Pathogen induced inflammatory environment controls effector and memory CD8⁺ T cell differentiation

Abstract

In response to infection CD8⁺ T cells integrate multiple signals and undergo an exponential increase in cell numbers. Simultaneously, a dynamic differentiation process occurs, resulting in the formation of short-lived (SLEC; CD127^lowKLRG1^high) and memory-precursor (MPEC; CD127^highKLRG1^low) effector cells from an early-effector cell (EEC) that is CD127^lowKLRG1^low in phenotype. CD8⁺ T cell differentiation during vesicular stomatitis virus (VSV) infection differed significantly than during Listeria monocytogenes infection with a substantial reduction in EEC differentiation into SLECs. SLEC generationwas dependent on Ebi3 expression. Furthermore, SLEC differentiation during VSV infection wasenhanced by administration ofCpG-DNA, through an IL-12 dependent mechanism. Moreover, CpG-DNAtreatment enhanced effector CD8⁺ T cell functionality and memory subset distribution, but in an IL-12 independent manner. Population dynamics were dramatically different during secondary CD8⁺ T cell responses, with a much greater accumulation of SLECs and the appearance of a significant number of CD127^highKLRG1^highmemory cells, both of which were intrinsic to the memory CD8⁺ T cell. These subsets persisted for several months, but were less effective in recall than MPECs. Thus, our data shed light on how varying the context of T cell priming alters downstream effector and memory CD8⁺ T cell differentiation.

Introduction

Pathogen-specific CD8⁺ T cells are activated after interaction with their cognate antigen presented by antigen-presenting cells, such as dendritic cells, in secondary lymphoid organs. This activation results in the clonal expansion and differentiation of the minute naïve antigen-specific CD8⁺ T cell population into a larger pool of effector cytotoxic T lymphocytes necessary for the clearance of intracellular pathogens. During this processthe antigen-presenting cells canactively shape the CD8⁺ T cell response by their expression of co-stimulators and secretion of cytokines.By the peak of the CD8⁺ T cell response both memory-precursors and terminally differentiated CTLs can be identified. Originally, these two subsets were solely identified based on CD127 (IL-7Rα expression levels (1,2), but more recent studies haveused CD127 expression in concert with killer cell lectin-like receptor G1 (KLRG1)² expression (3,4). In these studies, memory-precursor effector cells (MPEC) were shown to be CD127^high KLRG1^low, while short-lived effector cells (SLEC) were CD127^low KLRG1^high in phenotype (3,4).

Interestingly, a single naïve antigen-specific CD8⁺ T cell can give rise to all the different effector and memory cell lineages observed after infection (5,6). Only recently have the factors regulating the differentiation of these subsets begun to be identified. Early work demonstrated that neither TCR- nor cytokine-mediated signals alone were sufficient for expression of KLRG1 on CD8⁺ T cells (7). More recent studies have shown that early inflammatory mediators in conjunction with TCR engagement can regulate the differentiation of the SLEC population (8). Two inflammatory mediators shown to be important in the differentiation of the SLEC population are IL-12 (3,8)and IL-2 (9-14). These cytokines work by regulating the levels of transcription factors (i.e. T-bet, Eomes, Blimp1, Bcl6) important in regulating effector and memory CD8⁺ cell differentiation (3,11,15). However, the role of other cytokines, such as IL-27 and type I interferons(16), that regulate these transcription factors in SLEC/MPEC differentiation remains unknown. Furthermore, the balance between the SLEC and MPEC differentiation seems to teeter on the metabolic status of the cells, because modulation of both mTOR and AMPK activity alters the differentiation pathway of effector CD8⁺ T cells (17-19). The mTOR pathway is crucial for integrating signals from the TCR, co-stimulatory receptors, and cytokines (20). This integration of signals seems to play a dominant role in regulating the expression pattern of transcription factors important for effector and memory CD8⁺ T cell differentiation.

During the development of vaccines an additional layer of complexity exists because in most situations a prime-boost regimen has been proposed to enhance T cell potency (21-24). This regimen works by greatly enhancing the absolute number of pathogen-specific T cells. Only recently have we begun to explore the functional consequences of multiple encounters with the same antigen. In these studies, it was demonstrated that secondary memory CD8⁺ T cells had elevated levels of granzyme B and decreased levels of CD62L and CD27 (25,26). Furthermore, global genetic analysis revealed drastic differences in memory T cells after primary through quaternary antigenic stimulation (27). However, the effector/memory differentiation dynamics in these situations has remained understudied.More importantly, whether all pathogens and vaccine vectors induce similar effector CD8⁺ T cell differentiation remains an open question. Here we demonstrate that effector CD8⁺ T cell differentiation differs substantially after vesicular stomatitis virus (VSV) and L. monocytogenes (LM) infections. These differences were tied to the composition of the inflammatory milieu induced by each infection. Inflammation not only altered SLEC/MPEC differentiation, but also had a striking effect on the functionality of the effector CD8⁺ T cell population and composition of the MPEC population by limiting the differentiation of CD62L^low T_EM cells. Additionally, multiple encounters with antigen dramatically altered SLEC/MPEC differentiation in a memory cell intrinsic manner. Thus, our data shed light on the fact that effector and memory CD8⁺ T cell differentiation is dynamically controlled and varies depending on the context of the activation, i.e. the type of priming pathogen or the number of times the cell is simulated with the same antigen.

Materials and Methods

Mice

Female C57BL/6 and B6-Ly5.2 mice between 5-8 weeks old were purchased from the National Cancer Institute, while female B6.129S1-Il12a^tm1Jm/J (p35^−/−) and B6.129S7-Ifng^tm1Ts/J (IFNγ^−/−) mice between 6-8 weeks old were purchased from Jackson Laboratories. Female B6.129X1-Ebi3^tm1Rsb/J (Ebi3^−/−) mice were provided by Dr. David Pascual (Montana State University). Female B6.129P2.Il18^tm1Aki/J (IL-18^−/−) mice were provided by Dr. Mark Jutila (Montana State University). OT-I Rag^−/− mice were bred in the University of Connecticut Health Center animal facility. OT-I mice deficient for IFNAR were provided by Dr. John Harty (University of Iowa), whereas OT-I mice deficient for IL12rβ or both IL12rβ and IFNAR were provided by Dr. Matthew Mescher (University of Minnesota). All animal protocols were approved by either the University of Connecticut Health Center or Montana State University Institutional Animal Care and Use Committee.

