IFN-γ Receptor Signaling Regulates Memory CD8⁺ T Cell Differentiation Free

Foreign and self-Ags can induce CD8⁺ T cell responses and subsequent memory formation. For example, pathogen-specific effector CD8⁺ T cells reach their maximum number 6–8 d after cognate Ag (cAg) contact (1). In the following contraction phase, most effector CD8⁺ T cells die; few cells survive to form a long-lived memory pool (1, 2). However, under lymphopenic conditions, low-affinity self-peptide–MHC complexes are sufficient to trigger CD8⁺ T cell activation, lymphopenia-induced proliferation (LIP), and subsequent differentiation into memory-like cells (3–8). Importantly, the comparison of their gene-expression profiles suggests that the molecular pathways guiding the differentiation of LIP- and cAg-induced memory are redundant (9).

The effector CD8⁺ T cell pool contains memory cell precursors (10–12) that differentiate into central memory T cells (T_CMs) or effector memory T cells (T_EMs) (13). T_CMs are mainly located in lymphoid tissues and express high levels of lymph-node homing receptors, such as CCR7 and CD62L (14, 15). In contrast, T_EMs express low levels of CCR7 and CD62L (13) and home to nonlymphoid tissues, where they exert immediate effector functions (16). However, the mechanisms regulating T_CM/T_EM differentiation, in particular under lymphopenic conditions, are poorly understood.

We previously showed that host IFN-γR signaling counterregulates CD8⁺ T cell expansion and limits the size of the memory pool (17). However, it remained unknown whether and how IFN-γR signaling affects T_CM/T_EM differentiation. With the help of peptide immunization, we show in this study that cAg-induced CD8⁺ T cell expansion and subsequent T_EM generation are increased in the absence of host IFN-γR signaling. However, IFN-γR signaling in CD8⁺ T cells does not affect their expansion and memory fate decision. Similar to peptide vaccination, the lack of IFN-γR signaling in host cells is associated with elevated rates of LIP and T_EM generation. In contrast, however, the lack of IFN-γR signaling in CD8⁺ T cells blocks lymphopenia-driven T_EM differentiation and promotes the generation of IFN-γ–producing T_CM-like cells. This can be prevented by peptide immunization, suggesting that IFN-γR signaling in CD8⁺ T cells regulates memory fate decision in a TCR ligand-dependent fashion.

Wild type (WT) C57BL/6J mice (B6), IFNγ^−/− (B6.129S7-Ifng^tm1Ts) mice, IFNγR^−/− (B6.129S7-Ifngr^tm1Agt) mice (all homozygous for Thy1.2) and congenic B6.PL-Thy1a/Cy (homozygous for Thy1.1) and Rag^−/− (B6.129S7-Rag1^tm1Mom/J) mice were purchased from Charles River Laboratories (Sulzfeld, Germany) and bred in our animal facility. C57BL/6 OT-I cells express a transgenic TCR (Vα2Vβ5) specific for the chicken OVA-derived, H2-K^b–restricted peptide OVA_257–264 (SIINFEKL). All mice were housed under specific pathogen-free conditions, and animal experiments were performed according to institutional guidelines.

CD8⁺ OT-I T cells were purified using CD8α-specific microbeads and AutoMACS (both from Miltenyi Biotec, Bergisch Gladbach, Germany). At day 1, 1 × 10⁶ to 3 × 10⁶ CD8⁺ OT-I cells were injected i.v. into the tail vein. At day 0, mice were immunized i.v. with 250 μg SIINFEKL and 20–50 μg LPS (Escherichia coli, serotype 0111: B4; Sigma, Schnelldorf, Germany). Control animals were injected with Dulbecco’s PBS (DPBS) instead. Single-cell suspensions from the indicated organs were stained with mAbs for CD8α (53-6.7), Thy1.1 (CD90.1; OX-7), Thy1.2 (CD90.2; 53-2.1), CD44 (IM7), CD62L (MEL-14) (BD Biosciences, San Jose, CA) or H2-K^b:SIINFEKL pentamers (ProImmune, Oxford, U.K.). All samples were analyzed individually on a FACSCalibur flow cytometer (BD Biosciences). The specificity of staining was confirmed with isotype-matched control Abs.

