Akt signaling is critical for memory CD8⁺ T-cell development and tumor immune surveillance

Significance

Immunotherapy has emerged as an important modality for the treatment of cancer, and T-cell vaccination provides an opportunity to generate a long-lasting anticancer response. Critical to this response is the generation of memory CD8⁺ T cells, but the signaling pathways that regulate their development are incompletely defined. In the current report we focused on the serine/threonine protein kinase Akt. Although Akt is known to be activated by the T-cell antigen receptor and the cytokine IL-2, its role in T-cell immunity remains unknown. In this study we show that Akt signaling profoundly impacts memory T-cell development and the antitumor response during the memory phase. Optimizing Akt activity therefore should maximize the therapeutic effect of anticancer vaccines.

Abstract

Memory CD8⁺ T cells confer long-term immunity against tumors, and anticancer vaccines therefore should maximize their generation. Multiple memory CD8⁺ T-cell subsets with distinct functional and homing characteristics exist, but the signaling pathways that regulate their development are ill defined. Here we examined the role of the serine/threonine kinase Akt in the generation of protective immunity by CD8⁺ T cells. Akt is known to be activated by the T-cell antigen receptor and the cytokine IL-2, but its role in T-cell immunity in vivo has not been explored. Using CD8⁺ T cells from pdk1^K465E/K465E knockin mice, we found that decreased Akt activity inhibited the survival of T cells during the effector-to-memory cell transition and abolished their differentiation into C-X-C chemokine receptor 3 (CXCR3)^loCD43^lo effector-like memory cells. Consequently, antitumor immunity by CD8⁺ T cells that display defective Akt signaling was substantially diminished during the memory phase. Reduced memory T-cell survival and altered memory cell differentiation were associated with up-regulation of the proapoptotic protein Bim and the T-box transcription factor eomesodermin, respectively. These findings suggest an important role for effector-like memory CD8⁺ T cells in tumor immune surveillance and identify Akt as a key signaling node in the development of protective memory CD8⁺ T-cell responses.

In a typical immune response to an acute infection, naive CD8⁺ T cells expand and differentiate into effector cells that target infected cells for destruction. Following elimination of the pathogen, the majority (90–95%) of effector cells die by apoptosis, and the remaining cells mature into memory cells. The mechanisms that regulate memory CD8⁺ T-cell generation are not fully defined. However, several lines of evidence indicate that increasing the intensity of inflammation skews differentiation toward short-lived effector cells (SLECs) at the expense of memory precursor effector cells (MPECs) (1–4). The memory CD8⁺ T-cell pool consists of populations of cells that vary in a number of characteristics, including tissue trafficking, effector function, and recall responses. Initially two memory cell subsets were identified based on the expression of CD62L (l-selectin) and C-C chemokine receptor 7 (CCR7) (5–7). Thus, central memory T cells (T_CM), which express CD62L and CCR7, localize preferentially to lymph nodes (LNs) and exhibit strong recall responses as well as robust IL-2 production. In contrast, effector memory T cells (T_EM), which lack CD62L and CCR7, traffic preferentially to nonlymphoid tissues, possess heightened cytotoxicity toward target cells, and display a limited proliferative recall capacity (5–7). More recently, Hikono et al. (8) identified distinct memory CD8⁺ T-cell subsets based on the expression of C-X-C chemokine receptor 3 (CXCR3), CD27, and an activation-associated glycoform of CD43. These markers further divide circulating T_CM and T_EM cells into subsets that differ in homeostatic proliferation, persistence, granzyme B expression, production of IL-2, and recall responses. CD27^hiCXCR3^hiCD43^lo memory CD8⁺ T cells persist longer and generate better recall responses than CD27^loCXCR3^loCD43^lo memory cells, but the latter subset confers the most efficient protective immunity against infection with Listeria monocytogenes or vaccinia virus (9).

The signaling pathways that regulate the development of memory T-cell subsets are not well understood, but knowledge of these pathways is important for the development of more efficacious vaccines against tumors and intracellular pathogens. In the current report we have focused on the serine/threonine protein kinase Akt. Akt regulates nutrient uptake and cellular metabolism in many cell types but appears to be dispensable for these processes in mature CD8⁺ T cells (10). Akt regulates the T-cell receptor (TCR) and IL-2 transcriptional programs that control the expression of cytolytic molecules, adhesion receptors, and cytokine and chemokine receptors that distinguish effector cells from naive and memory cells (11, 12). It therefore has been suggested that Akt activity promotes the differentiation of naive CD8⁺ T cells into cytotoxic T cells (CTLs) at the expense of memory cells (11, 12). However, this view is based on findings derived from in vitro studies, and therefore the true impact of physiological Akt activation on CD8⁺ T responses in vivo remains unknown. A critical step for Akt activation is the phosphorylation of Thr308 by phosphoinositide-dependent kinase 1 (PDK1). This event is mediated by the tethering of Akt and PDK1 to the plasma membrane via their pleckstrin homology (PH) domains (13). The PH domains of Akt and PDK1 bind to the lipid second messenger phosphatidylinositol 3,4,5-trisphosphate (PIP₃), the product of the class I PI3Ks. Hence, knockin mice that express a mutant form of PDK1 (PDK1-K465E) that cannot bind to PIP₃ have strongly reduced Akt activation, but other PDK1 targets remain unaffected (13, 14). Because low levels of Akt activity remain in PDK1-K465E mice, thymic development of T cells, which is absent in Akt knockout mice (15), proceeds normally (14). Thus, PDK1-K465E knockin mice provide a powerful model for investigating the role of Akt signaling in mature T cells. Using CD8⁺ T cells from PDK1-K465E knockin mice we now show that, contrary to the currently held view, Akt signaling is critical for the development of CD8⁺ memory T cells.

Results

Reduced Akt Signaling Does Not Impede the Generation of Primary Effectors.

Analysis of TCR-mediated phosphorylation of Akt at Thr308 in PDK1-K465E and PDK1-WT CTLs confirmed the reduced ability of PDK1-K465E OT-I cells to activate Akt (Fig. S1A). In addition, mechanistic target of rapamycin complex 1 (mTORC1) activity, as assessed by the levels of phospho-S6 at Ser235/236, was still detectable in PDK1-K465E CTLs (Fig. S1B), consistent with previous findings demonstrating that mTORC1 activation in CD8⁺ T cells can be induced independently of Akt activation (16). Next we investigated the effect of reduced Akt signaling on the CD8⁺ T-cell response following vaccination with ovalbumin (OVA) protein coadministered with agonist anti-CD40 mAb and lipopolysaccharide (LPS). This vaccination approach induces robust expansion of cytotoxic effectors and long-lived lymphoid and mucosal memory CD8⁺ T cells (17, 18). We adoptively transferred a small number of PDK1-K465E or PDK1-WT OT-I cells, which bear a TCR specific for the OVA_257–264/H-2K^b complex, into C57BL/6 mice before vaccination. At the peak of the response similar numbers of PDK1-WT and PDK1-K465E donor OT-I cells were recovered from the spleen, LNs, and blood, indicating that maximal Akt activation is not required for the expansion of CD8⁺ T cells in vivo (Fig. 1A). The impact of reduced Akt activity on OT-I T-cell effector function was also minimal, with a small but reproducible decrease in the proportion of granzyme B⁺ cells in the PDK1-K465E OT-I population compared with PDK1-WT effectors (Fig. 1B). Furthermore, frequencies of IFN-γ– and TNF-α–producing cells were similar in PDK1-WT and PDK1-K465E OT-I effectors, whereas the proportion of cells that produced IL-2 was marginally higher in PDK1-K465E effectors (Fig. 1C and Fig. S1C). Moreover, cytokine production on a per-cell basis was indistinguishable between the two cell types (Fig. S1D). Similar results were obtained when the anti-CD40/LPS adjuvant was replaced with agonist anti–4-1BB mAb/LPS or polyinosinic:polycytidylic acid (polyI:C) (Fig. S1 E–H). Finally, to determine the broader impact of the PDK1-K465E mutation on effector function and protective immunity in vivo, we compared the ability of PDK1-WT and PDK1-K465E OT-I T cells to control the growth of OVA-expressing E.G7 tumor cells. Four days after E.G7 inoculation, mice received either naive PDK1-WT or PDK1-K465E OT-I T cells followed by OVA/anti-CD40/LPS. As shown in Fig. 1D, PDK1-K465E and PDK1-WT OT-I cells were equally effective in controlling tumor growth. Thus our data collectively demonstrate that the development of primary effectors following protein vaccination is largely insensitive to suboptimal Akt activity.