Pathogens, infections, and treatments

Both the recombinant VSV expressing ovalbumin (28) and recombinant LM expressing ovalbumin (29) have been previously described. For primary infections, mice were infected i.v. with either 10⁵ plaque-forming units (PFU) of VSV-ova (Indiana) or 10³ colony-forming units (CFU) of LM-ova. To study the recall response, mice were challenged with 2×10⁶ PFU of VSV-ova (New Jersey) or 5×10⁴ CFU of LM-ova. Some VSV infected mice were treated i.p. with 50μg of ODN1826 (Invivogen) given 1 hour after infection. Furthermore, some VSV infected mice were treated i.p. with two doses of 0.5μg of recombinant murine IL-12 (Peprotech) given 1 and 24 hours post-infection.

Systemic cytokine/chemokine analysis

Three groups of mice were analyzed: 1) LM infected, 2) VSV infected, and 3) VSV infected and treated with 50μg ODN1826. Peripheral blood was collected at 6, 12, 24, 48, and 72 hours. Serum was collected 3 hours later by centrifugation. Serum was then frozen at −80°C until used. Three different Luminex™-based multiplex bead assay plates were used to quantify: 1) IFNα and IFNβ (Affymetrix); 2) G-CSF, GM-CSF, IFNγ, IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12p40, IL-12p70, IL-13, IL-17, IL-23, IP-10, RANTES, and TNFα (Affymetrix); 3) IL-16, IL-21, IL-22, IL-25, IL-28B, MCP-5, MIP-3α, and MIP-3β (Millipore).

Tissue sample preparation and flow cytometric analysis

The H-2K^b tetramer containing the ovalbumin derived peptide SIINFEKL was generated as previously described (30,31). Analysis of endogenous antigen-specific CD8⁺ T cells throughout the entire immune response was conducted as has previously been described (32).

Intracellular cytokine staining

Cells were incubated with 1 μg/ml of the Ova or VSV-N peptideplus 1 μl GolgiPlug (BD Biosciences) in medium at 37°C for 4.5 h. Cells were stained for cell surface antigens on ice for 15 minutes. Cells were then fixed and rendered permeable using BD Cytofix/Cytoperm (BD Biosciences) before stainingwith antibodies to IFNγ and TNFα.

Secondary memory adoptive transfer

To generate secondary memory mice C57BL/6 mice were initially infected with 10³ CFU of LM-ova then >200 days later mice were rechallenged with 5×10⁴ CFU of LM-ova. Approximately 30 days after secondary infection mice were sacrificed and their spleens collected. CD8⁺ T cells were enriched using magnetic beads and the CD8⁺ enrichment kit and AutoMACS (Miltenyi Biotech). Cells were then stained with anti-CD11a, anti-CD44, anti-KLRG1, anti-CD127, anti-classII, and anti-CD4. Memory phenotype cells (ClassII^neg CD4^neg CD11a^high CD44^high) were sorted into three populations: 1) CD127^high KLRG1^low (MPEC), 2) CD127^high KLRG1^high (DPEC), and 3) CD127^low KLRG1^high (SLEC). Cells were sorted using a FACS Aria (BD Bioscience). After sorting, an aliquot of cells was further stained with anti-CD8α and Ova/K^b in order to enumerate the number of Ova/K^b-specific CD8⁺ T cells in each memory population. For adoptive transfer 10⁴ Ova/K^b-specific CD8⁺ T cells were transferred to naïve CD45.1⁺ C57BL/6 mice. One day later mice were infected with 10⁴ CFU of LM-ova.

Statistical analysis

Statistical significance was determined by either a paired Student’s t-test or one-way ANOVA using Prism 5 (Graphpad Software). Significance was set as any p-value less than 0.05 (p<0.05).

Results

Effector CD8⁺ T cell subset generation differs between L. monocytogenes and VSV infection

During the CD8⁺ T cell response, a large number of cytotoxic effector cells are generated. Recent studies have identified subsets of effector cells generated after lymphocytic choriomeningitis virus (LCMV) infection based on CD127 and KLRG1expression(3,4). At the peak of the response to LCMV infection the predominant CD8⁺ T cell population is CD127^low KLRG1^high, representing SLECs, while a small population of CD127^high KLRG1^low MPECs is also found. To examine whether the effector cell subsets present following VSV and LM infections were similar to those found after LCMV infection, mice were infected intravenously with 10⁵ PFU of a recombinant VSV expressing ovalbumin (VSV-ova) or 10³ CFU of a recombinant LM expressing ovalbumin (LM-ova). Subsequently, Ova/K^b-specific CD8⁺ T cells were monitored longitudinally using a recently described tetramer enrichment protocol(32,33) to examine the phenotype of early responding populations. On day 3 post-infection with either VSV or LM, the early effector CD8⁺ T cells (EEC) were primarily CD127^low KLRG1^low. Two days later, the SLEC and MPEC populations began to emerge after either infection (Figure 1A). Notably, approximately twice as many EEC remained after VSV infection as compared to LM infection.By day 8 the effect was even more dramatic, with ~75% of tetramer⁺ cells being SLECs, while only a small population of MPEC (~5%)emerged after LM infection. Comparatively, 8 days after VSV infection, 40-50% of tetramer⁺ cells remained as EEC with ~30% and ~20% of the cells being SLECs and MPECs, respectively. Over time, CD127^highKLRG1^low memory CD8⁺ T cells gradually became the dominant population after either infection, but this occurred much more rapidly following VSV infection (Figure 1A&B). Fifteen days after VSV infection, approximately 50% of the population was MPEC, while this occurred only at ~100 days after LM infection. A notable population of “double positive” effector cells (DPEC), i.e. CD127^high KLRG1^high cells, was also observed after LM infection, which was less evident during VSV infection (Figure 1A&B). In fact, ~50 days after LM-ova infection, the antigen-specificpool was distributed roughly equally between MPEC, SLEC and DPEC subsets. Quantification of the effector populations at the peak of the CD8⁺ T cell response revealed that the number of MPEC/EEC, rather than the overall number of Ova/K^b-specific CD8⁺ T cells more closely correlated to the size of the memory population (Figure S1). Thus, effector CD8⁺ T cell populations generated by VSV and LM infection substantially differed in kinetics and composition.

Figure 1. Effector cell subset differentiation differs following VSV and LM infection.

Mice were infected i.v. with either VSV-ova or LM-ova. At the indicated times the Ova/K^b-specific CD8⁺ T cells in the spleen were monitored for expression of KLRG1 and CD127. (A) Representative plots show the expression of KLRG1 and CD127 on the Ova/K^b-specific CD8⁺ T cells. Plots are representative of 4-5 mice per time-point and three independent experiments. In the upper right corner of each plot is the mean percentage of the population in each quadrant. (B) Graphical depiction of the Ova/K^b-specific effector CD8⁺ T cell sub-population dynamics following LM (left graph) or VSV (right graph) infection. Each dot represents the mean of 4-5 individual mice ± one SEM. Each color represents an effector cell sub-population: CD127^low KLRG1^low, early effector cells (black); CD127^low KLRG1^high, short-lived effector cells (red); CD127^high KLRG1^low, memory-precursor effector cells (blue); CD127^high KLRG1^high (green). These data are representative of 3 independent experiments.