Isolation of splenic CD11b⁺ cells and CFSE-labeling of IFNγR^−/− OT-I cells were done as described previously (17). CD11b⁺ cells and IFNγR^−/− OT-I cells were cocultured at a ratio of 5:1 in the presence of 100 nM SIINFEKL and 100 ng/ml LPS with or without 50 ng/ml recombinant mouse IFN-γ (R&D Systems, Wiesbaden-Nordenstadt, Germany). After 3 d, cells were stained with 7-aminoactinomycin D (7-AAD; BD Biosciences) and mAbs for CD8α (53-6.7) or CD62L (MEL-14). Dead 7-AAD⁺ cells were excluded from analysis.

Splenic single-cell suspensions from individual recipient mice were cultured in 96-well tissue culture plates for 3–4 h at 37°C in RPMI 1640 + 10% FCS, 1% penicillin/streptomycin, and 2-ME (50 μM), with or without 10 μM SIINFEKL, in the presence of brefeldin A. For intracellular staining, IFN-γ–specific mAbs (XMG1.2) and the Intracellular Cytokine Staining Kit (all from Pharmingen, Hamburg, Germany) were used according to the manufacturer’s recommendations.

Mice were injected with 50 μg LPS, and serum was prepared 6 h later. IFN-γ levels were determined for individual serum samples by fluorescent bead immunoassay (mouse IFN-γ simplex kit, Bender MedSystems, Vienna, Austria), according to the manufacturer’s recommendations. Data were acquired on a FACSCalibur flow cytometer and analyzed with FlowCytomix Pro software (Bender MedSystems).

We previously showed that IFN-γR signaling in host cells controls the magnitude of CD8⁺ T cell responses and the size of the resulting memory pool (17). However, this did not exclude a contribution of IFN-γR signaling in CD8⁺ T cells. To test this possibility, equal numbers of Thy1.1-homozygous CD8⁺ TCR-transgenic WT OT-I cells and Thy1.1/1.2-heterozygous IFNγR^−/− OT-I cells were transferred simultaneously into Thy1.2-homozygous WT mice (Fig. 1A). One day after transfer, recipient mice were injected with cAg for OT-I (SIINFEKL) and LPS (immunized) or PBS (naive). OT-I cell numbers in the spleen were determined by flow cytometry in the priming, contraction, and memory phase at days 7, 14, and 30 after immunization. As shown in Fig. 1B, WT and IFNγR^−/− OT-I cells were equally abundant at the indicated time points and irrespective of whether the recipient had been immunized. After restimulation with SIINFEKL in vitro, the percentages of IFN-γ–producing WT and IFNγR^−/− OT-I cells were nearly identical at any given time point (Fig. 1C). Thus, IFN-γR signaling in OT-I cells did not affect their frequency (Fig. 1B) or function (Fig. 1C) in the priming, contraction, and memory phase in WT mice.

We previously showed that OT-I responses are more efficient in LPS/SIINFEKL-immunized IFNγR^−/− mice (17). Because of the lack of its consumption, elevated levels of IFN-γ are found in the serum of LPS-injected IFNγR^−/− mice (Fig. 2A). However, increased IFN-γ abundance did not affect the number of OT-I cells that were recruited into the response. As shown in Fig. 2B, CFSE-labeled naive WT OT-I cells primed in WT and IFNγR^−/− mice had completely lost CFSE 7 d after immunization, indicating that they had undergone at least seven cell divisions. Furthermore, all primed OT-I cells expressed high levels of CD44, irrespective of the host. Hence, the lack of host IFN-γR expression and subsequent IFN-γ accumulation do not affect the number of naive OT-I cells that are primed after peptide vaccination.

Next, we analyzed whether elevated levels of IFN-γ in IFNγR^−/− mice were sufficient to promote the expansion of IFN-γR–competent OT-I cells. For this purpose, equal numbers of WT and IFNγR^−/− OT-I cells were transferred into WT and IFNγR^−/− mice. As shown in Fig. 2C and 2D, the relative abundance of both OT-I populations was similar in WT and IFNγR^−/− mice at days 7 and 14 after immunization. However, at day 30, IFNγR^−/− OT-I cells were more frequent in IFNγR^−/−, but not WT, recipients (Fig. 2D). In summary, these results show that despite high levels of IFN-γ, IFN-γR signaling in OT-I cells is dispensable for expansion and memory formation in IFNγR^−/− mice. In contrast, the preferential survival of IFNγR^−/− memory OT-I cells (Fig. 2D) suggests their protection from the proapoptotic effects of IFN-γ (18–20), which might be more pronounced in IFNγR^−/− mice bearing elevated levels of IFN-γ after peptide vaccination (Fig. 2A).