Fig. S1.

(Related to Fig. 1) (A and B) PDK1-WT and PDK1-K465E OT-I splenocytes were stimulated for 2 d with OVA_257–264 and cultured in IL-2 for three additional days before restimulation with OVA_257–264 for 5, 10, or 15 min. Control PDK1-WT cells were treated with the Akt1/2 inhibitor for the whole culture. Data show levels of Akt phosphorylation at Thr308 (A) and S6 ribosomal subunit phosphorylation at Ser235/236 (B). (C–H) CD45.1 PDK1-WT or PDK1-K465E OT-I cells (10⁴ in C–E and G; 10⁵ in F and H) were adoptively transferred to CD45.2 C57BL/6 recipient mice followed by vaccination with OVA/anti-CD40 mAb/LPS (C and D), OVA/anti–4-1BB mAb/LPS (E and G), or OVA/polyI:C (F and H). Spleens were harvested on day 6 (C–E and G) or on day 5 (F and H) postpriming. C and D show representative dot plots of IFN-γ, TNF-α, and IL-2 intracellular staining (C) and IFN-γ, TNF-α, and IL-2 MFI of cytokine-expressing PDK1-WT and PDK1-K465E cells (D) following ex vivo restimulation of splenocytes with 10 pM or 1 nM of OVA_257–264. (E and F) Frequency of granzyme B⁺ cells among PDK1-WT and PDK12-K465E OT-I cells in spleen and (G and H) frequency of PDK1-WT and PDK1-K465E cells producing IFN-γ, TNF-α, or IL-2 following ex vivo restimulation of splenocytes with 1 nM of OVA_257–264 after priming with OVA/anti–4-1BB/LPS (E and G) or OVA/polyI:C (F and H). Data show the mean ± SEM of two combined experiments with n = 4 mice per group in each experiment (D and F) or one experiment with n = 3 or 4 mice per group (E, G, and H). *P ≤ 0.05, **P ≤ 0.01.

Fig. 1.

Reduced Akt signaling does not impede the generation of primary effectors. Following adoptive transfer of CD45.1 PDK1-WT or PDK1-K465E OT-I cells, congenic recipient mice were vaccinated with OVA/anti-CD40 mAb/LPS. Spleens, inguinal LNs, and blood samples were harvested 6 d later. (A) Numbers of CD45.1 tetramer⁺ PDK1-WT and PDK1-K465E cells in spleen, LNs, and blood. (B, Left) Frequency of splenic CD45.1 OT-I cells expressing granzyme B. (Right) Representative histogram overlays of the expression of granzyme B in PDK1-WT (shaded histograms) and PDK1-K465E OT-I (black line). Dotted lines show the corresponding isotypes. (C) Frequency of splenic CD45.1 OT-I cells producing IFN-γ, TNF-α, or IL-2 following ex vivo restimulation with 10 pM or 1 nM of OVA_257–264. Data in A–C show the mean ± SEM of two combined experiments with n = 4 mice per group. (D) Four days before adoptive transfer and vaccination with OVA/anti-CD40/LPS, mice received 2.5 × 10⁵ E.G7 tumor cells s.c. (Left) Mean tumor diameter ± SEM. Data are from one representative experiment of two with n = 5 or 6 mice per group. (Right) Mean tumor diameter ± SEM of two combined experiments on day 18 after tumor injection. Each symbol represents an individual mouse. **P ≤ 0.01, ***P ≤ 0.001.

Generation of Memory CD8⁺ T Cells Following Protein Vaccination Requires Akt Signaling.

To assess the impact of reduced Akt signaling on the memory CD8⁺ T-cell response, we monitored the contraction of OT-I T cells in blood following vaccination with OVA/anti-CD40/LPS. As shown in Fig. 2A, PDK1-K465E OT-I cells displayed an accelerated contraction phase compared with PDK1-WT cells. Furthermore, when equal numbers of PDK1-WT and PDK1-K465E OT-I cells were cotransferred into congenic mice, PDK1-WT cells did not rescue the rapid contraction of PDK1-K465E cells, demonstrating a cell-intrinsic defect in the ability of PDK1-K465E cells to transition from the effector to the memory stage (Fig. 2B). Despite reaching similar frequencies at the peak of the primary response, fewer PDK1-K465E OT-I T cells were recovered at the resting memory phase in both lymphoid and nonlymphoid tissues (Fig. 2C). To discriminate between tissue-localized and vasculature-associated memory CD8⁺ T cells, we performed i.v. antibody labeling as described by Anderson et al. (19). The majority of CD45.1⁺CD8⁺ T cells recovered from the lungs and liver were not protected from i.v. anti-CD8α mAb (Fig. S2A). In contrast, almost all CD45.1⁺CD8⁺ T cells isolated from the intestine were protected from i.v. antibody labeling (Fig. S2A). These experiments revealed that reduced Akt signaling had the greatest effect on vasculature-associated memory CD8⁺ T cells in the lungs and liver (Fig. S2B). Importantly, the reduction in the frequency of i.v. mAb-protected PDK1-K465E OT-I memory cells in the intestine demonstrated that Akt signaling is required for optimal accumulation of memory cells in the intestine parenchyma following vaccination with OVA/anti-CD40/LPS (Fig. S2B).

Fig. 2.

Akt signaling is required for the generation of memory cells following protein vaccination. Mice received either a single transfer (A, C, and D) or a 1:1 cotransfer (B) of PDK1-WT and PDK1-K465E OT-I cells and then were vaccinated with OVA/anti-CD40/LPS. (A) Frequency of CD45.1 tetramer⁺ OT-I cells in blood. (B) Proportion of PDK1-WT and PDK1-K465E OT-I cells among CD45.1 donor cells following cotransfer. (C) Absolute numbers of CD45.1 tetramer⁺ OT-I cells in various organs 12 wk postvaccination. LP, lamina propria. (D) Pie charts depicting the frequency of CXCR3^hiCD43^lo, CXCR3^hiCD43^hi, and CXCR3^loCD43^lo cells among donor OT-I cells 6–7 wk postpriming. Statistical differences between PDK1-WT and PDK1-K465E OT-I cells are given for each subset. Data show the mean ± SEM of one representative experiment of four with n = 5–10 mice per group (A), one experiment with n = 5 mice (B), one representative experiment of two with n = 5 mice per group (C), and combined data from two independent experiments with n = 3 mice per group (D). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001; ns; not significant.

Fig. S2.