Developmental potential of effector CD8⁺ T cell subsets

Although the population kinetics just described suggested that EEC may be progenitors to the other effector subsets, we wished to test the developmental potential of each of the populations. To do so, a small number of OT-I cells was transferred and the mice were infected with VSV-ova (Figure 2). Five days later EEC, SLEC, and MPEC were purified by sorting and transferred to infection-matched mice. Two days after transfer,the purified SLECs largely remained as SLEC and did not produce any MPECs. In contrast,the sortedEEC population not only generated both MPECs and SLECs,but also additional EECs. Purified MPECs developed into MPECs, as well as EECs. Thirty five days after transfer, SLEC-derived cells were undetectable; in contrast,sorted EECs primarily developed into CD127^high KLRG1^low memory CD8⁺ T cells, while transferred MPECs focused exclusively to CD127^high KLRG1^lowmemory CD8⁺ T cells. Thus, EECs have the greatest developmental potential with the ability to generate all effector CD8⁺ T cell lineages.

Figure 2. Early effector CD8⁺ T cells have the greatest development potential.

Five days after VSV-ova infection CD11a^high CD45.1⁺ OT-I CD8⁺ T cells were sorted into three populations: SLEC (CD127^low KLRG1^high), EEC (CD127^low KLRG1^low), and MPEC (CD127^high KLRG1^high). After which, 10⁴ CD45.1⁺ OT-I CD8⁺ T cells were transferred into infection matched CD45.2⁺ mouse. Either 2 or 35 days later expression of KLRG1 and CD127 was assessed on the transferred CD45.1⁺ OT-I CD8⁺ T cells in the spleen. These data are representative of 2 independent experiments, each containing 3-4 mice per group.

Universal granzyme B expression by effector CD8⁺ T cell subsets, but hierarchal loss of expression

Although the EEC population appears to be highly activated, we wanted to confirm whether all such cells had lytic potential. To do this, mice were infected with VSV-ova or LM-ova and subsequently granzyme B levels in the responding CD8⁺ T cell subsets were assessed by flow cytometry. Five days after infection each of the antigen-specific CD8⁺ T cell subpopulations contained high levels of granzyme B (Figure S2).Interestingly, only two days later (day 7 post-infection) certain effector CD8⁺ T cell subpopulations began to lose granzyme B expression in a hierarchical manner: first CD62L^high MPEC followed by CD62L^low MPEC and then the EEC population followed by the SLEC subset (Figure S2). Furthermore, there were notable differences in granzyme B expression depending on the infection. In general, VSV-specific effectors expressed less granzyme B as compared to LM-specific effectors, and the VSV-specific MPEC populations rapidly lost granzyme B expression.Thus, there appeared to be a hierarchical programming of at least granzyme B expression in the effector subpopulations.

IL-12 and IFNα/β differentially regulateSLEC differentiation after L. monocytogenes and VSV infection

Optimal activation of the CD8⁺ T cell response has been shown to be controlled by three distinct signals during priming: 1) pMHC:TCR engagement, 2) co-stimulation, and 3) cytokines (34). Recent work demonstrates that SLEC differentiation following LCMV and L. monocytogenes infection is controlled at the time of priming by the inflammatory milieu (3,9,35). Two of the cytokines shown to beintimately involved in activation, differentiation, expansion, and survival of CD8⁺ T cells are IL-12 and IFNα/β(3,36-41).To study the effects of IL-12 and IFNα/β signaling on the differentiation of SLEC and MPEC populations, equal numbersof WTand either IL12rβ^−/−, IFNAR^−/−, IL12rβ^−/−/IFNAR^−/− OT-I cellswere co-transferred into WT C57BL/6 mice (Figure 3A). Effector CD8⁺ T cell differentiation following either VSV-ova or LM-ova infection were measured in the peripheral blood. Following VSV-ova infection there was no difference observed between WT and IL12rβ^−/− OT-I cells in their ability to generate the effector subsets (Figure 3B), but a ~2-fold reduction in expansion was noted (data not shown). In contrast, following LM-ova infection IL-12rβ^−/− OT-I cells did not undergo differentiation to the SLEC population as efficiently as WT OT-I cells did, which resulted in a relative increasein EEC and MPEC (Figure 3B) and a ~5-fold reduction in expansion was observed (data not shown). Moreover, following either LM-ova or VSV-ova infection IFNAR^−/−and IL12rβ^−/−/ IFNAR^−/−OT-I cells failed to generate the SLEC population as well as WT OT-I cells (Figure3C & 3D). In all cases, EEC and MPEC subsets were increased suggesting that inflammatory signals acting at the EEC stage were driving SLEC development and that MPEC generation was the ‘default’ pathway. Furthermore, IFNAR^−/− and IL12rβ^−/− / IFNAR^−/− OT-I cells failed undergo as robust expansion after both VSV and LM infection when compared to WT OT-I cells; during VSV infection effector CD8⁺ T cell expansion was ~9-fold and ~100-fold reduced, respectively, while during LM infection effector CD8⁺ T cell expansion was ~6-fold and ~400-fold reduced, respectively (data not shown). Thus, our results indicated that both IL-12 and IFNα/βwere important factors driving SLEC differentiation following LM infection, while only IFNα/βwas important following VSV infection. However, in either case some SLEC differentiation was apparent indicating that additional factors were involved in the SLEC differentiation pathway, such as perhaps IL-2 (9,10).

Figure 3. Differential role of IFNAR and IL-12rβin effector CD8⁺ T cell differentiation following VSV and LM infection.