However, it is important to emphasize that both OT-I populations were more abundant in IFNγR^−/− mice than in WT mice at any given time point (Fig. 2C). Because recruitment of OT-I cells into the response did not differ between WT and IFNγR^−/− mice (Fig. 2B), more pronounced expansion and/or survival of WT and IFNγR^−/− OT-I cells seems to be responsible for their elevated abundance in IFNγR^−/− recipients.

The frequency of naive CD8⁺ T cell precursors affects the size of the effector cell pool and subsequent memory fate decision (21, 22). Having shown that OT-I cell numbers are elevated in IFNγR^−/− mice, we asked next whether this is associated with alterations in T_CM/T_EM differentiation. WT OT-I cells were transferred into WT and IFNγR^−/− recipients. Four to five weeks after immunization, all OT-I cells had developed a memory phenotype, as judged by their high levels of CD44 expression (data not shown). In spleens of WT recipients, 32% of OT-I cells expressed low levels of CD62L, indicating their T_EM phenotype (Fig. 3A). However, in IFNγR^−/− spleens, the frequency of T_EM cells was increased to 54% (Fig. 3A). Alterations in T_CM/T_EM ratios were also apparent in the bone marrow (BM) of IFNγR^−/− mice (Fig. 3A): 70% of OT-I cells had a T_EM phenotype as opposed to 54% in WT BM (Fig. 3A). T_CM/T_EM ratios in IFNγR^−/− mice were not altered because of the loss of T_CMs. As shown in Fig. 3B, more T_CMs were found in spleen and BM of IFNγR^−/− mice compared with WT recipients. However, the number of T_EMs in IFNγR^−/− mice was greatly increased (Fig. 3B), demonstrating that altered T_CM/T_EM ratios in IFNγR^−/− mice (Fig. 3A) mainly resulted from the more efficient generation of T_EMs rather than the loss of T_CMs (Fig. 3B). This was not affected by IFN-γR signaling in OT-I cells. When WT and IFNγR^−/− mice were reconstituted with equal numbers of WT and IFNγR^−/− OT-I cells, immunized, and analyzed 36 d later, T_CM/T_EM ratios in spleen and BM (Fig. 3C) and splenic IFN-γ production (Fig. 3D) were nearly identical for both OT-I populations, irrespective of the host. However, the frequencies of WT and IFNγR^−/− T_EM cells were increased in spleen and BM of IFNγR^−/− mice (Fig. 3C). It is important to emphasize that the difference in CD62L expression on OT-I cells was a vaccination-induced effect rather than a non–Ag-specific phenomenon. As shown in Fig. 3E, CD62L was expressed at high levels on freshly isolated WT and IFNγR^−/− OT-I cells prior to as well as 28–31 d after transfer into nonvaccinated WT mice. Similarly, WT OT-I cells maintained their naive phenotype in nonvaccinated WT and IFNγR^−/− recipients (Supplemental Fig. 1).

This point was further confirmed when memory differentiation was studied in OT-I–reconstituted IFNγ^−/− mice. WT OT-I cells primed in IFNγ^−/− mice contained 28% of T_EMs 4–5 wk after immunization (Fig. 3F), which was similar to WT hosts reconstituted with WT and IFNγR^−/− OT-I cells (Fig. 3A). In contrast, IFNγ^−/− OT-I cells preferentially differentiated into T_EMs (Fig. 3F). Thus, T_CM/T_EM differentiation in the absence of host- and OT-I–derived IFN-γ recapitulates our findings in IFNγR^−/− mice.

