection by multispectral imaging. (Left) Representative images showing cells containing 1, 2, 3, and >4 GrzB spots per cell [bright field (BF), CD8, CD45.2, GrzB and composite of CD45.2 and GrzB (merge)]. (Right) Bar graph showing the average distribution of spot counts (mean ± SEM) in P14 Bcl2 (black bars) and P14 Bcl2 γ_c^−/− (white bars) CD8⁺ T cells. (C) IFNγ, TNFα, and IL-2 production by intracellular cytokine staining. Bar graph shows the average percent cytokine production (mean ± SEM) by day 7 P14 Bcl2 (black bars) and P14 Bcl2 γ_c^−/− (white bars) CD8⁺ T cells (n = 5–6 per genotype). (D and E) The 10⁵ P14 Bcl2 or P14 Bcl2 γ_c^−/− CD8⁺ T cells were adoptively transferred into naive perforin knock-out (Pfp^−/−) mice subsequently infected. Killing function and development of hemophagocytic lymphohistiocytosis were followed over time. (D) In vivo CTL assay comparing day 8 P14 Bcl2 γ_c^+/+ (filled circle) and P14 Bcl2 γ_c^−/− (opened circle) CD8⁺ effector T cells, to infected (filled square) and uninfected/naive (opened square) Pfp^−/− cells. (E) Survival of Pfp^−/− (filled square) chimeric mice transferred with P14 Bcl2 (filled circle) and P14 Bcl2 γ_c^−/− (opened circle) CD8⁺ T cells. (* P < 0.05, ** P < 0.005, *** P < 0.0005, NS P ≥ 0.05)

We next assessed the killing capacity of γ_c^−/− CD8 T cells. As perforin-deficient (Pfp^−/−) mice show defective cytotoxic T lymphocyte (CTL) and natural killer (NK) cell killing (25), we generated chimeric mice by adoptive transfer of naïve γ_c⁺ or γ_c^−/− P14 Bcl2 CD8⁺ T cells into Pfp^−/− recipients. Mice were infected with LCMV, and, on day 7, we performed an in vivo killing assay using GP_33–41-loaded target cells. We found that γ_c^−/− CD8⁺ T cells were as potent as their γ_c^+/+ counterparts in eliminating target cells; a surprising result considering the 10-fold reduction in the number of γ_c^−/− CD8⁺ effector T cells at the peak of the response (Fig. 2D). We next determined whether γ_c^−/− P14 Bcl2 T cells could correct or prevent LCMV-induced hemophagocytic lymphohistiocytosis syndrome in Pfp^−/− hosts. In this model, dysregulated cytotoxic function in response to LCMV infection leads to subsequent macrophages activation, hypercytokinemia, and multiorgan infiltration, resulting in hepatosplenomegaly, pancytopenia, fever/hypothermia, weight loss and death (26). Pfp^−/− recipients receiving either γ_c⁺ or γ_c^−/− CD8⁺ T cells survived equally well through the period following LCMV infection and remained healthy (Fig. 2E). Both groups maintained their weight, had stable body temperature (Fig. S4 A and B), and failed to develop organomegaly or pancytopenia (Fig. S4 C and D). Similar results were obtained at lower precursor cell frequencies (transfer of 2 × 10⁴ γ_c⁺ or γ_c^−/− CD8⁺ P14 T cells; Fig. S5). Our data indicate that γ_c^−/− CD8⁺ T cells have potent in vivo effector functions, despite lower number of KLRG1^high CD127^low SLEC and decreased granzyme B levels.

Defective CD8⁺ T Cell Expansion Due to Lack of Responsiveness to γ_c Cytokines in the Late Proliferative Phase.