(A) Wild-type (CD45.1/2⁺) OT-I CD8⁺ T cells and cytokine receptor deficient (CD45.2⁺) OT-I CD8⁺ T cells were co-transferred to CD45.1⁺ mice. Mice were infected 18 hours later i.v. with either VSV-ova or LM-ova. Seven days later the Ova/K^b-specific CD8⁺ T cells in the PBL were monitored for differentiation of effector CD8⁺ T cell subsets based on KLRG1 and CD127 expression. Three sets of mixed OT-I adoptive transfers were analyzed: WT:IL-12rβ^−/−(B), WT:IFNAR^−/− (C), and WT:IL-12rβ^−/− IFNAR^−/− (D).Each bar represents the median ± one SEM. Each graph represents the mean of 4-5 mice per group and at least 2 independent experiments. Statistical significance was determined using a paired Student’s t-test (*p<0.05; **p<0.01)

Overt inflammation mediated through IL-12 enhances SLEC differentiation during VSV infection

Our previous result indicated that the inflammatory milieu induced by LM and VSV infection differed significantly.Specifically, IL-12 mediated signals appearedto be lacking during VSV infection because IL-12rβ-deficiency had minimal effect on CD8⁺ T cell differentiation (Figure 3B). To examine the early systemic inflammatory milieu induced during VSV and LM infection serum was collected 6, 12, 24, 48, and 72 hours post-infection and analyzed using a multiplex cytokine/chemokine panel (Figure S3). VSV infection induced high levels of IFNα, while LM infection did not. In contrast, LM infection induced a low level of IFNβwhich was not evident after VSV infection. LM infection was a potent inducer of Th1 pro-inflammatory cytokines, as evidenced by IL-6, IL-12p70, and IFNγ production, whereas VSV infection failed to elicit these cytokines, at least to levels that were detectable in the blood.Additionally, both IL-9 and IL-22 were increased following LM infection, but were not after VSV infection. Following both LM and VSV infection the expression of IP-10 (CXCL10), G-CSF, and IL-16 was observed.

Considering the major differences in the cytokine milieu generated in response to the different infections, it wasof interest to test whether altering the inflammatory conditionsresulted in alterations in effector cell subset development. To this end, we treated VSV infected mice with 50μg of ODN1826, a type B CpG-containing oligonucleotide which resulted in induction of IL-12p70, IL-9, and IL-22, as well as enhanced expression of IP-10 and G-CSF (Figure S3, blue lines).Treatment of mice with ODN1826 at the time of VSV infection did not alter effector CD8⁺ T cell expansion (Figure 4A). However, ODN1826 treatment didresult in significantly greater SLEC differentiation as compared to control mice (Figure 4B&C), which the result of a loss of approximately half the EEC and MPEC subsets. Treatment ofmice with a control oligonucleotide lacking the CpG motif or ODN1584, a type A CpG-containing oligonucleotide that does not induce IL-12p70 expression, did not alter the effector CD8⁺ T cell differentiation pathway (data not shown). Since IL-12 has been shown to be important in the induction of T-bet and subsequent SLEC differentiation(3), we next tested whether IL-12 was necessary for the ODN1826-induced SLEC differentiation, as well as whether IL-12 was sufficient to promote SLEC formation. To test this hypothesis C57BL/6 and p35^−/− mice were either left untreated, treated with 50μg of ODN1826, or treated with recombinant murine IL-12 at the time of VSV-ova infection. Infection of C57BL/6 and p35^−/− mice resulted in similar effector CD8⁺ T cell expansion and differentiation (Figure 4A-C). Treatment of C57BL/6 mice with ODN1826 significantly enhanced SLEC differentiation (Figure 4B&C). However, ODN1826 treatment of p35^−/− mice did not result in augmentation of SLEC differentiation (Figure 4B&C), indicating that IL-12 was necessary for the enhanced SLEC differentiation following ODN1826 administration. Treatment of C57BL/6 mice with recombinantmurine IL-12 at the time of VSV infection resulted in an enhancement of SLEC differentiation (Figure 4B&C), demonstratingthat IL-12 was sufficient for enhancing SLEC differentiation after VSV infection.Thus, the limited SLEC differentiation observed during VSV infection appears to be due to a lack of IL-12 mediated events which can be mimicked with IL-12-inducing adjuvants.

Figure 4. ODN1826 enhances SLEC differentiation through an IL-12 and Ebi3 dependent mechanism.

(A-C) C57BL/6 and p35^−/− mice were infected i.v. with VSV-ova. After viral infection, some mice were treated with either 50μg ODN1826 (within 2 hours) or 0.5μg rIL-12 (two doses, 1 hour and 24 hours). Seven days later the VSV-N/K^b-specific CD8⁺ T cells in the spleen were identified by tetramer staining. Absolute number of VSV-N/K^b-specific CD8⁺ T cells (A), while effector CD8⁺ T cell differentiation was then assessed based on KLRG1 and CD127 expression by flow cytometry (B) and each effector sub-population was quantified as a percentage of the tetramer⁺ CD8⁺ T cell population (C). (D-E) C57BL/6 and Ebi3^−/− mice were infected i.v. with VSV-ova. After viral infection, some mice were treated with either 50μg ODN1826 (within 2 hours). Seven days later the Ova/K^b-specific CD8⁺ T cells in the spleen were identified by tetramer staining. Absolute number of Ova/K^b-specific CD8⁺ T cells (D) while effector CD8⁺ T cell differentiation was then assessed based on KLRG1 and CD127 expression by flow cytometry (E) and each effector sub-population was quantified as a percentage of the tetramer⁺ CD8⁺ T cell population (F). For panels B &E representative contour plots show the expression of KLRG1 and CD127 on the antigen-specific CD8⁺ T cells. In the upper right corner of each plot is the mean percentage of the population in each quadrant, which is graphed in panels C & F. Statistical significance was determined using a one-way ANOVA (*p<0.05 and **p<0.01). These data are representative of 4-5 mice per group and two independent experiments.

IL-27/Ebi3 contributes to SLEC differentiation during VSV infection

Previous studies have shown that SLEC differentiation is highly dependent on T-bet (3). Infection of Tbx21^−/− mice with VSV resulted in severely impaired SLEC differentiation (data not shown), but SLEC differentiation during VSV infection occurs independently of IL-12 (Figure 4B&C), which is the prototypic inducer of T-bet expression (42). Thus, it was of interest to test whether other cytokines that induced T-bet expression could regulate effector cell subset development. Other cytokines known to regulate T-bet expression include IL-18 (43), IL-27 (44), and IFNγ(45,46). To this end, we infected C57BL/6, IL-18^−/−, IFNγ^−/−and Ebi3^−/− mice with VSV-ova. IL-18 had minimal effect on the differentiation of effector CD8⁺ T cells during VSV infection with or without ODN1826 treatment (Figure S4A). Also, IFNγ deficiency had only a slight impact on SLEC differentiation after ODN1826 treatment (Figure S4B). More interestingly, while effector CD8⁺ T cell expansion was largely independent of Ebi3 (Figure 4D), differentiation of SLECs was markedly impaired in Ebi3^−/− mice, with nearly a 2-fold reduction in SLEC differentiation (Figure 4E&F). Simultaneously, there was an increase in the proportion of MPECs found, while EECs remained largely unchanged (Figure 4E&F). Next, we examined whether Ebi3 expression was necessary for the ODN1826-mediated enhancement of SLEC differentiation after VSV infection. To this end, C57BL/6 and Ebi3^−/− mice were infected with VSV-ova and one day later treated with 50μg of ODN1826. Like our previous results, treatment of C57BL/6 mice with ODN1826 significantly enhanced SLEC differentiation (1.73-fold increase), while decreasing the proportion of EECs and MPECs (Figure 4E&F). Similarly, ODN1826 treatment enhancedSLEC differentiationwas still observed in Ebi3^−/− miceas compared to controls (1.79-fold increase), but the overall proportion of SLECs will still significantly decreased compared to ODN1826 treated C57BL/6 mice(Figure 4E&F). Thus, Ebi3, and therefore likely IL-27, playeda prominent role in the generation of SLECs and furthermore appeared to actindependently of IL-12.