IFN-γ regulates the magnitude of CD8⁺ T cell responses prior to the contraction phase (23), partially via its action on suppressive CD11b⁺ cells (17). In this early phase of the response, the frequency of CD62L^loCD8⁺ effector T cells indicates the relative abundance of T_EMs in the ensuing memory pool (22). Indeed, the frequency of CD62L^lo OT-I cells was already elevated in IFNγR^−/− mice 7 d after immunization (Fig. 4A), suggesting that the accumulation of CD62L^lo T_EMs in these mice is already determined in the early phase of the response. CD11b⁺ cells lacking IFN-γR fail to counterregulate OT-I responses (17). Whether the lack of CD11b⁺-mediated suppression is associated with an increased rate of CD62L downregulation during priming was tested in vitro. To avoid any possible effects of IFN-γ on OT-I cells, they were isolated from IFNγR^−/− OT-I mice and cocultured for 3 d with WT or IFNγR^−/− CD11b⁺ cells in the presence of SIINFEKL. As shown in Fig. 4B (upper left panel), a considerable number of OT-I cells had not divided and maintained high levels of CD62L in the presence of WT CD11b⁺ cells. In contrast, nearly all OT-I cells had divided and downregulated CD62L in IFNγR^−/− CD11b⁺ cocultures (Fig. 4B, upper right panel). To mimic our vaccination protocol, LPS was added to the cocultures (Fig. 4B, middle panels). In the presence of WT CD11b⁺ cells, nearly all OT-I cells had divided, and CD62L downregulation was slightly more pronounced (Fig. 4B, middle left panel). Hence, LPS promotes rather than suppresses OT-I activation/proliferation in the presence of WT CD11b⁺ cells. However, LPS had no apparent effect on IFNγR^−/− CD11b⁺ cocultures (Fig. 4B, middle right panel). When rIFN-γ was added to SIINFEKL+LPS cocultures, OT-I responses were strongly suppressed in the presence of WT CD11b⁺ cells, as shown by the fact that only few OT-I cells had divided and downregulated CD62L (Fig. 4B, lower left panel). In contrast, OT-I cell cycle progression and CD62L downregulation still remained unaffected in IFNγR^−/− CD11b⁺ cocultures (Fig. 4B, lower right panel). Hence, the lack of IFN-γR signaling in CD11b⁺ cells allows more efficient OT-I proliferation and CD62L downregulation shortly after OT-I priming.

It is important to stress that OT-I cells that underwent the same number of cell divisions in WT and IFNγR^−/− CD11b⁺ cocultures always expressed less CD62L in the latter (Fig. 4C). This suggests that IFN-γR signaling triggers different effector mechanisms in CD11b⁺ cells, independently regulating cell cycle progression and CD62L expression of OT-I cells.

Having shown that IFN-γR signaling in CD11b⁺ cells is sufficient to block CD62L downregulation by OT-I cells, we asked next whether this would be sufficient to prevent the accumulation of CD62L^lo T_EMs in IFNγR^−/− mice. To test this, WT OT-I cells were transferred into WT and IFNγR^−/− mice. A group of IFNγR^−/− recipients was reconstituted with WT CD11b⁺ cells simultaneously. Four to five weeks after vaccination, the frequency and phenotype of memory OT-I cells were determined in the spleen. In accordance with previous experiments (17), the number of memory OT-I cells was reduced to WT levels in IFNγR^−/− mice reconstituted with WT CD11b⁺ cells (Fig. 4D). Surprisingly, however, suppression was not associated with a blockade of T_EM generation (Fig. 4E). Hence, IFN-γR signaling in CD11b⁺ cells counterregulates CD8⁺ T cell expansion (Fig. 4D) and blocks CD62L downregulation in the priming phase (Fig. 4B), but it is not sufficient to prevent the preferential generation of T_EMs in IFNγR^−/− mice (Fig. 4E).