Because the significant differences in CD8⁺ T cell numbers at the peak of proliferation could be the result of either abnormal proliferation or increased apoptosis, we analyzed expansion kinetics in the early postinfection period by CFSE labeling the CD8 T cells before adoptive transfer and infection. The initial precursor frequency was similar in both groups with approximately 10% surviving cells after transfer (27). Furthermore, as shown in Fig. 3A, the initial kinetics of T cell proliferation was unaffected in the absence of γ_c, with cells dividing at least 4–5 times in the first 3 days postinfection. Although the percentage of cycling cells is identical at day 5 between both groups, γ_c^−/− cells proliferated less at day 7, with 2.5-fold fewer cells in the S/G2/M phase of the cell cycle and a 1.3-fold increase in nonproliferating cells (KI-67^low) (Fig. 3B). Flow cytometric analysis showed no abnormalities in the apoptotic pathway during the contraction period, with normal expression of Fas, TNFRI, and caspase 3 (Fig. 3C). Finally, γ_c^−/− P14 Bcl2 CD8⁺ T cells did not demonstrate a survival defect in culture (Fig. 3D). Collectively, these results demonstrate the requirement for γ_c-dependent signaling for the late phase of the proliferative response (day 5–7) to maintain an elevated number of effector CD8⁺ T cells.

Fig. 3.

Defective CD8 T cell expansion is related to lack of responsiveness to γ_c cytokines in the late phase of the proliferative response. (A) CFSE-labeled P14 Bcl2 or P14 Bcl2 γ_c^−/− CD45.2 CD8⁺ T cells were adoptively transferred into naive CD45.1 mice that were subsequently infected. Numbers indicate the percentage of divided cells at day 3 postinfection. (B–D) P14 Bcl2 and P14 Bcl2 γ_c^−/− chimeric mice were generated as described in Fig. 1. (B) Five (Left) and seven (Right) days postinfection, P14 Bcl2 (Upper) and P14 Bcl2 γ_c^−/− (Lower) CD8⁺ T cells were analyzed for KI-67 and DAPI (n = 3). Numbers indicate the percentage of cells in each boxed gate. (C) Cell surface expression of the indicated apoptotic-related molecules by day 7 P14 Bcl2 (shaded) and P14 Bcl2 γ_c^−/− (line) CD8⁺ T cells (n = 3). (D) Day 5 postinfection, splenocytes were recovered from the appropriate mice and kept in culture for 48 h. Bar graph shows the corrected percentage of CD8⁺ CD45.2⁺ T cells alive based on a propidium iodide staining.

Memory Cell Generation Requires γ_c-Dependent Bcl2-Independent Signals.

To evaluate whether perturbed differentiation of γ_c^−/− CD8⁺ T had an impact on the development of memory T cells, we studied LCMV-infected chimeric mice more than 90 days after infection. Surprisingly, γ_c^−/− P14 Bcl2 CD8⁺ T cells were not detected after the contraction phase (from day 13 onward, n > 15 mice; Fig. 4A), despite the presence of KLRG1^low CD127^high MPEC at the peak of the response (Fig. 1D). We failed to detect γ_c^−/− P14 Bcl2 CD8⁺ T cells in the spleen or the bone marrow of chimeric mice 90 days postinfection, and reinfection with LCMV Armstrong did not elicit memory responses from γ_c-deficient P14 Bcl2 CD8⁺ T cells (Fig. 4B). These results identify a critical γ_c cytokine-dependent but Bcl2-independent signaling pathway for memory T cell generation.

Fig. 4.

Memory cell generation is established by γ_c-dependent Bcl2-independent signals. (A) P14 Bcl2 and P14 Bcl2 γ_c^−/− chimeric mice were generated as described in Fig. 1 and were followed longitudinally. Frequency of GP_33–41⁺ CD45.2⁺ cells from P14 Bcl2 (filled circle) and P14 Bcl2 γ_c^−/− (opened circle) CD8⁺ T cells over a 90-day time course (n = 6–10). (B) Absolute number of GP_33–41⁺ CD45.2⁺ T cells in the spleen and bone marrow of P14 Bcl2 (black bars) and P14 Bcl2 γ_c^−/− (white bars) chimeric mice at 90 days postinfection (n = 4, N.D. not detected).

Discussion

Differentiation of CD8⁺ T cells from the naive state to the fully competent effector cell stage is a progressive process involving both intrinsic and extrinsic factors. It has been proposed that very early in the immune response, CD8⁺ T cells are imprinted to become either short-lived terminal effectors or long-lived memory cells (9, 28). This process is influenced by several parameters, including the duration of antigen exposure and the presence of costimulatory molecules and soluble factors (IL-2, IL-21) derived from CD4⁺ T cells (1). Stromal cells also play critical roles by elaborating nutritive factors (including IL-7 and IL-15) that promote T cell survival and proliferation (29). As such, diverse γ_c cytokines coordinate the T cell differentiation process (12, 14, 15, 30).