ODN1826 treatment enhances effector CD8⁺ T cell cytokine production during VSV infection

Generating polyfunctional effector and memory CD8⁺ T cells provides optimal protective capabilities (47).Previous results show that after LM infection the effector CD8⁺ T cell population is largely polyfunctional, producing both IFNγ and TNFα and producing high levels of IFNγ on a per cell basis (48).In contrast, after VSV infection the effector CD8⁺T cell population appeared to be of lower quality (data not shown). Therefore, we tested whether enhancing the inflammatory milieu during VSV infection by treatment with ODN1826 would enhance the functionality of the responding CD8⁺ T cells. To test this we treated C57BL/6 or p35^−/− mice with PBS, 50μg ODN1826, or 1μg rIL-12. Seven days later we analyzed cytokine production from the effector CD8⁺ T cells found in the spleen by intracellular cytokine staining after a 4.5h re-stimulation with peptide. VSV infection of C57BL/6 or p35^−/− mice induced a substantial population of IFNγproducers but only ~25% of these produced TNFα (Figure 5). When C57BL/6 mice were treated with either ODN1826 or rIL-12 the quality of the effector CD8⁺ T cells improved as judged by production of higher levels of IFNγ on a per cell basis (Figure 5A) and ~60% of the IFNγ⁺ cells producing TNFα (Figure 5B). Surprisingly, the functionality of the effector CD8⁺ T cells was similarly enhanced in ODN1826-treated p35^−/− mice. Thus, while IL-12 was partially sufficient for increasing the functionality of the effector CD8⁺ T cells induced by VSV infection, IL-12 was not necessary for the enhancement of effector CD8⁺ T cells function following ODN1826.

Figure 5. ODN1826 enhances effector CD8⁺ T cell cytokine functionality through an IL-12 independent mechanism.

C57BL/6 and p35^−/− mice were infected i.v. with VSV-ova. After viral infection, some mice were treated with either 50μg ODN1826 (within 2 hours) or 0.5μg rIL-12 (two doses, 1 hour and 24 hours). Seven days after infection the potential of CD8⁺ T cells to produce IFNγ and TNFα was assessed by intracellular cytokine staining. Either IFNγ mean fluorescent intensity (A) or the percentage of IFNγ⁺ TNFα⁺ cells (B) was examined. For each panel a representative FACS plot showing the typical cytokine profile and gating is shown. Graphs show the median for 4-5 mice per group and are representative of at least three independent experiments. Statistical significance was determined using a one-way ANOVA and a Newman-Keuls post-test (*p<0.05; **p<0.01; ***p<0.001)

ODN1826 treatment limits formation of CD62L^loweffector-memory precursors during VSV infection

Memory CD8⁺ T cells are a heterogeneous population. Based on CD62L and CCR7 expression at least two populations of memory cells have been described (49,50). These populations are centralmemory cells that express both CD62L and CCR7, while effector memory cells lack expression of these proteins(50). Our previous results show that following VSV infection many more effector-memory CD8⁺ T cell precursors are generated as compared to LM infection(51). Here we examined whether treatment of VSV infected mice with ODN1826 altered the distribution of T_CM and T_EM precursors, as well as if IL-12 was necessary and sufficient for such alterations. To test this we treated C57BL/6 or p35^−/− mice with PBS, 50μg ODN1826, or 1μg rIL-12.Seven days later we analyzed the proportion of antigen-specific CD62L^highMPECswithin the lymph nodes (Figure 6A & B), but similar effects were also observed in the spleen (data not shown). The percentage of CD62L^high MPECs was significantly enhanced after CpG treatment and this effect was partially IL-12 independent (Figure 6B). IL-12 injection did not enhance CD62L^highMPEC development. However, the absolute number of CD62L^highMPECs was not significantly changed irrespective of the treatment (Figure 6C), while the absolute number of CD62L^low MPECs after VSV infection was markedly lower in C57BL/6 mice treated with ODN1826, but not rIL-12 (Figure 6D).The effect of CpG treatment was again largely IL-12 independent. The differences in total numbers of cells compared to percentages can be explained by our findings showing that CpG treatment reduced the overall MPEC population (Figure 4). Nevertheless, within the MPEC compartment, CpG treatment enhanced CD62L expression in an IL-12 independent manner, similar to the enhancement of cytokine production by CpGtreatment (Figure 5).

Figure 6. ODN1826 limits CD62L^low MPEC differentiation in an IL-12 independent mechanism.

C57BL/6 and p35^−/− mice were infected i.v. with VSV-ova. After viral infection, some mice were treated with either 50μg ODN1826 (within 2 hours) or 0.5μg rIL-12 (two doses, 1 hour and 24 hours). Seven days later the Ova/K^b-specific CD8⁺ T cells in the spleen were identified by tetramer staining. Memoryprecursor CD8⁺ T cells were identified based on KLRG1 and CD127 expression. The proportion of CD62L^highMPECs was assessed (A&B). Additionally, the absolute number of CD62L^high MPEC (C) and CD62L^low MPEC (D) were calculated. Each dot represents an individual mouse from three independent experiments. Statistical significance was determined using a one-way ANOVA and a Newman-Keuls post-test (*p<0.05; **p<0.01; ***p<0.001)

Effector cell population dynamics differ between primary and secondary CD8⁺ T cell responses

One strategy that has commonly been used to improve vaccine efficacy is a prime-boost regimen, in which the same antigen is given repeatedly over time typically in different vaccine vectors(21-24). Only recently have the functional outcomes of this prime-boost protocol been explored; repeated encounters with antigen resulted in memory populations that were more cytotoxic and less able to proliferate to subsequent antigen encounter (25-27). The reason(s) for these differences have not been determined. To address how repeated encounters with the same antigen impact effector and memory differentiation, we compared the CD8⁺ T cell response in mice infected with 10³ CFU of LM-ova (1° LM) or 10⁵ PFU of VSV-ova (1° VSV) to that of memory mice (>90 days after initial infection) challenged with 5×10⁴ CFU of LM-ova (2° LM-LM or 2° VSV-LM) or 2×10⁶ PFU VSV-ova (2° LM-VSV or 2° VSV-VSV). After infection, mice were serially bled and the Ova/K^b-specific CD8⁺ T cell response was analyzed. Surprisingly, the composition of the secondary responding populations was drastically different than what was observed during the primary response. This was especially evident during secondary VSV infection (Figure 7A). In all cases, the KLRG1^high population predominated for substantially longer in the secondary response(Figure 7A). This resulted in a much more protracted appearance of MPEC in the secondary response. As compared to the primary response, the recall response was comprised of a much greater population of CD127^high KLRG1^high (DPEC) cells. Even at ~150 days aftersecondary infection, ~60-70% of the memory cells expressed KLRG1 and ~40% of those cells co-expressed CD127(data not shown). Thus, in the secondary response CD127 expression and the absence of KLRG1 did not correlate with survival.

