FOXO3 Regulates CD8 T Cell Memory by T Cell-Intrinsic Mechanisms

CD8 T cell responses have three phases: expansion, contraction, and memory. Dynamic alterations in proliferation and apoptotic rates control CD8 T cell numbers at each phase, which in turn dictate the magnitude of CD8 T cell memory. Identification of signaling pathways that control CD8 T cell memory is incomplete. The PI3K/Akt signaling pathway controls cell growth in many cell types by modulating the activity of FOXO transcription factors. But the role of FOXOs in regulating CD8 T cell memory remains unknown. We show that phosphorylation of Akt, FOXO and mTOR in CD8 T cells occurs in a dynamic fashion in vivo during an acute viral infection. To elucidate the potentially dynamic role for FOXO3 in regulating homeostasis of activated CD8 T cells in lymphoid and non-lymphoid organs, we infected global and T cell-specific FOXO3-deficient mice with Lymphocytic Choriomeningitis Virus (LCMV). We found that FOXO3 deficiency induced a marked increase in the expansion of effector CD8 T cells, preferentially in the spleen, by T cell-intrinsic mechanisms. Mechanistically, the enhanced accumulation of proliferating CD8 T cells in FOXO3-deficient mice was not attributed to an augmented rate of cell division, but instead was linked to a reduction in cellular apoptosis. These data suggested that FOXO3 might inhibit accumulation of growth factor-deprived proliferating CD8 T cells by reducing their viability. By virtue of greater accumulation of memory precursor effector cells during expansion, the numbers of memory CD8 T cells were strikingly increased in the spleens of both global and T cell-specific FOXO3-deficient mice. The augmented CD8 T cell memory was durable, and FOXO3 deficiency did not perturb any of the qualitative attributes of memory T cells. In summary, we have identified FOXO3 as a critical regulator of CD8 T cell memory, and therapeutic modulation of FOXO3 might enhance vaccine-induced protective immunity against intracellular pathogens.


Introduction
The ability of the immune system to respond rapidly and vigorously to antigen re-exposure is termed immunological memory, which is one of the tenets of adaptive immunity [1,2]. Induction of memory B and T cells is the basis of immunological memory induced by infections or vaccinations [2][3][4]. As compared to naïve T cells, memory T cells are hyper-reactive to antigenic stimulation and swiftly proliferate and/or differentiate into effector cells to confer protective immunity expeditiously [5][6][7][8]. The ability of memory T cells to confer protective immunity depends upon the number and quality of memory T cells [5,[9][10][11][12][13]. Understanding the mechanisms that regulate the quantity and quality of T cell memory is fundamentally important for the development of effective vaccines.
During a CD8 T cell response, engagement of the TCR, along with appropriate co-stimulatory and inflammatory signals, activate naïve T cells to proliferate and differentiate into effector cells [1,4,8,13,14]. In the case of LCMV infection, the peak of T cell expansion is reached at 8-10 days after infection, and the majority of the newly generated effector cells present at the peak of the response are short-lived and fated for deletion [15][16][17]. But, a small subset of the effectors, termed memory precursor effector cells (MPECs), possesses the potential to survive and differentiate into long-lived memory cells [16,17]. The number of memory CD8 T cells generated depends largely upon the magnitude of the expansion of MPECs during the T cell response. Substantial progress has been made in deciphering the extracellular signals and transcription factors that regulate the differentiation of MPECs [1], but the signaling pathways that govern the number of MPECs, their differentiation into memory CD8 T cells, and the maintenance of CD8 T cell memory are not fully understood.
The FOXO family of transcription factors plays a crucial role in governing cellular proliferation, apoptosis, energy metabolism, and stress resistance in response to dynamic alterations in stress and abundance of nutrients and growth factors in many cell types [18][19][20][21][22][23][24][25][26]. In mammals, the FOXO family consists of at least four members: FOXO1, FOXO3, FOXO4, and FOXO6 [22,27,28]. The activity of FOXOs is regulated by post-translational modifications, most notably, phosphorylation [25,26,[29][30][31]. In resting, quiescent cells, FOXOs exist in a hypo-phosphorylated state and localize to the nucleus to control the transcription of their target genes such as p27 Kip1 , p21 Cip1 , p300, BIM, Fas ligand that are involved in regulating cellular proliferation or apoptosis [25, [32][33][34][35][36][37][38][39][40][41][42][43]. However, in response to stimulation with growth factors or cytokines that stimulate the PI3K/AKT signaling pathway, activated Akt phosphorylates FOXO leading to its exclusion from the nucleus and degradation by proteolysis in the cytoplasm [18,25,29,44,45]. As a result, the transcription of FOXO target genes is diminished, which in turn facilitates cell cycle entry and/or survival. T cells express FOXO1 and FOXO3, and there has been a recent surge in interest to elucidate the importance of FOXOs in regulating T cell biology [46][47][48][49][50][51][52][53]. Phosphorylation-mediated inactivation of FOXOs and downregulation of p27 Kip1 appear to be obligatory steps for T cells to enter the cell cycle in response to TCR engagement [24,41,53]. FOXO1 has been shown to control several aspects of T cells including the expression of adhesion molecules like L-selectin and CCR7, cytokine receptors like the IL-7 receptor, development of regulatory T cells, and protection against autoimmunity [51,54]. Unlike FOXO1, the role of FOXO3 in T cell homeostasis is less well understood.
It was first reported that global deletion of FOXO3 results in lymphadenopathy and spontaneous activation of T cells, but other independent studies have failed to confirm these results, possibly due to differences in the genetic background of mutant mice [19,40,49]. Nonetheless, elegant studies from the Hedrick group have showed that FOXO3 inhibits the primary expansion of CD8 T cells in the spleen by regulating IL-6 production by dendritic cells [49]. However, the T cell intrinsic role of FOXO3 in regulating various phases of the polyclonal multi-epitopespecific CD8 T cell response to an acute viral infection in lymphoid and non-lymphoid organs remains to be determined. In this study, using global and conditional T cell-specific FOXO3 knockout mice, we have systematically examined the role of FOXO3 in regulating the: (1) expansion and function of polyclonal antigen-specific CD8 T cells in lymphoid and nonlymphoid organs; (2) antigen-driven in vivo proliferation of virusspecific CD8 T cells; (3) differentiation of CD8 T cells into shortlived effector cells (SLECs) and MPECs; (4) contraction of antigen-specific CD8 T cells in lymphoid and non-lymphoid organs; (5) the numbers and function of memory CD8 T cells; (6) proliferative renewal of memory CD8 T cells; (7) secondary CD8 T cell responses and protective immunity. These studies show that FOXO3 regulates the clonal expansion of polyclonal CD8 T cells in a tissue-specific fashion by T cell intrinsic mechanisms. The enhanced expansion of CD8 T cells was clearly not due to an increased proliferation rate, but was instead associated with reduced cellular apoptosis. Furthermore, we show that FOXO3 deficiency markedly enhances the size of the memory CD8 T cell compartment without affecting the phenotype or quality of memory CD8 T cells. These findings have implications in the design of effective vaccines that engender potent and effective protective cellular immunity against intracellular pathogens and tumors.