(Related to Fig. 2) Following adoptive transfer of CD45.1 PDK1-WT or PDK1-K465E OT-I cells, recipient mice were vaccinated with OVA/anti-CD40/LPS, and spleen, inguinal LNs, blood, liver, lungs, and colon were harvested 6–7 wk later. (A, B, and E) Mice were injected i.v. with anti-CD8α APC-mAb and were culled 3 min later. Lymphocytes were isolated, stained, and analyzed by flow cytometry. (A) Representative dot plots of intravascular staining with anti-CD8α mAb, gated on CD8β⁺ CD45.1 PDK1-WT cells. Numbers in dot plots indicate the proportion of cells stained with the anti-CD8α mAb. LP, lamina propria. (B) Frequency of PDK1-WT and PDK1-K465E OT-I in the CD8β⁺ CD8α mAb i.v.-positive and the CD8β⁺ CD8α mAb i.v.-negative fractions in lungs, liver, and colon lamina propria. Each data point represents an individual mouse. N.D., not detected. (C) Frequency of CD62L^hi cells among PDK1-WT and PDK1-K465E donor OT-I cells. (D) Expression of CD62L, KLRG1, and granzyme B by PDK1-WT CXCR3^hiCD43^lo, CXCR3^hiCD43^hi, and CXCR3^loCD43^lo memory cells in spleen. Dotted lines are set against matching isotypes. (E) Representative histogram overlays of CXCR3 (Upper) and KLRG1 (Lower) expression on PDK1-WT cells stained with anti-CD8α mAb i.v (black line) or protected from i.v. staining (shaded histogram). Data show the mean ± SEM of one experiment with n = 3 mice per group, with each data point representing an individual mouse (B) or two combined experiments with n = 3 mice per group in each experiment (C). Data in A, D, and E are from one representative experiment. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

We next determined whether reduced Akt signaling selectively impairs the development of a particular subset of memory cells. This analysis was conducted 6–7 wk after immunization when PDK1-K465E OT-I cells were still readily detectable. Initially we analyzed the expression of CD62L, a marker classically used to discriminate T_CM from T_EM (5, 6). The frequency of CD62L^hi cells was higher among PDK1-K465E OT-I cells in both the spleen and LNs, suggesting that Akt activity normally restrains the generation of T_CM (Fig. S2C). However, CD62L is transcriptionally induced by Forkhead box protein O1 (Foxo1), which is inactivated by Akt phosphorylation (14, 20, 21). Thus, the increased expression of CD62L on PDK1-K465E memory cells may not truly reflect their differentiation state. We therefore used a complementary strategy based on the expression of CXCR3 and an activation-associated glycoform of CD43 to delineate the various memory T-cell subsets (8, 9). In the memory phase, PDK1-WT OT-I cells gave rise to subsets that conformed to previously defined phenotypic features obtained in disparate models, with CXCR3^loCD43^lo cells displaying the lowest amount of CD62L and the highest levels of killer cell lectin-like receptor G1 (KLRG1) and granzyme B (Fig. S2D). As depicted in Fig. 2D, the frequency of CXCR3^loCD43^lo cells was highest in nonlymphoid organs such as liver and lungs and was lowest in LNs. Intravascular staining showed that the CXCR3^lo memory cells recovered from the lungs and liver are associated with the vasculature, whereas CXCR3^lo memory cells recovered from the intestine are protected from i.v. anti-CD8α mAb and therefore localized in the parenchyma (Fig. S2E). Importantly, the percentage of CXCR3^loCD43^lo cells was noticeably reduced in the PDK1-K465E OT-I population, and this diminution was associated with a reciprocal increase in the frequency of CXCR3^hiCD43^lo cells (Fig. 2D). These data indicate that the generation of CXCR3^loCD43^lo effector-like memory cells is particularly sensitive to the reduction in Akt signaling.

To address how the diminution in resting memory cells influenced the recall response, mice were rechallenged with OVA_257–264. The magnitude of the secondary response was significantly reduced when memory T cells were derived from PDK1-K465E OT-I T cells (Fig. 3A). Additionally, PDK1-K465E secondary effectors expressed less granzyme B than with PDK1-WT cells (Fig. 3B and Fig. S3A). In the few mice that had detectable memory PDK1-K465E cells the fold-expansion of PDK1-K465E memory cells was similar to that of PDK1-WT cells, indicating that decreased Akt activity affected memory cell generation rather than the cell’s capacity to mount a secondary proliferative response (Fig. S3B). To assess if the reduced recall response impacted immune protection, we inoculated mice with E.G7 tumor cells 3 d after secondary immunization. As shown in Fig. 3C, mice that possessed memory PDK1-WT OT-I cells were better protected than those that had PDK1-K465E OT-I memory T cells. Finally, we showed that reduced Akt signaling abolished memory CD8⁺ T-cell formation in two additional protein vaccines that encompass alternative adjuvants (Fig. S3 C and D). These results demonstrate that Akt signaling is required for the effector-to-memory cell transition after protein/adjuvant vaccination.

Fig. 3.

Loss of PDK1-K465E OT-I memory cells following vaccination results in a diminished recall response. Adoptive cell transfer and vaccination with OVA/anti-CD40/LPS were performed as in Fig. 1. (A) Frequency of CD45.1 tetramer⁺ OT-I cells in blood following priming with OVA/anti-CD40/LPS and rechallenge (arrow) with OVA_257–264. (B) Frequency of CD45.1 OT-I cells expressing granzyme B cells in blood on day 4 and day 7 after rechallenge. Each data point represents an individual mouse. (C) Mice received 5 × 10⁵ E.G7 tumor cells s.c. on day 3 after rechallenge. (Left) Mean tumor diameter ± SEM. (Right) Mean tumor diameter ± SEM on day 19 after tumor injection. Each symbol represents an individual mouse. Data show the mean ± SEM of one representative experiment of three with n = 5–10 mice per group (A), one experiment with n = 3–8 mice per group (B), or one experiment with n = 10 mice per group (C). *P ≤ 0.05, **P ≤ 0.01.

Fig. S3.

(Related to Fig. 3) (A) Mice received cotransfer of PDK1-WT and PDK1-K465E OT-I cells in a 1:1 ratio followed by OVA/anti-CD40/LPS vaccine and were rechallenged at the memory phase with OVA_257–264. Data show the frequency of granzyme B⁺ cells in blood among PDK1-WT and PDK1-K465E OT-I cells on day 4 and day 7 after rechallenge. (B) Adoptive transfer of 10⁴ PDK1-WT and PDK1-K465E OT-I, vaccination with OVA/anti-CD40/LPS, and rechallenge with OVA_257–264 were performed as in Fig. 3. Data show the numbers of CD45.1 tetramer⁺ OT-I cells/mL blood before and 4 d (Left) and 7 d (Right) after rechallenge. Each symbol represents an individual mouse. Mean fold expansion ± SEM is given. Data are from one representative experiment of three. (C and D) Mice received 10⁵ (C) or 10⁴ (D) PDK1-WT or PDK1-K465E OT-I cells followed by vaccination with OVA/polyI:C (C) or OVA/anti–4-1BB/LPS (D) and then were rechallenged (arrow) at the memory phase with OVA_257–264/anti-CD40 mAb. Data show the mean ± SEM of the frequency of donor OT-I cells among CD8⁺ T cells in blood and are representative of two experiments with n = 3 or 4 mice per group in each experiment (C) or one experiment with n = 4 mice per group (D). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Vaccine-Induced PDK1-K465E OT-I Effectors Express Increased Levels of the Proapoptotic Protein Bim.