Figure 7. CD8⁺ T cell-intrinsic enhancement of KLRG1 expression during secondary infection.

To compare the primary and secondary immune responses, naïve and memory C57BL/6 mice were infected with either VSV-ova or LM-ova. At either the peak of the CD8⁺ T cell response (d5 secondary infection or d7 primary infection) or 2 months post-infection, mice were bled and the Ova/K^b-specific CD8⁺ T cells were identified by tetramer staining and their expression of KLRG1 and CD127 was assessed (A). To determine if the enhanced KLRG1 expression was intrinsic to the memory CD8⁺ T cell, we co-transferred VSV-induced memory OT-I CD8⁺ T cells (CD45.1/2⁺) with naïve OT-I CD8⁺ T cells (CD45.2⁺). One day later those mice were infected with 10⁶ PFU of VSV-ova and both the 1° and 2° effector CD8⁺ T cell populations were examined for expression of KLRG1, CD127, and CD62L (B).These data are representative of 2 independent experiments, each containing 4-5 mice per group.

We next tested whether the enhanced SLEC differentiation observed in the secondary response was cell intrinsic or the result of possible changes in the inflammatory milieu during the secondary infection. For these experiments we utilized the VSV model, as the difference between the primary and secondary responses was mostdramatic (Figure 7A). Experimentally, 10⁴ memory OT-I cells (CD45.1/2⁺) were co-transferred with 10⁴ naïve OT-I (CD45.1⁺) to naïve C57BL/6 mice that were infected one day later with VSV-ova. Analysis of both the naïve and memory OT-I cells revealed that a much greater proportion of the secondary CD8⁺ T cells expressed KLRG1 than did the cells responding for the first time (Figure 7B). In addition, many more primary cells expressed CD62L than did reactivated memory cells (Figure 7B). These data implied that memory CD8⁺ T cells had undergone genetic modifications that predisposed the cells to differentiate toward a SLEC phenotype upon secondary encounter with antigen, while the primary responders were largely influenced by environmental factors.

Recall and differentiation potential of the individual secondary memory CD8⁺ T cell subsets

Relatively little is known about the role of the effector/memory subsets in recall responses, especially with respect to the CD127^high KLRG1^high(DPEC) population. To test the capabilities of these three populations of secondary memory cells, memory phenotype CD8⁺ T cells (CD11a^high CD44^high) from CD45.2⁺ C57BL/6 mice previously infected and challenged with LM, were separated into CD127^high KLRG1^low, CD127^high KLRG1^high, and CD127^low KLRG1^high subsets. After sorting, 10⁴Ova/K^b-specific CD8⁺ T cells of each population were transferred individually into naïve CD45.1⁺ C57BL/6 mice (Figure 8A). One day later mice were infected with 10⁴ CFU of LM-ova and were subsequently bled at the indicated times to monitor expansion and effector cell differentiation. Each memory subsetgenerated a readily detectible effector CD8⁺ T cell population in the blood, indicating that each had undergone substantial expansion. CD127^high KLRG1^low memory cells mounted a robust tertiary expansion, while CD127^low KLRG1^high memory cells had a substantially diminished ability to undergo expansion (Figure 8B). Interestingly, CD127^high KLRG1^highmemory cells exhibited anintermediate abilityto undergo expansion upon re-encounter with antigen (Figure 8B).

Figure 8. Distinction recall potential and differentiation of secondary memory CD8⁺ T cell populations.

Secondary memory CD8⁺ T cells (CD45.2⁺) were sorted into SLEC (KLRG1^high CD127^low), DPEC (KLRG1^high CD127^high), and MPEC (KLRG1^low CD127^high) to >98% purity and 10⁴ cells of each phenotype were adoptively transferred to naïve CD45.1⁺ mice. One day later those mice were challenged with 10⁴ CFU of LM-ova (A). The recall kinetics (B) and subsequent tertiary effector CD8⁺ T cell differentiation (C) of the Ova/K^b-specific memory CD8⁺ T cell sub-populations was tracked in the PBL using CD45.2 expression to identify the adoptively transferred cells and KLRG1 and CD127 expression to monitor effector cell differentiation. In the upper right corner of each plot is the mean percentage of the population in each quadrant. These data are representative of 2 independent experiments, each containing 4-5 mice per group.

More intriguing was the subsequent effector cell differentiation from each purified memory subset. Recall of the CD127^low KLRG1^high population resulted in expansion of effector CD8⁺ T cells that exclusively retained the same phenotype (CD127^low KLRG1^high) even into memory (Figure 8C). When the CD127^high KLRG1^high memory population was recalled effector CD8⁺ T cells were largelyCD127^low KLRG1^high in phenotype; with time a CD127^high KLRG1^highmemory population emerged, but CD127^high KLRG1^lowcells never emerged (Figure 8C). Strikingly, only the CD127^high KLRG1^low memory population appeared to contain true ‘memory stem cells’ which were capable of re-generating all effector/memory cell subsets following re-challenge. The effector CD8⁺ T cell response generated by the sorted CD127^high KLRG1^low population was largely dominated by cells of CD127^low KLRG1^high phenotype, but small numbers of the other three effector CD8⁺ T cell populations could be found. With time CD127^high KLRG1^high and CD127^high KLRG1^lowphenotype cells increased in frequency (Figure 8C), similar to what was observed during the secondary immune response in intact mice (Figure 7A). Nonetheless, even by ~200 days after challenge only a small fraction of the cells were of the IL-7R^highKLRG1^low phenotype (~10%). Thus, we have demonstrated the importance of generating large numbers of CD127^low KLRG1^high CD8⁺ T cells during prime-boost vaccination protocols in order to generate a memory pool which has ‘stem-cell’-like properties

Discussion

Immunological memory is the foundation of vaccination. Understanding the factors governing and regulating the differentiation of the memory pool are important for understanding vaccine efficacy or failure. Here we demonstrate that early after infection, naïve antigen-specific CD8⁺ T cells become activated and form an EEC pool that is CD127^low KLRG1^low in phenotype. These EECs then differentiate into MPECs (CD127^high KLRG1^low) or SLECs (CD127^low KLRG1^high). Further differentiation of the EECs was dependent on the context of the immune response. First, the inflammatory environment induced during thedifferent infections dramatically altered effector CD8⁺ T cell differentiation, effector cytokine production, and memory subset development. Second, multiple encounters with antigen altered effector CD8⁺ T cell population dynamics through a memory cell-intrinsic manner. Lastly, the make-up of the effector/memory population at the time of re-challenge dramatically impacted the dynamics of the subsequent immune response.