We previously used γ_c-deficient mice to assess the unique and redundant roles for γ_c cytokines in CD4⁺ T cell differentiation (18). γ_c cytokines conditioned the progressive differentiation of CD4⁺ T cells, and in the absence of γ_c, activated T cells were generated but essentially lacked effector functions. In contrast, many aspects of CD8⁺ T cell differentiation proceed normally in the absence of γ_c with the generation of robust effectors. Thus, γ_c cytokine requirements for CD4⁺ versus CD8⁺ T cell differentiation diverge remarkably.

What could account for this difference? The inflammatory cytokine milieu, that can include IL-12/IL-23 and type I/II IFN, has been shown to influence the T cell differentiation process (11). In the LCMV model, type I IFN is abundantly produced and may provide accessory cell-dependent signals that can functionally replace the signals provided by γ_c cytokines via functionally redundant JAK/STAT activation pathways (31). An alternative explanation would imply that CD4⁺ T cell differentiation would be more dependent on γ_c cytokines that are elaborated during interactions with accessory cells (dendritic cell, stroma) than CD8⁺ T cells. It is known that CD4⁺ T cells have a strong requirement for costimulatory signals during their differentiation (32) that likely extends to γ_c cytokines (18). In contrast, CD8⁺ T cell differentiation appears intrinsically programmed following antigen encounter (28, 33) that would obviate requirements for prolonged γ_c cytokine stimulation.

Although robust CD8⁺ T cell differentiation was observed in the absence of γ_c, γ_c-dependent signals were necessary for the transition from effector to memory cells, affecting the differentiation and late proliferation of KLRG1^high CD127^low SLEC. This is consistent with the role of IL-2 in sustaining the expansion of CD8⁺ T cells (34). Moreover, IL-15 influences the survival of KLRG1^high CD127^low CD8⁺ T cells during the con-traction phase (8, 35). Here, we demonstrated that γ_c signals were not only essential for the survival and proliferation of CD8⁺ T cells but were also crucial for their differentiation into granzyme B⁺ SLECs, through the regulation of T-bet. Furthermore, despite the presence of Bcl2⁺ KLRG1^low CD127^high long-lived MPEC, γ_c signals were fundamental for the generation and maintenance of memory cells. Together, our results define the critical stages for γ_c cytokines in the programming of terminal effector CD8⁺ T cells and in the Bcl2-independent survival and homeostatic proliferation of memory CD8⁺ T cells.

Previous studies demonstrated that γ_c-dependent cytokines are important determinants of CD4⁺ and CD8⁺ T cell fate, with IL-2, -7, -15, and -21 impacting not only on the differentiation process but also the ability to generate and sustain memory responses (5, 12, 14, 15, 30). In all cases of single γ_c cytokine deficiency, memory CD8⁺ T cells were detected, albeit at reduced levels (7, 13, 16, 17, 36–39). This reduced memory formation, however, might have been secondary to reduced T cell survival, as γ_c cytokines promote homeostasis through enhanced expression of antiapoptotic Bcl2 family members (40). Here we find that in the absence of all γ_c cytokine signals, memory CD8⁺ T cell formation is completely abolished. Moreover, this γ_c-dependent role in memory formation is Bcl2-independent.

What γ_c cytokines are involved in this process? Antigen-specific memory CD8⁺ T cells are detected in the absence of IL-2, IL-7, IL- 5, or IL-21 (16, 17, 36–39). Moreover, adoptive transfer of P14 CD122^−/− cells (lacking IL-2Rβ) generates a pool of memory CD8⁺ T cells indicating that a combination of γ_c cytokines are involved. IL-7 and IL-15 have overlapping roles in T cell homeostasis (12, 41, 42) and a similar synergy may operate during memory generation. The downstream targets of IL-21 in memory T cells are poorly defined and could promote memory formation through distinct pathways.