To probe the mechanism underpinning the diminished ability of PDK1-K465E OT-I effector cells to transition into the memory phase, we first assessed if reduced Akt activity affected the ratio of SLECs to MPECs as defined by differential expression of CD127 and KLRG1 (1). The frequency of CD127^hiKLRG1^lo cells, defined by this criterion as MPECs, was higher among PDK1-K465E effectors (Fig. 4 A and B). However, the use of CD127 to demarcate MPECs in this case is complicated by the fact that Il7ra is a target of Foxo1, a transcription factor inactivated by Akt (20–22). Therefore, the frequency of PDK1-K465E MPECs as identified by CD127 expression may be artificially inflated. This possibility is supported by the observation that PDK1-K465E effectors exhibited higher expression of CD127 on both KLRG1^hi and KLRG1^lo cells than did PDK1-WT effectors (Fig. S4).

Fig. 4.

Bim expression is increased in PDK1-K465E OT-I effectors. Adoptive cell transfer and vaccination with OVA/anti-CD40/LPS were performed as in Fig. 1. (A) Representative dot plots of CD127 and KLRG1 expression. (B) Frequency of CD127^hiKLRG1^lo and CD127^loKLRG1^hi cells among PDK1-WT and PDK1-K465E OT-I cells in spleen on day 6 postpriming. (C) On day 7 postvaccination, Bim and actin expression levels in purified splenic PDK1-WT and PDK1-K465E OT-I cells were analyzed by Western blotting (Left) and densitometry (Right). Bar graphs represent the ratio of Bim to actin expressed as the percent of PDK1-WT values. Data show the mean ± SEM of two combined experiments with n = 4 mice per group (B) or three combined experiments with n = 2 mice per group (C). **P ≤ 0.01.

Fig. S4.

(Related to Fig. 4) Following adoptive transfer of PDK1-WT or PDK1-K465E OT-I cells, recipient mice were vaccinated with OVA/anti-CD40 mAb/LPS. Spleens were harvested 6 d later, and CD127 expression on KLRG1^hi and KLRG1^lo PDK1-WT and PDK1-K465E OT-I cells was assessed. Data show the mean ± SEM of two combined experiments with n = 4 mice per group in each experiment. **P ≤ 0.01, ***P ≤ 0.001.

Because Bim-deficient effector CD8⁺ T cell are largely resistant to apoptosis during the contraction phase (23), we hypothesized that increased Bim protein in PDK1-K465E OT-I effectors accelerates cell death and limits the generation of memory cells. Comparing the amount of Bim protein in PDK1-WT and PDK1-K465E OT-I effectors isolated 7 d postvaccination revealed a twofold increase in its expression in PDK1-K465E OT-I cells (Fig. 4C). Thus, our findings support the notion that physiological Akt activity in effector T cells exerts prosurvival effects by limiting the expression of the proapoptotic protein Bim.

The Role of Akt Signaling in CD8⁺ T-Cell Differentiation After Infection with L. monocytogenes.

To understand the role of Akt signaling in effector CD8⁺ T-cell differentiation in a more inflammatory setting, we compared the responses of adoptively transferred PDK1-WT and PDK1-K465E OT-I cells following infection with OVA-expressing L. monocytogenes (Lm-OVA). On day 6 postinfection, the frequency of PDK1-K465E OT-I effectors in the spleen and blood was lower than that of PDK1-WT cells (Fig. 5A). This decrease was not the result of reduced proliferation of PDK1-K465E OT-I cells, because the frequency of Ki67⁺ cells in spleen was comparable in PDK1-WT and PDK1-K465E effectors (Fig. S5A), but instead was caused by preferential localization of PDK1-K465E OT-I cells in LNs (Fig. 5 A and B). CD62L expression was higher on PDK1-K465E OT-I cells (Fig. 5C) and likely contributed to redirecting their trafficking to LNs. Because reduced Akt activity results in high CD127 expression on both KLRG1^hi and KLRG1^lo cells (Fig. S5 B and C), we used KLRG1 expression alone to identify the various effector cell subsets. Disparate levels of KLRG1 subdivide the effector cell population into subsets that are enriched in either terminal effectors or precursors of T_CM (24). As expected, the frequency of KLRG1^hi cells generated after Lm-OVA infection (Fig. 5D and Fig. S5 B and D) was greater than following protein vaccination (Fig. 4A) (25). The proportion of KLRG1^hi cells was lower in the PDK1-K465E OT-I population in the spleen (Fig. 5D) and blood (Fig. S5 B and D), leading to a twofold reduction in the numbers of PDK1-K465E KLRG1^hi cells compared with PDK1-WT KLRG1^hi cells (Fig. 5D). Furthermore, although an increase in the proportion of CD127^hi cells among PDK1-K465E OT-I effectors was evident in the spleen, the numbers of PDK1-K465E and PDK-WT CD127^hi T cells were similar (Fig. 5E).

Fig. 5.

Akt signaling regulates the differentiation of KLRG1^hi effectors following L. monocytogenes infection. Mice received either a single transfer (filled symbols) or a cotransfer (open symbols) of PDK1-WT and PDK1-K465E OT-I cells and were challenged 1 d later with ΔActA-Lm-OVA. Spleen, inguinal LNs, and blood samples were harvested 6 d later, and the frequency and phenotype of OT-I cells were analyzed by flow cytometry. (A) Frequency of OT-I cells among CD8⁺ T cells. (B) PDK1-WT:PDK1-K465E OT-I cell ratio in the different tissues. (C) Frequency of CD62L^hi cells among OT-I cells in spleen. (D and E) Frequency and numbers of KLRG1^hi (D) and CD127^hi (E) cells in spleen. (F) Frequency of splenic PDK1-WT and PDK1-K465E OT-I cells producing IL-2 following ex vivo restimulation with 10 pM or 1 nM of OVA_257–264. (G) Representative histogram overlays of Eomes and T-bet expression in PDK1-WT (shaded histogram) and PDK1-K465E OT-I cells (black line). Dotted lines show the corresponding isotypes. (H) Frequency of Eomes^hi and T-bet⁺ cells among OT-I cells in spleen. Data are from one representative experiment of two with n = 3–5 mice per experiment (A and B) or are combined from two independent experiments, with each symbol representing an individual mouse (C–F and H). Mean ± SEM is shown. *P ≤ 0.05, ***P ≤ 0.001.

Fig. S5.

(Related to Fig. 5) Single transfer (filled symbols) or cotransfer (open symbols) of PDK1-WT and PDK1-K465E OT-I cells and infection with ΔActA-Lm-OVA were performed as in Fig. 5. Spleens and blood samples were harvested on day 6 postinfection. (A) Frequency of Ki67⁺ cells among donor PDK1-WT and PDK1-K465E OT-I cells in spleen following cotransfer. (B) Representative dot plot of CD127 and KLRG1 expression on PDK1-WT and PDK1-K465E donor OT-I cells in blood. (C) CD127 expression on KLRG1^hi and KLRG1^lo PDK1-WT and PDK1-K465E OT-I cells in spleen. (D) Frequency of KLRG1^hi cells among donor PDK1-WT and PDK1-K465E OT-I cells in blood. (E, Right) Frequency of granzyme B⁺ cells among donor PDK1-WT and PDK1-K465E OT-I cells. (Left) Representative histogram overlays of the expression of granzyme B in PDK1-WT (shaded histograms) and PDK1-K465E OT-I cells (black line). Dotted lines show the corresponding isotypes. (F) Frequency of CD107a⁺ cells among donor PDK1-WT and PDK1-K465E OT-I cells following ex vivo restimulation with 1 nM of OVA_257–264. (G–I) Splenocytes were restimulated ex vivo with 10 pM or 1 nM of OVA_257–264. Data show the frequency of CD45.1 PDK1-WT and PDK1-K465E OT-I producing IFN-γ or TNF-α (G), representative dot plots of IFN-γ, TNF-α, and IL-2 intracellular staining (H) and IFN-γ, TNF-α, and IL-2 mean fluorescence intensity of cytokine-expressing PDK1-WT and PDK1-K465E cells (I). Data show one representative experiment of two with n = 3–5 mice/experiment (A), the mean ± SEM of one representative experiment of two (one single transfer with n = 4 mice per group and one cotransfer with n = 5 mice) (C, D, and I), the mean of two combined experiments (one single transfer with n = 4 mice per group and one cotransfer with n = 5 mice) with each symbol representing an individual mouse (E and G), and the mean ± SEM of one experiment with n = 4 mice per group (F). *P ≤ 0.05, ***P ≤ 0.001.