The origin of the memory CD8⁺ T cell population has long been debated. It is thought that memory cells must go through an effector cell stage (52-54). This concept is further supported by single cell adoptive transfer (5) and cellular bar-coding (6) experiments, which demonstrate that each naïve CD8⁺ T cell has the potential to generate all the effector and memory CD8⁺ T cell lineages described. Our data further support that memory CD8⁺ T cells pass through an obligatory effector state as all effector CD8⁺ T cell subsets expressed granzyme B on day 5 post-infection. However, with time effector CD8⁺ T cells lost granzyme B expression in a hierarchal manner (T_CM MPEC → T_EM MPEC → EEC → SLEC), which supports the progressive differentiation model of CD8⁺ T cell differentiation (55-57).The hierarchal loss of granzyme B expression is likely tied to the responsiveness of those populations to IL-2. IL-2 has been shown to stabilize granzyme B and perforin expression in effector CD8⁺ T cells (11); furthermore, we have shown that CD25/IL-2rα is expressed at higher levels on EECs and SLECs when compared to MPECs (9). Additionally, granzyme B levels were substantially lower after VSV infection when compared with LM infection. The lack of robust Th1 inflammation during VSV infection might explain why granzyme B levels were lower, because IL-6 and IL-12 enhance granzyme B expression in CD8⁺ T cells (11,58). Decreased expression of granzyme B could be one of the reasons why VSV is able to persist for up to 60 days in the host (59).

The decision to become either a SLEC or MPEC teeters on the transcription factor profile of the responding CD8⁺ T cells, with T-bet and Blimp-1 enhancing SLEC differentiation, while Eomesodermin(Eomes) and Bcl6 are important in MPEC differentiation (60-62). Recent work with Th1 CD4⁺ T cells demonstrated that T-bet repressed Tcf7 (TCF-1) expression (63); furthermore Tcf7 is an important transcription factor for memory CD8⁺ T cell development (64,65).Recent work has begun to identify the factors regulating these critical transcription factors. We have demonstrated that IL-2 signaling is important for SLEC differentiation (9); furthermore, IL-2 enhances Blimp-1 expression while repressing Bcl6 expression (11). IL-12 has been shown to induce the expression of Tbx21 (T-bet), while repressing Eomesin CD8⁺ T cells resulting in the accumulation of effector cells and limiting memory development (61). Other studies have shown that IL-12 is necessary for the accumulation of KLRG1^high CD8⁺ T cells after Toxoplasma gondii infection (66) and vaccination with peptide pulsed dendritic cells (3,8), but not in other systems (67). Our data support the role of IL-12 in SLEC differentiation during strong Th1-like inflammation, i.e. LM infection or ODN1826 treatment. However, during VSV infection there was no role for IL-12 in mediating SLEC differentiation. This resulted in a large proportion of the effector CD8⁺ T cells retaining a CD127^low KLRG1^low phenotype, reminiscent of the EEC population. This was not unique to the VSV model, as intranasal infection with influenza A virus or infection with vaccinia virus by skin scarification resulted in a similar retention of early effector CD8⁺ T cells with the CD127^low KLRG1^low phenotype (data not shown). These data fit well with a previous report demonstrating that the CD8⁺ T cell response during LM infectionwas highly dependent on IL-12Rβ signal, while responses to viral infections were not (41).Nevertheless, after VSV infection ~40% of the effector CD8⁺ T cell population stillexpressed KLRG1.Here we showed that Ebi3, a component of IL-27 together with p28(68), regulated SLEC differentiation during VSV infection. IL-27 is an IL-12 cytokine family member, which is an important inducer of T-bet and IL-12rβ in both CD4⁺ and CD8⁺ T cells via a STAT1-dependent mechanism (16,44). Interestingly, IFNαcan also induce T-bet and IL-12rβ expression in naïve T cells through a STAT1-dependent mechanism (16), which fits with our observation that IFNAR^−/− CD8⁺ T cells had a reduced differentiation of SLECs. Our results might explain why after dendritic cell vaccination CpG DNA induced a more pronounced SLEC differentiation than IL-12 or IL-12/IFNγ treatment (3,8), because this reductionist system lacked IL-27 and/or type I interferons. In our system, we demonstrated that IFNγplayed only a minimal role, which contradicts previous work which suggests IFNγ enhances SLEC differentiation (8,69,70). This difference is likely due to the magnitude of IFNγ expression in those systems.In our hands use of ODN1826 as an adjuvant induced minimal IFNγ expression, whereas the other groups utilized live LM as an inflammatory adjuvant which induces robust levels of IFNγ.

Thus, our data together with published results support a modelin which high levels of inflammatory cytokines drive maximal SLEC differentiation while limiting MPEC generation. However, depending on the pathogen or vaccination protocol only a subset of these inflammatory cytokines may be expressed thus dampening SLEC differentiation and enhancing the number of MPECs generated (Figure 9A). Specifically,early IL-27 and IFNα expressionenhances T-bet and IL-12rβ expression in responding CD8⁺ T cells, beginning the SLEC differentiation process. Activated CD8⁺ T cells then respond to IL-12, if present. IL-12 signaling further augments T-bet expression in the responding CD8⁺ T cells enhancing SLEC differentiation. IL-12 signaling also amplifies CD25 expression on the responding CD8⁺ T cells (11), which promotes the expansion and differentiation of more SLECsthrough IL-2 signaling. Furthermore, IL-12 enhances IFNγproduction leading to a positive feedback loop of IFNγ on T-bet expression (71); thus, IFNγ can enhance SLEC differentiation (8) and/or expansion (72)in CD8⁺ T cells if IFNγ is highly expressed during the immune response. Subsequent survival of the SLEC population is highly sensitive to the balance of IL-15 and TGFβ, which promote their survival and death, respectively (73). A shift of this balance towards TGFβ might explain why CD8⁺ T cells induced by VSV infection equilibrate to a memory phenotype more quickly.

Figure 9. Model of SLEC/MPEC differentiation under different inflammatory conditions.