Results
Dynamic, in vivo phosphorylation of Akt, FOXO, and mTOR in antigen-specific CD8 T cells FOXO3 has emerged as a key regulator in a number of physiological outcomes, including metabolism, ageing, and vascular reactivity [21][22][23]27,46,55]. More recently, in the immune system, there is increasing evidence that FOXO3 is a critical regulator of T cell homeostasis [46,47,52,53,55]. While a number of signaling pathways and post translational modifications may be involved in controlling the activity of FOXO3, phosphorylation mediated through a PI3K-Akt centric signaling module primarily regulates FOXO3 function in T cells [29,45,56,57]. In order to fully understand the role of the Akt/ FOXO3 axis in the control of CD8 T cell homeostasis, we developed a phospho-specific, flow cytometric method to quantify in vivo phosphorylation levels of key proteins implicated in FOXO3 activity: the upstream kinase Akt, FOXO3 itself, as well as a potential downstream substrate of Akt, the kinase, mammalian target of Rapamycin (mTOR) ( Figure 1A). Figure 1A illustrates the kinetics of phosphorylation of Akt (Thr308), FOXO1/3 (T24/T32), and mTOR (S2448) in antigenspecific CD8 T cells during an acute LCMV infection. The phosphorylation of Akt (Thr308) and mTOR (S2448) was highest at day 5 post-infection (PI), and subsided to steady-state levels by days 10 and 8 PI respectively. Interestingly, the phosphorylation dynamics for FOXO1/O3 were different from that of Akt and mTOR; the phosphorylation levels for FOXO1/O3 dropped between days 5 and 8 PI, but gradually increased back to steadystate levels by days 10-15 PI. Note that the phosphorylation kinetics of NP396-specific CD8 T cells after day 8 PI was slightly delayed, as compared to those in GP33-specific CD8 T cells. In summary, these findings demonstrate that the in vivo phosphorylation levels of Akt, FOXO1/O3, and mTOR are highly dynamic, and of note, the phosphorylation kinetics of Akt correlated with mTOR phosphorylation but not with phosphorylation of FOXO1/O3 during a T cell response to LCMV. However, the specific role of FOXO3 in regulating different phases of the CD8 T cell response has not been carefully examined.

Regulation of CD8 T cell expansion by FOXO3 is tissue specific
Infection of mice with LCMV elicits a potent, multiple epitopespecific, CD8 T cell response wherein virus-specific CD8 T cells are distributed to both lymphoid and non-lymphoid organs [58,59]. To examine the role of FOXO3 in regulating CD8 T cell responses to an acute viral infection, groups of wild type (+/+) and FOXO3-deficient (FOXO32/2) mice were infected with LCMV. At day 8 PI, we quantified CD8 T cells that are specific to three immuno-dominant LCMV epitopes in lymphoid (spleen and lymph nodes) and non-lymphoid (liver) tissues. Figure 1B shows that the numbers of LCMV-specific CD8 T cells in spleens of FOXO32/2 mice were significantly (P,0.05) higher than in +/+ mice. Surprisingly, the numbers of virus-specific CD8 T cells in Author Summary CD8 T cells are vital for controlling infections with viruses, intracellular bacteria and protozoa. Induction of T and B cell memory is the basis of vaccinations and cellular immunity to intracellular pathogens depends upon the number and quality of memory CD8 T cells. Understanding the mechanisms that control various facets of CD8 T cell memory is of fundamental importance for development of effective vaccines. In this study, we have identified the transcription factor FOXO3 as a crucial regulator of the magnitude of CD8 T cell memory. During a T cell response, FOXO3 limits the number of memory CD8 T cells by inhibiting the accumulation of memory precursor effector cells that give raise to long-lived CD8 T cells. Loss of FOXO3 activity in T cells led to a durable increase in the number of memory CD8 T cells, and the functional quality of FOXO3-deficient memory CD8 T cells was unaffected by FOXO3 deficiency. Thus, our studies suggest that targeting FOXO3 activity may be a fruitful strategy to augment vaccine-induced CD8 T cell memory and protective immunity.
the lymph nodes and liver of FOXO32/2 mice were comparable to those in +/+ mice ( Figure 1B). These data suggested that FOXO3 downregulates the accumulation of CD8 T cells in a tissue-specific fashion during an acute LCMV infection.
Next, we assessed whether FOXO3 deficiency affected the cell surface phenotype of LCMV-specific effector CD8 T cells at day 8 PI. The expression levels of CD44, CD62L, CD27, and CD122 on FOXO32/2 LCMV-specific CD8 T cells were comparable to those on +/+ CD8 T cells (Figure 2A). The population of LCMV-specific CD8 T cells in the spleen is comprised of at least two subsets of effector cells based on the cell surface expression of IL-7 receptor a (CD127) and KLRG-1: the SLECs (KLRG-1 HI / CD127 LO ) a majority of which are destined for apoptosis, and the MPECs (CD127 HI /KLRG-1 LO ), which are poised to differentiate into memory CD8 T cells [16,17,60]. Figure 2B shows that FOXO3 deficiency significantly enhanced the absolute numbers of both SLECs and MPECs in the spleen at day 8 PI. Of note is the marked increase in the total number of MPECs in FOXO32/2 mice. In summary, these data suggested that FOXO3 deficiency increased the clonal expansion of CD8 T cells without disrupting the differentiation of SLECs and MPECs.