Triggering of γ_c receptors is linked to enhanced survival. Antiapoptotic molecules other than Bcl2 are targets of γ_c cytokines, including Mcl-1 that is downstream of IL-7 (43). γ_c-dependent cytokines also regulate transcriptional profiles, partly through activation of the mammalian target of rapamycin (mTOR) kinase that modulates the expression of both T-bet and eomesodermin (44). Another key transcription factor, the B cell transcription repressor Blimp-1, is involved in the terminal differentiation of CD8⁺ T cells (45). Very recently, IL-2 signals have been shown to favor the terminal differentiation of cytolytic CD8 T cells, in part through regulation of Blimp-1 and Bcl6 (46, 47). Interestingly, Blimp1^−/− CD8 T cells fail to up-regulate granzyme B and KLRG1 upon activation (48, 49).

Collectively, our results reveal an unexpected divergence in the requirement for γ_c cytokines in the differentiation of CD4⁺ versus CD8⁺ cytotoxic T lymphocytes. Nevertheless, γ_c cytokines remain critical determinants of T cell memory. These observations suggest that selective modulation of γ_c cytokines could im-pact strongly on immunotherapies and should be taken into account when optimizing vaccine protocols.

Materials and Methods

Mice.

CD45.2 Rag2^−/− P14 TCR Tg mice (P14; expressing a TCR specific for the LCMV GP_33–41 epitope) were provided by A. Freitas (Institut Pasteur, Paris, France) and backcrossed onto the C57BL/6 (B6) background. P14 Bcl2⁺ mice with or without the γ_c chain were then produced as previously described (22). P14, P14 Bcl2⁺, and P14 Bcl2⁺ γ_c^−/− chimeric mice were generated by adoptive transfer of 10⁵ or 2 × 10⁴ MACS-purified naive CD45.2 TCR Tg CD8⁺ T cells into 4- to 6-week-old naive CD45.1⁺ B6 recipients. To minimize rejection, CD45.1⁺ B6 recipients were generated by crossing Tg^- mice with CD45.1⁺ B6 mice (Charles River). B6^Prf1tm1sdz/J (Pfp^−/−) mice were provided by G. De Saint-Basile (Institut National de la Santé et de la Recherche Médicale U768, Paris, France). All mice were housed in specific pathogen-free facilities at the Institut Pasteur.

Virus.

Stocks of the Armstrong strain of LCMV were plaque purified on Vero cells and grown in BHK-21 cells as described previously (50). Infectious LCMV was quantified by plaque assay. Twenty-four hours after T cell transfer, mice were injected i.p. with 2 × 10⁵ PFU of LCMV.

Cell Isolation and Flow Cytometry Analysis.

Single cell suspensions from thymus, spleen and bone marrow were analyzed as previously described (22). Antibodies were purchased from eBioscience and BD Bioscience except for Granzyme B (Caltag), TNFR (Biolegend), T-bet (Santa Cruz Biotechnology), and ultra-avidin-R-phycoerythrin (Leinco). MHC class I peptide tetramers were used as described (51). Dead cells were excluded using Live/Dead Fixable Aqua Dead Cell stain kit (Molecular Probes; Invitrogen), except for the in vitro survival assay where propridium iodide (3 μM) was used. Data were acquired using a FACSCanto flow cytometer (BD Biosciences) and analyzed using FlowJo software (Treestar).

Intracellular Staining and in Vivo Cytotoxicity Assay.

Intracellular staining was done on freshly isolated splenocytes according to manufacturer's instructions, with the Fixation and Permeabilization Kits from eBioscience or BD Bioscience. IFNγ, TNFα, and IL-2 quantification were performed after in vitro restimulation with GP_33–41 peptide (0.2 μg/mL) in the presence of brefeldin A (10 μg/mL) for 4 h. For cell cycle analysis, DAPI (Sigma) was added at the time of analysis, after intracellular staining for K_i-67. CFSE labeling was done by incubating cells for 10 min at 37 °C with 5 μM CFSE in PBS 2% FCS. For the in vivo cytotoxicity assay, the fate of GP_33–41 peptide pulsed CFSE^high-labeled (1 μM) and nonpulsed CFSE^low-labeled (0.1 μM) splenocytes in Pfp^−/− chimeric mice was analyzed (50).

Multispectral Imaging.

Digital imaging was performed on an ImageStream100 (Amnis Corporation). 15,000–25,000 cells were imaged for each sample and analyzed using the manufacturer's software. The spot count algorithm calculated the number of granzyme b spots of more than one pixel in radius and twice the intensity of background. Spot counting accuracy was confirmed by manual verification of each image.