In addition to vaccination, memory formation can be induced in response to lymphopenia. Under lymphopenic conditions, the recognition of self-peptide–MHC complexes is sufficient to activate naive CD8⁺ T cells and induce their LIP and subsequent differentiation into IFN-γ–producing memory-like cells (3–8). In lymphopenic mice, innate immune cells produce IFN-γ in response to the commensal microflora (24). Hence, LIP and subsequent memory differentiation occur in the presence of host-derived IFN-γ. Having shown that cAg-induced CD8⁺ T cell responses are regulated by IFN-γ, we asked next whether its action on host and/or CD8⁺ T cells also regulates expansion and subsequent memory differentiation in response to lymphopenia. To test this, WT and IFNγR^−/− OT-I cells were cotransferred into Rag^−/− and IFNγR^−/−Rag^−/− mice. Eight to 12 d after transfer, WT and IFNγR^−/− OT-I cells expressed high levels of CD44, indicating their memory phenotype (data not shown). Irrespective of the recipient, WT and IFNγR^−/− OT-I cells were equally abundant in the spleen (Fig. 5A). However, OT-I cell numbers were increased 4-fold in IFNγR^−/−Rag^−/− mice (Fig. 5A). The cellular composition of the spleen (Fig. 5C) and the activation state of dendritic cells (DCs) (Fig. 5D) were nearly identical in Rag^−/− and IFNγR^−/−Rag^−/− mice. This indicates that the lack of host IFN-γR signaling during LIP, rather than developmental defects in the hematopoietic system, promotes the accumulation of OT-I cells in IFNγR^−/−Rag^−/− mice.

Next, the phenotype of LIP-derived memory-like OT-I cells was analyzed. In Rag^−/− mice, 47% of WT OT-I cells had a T_EM phenotype (Fig. 6A, upper left panel). This was the case for 81% of OT-I cells in IFNγR^−/−Rag^−/− mice (Fig. 6A, upper right panel). Although slightly less pronounced, the preferential generation of T_EM-like OT-I cells in IFNγR^−/−Rag^−/− mice was still visible at day 30 (Fig. 6E, upper row). After restimulation in vitro, the fraction of WT OT-I cells producing IFN-γ was similar for Rag^−/− and IFNγR^−/−Rag^−/− mice at day 12 (Fig. 6B, upper row) and day 30 (Fig. 6F, upper row). Hence, the lack of IFN-γR signaling in lymphopenic recipients is associated with the more pronounced expansion of OT-I cells (Fig. 5A) and the preferential differentiation of T_EMs (Fig. 6A, 6E, upper row), without affecting IFN-γ production (Fig. 6B, 6F, upper row).

In contrast to cAg immunization of lymphoreplete mice (Fig. 3), the lack of IFN-γR signaling in OT-I cells directly affected T_CM/T_EM differentiation in lymphopenic mice. Eight to 12 d after transfer, 14% and 54% of IFNγR^−/− OT-I cells had a T_EM phenotype in Rag^−/− and IFNγR^−/−Rag^−/− mice, respectively (Fig. 6A, lower row). At day 30, the frequency of IFNγR^−/− OT-I T_EM cells was still reduced in both hosts compared with WT OT-I cells (Fig. 6E, lower row). Importantly, the percentage of IFN-γ–producing IFNγR^−/− OT-I cells was greatly elevated at day 12 (Fig. 6B, lower row) and day 30 (Fig. 6F, lower row). Thus, the lack of IFN-γR signaling in CD8⁺ T cells leaves cell numbers unaffected (Fig. 5A), blocks T_EM differentiation (Fig. 6A, 6E, lower row) and promotes the generation of IFN-γ–producing T_CM-like cells in lymphopenic mice (Fig. 6B, 6F, lower row).

In contrast to SIINFEKL, the self-peptides driving LIP of OT-I cells are supposed to be weak TCR agonists. We have shown that IFN-γR signaling in OT-I cells affects lymphopenia- but not SIINFEKL-driven memory cell differentiation. Therefore, we hypothesized that TCR signal strength determines whether IFN-γR signaling in OT-I cells contributes to memory fate decision. To test this, Rag^−/− and IFNγR^−/−Rag^−/− mice were reconstituted with equal numbers of WT and IFNγR^−/− OT-I cells and immunized with SIINFEKL. Both OT-I populations expanded equally well in either host (Fig. 5B). Nevertheless, T cell numbers were elevated 6-fold in IFNγR^−/−Rag^−/− mice (Fig. 5B). As judged by CD62L expression, WT and IFNγR^−/− OT-I cells contained similar percentages of T_EM-like cells in both hosts at day 12 (Fig. 6C) and day 30 (Fig. 6G) after peptide vaccination. Nevertheless, the generation of T_EM was again more pronounced in IFNγR^−/−Rag^−/− mice (Fig. 6C, 6G). Furthermore, both OT-I populations contained similar frequencies of IFN-γ–producing cells, irrespective of the host and time point (Fig. 6D, 6H). It is important to note that IFN-γ production was less pronounced in immunized IFNγR^−/−Rag^−/− mice (Fig. 6D, 6H). Hence, increased rates of CD62L downregulation in immunized IFNγR^−/−Rag^−/− mice coincide with impaired effector function but are not affected by IFN-γR signaling in OT-I cells. In conclusion, IFN-γR signaling in CD8⁺ T cells does not affect memory cell numbers and differentiation in lymphopenic recipients if a high-affinity TCR ligand is provided. However, if only low-affinity self-peptide–MHC complexes are available, the lack of IFN-γR signaling in CD8⁺ T cells is associated with the more efficient generation of IFN-γ–producing T_CM-like cells without affecting cell numbers.