FOXO3 deficiency does not affect the function of effector CD8 T cells
To detect possible differences in the functionality of antigenspecific CD8 T cells from +/+ and FOXO32/2 mice, we assessed their ability to produce the cytokines IFNc, TNFa, and IL-2 in response to antigenic stimulation directly ex vivo ( Figure 2C). The MFIs of IFNc staining for FOXO32/2 and +/+ effector CD8 T cells were similar ( Figure 2C). Additionally, the percentages of epitope-specific, CD8 T cells that produced two cytokines (IFNc and TNFa) or three cytokines (IFNc, TNFa, and IL-2) in FOXO32/2 mice were similar to those in +/+ mice ( Figure 2C). The MFIs for TNFa and IL-2 were also comparable between +/+ and FOXO32/2 CD8 T cells (data not shown). As a surrogate marker of the lytic function of effector CD8 T cells, we compared granzyme B expression between +/+ and FOXO32/2 LCMV-specific effector CD8 T cells directly ex vivo. The levels of granzyme B in FOXO32/2 CD8 T cells were similar to those in +/+ CD8 T cells ( Figure 2D). Taken together, data in Figures 1  and 2 suggested that FOXO3 deficiency increased the expansion of virus-specific CD8 T cells without affecting their phenotype or function. Consistent with normal CD8 T cell effector function, Figure 1. Loss of FOXO3 leads to a tissue-specific increase in expansion of effector CD8 T cells. (A) Dynamic, in vivo alterations in phosphorylation of FOXO-associated signaling proteins in LCMV-specific CD8 T cells. C57BL/6 mice were infected with LCMV, and at the indicated days PI, splenocytes were stained with anti-CD8, MHC-I tetramer (D b /NP396 or D b /GP33) and either the anti-P-Akt (T308), P-FOXO-1/3a (T24/T32), or P-mTOR (S2448) antibodies. As controls, these antibodies were pre-incubated with their specific antigenic peptide before adding on to the cells. Representative flow plots (left) gated on tetramer-binding CD8s from day 5 PI mice indicate specific antibody staining in relation to the peptide blocked controls. Plotted data (Corrected MFI) is expressed as the difference of observed MFI for the phospho-specific protein and peptide-blocked control (right), divided by the peptide-blocked control. Data are representative of at least three independent experiments. (B) +/+ and FOXO32/2 mice were infected with LCMV, and at 8 days PI, LCMV specific CD8 T cells were quantified in spleen, liver and lymph nodes by staining with anti-CD8 and MHC-I tetramers. Data are representative of 4 independent experiments with 4-6 mice/group/experiment. doi:10.1371/journal.ppat.1002533.g001 LCMV control in FOXO32/2 mice was similar to that in +/+ mice (not shown).

Increased accumulation of CD8 T cells in FOXO32/2 mice is linked to reduced apoptosis and not enhanced proliferation
To rigorously address whether the greater number of LCMVspecific CD8 T cells in the spleen of FOXO32/2 mice was due to increased proliferation, we utilized two different approaches, Ki67 staining and BrdU incorporation. First, we quantified proliferation of LCMV-specific CD8 T cells by staining for the nuclear antigen Ki67 directly ex vivo ( Figure 3A) during the peak clonal expansion phase (days 6 and 8 PI) of the CD8 T cell response. For the second approach, we measured BrdU incorporation by LCMV-specific CD8 T cells in vivo from days 6-8 PI ( Figure 3B). As shown in Figures 3A and 3B, the percentages of of Ki67 +ve LCMV-specific CD8 T cells and BrdU +ve LCMV epitope-specific CD8 T cells in spleens, liver, and lymph nodes of FOXO32/2 mice were similar to those in +/+ mice. Measurement of Ki67 expression at day 5 PI also showed that FOXO3 deficiency did not alter the proliferation of LCMV-specific CD8 T cells ( Figure S1) Taken together, these data indicated that enhanced proliferation was not sufficient to explain the increase in antigen-specific CD8 T cells in FOXO32/2 mice during LCMV infection.
CD8 T cell homeostasis is not simply determined by alterations in proliferation, it is also regulated by alterations in cell death. During the clonal expansion phase of the CD8 T cell response, there is concomitant proliferation and apoptosis [61], and therefore the magnitude of clonal expansion is dependent upon the relative rates of proliferation and apoptosis. To assess cellular apoptosis, we determined the percentages of Annexin V HI LCMVspecific CD8 T cells in spleens of +/+ and FOXO32/2 mice at day 6 PI, directly ex vivo. Data in Figure 4A shows that the percentages of Annexin V HI LCMV-specific CD8 T cells in spleens of FOXO32/2 mice were significantly lower than in spleens of +/+ mice. These data suggested that FOXO3 controls the accumulation of effector CD8 T cells by promoting cellular apoptosis during an acute LCMV infection. During antigen-driven proliferation, the competing pro-apoptotic effects of TGF-b and the anti-apoptotic effects of IL-15 regulate the apoptotic rate of CD8 T cells [61]. Because apoptosis of LCMV-specific CD8 T cells was dampened by FOXO3 deficiency ( Figure 4A), and IL-15 has been reported to induce phosphorylation of FOXO3 [62], we hypothesized that FOXO3 might downregulate clonal expansion by reducing the viability of CD8 T cells that are deprived of IL-15. The balance in the levels of the anti-apoptotic molecule Bcl-2 and the pro-apoptotic molecule BIM controls the susceptibility of a T cell to apoptotic stimuli [36,38,39,[63][64][65]. Since BIM is a target gene for FOXO3, we tested whether deficiency of FOXO3 might lead to lower expression of BIM in proliferating CD8 T cells (day 6 PI), by comparing the levels of BIM in +/+ and FOXO32/2 LCMV-specific CD8 T cells after culture with or without IL-15. As shown in Figure 4B and Figure S2, BIM expression in +/+ CD8 T cells was higher than in FOXO32/2 CD8 T cells, when cultured in media without IL-15. However, IL-15 reduced BIM expression in +/+ CD8 T cells to levels seen in FOXO32/2 CD8 T cells; IL-15 did not affect BIM expression in FOXO32/2 CD8 T cells. These data suggested that FOXO3 might control BIM expression in IL-15deprived CD8 T cells. When analyzing for BIM in direct relation to Bcl-2, we observed that after 6 days of infection, LCMV-specific CD8 T cells from +/+ mice exhibit an increased BIM to Bcl-2 ratio, as compared to FOXO32/2 CD8 T cells ( Figure 4B). Thus, we propose that FOXO3 might downregulate the accumulation of proliferating CD8 T cells by inducing BIM expression.