Next we assessed the effect of reduced Akt signaling on effector function. Consistent with our findings in the protein vaccination setting, PDK1-K465E OT-I exhibited only a modest reduction in granzyme B expression compared with PDK1-WT effectors (Fig. S5E). Mobilization of the degranulation marker CD107a at the cell surface and IFN-γ and TNF-α production were equivalent in PDK1-WT and PDK1-K465E cells (Fig. S5 F–I). Although MPECs and SLECs exhibit equivalent cytotoxicity and a similar capacity to produce TNF-α and IFN-γ, MPECs produce more IL-2 than SLECs (1, 24). We found that the proportion of IL-2–producing cells was significantly higher among PDK1-K465E effectors, but IL-2 production on a per-cell basis was equivalent (Fig. 5F and Fig. S5 H and I), indicating that PDK1-K465E CD8⁺ T effectors were enriched in MPECs.

The T-box transcription factors T-bet and eomesodermin (Eomes) regulate the differentiation of CD8⁺ T cells into effector and memory cells (4). T-bet expression is required for the generation of KLRG1^hi cells (1), and Eomes promotes the generation and persistence of T_CM (26). Examination of the expression of T-bet and Eomes revealed that PDK1-K465E OT-I effectors express similar levels of T-bet but higher levels of Eomes as compared with PDK1-WT cells, suggesting that PDK1-K465E effectors have a superior potential to give rise to T_CM (Fig. 5 G and H).

The Role of Akt Signaling in the Generation of Memory Responses After Infection with L. monocytogenes.

To assess the impact of reduced Akt activity on the persistence of CD8⁺ T cells following infection, we first monitored the frequency of adoptively transferred PDK1-WT and PDK1-K465E OT-I cells in blood. In contrast to our observations following protein vaccination, the frequency of total memory PDK1-K465E OT-I cells generated by infection was similar to that of PDK1-WT OT-I cells (Fig. 6A). However, when the expression of KLRG1 was taken into account, we noticed a striking reduction in the generation and persistence of KLRG1^hi effector and memory CD8⁺ T cells in blood, whereas the accumulation of KLRG1^lo cells was minimally affected by the diminished Akt activity (Fig. 6B). Assessment of the ratios of PDK1-WT to PDK1-K465E memory cells in different organs revealed similar numbers of PDK1-WT and PDK1-K465E cells in the spleen, preferential localization of PDK1-K465E OT-I cells in the LNs, and decreased accumulation of PDK1-K465E cells in the liver and lungs (Fig. S6A). Furthermore, using CXCR3 and CD43 as markers that denote functionally distinct memory cell subsets, we found marked differences in the composition of PDK1-K465E and PDK1-WT OT-I memory cells. A significant reduction in the proportion of CXCR3^loCD43^lo cells was noted in the PDK1-K465E memory pool and was accompanied by a reciprocal increase in the proportion of the CXCR3^hi cells, in particular the CXCR3^hiCD43^lo subset (Fig. 6C). As a result, we observed a significant decrease in the absolute numbers of PDK1-K465E CXCR3^loCD43^lo effector-like memory cells in the spleen, liver, and lungs and an increase in the numbers of PDK1-K465E CXCR3^hiCD43^lo cells, and to a lesser extent of PDK1-K465E CXCR3^hiCD43^hi cells in spleen and LNs (Fig. 6D). Accordingly, lower numbers of PDK1-K465E cells expressing KLRG1, which is expressed predominantly on CXCR3^loCD43^lo memory cells (Fig. S2D and refs. 8 and 9), were recovered from the spleen, liver, and lungs (Fig. S6B). Similar to our observations for memory cells generated by protein vaccination, CXCR3^loCD43^loKLRG1^hi memory cells recovered from the lungs and liver were largely unprotected from i.v. antibody labeling, suggesting that these cells were associated with the vasculature of these organs (Fig. S7). When restimulated ex vivo, PDK1-K465E memory cells and PDK1-WT cells were equally capable of producing IFN-γ and TNF-α, but PDK1-K465E cells comprised a larger proportion of IL-2–producing cells (Fig. 6E and Fig. S8 A and B), consistent with the altered composition of the PDK1-K465E memory population, which is skewed toward the CXCR3^hi subset (8). Furthermore, the remaining PDK1-K465E CXCR3^loCD43^lo memory cells expressed substantially less granzyme B than PDK1-WT cells (Fig. 7A and Fig. S8C). To determine whether the loss of CXCR3^loCD43^lo cells from the memory population of PDK1-K465E T cells impacts the killing capacity of memory cells, we compared PDK1-WT and PDK1-K465E memory T cells in an ex vivo cytotoxicity assay. PDK1-WT cells were more efficient than PDK1-K465E cells on a per-cell basis in killing OVA peptide-pulsed target cells (Fig. S8D), suggesting that PDK1-WT memory cells are potentially better in conferring protective immunity in vivo.

Fig. 6.

Akt signaling in CD8⁺ T cells regulates the composition of memory cells after L. monocytogenes infection. Cotransfer of PDK1-WT and PDK1-K465E OT-I cells and infection with ΔActA-Lm-OVA were performed as in Fig. 5. (A and B) Blood samples were analyzed at the indicated time points. (C–E) Spleen, inguinal LNs, liver, and lungs were harvested 6–9 wk following infection. (A) Frequency of OT-I cells among CD8⁺ T cells in blood. (B) Numbers of KLRG1^hi (Left) or KLRG1^lo (Right) OT-I cells/mL of blood. (C) Pie charts depicting the frequency of CXCR3^hiCD43^lo, CXCR3^hiCD43^hi, and CXCR3^loCD43^lo cells among OT-I cells. Statistical differences between PDK1-WT and PDK1-K465E OT-I are given for each subset. (D) Numbers of CXCR3^hiCD43^lo, CXCR3^hiCD43^hi, and CXCR3^lo CD43^lo memory cells. (E) Frequency of PDK1-WT and PDK1-K465E OT-I cells producing IFN-γ, TNF-α, or IL-2 in spleen following ex vivo restimulation with 1 nM of OVA_257–264. Data show the mean ± SEM of one representative experiment of three, with n = 5–10 mice per experiment (A and B) or the mean of two combined experiments with n = 4–6 mice per experiment, with each symbol representing an individual mouse (E). Spleen data in C and D are representative of three independent experiments with n = 4–6 mice per experiment; data for LN, liver, and lungs in C and D are representative of two experiments with n = 4 mice per experiment. Data show the mean ± SEM; each symbol represents an individual mouse. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ns; not significant.

Fig. S6.

(Related to Fig. 6) Cotransfer of PDK1-WT and PDK1-K465E OT-I cells was performed as described in Fig. 5. Mice were challenged with ΔActA-Lm-OVA, and spleens, blood samples, inguinal LNs, livers, and lungs were harvested 6–9 wk later. (A) PDK1-WT:PDK1-K465E OT-I memory cell ratios. (B) Numbers of KLRG1^hi PDK1-WT and PDK1-K465E OT-I cells in the different tissues. Data show the mean of two combined experiments with n = 4–6 mice per experiment (A) or the mean ± SEM of one representative experiment of two with n = 4–6 mice per experiment, with each data point representing an individual mouse (B). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001.