(A) The proportion of effector CD8⁺ T cells that differentiate into SLECs is highly dependent on the inflammatory milieu induced by infection or vaccination. Under conditions of high inflammation, such as after LM infection, SLEC differentiation is significantly enhanced by the expression of inflammatory cytokines, which include type I interferons (IFNα/β), Ebi3 (and therefore likely IL-27), IL-12 and IFNγ (top). In contrast, when only low levels of inflammation are induced, such as following DC vaccination (70), there is limited SLEC differentiation due to limited expression of inflammatory cytokines (bottom left). Mild inflammation results in fewer SLECs being generated, due to the limited expression of inflammatory cytokines. For example, during VSV infection type I interferons (IFNα/β) and Ebi3 are expressed, but IL-12 and IFNγare not expressed early. This inflammatory cytokine profile will result in significantly fewer SLECs being generated (middle). (B) ODN1826 treatment had significant effects on effector CD8⁺ T cell differentiation, which were either IL-12 –dependent or –independent. Enhancement of SLEC differentiation was highly dependent on IL-12, which is due to the ability of IL-12 to enhance T-bet expression (3). In contrast, enhancement of effector cytokine expression by ODN1826 treatment occurred through an IL-12-independent mechanism. Also, ODN1826 shifted the T_CM/T_EM ratio in favor of T_CM generation through an IL-12-independent mechanism.

In addition to enhancing SLEC differentiation, CpG-induced inflammation limited T_EM differentiation and enhanced CD8⁺ T cell poly-functionality. In contrast to the enhanced SLEC differentiation, both of these inflammation-induced alterationsto the CD8⁺ T cell population occurred through an IL-12 independent pathway (Figure 9B). Alterations in T_CM/T_EM differentiation could be the result of changes in the levels of IL-15 and IL-2, which we have previously shown to be important in T_CM and T_EM differentiation, respectively(51). CpG treatment has been shown to enhance IL-15 expression (74), which could maintain T_CM MPEC numbers while the CD62L^low cells are driven towards terminal effectors. In addition to altering inflammatory cytokine production, CpG treatment alters expression of co-stimulatory molecules(75). Specifically,co-stimulation through OX40:OX40L interactions during vaccinia virus infectionresulted in CD8⁺ T cellswith greater functionality(76). Thus, it is likely that multiple cytokines and co-stimulatory pathways simultaneously play a crucial role in regulating effector CD8⁺ T cell differentiation.

Importantly, our studies demonstrated that multiple encounters with the same antigen had dramatically different outcomes on the CD8⁺ T cell response even when the same priming vector was utilized. This observation has important implicationsfor the development of effective prime-boost vaccination protocols. Because of the ease of genetic manipulation and ability to induce strong cellular immune responses, both L. monocytogenes and VSV have been postulated to be effective vaccine vectors for the induction of strong T cell responses (77). Our data demonstrated that during secondary and tertiary infection more SLEC phenotype cells are generated and that these KLRG1^high cells survive for prolonged periods of time. This was most dramatic following VSV infection, in which during primary infection ~30% of the effector CD8⁺ T cells were of SLEC phenotype, while during secondary VSV infection ~80% of the effector CD8⁺ T cells were of SLEC phenotype. Indeed, it might be worth considering modifying the current nomenclature since in the secondary response a substantial number of long-lived memory cells express KLRG1 either with or without CD127 expression. Previous studies demonstrate that secondary and tertiary memory populations appear more terminally differentiated, as measured by high expression of KLRG1 and low expression of CD27 and CD62L (25-27,78). Furthermore, we demonstrated that this enhanced SLEC formation was intrinsic to the memory CD8⁺ T cell, as co-adoptive transfer of naïve and memory CD8⁺ T cells into the same recipient resulted in the same phenomenon. Supporting this is the observation that in human memory CD8⁺ T cells the KLRG1 promoter region is in an open configuration (79). An alternative explanation for our observation is that memory cells express a different panel of cytokine/chemokine receptors which enables them to sense different cues and could result in enhancement of this terminally differentiated phenotype. Using a transcriptome approach Wirth et al. have found that numerous cytokine and chemokine receptors are differentially regulated in naïve and memory CD8⁺ T cells, such as IL-12rβ2, IL-2rα, CXCR3, and CCR7 (27), which are known to play important roles in regulating SLEC/MPEC differentiation(3,9,10,80). Thus, naïve and memory CD8⁺ T cells may express different cytokine receptors which could alter their ability to sense the inflammatory environment; for example, memory CD8⁺ T cells are known to be more sensitive to IL-18 (81). Consequently, it is crucial that we understand the factors regulating effector and memory CD8⁺ T cell differentiation under a range of differentiation conditions.

The heterogeneous nature of the memory populations begs the question as to which of the subpopulations are involved in responding to secondary and tertiary challenges and at what level. Since multiple encounters with the same antigens result in KLRG1^high cells with prolonged life spans we examined how both the KLRG1^high CD127^low and KLRG1^high CD127^high phenotype memory CD8⁺ T cells may contribute to subsequent immune responses. Interestingly, the KLRG1^low CD127^high, classical memory cells, had the greatest recall potential and had the ability to generate all the effector CD8⁺ T cell populations, including more memory-precursors. In contrast, both KLRG1^high memory cell populations were unable to regenerate KLRG1^low CD127^high memory-precursors, but were able to contribute to the recall response to varying degrees. KLRG1^high CD127^highphenotype memory cells underwent robust expansion, only about 2-4 fold less than the KLRG1^low CD127^high memory population; in contrast, KLRG1^high CD127^low phenotype memory cells had relatively poor recall potential, about 20-fold less than the KLRG1^low CD127^high memory population. It has been reported that Bmi-1 is necessary for optimal CD8⁺ T cell expansion and TCR-mediated Bmi-1 expression is reduced in KLRG1^high CD8⁺ T cells (82), which might explain why both KLRG1^high memory CD8⁺ T cell populations underwent less expansion than CD127^high KLRG1^low memory CD8⁺ T cells. In addition, KLRG1 may actively suppress CD8⁺ T cell proliferation, because monoclonal antibody blockade of KLRG1:E-cadherin interactions led to enhanced phosphorylated Akt, which in turn resulted in increased cyclin D, increased cyclin E, and decreased p27 resulting in more proliferation (83). Why the CD127^highKLRG1^high population is more competent for expansion than the CD127^lowKLRG1^high population remains unknown. One possibility is that IL-7 mediated signaling might partially overcome the inhibitory signals mediated by KLRG1. These findings have important implications in the development of prime-boost vaccination regimens, where the timing and formulation of boosters may be a critical factor in the generation of long-term protective immunity.