Contraction of CD8 T cells in FOXO32/2 mice
Next, we examined whether FOXO3 deficiency regulated the contraction of CD8 T cells in lymphoid and non-lymphoid tissues during an acute LCMV infection. Virus-specific CD8 T cells were quantified in spleen, liver, and lymph nodes at days 8, 11, 15, and 30 PI ( Figure 5A). Overall, the slopes of the contraction curves for LCMV-specific CD8 T cells in FOXO32/2 mice were comparable to those of +/+ mice. Thus, the contraction of LCMV-specific CD8 T cells was minimally affected by FOXO3 deficiency. Next, we assessed whether FOXO3 regulated proliferation of LCMV-specific CD8 T cells during the contraction phase. In vivo BrdU incorporation studies showed that the percentages of LCMV-specific CD8 T cells that incorporated BrdU between days 8-11 or 12-15 PI in FOXO32/2 mice were similar to those in +/+ mice ( Figure 5B).

FOXO3 deficiency enlarges the size of the memory CD8 T cell compartment in a tissue-specific fashion
The number of memory CD8 T cells is a function of the magnitude of expansion and contraction during the T cell response [5]. Here, we determined whether increased clonal expansion of MPECs in FOXO32/2 mice ( Figure 2B) translated to inflation of LCMV-specific memory CD8 T cells in lymphoid and non-lymphoid organs. We observed that FOXO32/2 mice exhibit a substantial increase in the numbers of NP396-(P,0.001), GP33-(P,0.01), and GP276-(P,0.04) specific CD8 T cells in spleen at 180 days PI ( Figure 6A). Interestingly, there was no detectable increase in the numbers of LCMV-specific memory CD8 T cells in either the liver or lymph nodes at day 180 PI ( Figure 6A). It should be noted that the magnitude of increase in the number of memory CD8 T cells in FOXO32/2 mice reflected the increased accumulation of MPECs during the primary response ( Figure 2B). High numbers of memory CD8 T cells were maintained in FOXO32/2 mice stably until at least day 300 PI (data not shown). These data strongly imply that FOXO3 plays an important role in downregulating the magnitude of CD8 T cell memory in the spleen following an acute viral infection.
Phenotypic analysis of LCMV-specific memory CD8 T cells in FOXO32/2 and +/+ mice suggested that FOXO3 deficiency did not affect the expression of molecules that control T cell trafficking (CD44) or cytokine receptors (CD122 and CD127)( Figure 6B). In addition, assessment of CD62L levels on antigen specific CD8 T cells indicated that both central (CD62L high ) and effector (CD62L low ) memory frequencies were unaffected by FOXO3 deficiency ( Figure 6C). Furthermore, functional analysis of antigen-triggered cytokine production did not reveal alterations in cytokine producing ability of LCMVspecific memory CD8 T cells from FOXO32/2 mice when compared to their +/+ counterparts ( Figure 6D). In summary, data presented in Figure 6 showed that FOXO3 deficiency increased the quantity of CD8 T cell memory without affecting the quality.
It is well established that memory CD8 T cells are maintained for extended periods of time by proliferative renewal, driven by homeostatic cytokines IL-7 and IL-15 [11,56,[66][67][68]. Although it is known that IL-7 and IL-15 signaling triggers phosphorylation of FOXO3 [56,62], the effect of FOXO3 deficiency on proliferative renewal of memory CD8 T cells has not been examined. Using three approaches, we compared the cytokine-driven proliferative renewal of memory CD8 T cells in +/+ and FOXO32/2 mice. First, in vivo BrdU incorporation studies showed that the percentages of BrdU +ve LCMV-specific memory CD8 T cells in FOXO32/2 mice were comparable to those in +/+ mice ( Figure 7A). Likewise, the percentages of Ki67 +ve LCMV-specific CD8 T cells were unaffected by FOXO3 deficiency ( Figure 7A). To further examine the effect of FOXO3 deficiency on the proliferative renewal of memory CD8 T cells, CD8 T cells from the spleens of LCMV-immune +/+ and FOXO32/2 mice were labeled with CFSE and adoptively transferred into congenic uninfected mice. Thirty days after cell transfer, flow cytometric analysis of CFSE staining revealed that donor LCMV-specific CD8 T cells from +/+ and FOXO32/2 mice proliferated equally in the recipient mice ( Figure 7B). Taken together, data in Figure 7 provided strong evidence that FOXO3 deficiency did not alter the homeostatic turnover of LCMV-specific memory CD8 T cells.