Fig. S7.

(Related to Fig. 6) Cotransfer of PDK1-WT and PDK1-K465E OT-I cells and infection with ΔActA-Lm-OVA were performed as in Fig. 5. Fifty-three days postinfection, mice were injected i.v. with anti-CD8α APC-mAb and were culled 3 min later. Spleen, inguinal LNs, blood, liver, and lungs were harvested. Lymphocytes were isolated, stained, and analyzed by flow cytometry. (A) Representative dot plots of intravascular staining with anti-CD8α mAb gated on CD8β⁺ CD45.1 PDK1-WT or PDK1-K465E OT-I cells. Numbers in dot plots indicate the proportion of cells stained with the anti-CD8α mAb. (B and C) Representative histogram overlays of CXCR3 (B) and KLRG1 (C) expression on PDK1-WT and PDK1-K465E cells stained with anti-CD8α mAb i.v (black line) or protected from the i.v. staining (shaded histogram). (D) PDK1-WT:PDK1-K465E OT-I memory cell ratios in the CD8α mAb i.v.-positive or i.v.-negative fractions; each data point represents an individual mouse. Data are from one experiment with n = 4 mice. **P ≤ 0.01.

Fig. S8.

(Related to Figs. 6 and 7) Cotransfer of PDK1-WT and PDK1-K465E OT-I cells was performed as described in Fig. 5. Mice were infected with ΔActA-Lm-OVA, and spleens were harvested 6–9 wk postinfection. (A and B) Splenocytes were restimulated ex vivo with 1 nM of OVA_257–264. Data are shown as representative dot plots of IFN-γ, TNF-α, and IL-2 intracellular staining (A) and pie charts depicting the frequency of PDK1-WT and PDK1-K465E OT-I cells in spleen producing IFN-γ only, IFN-γ and TNF-α, or IFN-γ, TNF-α, and IL-2 (B). (C) Expression of granzyme B in PDK1-WT and PDK1-K465E memory subsets was analyzed directly ex vivo. (D) Cytotoxicity of PDK1-WT and PDK1-K465E memory cells against OVA_257–264-pulsed targets. Data are from two combined experiments with n = 4–6 mice per experiment (B), one representative experiment of two with n = 4 mice per experiment (C), and one experiment with n = 3 or 4 mice per group (D). **P ≤ 0.01, ***P ≤ 0.001.

Fig. 7.

PDK1-K465E memory cells formed after Listeria-OVA infection fail to provide long-term protection against E.G7 tumors. After adoptive transfer and infection with ΔActA-Lm-OVA, mice were challenged on day 62, without further boosting, with 5 × 10⁵ E.G7 tumor cells (s.c.). Control mice received tumor cells only. (A, Left) Dot plots show a representative examples of CXCR3 and CD43 expression on PDK1-WT and PDK1-K465E memory cells in spleen before tumor challenge. (Right) Histograms show the expression of granzyme B in CXCR3^loCD43^lo memory cells. (B) Mean tumor diameter ± SEM. (C) Kaplan–Meier analysis of survival of naive mice and mice bearing PDK1-WT or PDK1-K465E memory cells. Data are representative of two independent experiments with n = 7 or 8 mice per group. *P ≤ 0.05, ***P ≤ 0.001.

To assess the impact of the altered memory pool on protective immunity in vivo, we injected E.G7 tumor cells into mice that had been previously challenged with Lm-OVA. Delayed tumor growth was observed in mice harboring either PDK1-WT or PDK1-K465E OT-I memory cells as compared with naive mice (Fig. 7B). However, the tumor protection provided by PDK1-WT memory cells was markedly more sustained. As a result, mice bearing PDK1-WT cells survived significantly longer than mice bearing PDK1-K465E cells (Fig. 7C). These results highlight the importance of optimal Akt signaling in the generation of protective memory CD8⁺ T cells. We next assessed the frequency and phenotype of tumor-infiltrating CD8⁺ T cells (TILs). The frequencies of PDK1-WT and PDK1-K465E T cells among tumor-infiltrating CD8⁺ T cells were similar (Fig. S9A). Unexpectedly, CXCR3 was down-regulated on TILs as compared with splenic CD8⁺ T cells (Fig. S9B), thus preventing its use as a marker to delineate the memory CD8⁺ T-cell subsets. However, chemokine (C-X₃-C motif) receptor 1 (CX₃CR1), recently shown to be expressed on effector-like memory CD8⁺ T cells (27) and confirmed by us (Fig. S9C), was expressed on a subpopulation of TILs (Fig. S9D). Importantly, we show that the frequency of CX₃CR1⁺ cells was reduced among PDK1-K465E TILs compared with PDK1-WT TILs, suggesting that the optimal accumulation of tumor-associated effector-like memory cells is dependent on Akt activity (Fig. S9 D and E).

Fig. S9.

(Related to Fig. 7) Single transfer of PDK1-WT and PDK1-K465E OT-I cells and infection of recipient mice with ΔActA-Lm-OVA were performed as in Fig. 5. On day 48 postinfection mice were challenged, without further boosting, with 5 × 10⁵ E.G7 tumor cells (s.c.). When the humane end point was reached, tumors and spleens were harvested. (A) Frequency of CD45.1 PDK1-WT and PDK1-K465E cells among tumor-infiltrating CD8⁺ T cells. (B) Representative dot plots of CXCR3 and CD43 expression gated on PDK1-WT and PDK1-K465E cells in spleen and tumor. (C) Expression of CX₃CR1 on PDK1-WT memory subsets in spleen. (D) Representative dot plots of CX₃CR1 expression on PDK1-WT and PDK1-K465E cells in tumors. (E) Frequency of PDK1-WT and PDK1-K465E cells expressing CX₃CR1 in tumors. Each symbol represents an individual mouse. Data are from one experiment with n = 3–6 mice per group. *P ≤ 0.05.

Discussion

In the present study we explored how PIP₃-dependent activation of Akt impacts the CD8⁺ T-cell response elicited by vaccination or infection. The overarching conclusion from the current study is that CD8⁺ T-cell expansion and acquisition of effector function during the primary response are relatively insensitive to suboptimal Akt activity, whereas multiple aspects of protective memory cell development are highly dependent on Akt signaling.

The current study demonstrates that maximal Akt signaling is neither required for CD8⁺ T-cell expansion nor necessary for the generation of CTLs during the primary response. Although primary PDK1-K465E OT-I effector T cells express marginally lower levels of granzyme B than PDK1-WT cells, they produce similar levels of IFN-γ and TNF-α. Importantly, we show that maximal Akt activity is not required for the antitumor response during the primary effector phase, because PDK1-K465E OT-I and PDK1-WT OT-I effector cells provide equivalent protection against E.G7 tumors. Although we cannot exclude the possibility that the residual Akt activity in PDK1-K465E T cells was sufficient for the generation of primary effectors, an alternative explanation is that the primary T-cell response is controlled largely by other signaling pathways. Notably, mTORC1 activity has been shown to be necessary for optimal CD8⁺ T-cell expansion and differentiation into effector cells (28). In this context, activation of mTORC1 in CD8⁺ T cells is independent of Akt activity (Fig. S1 A and B and ref. 16). By controlling the expression of hypoxia-inducible factor 1 (HIF-1), mTORC1 enhances glucose metabolism and glycolysis and regulates the expression of perforin, granzymes, CD62L, and CCR7 (16). Therefore, there is some redundancy in the effects of Akt and mTORC1 signaling on the regulation of cytotoxicity and tissue trafficking.