IFN-γ regulates multiple aspects of CD8⁺ T cell homeostasis. On the one hand, it promotes CD8⁺ T cell expansion and memory formation (25, 26). On the other hand, it induces contraction of the effector CD8⁺ T cell pool (23, 27), partially via its proapoptotic effects (18–20). Hence, agonistic and antagonistic effects of IFN-γ must be balanced to generate functional CD8⁺ T cell responses during the course of infection.

However, the cellular interactions used by IFN-γ are controversial. It was suggested that the direct action of IFN-γ on CD8⁺ T cells promotes their expansion and subsequent memory formation (25, 26). This direct, growth-promoting effect of IFN-γ could not be observed in other experimental systems (28, 29). In line with the latter findings, we showed in this paper that IFN-γR signaling in OT-I cells is dispensable for their expansion and functional maturation in cAg-immunized lymphoreplete mice (Fig. 1A–C). From previous experiments we know that LPS induces the production of IFN-γ by NK cells (17), which is easily detectable in the serum (Fig. 2A). Nevertheless, we cannot exclude that in our experimental system, IFN-γ is produced at levels that are still too low to directly promote OT-I expansion and that, for example, LPS-induced cytokines circumvent the need for IFN-γR signaling in CD8⁺ T cells. Because the degree of IFN-γ induction and the cytokine pattern, in general, are pathogen-dependent (30), it is reasonable to assume that the relative impact of direct and indirect IFN-γ effects on CD8⁺ T cell expansion vary between experimental systems, similar to what was shown for type I IFNs (30). Nevertheless, it is important to stress that IFN-γ levels are high enough in our experimental system to counterregulate OT-I expansion via its action on host cells (Fig. 2). For example, CD11b⁺ cells counterregulate the generation of effector and memory OT-I cells in an IFN-γR–dependent fashion (Fig. 4B, 4D). The lack of IFN-γR signaling in CD11b⁺ cells directly affects CD8⁺ T cell priming. In cocultures with IFNγR^−/− CD11b⁺ cells, OT-I expansion and CD62L downregulation were more pronounced compared with cocultures with WT CD11b⁺ cells (Fig. 4B, 4C). Importantly, both processes seem to be uncoupled. For example, after three cell divisions, OT-I cells cocultured with IFNγR^−/− CD11b⁺ cells expressed less CD62L than their counterparts in WT CD11b⁺ cocultures (Fig. 4C). This suggests that IFN-γR signaling initiates different mechanisms in CD11b⁺ cells, which independently control cell-cycle progression and CD62L downregulation of recently primed CD8⁺ T cells. CD11b⁺ cells counterregulate T cell responses in several experimental systems (31, 32). For example, IFN-γ leads to NO production by CD11b⁺ cells, which, in turn, suppresses T cell expansion (33, 34). The fact that OT-I expansion was not impaired in LPS/SIINFEKL-immunized inducible NO synthase-deficient mice suggests that NO-mediated suppression plays only a negligible role in our system (17). It was shown recently that CD11b⁺ cells from tumor-bearing mice induce nitration of TCR-CD8 complexes on OT-I cells, thereby blocking their Ag responsiveness (35). This type of suppression was associated with direct cellular interactions and required reactive oxygen species and peroxinitrite (35). Hence, CD11b⁺ cells use multiple molecular pathways by which they influence the outcome of T cell responses (31, 32); it remains to be shown which are active in our system.