Secondary CD8 T cell responses and protective immunity in FOXO32/2 mice
To observe the effect of FOXO3 deletion on recall responses of memory CD8 T cells and protective immunity, LCMV-immune +/+ and FOXO32/2 mice were challenged with LCMV clone 13, a strain of LCMV that establishes a chronic infection in naïve immunocompetent mice. At day 5 after challenge, the numbers of NP396-(P,0.01), GP33-(P,0.01) and GP276-(P,0.03) specific CD8 T cells in FOXO32/2 mice ( Figure 8A) were significantly higher than in +/+ mice. The increased number of LCMVspecific CD8 T cells in spleens of LCMV clone 13-Challenged FOXO32/2 mice correlated with the increased numbers of memory CD8 T cells ( Figure 6A). As in a primary infection ( Figure 1B), we did not see a statistically significant increase in the numbers of LCMV-specific CD8 T cells in either the liver or lymph nodes ( Figure 8A). Staining for Ki67 illustrated that FOXO3 deficiency did not affect the proliferation of LCMVspecific CD8 T cells during a secondary response ( Figure 8B). Secondary effector CD8 T cells in spleens of FOXO32/2 mice produced comparable levels of IFNc, TNFa, and IL-2 ( Figure 8C). LCMV titers in tissues were comparable in +/+ and FOXO32/2 mice, which indicated that protective immunity was not compromised in the absence of FOXO3 ( Figure 8D).

FOXO3 regulates CD8 T cell Expansion by T cell intrinsic mechanisms
To address whether FOXO3 has a T cell intrinsic role in regulating polyclonal CD8 T cell responses to LCMV, we used a cre-loxP knockout strategy to generate the FOXO3L mice that lacked FOXO3 only in T cells [69,70]. T cell-specific loss of FOXO3 in FOXO3L mice was confirmed by western blot and flow cytometry ( Figure S3). FOXO3L and littermate +/+ mice were infected with LCMV and virus-specific CD8 T cell responses were analyzed at day 8 PI. In the spleen, FOXO3L mice exhibited a statistically significant increase in the numbers of LCMV-specific CD8 T cells (NP396 P,0.02; GP33 P,0.01; GP276 P,0.04) over their +/+ littermate controls ( Figure 9A). It should be noted that the observed increase in the expansion of CD8 T cells in global FOXO32/2 mice ( Figure 1B) was fully recapitulated in FOXO3L mice ( Figure 9A). To assess whether greater accumu-lation of effector CD8 T cells in spleens of FOXO3L mice was driven by increased proliferation, we measured Ki67 expression and BrdU incorporation during the clonal expansion phase of the CD8 T cell response to LCMV. At day 6 PI, the percentages of both Ki67 +ve and BrdU +ve LCMV-specific CD8 T cells in FOXO3L mice were comparable to those in +/+ mice, which suggested that enhanced accumulation of effector CD8 T cells in FOXO3L mice are not linked to an altered proliferation rate ( Figure 9B). Likewise, percentages of Ki67 +ve CD8 T cells at day 5 PI were comparable in +/+ and FOXO3L mice ( Figure S4). However, the percentages of Annexin V HI LCMV-specific CD8 T cells in spleens of FOXO3L mice were significantly lower than in +/+ mice ( Figure 9B). These data suggested that FOXO3 controls the accumulation of CD8 T cells during a primary response by regulating apoptosis of proliferating cells. Conditional deficiency of FOXO3 in T cells did not affect the cell surface phenotype (data not shown) or antigen-triggered cytokine production by LCMV-specific effector CD8 T cells at day 8 PI ( Figure 9C).
Since both CD8 and CD4 T cells lack FOXO3 activity in FOXO3L mice, it could be argued that increased clonal expansion of CD8 T cells might result from enhanced CD4 T cell help in FOXO3L2/2 mice. To address this issue, we depleted CD4 T cells in +/+ and FOXO3L mice and quantified CD8 T cell responses to LCMV in the absence of CD4 T cells. At day 8 PI, .95% of the CD4 T cells were depleted in spleen of both +/+ and FOXO3L mice (data not shown). The number of LCMV-specific CD8 T cells in spleen of CD4-depleted FOXO3L mice was substantially higher than in CD4 T cell-depleted +/+ mice ( Figure 9D). Thus, in the apparent absence of CD4 T cells, FOXO3 deficiency in CD8 T cells was sufficient to increase the accumulation of virus-specific CD8 T cells during an acute LCMV infection.