A striking finding of the current study is that Akt activity in CD8⁺ T cells is required for cell survival during the transition from the effector to the memory cell stage after protein vaccination. Reduced Akt activity had profound effects on the overall number of vaccine-generated memory cells so that the magnitude of the secondary response after rechallenge was significantly diminished. The prosurvival role of Akt in CD8⁺ T cells during the contraction phase is supported by our finding that the levels of the proapoptotic Bcl2 family protein Bim are elevated in PDK1-K465E effector T cells after protein vaccination. Bim is known to promote the contraction phase of CD8⁺ effector T cells (23), and its expression in T-cell lines is up-regulated by Foxo3 (29). In this respect, recent studies have shown that deletion of Foxo3 in T cells results in an increase in T-cell survival during either the primary response or the contraction phase concomitant with reduced expression of Bim (30–32). Foxo3 is phosphorylated by Akt, a posttranslational modification that results in Foxo3 nuclear exclusion and translocation to the cytosol (29). Hence, the finding that PDK1-K465E effectors express elevated levels of Bim is consistent with the notion that reduced Akt signaling results in Foxo3-dependent up-regulation of Bim.

By extending our analysis to an infection model, we found that Akt signaling is primarily required for the differentiation of CD8⁺ T cells into CXCR3^loCD43^loKLRG1^hi effector-like memory CD8⁺ T cells. The reduction in the number of effector-like memory PDK1-K465E OT-I cells followed a decline in KLRG1^hi effectors, suggesting that the lack of Akt signaling had an early impact on the generation of this population. In addition, the frequency of IL-2–secreting cells among effector and memory cells mirrored the phenotypic changes, consistent with previous studies demonstrating that KLRGl^hi effectors and effector-like memory cells produce less IL-2 than KLRG1^lo effectors and T_CM (1, 8, 24). Because the transcription factors T-bet and Eomes have crucial roles in the generation of KLRG1^hi effectors and T_CM, respectively (4), we assessed whether reduced Akt signaling impacts their expression. T-bet expression within effectors was not altered by the reduction in Akt signaling, suggesting that Akt promotes the generation of KLRG1^hi effectors and effector like-memory T cells by an independent mechanism. In contrast, Eomes expression was increased among PDK1-K465E OT-I effectors as compared with its levels in PDK1-WT OT-I effectors. Previous work has shown that CD8⁺ effector T cells that lack Eomes generate fewer T_CM (26). However, Eomes does not appear to play a role in the generation of MPECs, because MPECs and SLECs express similar amounts of Eomes, and Eomes deficiency does not alter the SLEC/MPEC ratio (26). Thus, the higher expression of Eomes among PDK1-K465E OT-I effectors as compared with PDK1-WT OT-I cells does not simply reflect the greater frequency of MPECs but instead indicates that Akt signaling actively suppresses Eomes expression. This finding suggests that Akt signaling normally limits T_CM generation and skews differentiation toward effector-like memory T cells, at least in part through down-regulation of Eomes. Eomes expression is dependent on the transcription factor Foxo1, which is a known target for Akt-mediated phosphorylation and inactivation (20, 33). Thus, the data presented here place Akt at the apex of the CD8⁺ T-cell differentiation program that controls effector-like memory cells and T_CM.

Generation of protective, long-lasting CD8⁺ T-cell–dependent immunity is an important goal for anticancer vaccines. Here we show that Akt signaling is required for long-term protective antitumor immunity in two different settings. In the protein vaccination approach, the diminished antitumor protection was primarily the result of the vastly reduced numbers of memory PDK1-K465E OT-I cells. Reduction in secondary effector differentiation, exemplified by lower granzyme B expression upon peptide boosting, also may have contributed to the lack of antitumor activity of memory PDK1-K465E T cells. In the second approach, T cells were primed with Lm-OVA. This setting generated similar numbers of total PDK1-K465E and PDK1-WT OT-I memory cells, but the PDK1-K465E memory population was largely devoid of CXCR3^loCD43^lo effector-like memory cells, and the remaining CXCR3^loCD43^lo cells expressed diminished amounts of granzyme B. Consequently when we compared the cytotoxicity of PDK1-K465E and PDK1-WT OT-I memory cells, we found that on a per-cell basis PDK1-K465E T cells were less efficient in killing than PDK1-WT T cells. In addition, assessment of their antitumor activity in vivo without further boosting showed that memory PDK1-K465E T cells were less effective in providing long-term protection than PDK1-WT T cells. These findings indicate that Akt has an important function in tumor immune surveillance by memory CD8⁺ T cells. Recently, Crompton et al. (12) showed that pharmacological inhibition of Akt during in vitro expansion of tumor-reactive T cells results in enhanced persistence and antitumor activity upon adoptive transfer and boosting. T-cell expansion protocols for adoptive T-cell therapy typically use anti-CD3 antibody and high-dose IL-2, conditions that promote sustained Akt signaling and the terminal differentiation of effector CD8⁺ T cells at the expense of T_CM generation (14, 34). Under these conditions the inhibition of Akt is expected to enhance the efficacy of adoptive T-cell therapy by reducing the generation of terminally differentiated effector cells (12). However, our findings suggest that the physiological role of Akt in immune responses is to promote the generation of effector-like memory cells, which, along with other memory subsets, ensure maximal protection of the host against antigen reencounter.

The data presented in the current study also demonstrate that the type of immune challenge dictates which facet of memory cell development is affected by Akt signaling. Although differentiation into effector-like memory T cells was similarly affected in the protein vaccination and infection models, its prosurvival effects during the contraction phase were confined to responses elicited by protein vaccination. Interestingly, Bim was expressed similarly in PDK1-K465E OT-I and PDK1-WT OT-I effectors in the infection model (Fig. S10), in contrast with the results obtained following protein vaccination (Fig. 4C). This divergence may explain why primary PDK1-K465E OT-I effectors underwent a more pronounced contraction than PDK1-WT OT-I effectors after protein vaccination but not after infection. Depending on the type of immune challenge, different mechanisms of Foxo3 regulation that subsequently influence Bim expression could be operating. Thus, in addition to Akt, Erk and IKK have been shown to inhibit Foxo3 activity, and various pathways are known to counter Foxo3 inhibition (35). In addition, Bim is regulated by Foxo3-independent transcriptional, posttranscriptional, and posttranslational mechanisms (36), which could differ depending on the strength/duration of TCR signaling, costimulation, and inflammation.

Fig. S10.

Single transfer of PDK1-WT and PDK1-K465E OT-I cells and infection with ΔActA-Lm-OVA were performed as in Fig. 5. On day 7 postinfection, Bim and actin expression levels in purified splenic PDK1-WT and PDK1-K465E OT-I cells were analyzed by Western blotting and densitometry. Bar graphs represent the ratio of Bim to actin expressed as percentage of PDK1-WT values. Data show the mean ± SEM of two combined experiments with n = 3 or 4 mice per group.

Finally, the effects of Akt signaling reported here are similar to those ascribed to IL-2 in the generation of memory cell subsets (2, 3), suggesting that Akt signaling is the primary mechanism responsible for these effects. In summary, the present study demonstrates a previously unappreciated role for Akt in the generation of protective memory CD8⁺ T-cell responses, notably against tumor recurrence. Optimizing Akt activity therefore should maximize the therapeutic effect of anticancer vaccines.

Materials and Methods

Mice and in Vivo Experiments.