The presence of WT CD11b⁺ cells efficiently suppressed OT-I expansion in IFNγR^−/− mice (Fig. 4D) but failed to normalize the composition of the memory pool (Fig. 4E). CD11b⁺Gr1⁺ myeloid cells are potent suppressors of T cell responses (31, 32), but they are only short-lived. After adoptive transfer, most of them die or differentiate within 10 d (35, 36). This suggests that the early action of IFN-γR–competent CD11b⁺ cells is sufficient to counterregulate the magnitude of CD8⁺ T cell responses (Fig. 4B). In contrast, T_CM/T_EM abundance may be determined later in the response. For example, 10–30 d after immunization, macrophages and DCs regulate the composition of the OT-I memory pool in an IL-15Rα–dependent fashion (37).

In contrast to peptide immunization, the lack of IFN-γR signaling in OT-I cells alters T_CM/T_EM differentiation under lymphopenic conditions (Fig. 6A, 6E). Independent of whether the lymphopenic recipient expressed IFN-γR, the frequency of T_EM-like IFNγR^−/− OT-I cells was less than that of WT OT-I cells (Fig. 6A, 6E). Additionally, the frequency of IFN-γ–producing IFNγR^−/− OT-I cells was greatly elevated compared with their WT counterparts (Fig. 6B, 6F). Hence, the lack of IFN-γR signaling in CD8⁺ T cells blocks their differentiation into T_EM-like memory cells and promotes the functional maturation of the remaining T_CM-like cells. With respect to lymphopenia-associated T_CM/T_EM differentiation, host IFN-γR signaling had opposing effects. In its absence, the abundance of T_EM like-cells was increased (Fig. 6A, 6E). This was also the case for OT-I cell numbers, which were elevated ∼3-fold in IFNγR^−/−Rag^−/− mice (Fig. 5A). Importantly, IFN-γR signaling in OT-I cells did not affect their abundance in either lymphopenic host (Fig. 5A). This was also true after cAg immunization (Fig. 5B). Surprisingly, however, the frequency of T_EM-like IFN-γ–producing cells did not differ between WT and IFNγR^−/− OT-I cells under these experimental conditions (Fig. 6C, 6D, 6G, 6H). Thus, IFN-γR signaling in CD8⁺ T cells has no impact on lymphopenia-associated T_CM/T_EM differentiation and subsequent functional maturation if SIINFEKL is provided. Therefore, we concluded that IFN-γR signaling in CD8⁺ T cells affects T_CM/T_EM differentiation in a TCR ligand-dependent fashion. It remains to be shown whether differences in TCR signaling strength explain why direct growth-promoting effects of IFN-γ on CD8⁺ T cells are detectable in some (25, 26), but not other, experimental systems (28, 29).

It is important to note that fewer WT and IFNγR^−/− OT-I cells produced IFN-γ in SIINFEKL-treated IFNγR^−/−Rag^−/− mice than in Rag^−/− controls (Fig. 6D, 6H). However, this defect was not observed in lymphoreplete IFNγR^−/− mice (Fig. 3D). Hence, the nature of the lymphopenic environment determines whether high-affinity TCR ligands promote or inhibit CD8⁺ T cell function. This may have important implications for adoptive T cell therapy of cancer. Prior to adoptive transfer, cancer patients are rendered lymphopenic to improve therapeutic T cell expansion/function (38). Our data suggest that the type of pretreatment regimen and the subsequent type of lymphopenia may critically determine whether Ag recognition by therapeutic T cells promotes therapy or is counterproductive.

In summary, we showed that host IFN-γR signaling controls the number of CD8⁺ T cells and their differentiation into memory cells in response to cAg and lymphopenia. In contrast, IFN-γR signaling in CD8⁺ T cells does not affect cell numbers under either condition. However, in conjunction with lymphopenia, the lack of IFN-γR signaling in CD8⁺ T cells blocks T_EM differentiation, an effect that is circumvented by high-affinity TCR ligands. Our data provide new insights into the regulation of CD8⁺ T cell homeostasis and memory differentiation and have implications for vaccine development and adoptive T cell therapy.

We thank S. Prokosch and D. Femerling for excellent technical assistance.

Disclosures The authors have no financial conflicts of interest.