T cell-specific deletion of FOXO3 enhances the magnitude of CD8 T cell memory
To determine whether deletion of FOXO3, exclusively from the T cell compartment, would affect CD8 T cell memory generation, FOXO3L and +/+ mice were infected with LCMV and virusspecific memory CD8 T cells were quantified in lymphoid and non-lymphoid tissues at 180 days PI. We observed a significant increase in the number of memory CD8 T cells that are specific to the three immuno-dominant LCMV epitopes (NP396 P,0.02; GP33 P,0.01; GP276 P,0.05) in spleens of FOXO3L mice over the +/+ mice ( Figure 10A). These data indicated that FOXO3 regulates the magnitude of CD8 T cell memory by T cell intrinsic mechanisms.
Next, we examined whether conditional deficiency of FOXO3 in T cells affected the phenotypic and functional attributes of memory CD8 T cells. The expressions of CD44, CD62L, CCR7 and LFA-1 on +/+ and FOXO3L memory CD8 T cells from spleen, liver and lymph nodes were similar ( Figure 10B). Additionally, the relative proportions of effector (CD62L Lo ) and central (CD62L HI ) memory CD8 T cells were unaffected by FOXO3 deficiency (Figure 10B). In response to antigenic stimulation, virus-specific memory CD8 T cells in LCMV-immune FOXO3L2/2 mice produced IFNc, TNFa, and IL-2 at levels comparable to those in +/+ mice ( Figure 10C). As was the case in our studies of global FOXO3-deficient mice (Figure 7), memory CD8 T cells in FOXO3L mice exhibited no difference in homeostatic turnover, as evidenced by BrdU incorporation from day 120 PI mice, pulsed for 8 days and by parallel Ki67 staining ( Figure 10D). Taken together, data presented in Figure 10 strongly suggested that FOXO3 regulates the quantity, with no apparent loss in quality, of CD8 T cell memory by T cell intrinsic mechanisms.