All procedures were conducted in accordance with UK Home Office guidelines and were approved by the University of Southampton’s ethical committee. C57BL/6 mice were obtained from Charles River. OT-I transgenic mice carrying the K465E knockin mutation in the PH domain of PDK1 (Pdpk1^K465E) have been described previously (13, 14). Unless otherwise specified, 10⁴ CD45.1 or CD45.1/CD45.2 OT-I cells isolated from spleen were transferred i.v. into CD45.2 C57BL/6 recipient mice. For cotransfer experiments, 5 × 10³ CD45.1 PDK1-WT OT-I cells were mixed with 5 × 10³ CD45.1/CD45.2 PDK1-K465E OT-I cells and were transferred i.v. to CD45.2 C57BL/6 mice. One day later, mice were challenged i.p. with 5 mg OVA (Sigma) plus 100 µg anti-CD40 mAb (clone 3/23) (37) and 10 µg LPS (Sigma) or OVA plus 200 µg anti–4-1BB mAb (clone Lob12.3) (38) and 10 μg LPS or OVA plus 50 μg polyI:C (Sigma). Alternatively, mice were infected i.v. with 10⁶ cfu of ΔActA–Lm-OVA (from H. Shen, University of Pennsylvania Perelman School of Medicine, Philadelphia). In some experiments, mice were rechallenged with 30 nmol OVA_257–264 peptide i.v. alone (in the case of OVA/anti-CD40/LPS priming) or mixed with 100 µg anti-CD40 mAb (when using OVA/anti-41BB/LPS and OVA/polyI:C for priming). Where indicated, mice were injected s.c. with OVA-expressing E.G7 tumor cells (ATCC). Tumor growth was monitored, and mice were killed when the humane end point was reached (15-mm mean tumor diameter when taking the two greatest perpendicular measurements).

Flow Cytometry.

Antibodies and methods are listed in SI Materials and Methods. Phycoerythrin (PE)-labeled H-2K^b/SIINFEKL tetramer was prepared in house. Samples were analyzed with a FACS Canto II and DIVA Software (BD Biosciences) or FCS Express (De Novo Software).

Lymphocyte Isolation from Nonlymphoid Tissues.

Colon lamina propria lymphocytes were isolated as described previously (39), except that 0.2 mg/mL collagenase VIII was used for enzymatic digestion (Sigma). PBS-washed livers were homogenized through a 100-μm cell strainer in PBS containing 0.5% FCS and 2 mM EDTA. After centrifugation, the cell pellet was resuspended in 40% Percoll solution (GE Healthcare) and layered on 70% Percoll solution. Cells recovered at the interface were washed twice in PBS. The same protocol was applied to lungs, except they first were cut into small pieces.

Western Blotting.

Detailed protocols of cell preparation are given in SI Materials and Methods. Cell lysates were prepared as previously described (16), and proteins were separated using NuPAGE Bis-Tris gels (Life Technologies) and were transferred to PVDF membranes (Immobilon-P; Millipore). Blots were probed with the antibodies listed in SI Materials and Methods. Densitometry analysis was performed using ImageJ software.

Statistical Analysis.

Where indicated, P values (*P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001) were determined by a two-tailed unpaired Student’s t test using Prism (GraphPad software), except for survival curves, in which a Mantel–Cox log-rank test was used.

Additional materials and methods are available in SI Materials and Methods.

SI Materials and Methods

Flow Cytometry.

Antibodies against CD8α (53-6.7), CD45.1 (A20), CD45.2 (104), CD127 (A7R34), KLRG1 (2F1), CXCR3 (CXCR3-173), CD62L (MEL-14), Eomes (DAN11MAG), T-bet (eBio4B10), IFN-γ (XMG1.2), TNF-α (MP6-XT22), IL-2 (JES6-5H4), and CD107a (eBio1D4B) were purchased from eBioscience. Antibodies against CD43 (activation glycoform, 1B11), CD8β (YTS156.7.7) and CX₃CR1 (SA011F11) were purchased from BioLegend, anti-granzyme B (GB11) antibody was purchased from Life Technologies, and anti-Ki67 (B56) was purchased from BD Biosciences. Intracellular staining for granzyme B, Eomes, and T-bet was performed using the Foxp3 staining buffer kit (eBioscience). For intracellular staining of cytokines, splenocytes were stimulated with 10 pM or 1 nM of OVA_257–264 peptide (PeptideSynthetics) at 37 °C for 4 h in the presence of GolgiPlug (BD Biosciences). After culture, cells were stained for cell-surface molecules, and intracellular staining was performed using the BD Biosciences Cytofix/Cytoperm buffers.

Cell Preparation for Western Blot Analysis.

For the analysis of Akt and S6 phosphorylation, PDK1-WT and PDK1-K465E OT-I splenocytes were stimulated with 10 pM OVA_257–264 for 2 d and then were expanded in IL-2 (10 ng/mL) for 3 d. CTLs (>95% CD8⁺) were washed and rested for 30 min before restimulation with OVA_257–264 for 5, 10, or 15 min. Control PDK1-WT cells were treated with 1 µM of Akt1/2 inhibitor VIII (Merck Millipore) for the entire culture period. For Bim analysis, splenocytes harvested from mice primed 7 d earlier with OVA/anti-CD40/LPS or ΔActA-Lm-OVA were stained with a biotinylated anti-CD45.1 antibody. CD45.1 cells were isolated by immuno-magnetic selection using anti-biotin Microbeads (Miltenyi) before protein extraction was performed.

Western Blot Antibodies.

Antibodies against phospho-S6 Ser235/236 (clone D57.2.2E), S6 ribosomal protein (clone 5G10), phospho-Akt Thr308 (clone 244F9), and Akt (polyclonal rabbit Ab) were purchased from Cell Signaling Technology. Anti-Bim antibody (clone 3C5) was purchased from Enzo Life Sciences, and anti-actin antibody (clone C-11) and HRP-linked anti-goat antibody were purchased from Santa Cruz Biotechnology. HRP-linked anti-rabbit antibody was purchased from GE Healthcare Life Sciences.

Intravascular Staining.

Intravascular staining was performed as described by Anderson et al. (19). Briefly, mice were injected i.v. with 3 µg of anti-CD8α APC-mAb (53-6.7; eBioscience) and were culled 3 min later. Lymphocytes were isolated, stained, and analyzed by flow cytometry.

In Vitro Killing Assay.

Eight weeks after PDK1-WT or PDK1-K465E OT-I adoptive transfer and infection with ΔActA-Lm-OVA, splenocytes were harvested and stained with a biotinylated anti-CD45.1 antibody. CD45.1 PDK1-WT and PDK1-K465E memory cells were enriched by immuno-magnetic selection using anti-biotin Microbeads (Miltenyi). Target cells were CD45.2 C57BL/6 naive splenocytes either loaded with 10 µM OVA_257–264 or left unpulsed for 1 h at 37 °C. Peptide-pulsed targets and unpulsed targets were stained for 10 min at 37 °C with 1 µM carboxyfluorescein succinimidyl ester (CFSE) and 0.1 µM CFSE, respectively. CFSE^high and CFSE^lo targets were mixed at a 1:1 ratio and were incubated alone or with PDK1-WT or PDK1-K465E memory cells for 6 h at 37 °C, in triplicate. The ratios of the surviving CFSE^high and CFSE^low cell populations were assessed by flow cytometry. Specific killing was calculated as follows: % killing = 100*[1 − (CFSE^high/CFSE^low)_effectors/(CFSE^high/CFSE^low)_control].

Analysis of Tumor-Infiltrating T Cells.

When the humane end point (15-mm mean tumor diameter) was reached, animals were culled, and tumors and spleens were digested for 30 min at 37 °C using 50 µg/mL DNase I (Roche) and 0.52 Wünsch unit/mL Liberase DL (Roche) and were forced through a 100-µm cell strainer before staining for flow cytometry analysis.