Discussion
The FOXO transcription factors are important regulators of cell cycle progression, apoptosis, and energy metabolism [21,23,26,28,56]. In the T cell compartment, FOXOs have been implicated in regulating homing of T cells, cytokine receptor expression, and development of regulatory T cells [48,51,53,54]. While it has been reported that FOXO3 may control CD8 T cell expansion, albeit through non T cell intrinsic mechanisms, a role for FOXO3 in memory T cell survival has also been posited [49]. What has not been thoroughly addressed, however, is the T cell intrinsic role of FOXO3 in governing different facets of the physiological polyclonal T cell response to foreign antigens, including the in vivo generation and maintenance of CD8 T cell memory. In the present study, we have systematically examined the T cell-intrinsic role of FOXO3 in controlling the expansion, contraction, and memory phases of the polyclonal CD8 T cell response to an acute viral infection. These studies have provided strong evidence supporting a T cell intrinsic role for FOXO3 in limiting the magnitude of expansion and the number of memory CD8 T cells in a tissue-specific fashion during a physiological response to an acute LCMV infection. These findings have advanced our mechanistic understanding of CD8 T cell homeostasis, and are expected to have implications in the development of effective vaccines.
FOXOs are known to maintain cellular quiescence by mechanisms including the induction of cell cycle inhibitors like p27 KIP1 and p21 Cip1 [30,34,37,38]. Downregulation of FOXO activity is believed to be an obligatory step for cell cycle entry in response to mitogenic stimuli [30]. The observed phosphorylation of FOXO1/O3 in LCMV-specific CD8 T cells was readily detectable at day 5 PI but exhibited a sharp decline by day 8 PI. The drop in the phosphorylation in FOXO1/O3 between days 5 and 8 PI coincides with declining viral load and decreased antigenic stimulation. However, we observed a rebound in the phosphorylation of FOXO1/O3 between days 8 and 11 PI. What controls the dynamics of FOXO1/O3 phosphorylation during a CD8 T cell response? In addition to TCR signaling, FOXO1/O3 phosphorylation is regulated by signaling via cytokine receptors such as IL-2, IL-7, and IL-15 [38,56,66]. It is possible that FOXO1/O3 is phosphorylated by different extracellular signals at different phases of the T cell response. For example, during the phase of antigen-driven proliferation, IL-7R expression is known to be very low [68], and TCR signaling along with IL-2/IL-15 might drive the phosphorylation of FOXO1/O3. However, after antigen clearance, IL-7 signaling might drive phosphorylation on the surviving IL-7 receptor-expressing MPECs, and eventually their resultant memory cells.
FOXO3 has been reported to suppress expansion of CD8 T cells indirectly by inhibiting IL-6 production by dendritic cells [49]. In this report, however, the T cell intrinsic role of FOXO3 was not assessed in polyclonal CD8 T cells, and monoclonal TCR transgenic CD8 T cells may not always mimic the responses of polyclonal CD8 T cells. In the present study, global FOXO3deficient mice exhibit increased expansion of polyclonal CD8 T cells specific to multiple epitopes during an acute LCMV infection. Furthermore, infection of T cell-specific conditional FOXO3deficient mice with LCMV, fully recapitulated the enhanced CD8 T cell expansion seen in global FOXO32/2 mice; even in the absence of CD4 T cells. Therefore, our data implies that FOXO3 suppresses CD8 T cell expansion in vivo by T cell intrinsic and extrinsic mechanisms. This inference is also supported by the reported T cell intrinsic regulation of regulatory T cell development by FOXO1 and FOXO3 [71,72]. One of the most interesting findings presented in this manuscript is that the effect of FOXO3 deficiency on CD8 T cell expansion and memory is observed in the spleen, but not in the liver or lymph nodes. The enhanced accumulation of effector CD8 T cells preferentially in spleens of FOXO32/2 mice could not be explained by tissuespecific differences in BIM expression in CD8 T cells directly ex vivo; regardless of the tissue (spleen, liver, or lymph nodes), BIM levels in FOXO32/2 CD8 T cells were slightly lower than in +/ + CD8 T cells ( Figure S5). Additionally, the selective increase in the number of memory CD8 T cells in spleens of FOXO32/2 mice could not be linked to alterations in the expression of molecules such as CD62L, LFA-1, CCR7, and CD44 that regulate T cell trafficking ( Figure 10B). Tissue-specific effects were observed in both global and T cell-specific conditional FOXO3 knockout mice, which suggests that the local immunological milieu influences the effects of FOXO3 in T cells. It is possible that FOXO3-deficient T cells are hyper-responsive to cellular and environmental cues unique to the spleen during or after cessation of antigen-driven proliferation. Tissue-specific alterations in CD8 T cell homeostasis are not unique to FOXO3 deficiency because lymph node-specific effects on CD8 T cell numbers have been reported in mice deficient for Fas and BIM [64]. Future experiments will address the mechanisms underlying the tissuespecific effects of FOXO3 in regulating CD8 T cell homeostasis.
FOXO3 is known to regulate both proliferation and apoptosis by controlling the transcription of genes like p27 Kip1 , p130, BIM, and Fas ligand [26,34,35,63,64]. Therefore, the observed increase in the accumulation of LCMV-specific CD8 T cells in FOXO32/2 mice could be attributed to altered proliferation and/or apoptosis. Analysis of in vivo proliferation by multiple strategies indicated that FOXO3 deficiency did not alter proliferation rates of LCMV-specific CD8 T cells in vivo. Clonal expansion of CD8 T cells is associated with concomitant proliferation and apoptosis, therefore, cellular accumulation is the result of the proliferation rate exceeding the apoptotic rate. The competing effects of TGF-b and IL-15 are known to dictate the apoptotic rate of proliferating CD8 T cells, but the signaling mechanisms involved are not well defined [61]. We theorized that IL-7/IL-15 deprivation during antigen-driven proliferation might diminish FOXO3 phosphorylation, and augment the expression of BIM. We find that at day 6 PI, apoptosis of LCMV-specific CD8 T cells was significantly reduced in spleens of FOXO32/ 2 mice. Additionally, IL-15 deprivation was indeed associated with higher BIM levels in +/+ CD8 T cells than in FOXO32/ 2 CD8 T cells, which suggested that proliferating FOXO32/2 CD8 T cells might be less susceptible to cytokine withdrawalinduced apoptosis during the expansion phase of the CD8 T cell response. During the early contraction phase (day 8-11 PI), a substantial number of LCMV-specific CD8 T cells are still in cycle ( Figure 5A), but during this interval the apoptotic rate presumably exceeds the proliferation rate resulting in a net loss of CD8 T cells. Interestingly, FOXO3 deficiency minimally altered the contraction of LCMV-specific CD8 T cells. These data suggest that mechanisms controlling apoptosis of CD8 T cells during expansion and contraction are likely distinct.
Remarkably, the numbers of memory CD8 T cells in the spleen of both FOXO32/2 and FOXO3L mice were substantially higher than in +/+ mice. The number of memory CD8 T cells is dictated by the magnitude of expansion (clonal burst size) and contraction of effector CD8 T cells [5]. The magnitude of increase in the number of memory CD8 T cells in FOXO32/2 or FOXO3L mice reflects enhanced expansion of MPECs ( Figure 2B) during the primary CD8 T cell response. Importantly, enhancement in the number of memory CD8 T cells induced by FOXO3 deficiency was not associated with detectable alterations in phenotype or effector function. Memory CD8 T cells in LCMV-immune FOXO32/2 mice exhibit strong recall responses and provide effective immunity against a persistent LCMV infection. Thus, FOXO3 deficiency increased the quantity of CD8 T cell memory without affecting their phenotype or effector functions.
Memory CD8 T cells are maintained by IL-7 and IL-15-driven proliferative renewal and phosphorylation of FOXO3 is an integral component of the signaling circuitry triggered by IL-7/ IL-15 signaling [11,54,66,67]. Additionally, we have previously shown that deficiency of the cell cycle inhibitor p27 Kip1 , a target gene for FOXO3, enhances the homeostatic turnover of memory CD8 T cells [43]. Surprisingly, despite the suggested importance of FOXO3 in regulating the homeostasis of memory T cells, FOXO3 deficiency exerted minimal effects on the proliferative renewal of antigen-specific memory CD8 T cells in vivo [48]. Studies of human memory T cells have ascribed a negative regulatory role for FOXO3 in the persistence of memory CD4 T cells and FOXO3 deficiency would be expected to increase the number of effector memory cells [48,73]. However, FOXO3 deficiency did not affect the relative proportions of central and effector memory CD8 T cells. It is plausible that FOXO3 might regulate the persistence of central/effector memory CD4 T cells, and not CD8 T cells. Alternatively, FOXO3 function may be redundant in maintaining fully differentiated memory CD4 and CD8 T cells.
In conclusion, this manuscript documents that FOXO3 plays a critical role in controlling the clonal burst size and the magnitude of CD8 T cell memory by T cell intrinsic mechanisms. Furthermore, the enhanced number of memory CD8 T cells induced by FOXO3 deficiency, is maintained for extended periods without compromising its quality. These findings have important implications in vaccine development, and suggest that modulation of FOXO3 activity during the expansion phase might be a fruitful

Mice and viral infection
The generation and characterization of the global FOXO3deficient (FOXO32/2) mice on the C57BL/6 (B6) background have been described previously [40]. The control wild type B6 (+/ +) mice were either littermates or purchased from the National Cancer Institute (Bethesda, MD). Derivation of mice carrying the floxed FOXO3 alleles has been described elsewhere [69,70]. Mice carrying the floxed FOXO3 alleles were bred with the CD4-Cre mice at UW-Madison to generate the T cell-specific FOXO32/2 (FOXO3L) mice. Littermate +/+ mice were used as controls with the FOXO3L mice. Mice used in experiments were between the ages of 6-8 weeks and all experiments were performed in accordance with the protocols approved by the University of Wisconsin School of Veterinary Medicine Institutional Animal Care and Use Committee (IACUC). The animal committee mandates that institutions and individuals using animals for research, teaching, and/or testing much acknowledge and accept both legal and ethical responsibility for the animals under their care, as specified in the Animal Welfare Act (AWA) and associated Animal Welfare Regulations (AWRs) and Public Health Service (PHS) Policy. Mice were infected with 2610 5 PFU of lymphocytic choriomeningitis virus (LCMV) Armstrong strain by intraperitoneal (IP) injection. Mice that have recovered from an infection with LCMV Armstrong were challenged with LCMV-Clone 13 (2610 6 PFU by intravenous injection). Tissue viral titers were quantified by plaque assay with Vero cell monolayers [74].

General flow cytometry
Single cell suspensions of splenocytes were stained with antibodies for surface markers including CD8, CD44, CD122, CD127, CD62L, CCR7, LFA-1 and KLRG-1 (BD Biosciences, Franklin Lakes NJ, eBIOSCEINCE, San Diego CA or Southern Biotech, Birmingham AL) in conjunction with MHC I tetramers (D b ) specific for the class I-restricted LCMV epitopes, NP396, GP33, and GP276 as previously described [15]. Cells were fixed in 2% paraformaldehyde (PFA) and acquired in a FACSCalibur or LSR II flow cytometer (BD Biosciences, Franklin Lakes NJ). To quantify intracellular cytokine production, splenocytes were incubated for 5 hours at 37uC with LCMV epitope viral peptides in the presence of Brefeldin A. After stimulation, cells were first incubated with antibodies for surface markers. Next, cells were permeabilized and stained for intracellular cytokines (IFNc, IL-2 and TNFa) using the Cytofix/Cytoperm kit (BD Biosciences, Franklin Lakes NJ). The percentages of cytokine-producing cells were quantified by flow cytometry.

Phospho-specific flow cytometry
Splenocytes were stained for cell surface markers as above. After cell surface staining, cells were fixed and permeabilized using Phosflow lysis and Phosflow PermWash I reagents (BD Biosciences, Franklin Lakes NJ) according to the manufacturer's recommendations. Next, cells were blocked for 30 minutes on ice in blocking buffer (10% normal goat serum in 2%BSA/PBS) and subsequently stained with either phospho-specific antibodies (Cell Signaling Technology, Danvers MA; P-Akt [T308], P-FOXO1/3 [T24,T32], P-mTOR [S2448]) or non-phospho state-specific antibodies (Akt, mTOR, FOXO3, BIM). As negative controls for staining, antibodies were pre-incubated/blocked with their specific antigenic peptide for 1 hr at room temperature before adding on to the cells. Following incubation with antibody or peptide blocked antibody, cells were washed twice and incubated with secondary antibody (Goat anti-Rabbit ALEXA488; Sigma-Aldrich, St. Louis MO) for 40 minutes. Cells were washed and fixed with 2% PFA. The levels of phospho-specific staining were quantified by flow cytometry. Specific levels of staining (Corrected Mean Fluorescence Intensity [MFI]) were calculated using the formula: difference of observed MFI for the phospho-specific protein and peptide-blocked control, divided by the peptideblocked control.

Assessment of in vivo proliferation
To assess in vivo proliferation of antigen specific cells, mice were administered an IP injection of 1.5 mg of 5-Bromo-29-deoxyuridine ([BrdU] MP Biomedicals, Solon OH) followed by exposure to 0.8 mg/ml of BrdU in drinking water for the rest of the pulse period. Splenocytes were stained for surface markers and determination of BrdU positive cells was performed using a BrdU staining kit (BD Biosciences).

Ki67, Bcl-2 and Granzyme B staining
Splenocytes were stained for surface markers and MHC I tetramers as described above. After surface staining, cells were fixed and permeabilized using FACS Lysing Solution and FACS Permeabilization Solution 2 reagents (BD Biosciences, Franklin Lakes NJ) and subsequently incubated with antibodies against Ki67, Bcl-2 or Granzyme B (BD Biosciences, Franklin Lakes NJ) for 45 minutes at room temperature. Virus-specific CD8 T cells staining positive for Ki67, Bcl-2 or Granzyme B were visualized using a FACSCalibur flow cytometer. Data is expressed as either a percentage of antigen-specific CD8 T cells positive for the respective protein or MFI for the indicated protein.

Assessment of apoptosis ex vivo
After 6 days PI, splenocytes from +/+ and FOXO32/2 or FOXO3L mice were isolated and stained with anti-CD8 and MHC-I tetramers as described above except no red blood cell lysis was performed. Annexin V staining (BD Biosciences, Franklin Lakes NJ ) was then carried out according to the manufacturer's protocol, with the exception that all staining was performed on ice. The percentage of Annexin V high cells amongst antigen-specific CD8 T cells was determined by flow cytometry and expressed accordingly.

CD4 depletion studies
Mice were depleted of CD4 T cells through IP administration of 100 mg of the monoclonal antibody GK1.5 (eBioscience, San Diego CA) at days 0 and 4 relative to LCMV infection.

Western analysis
T cells and non-T cells were purified from spleens of +/+ and FOXO3L mice using the anti-CD90.2 based MACS cell separation system (Miltenyi Biotec, Auburn CA). Purity of cells was .93%. Cells were subsequently lysed in buffer (50 mM HEPES, 100 Mm NaCl, 10 mM EDTA, 10 Mm NaF, 4 Mm Na(PO 4 ) 2 , 1% Triton X-100, 5 mg/ml Aprotinin, 1 Mm Phenylmethylsulfonylflouride), sonicated, and total protein levels in each lysate were determined by the Bicinchoninic Acid protein assay. (Sigma-Aldrich, St. Louis, MO). 20 mg samples were loaded and resolved on a 10% SDS-PAGE. Total levels of FOXO3 protein in each sample were detected using a Rabbit primary antibody specific for FOXO3 (Cell Signaling Technology, Danvers MA) followed by a Donkey anti Rabbit F(ab) 2 fragment HRPconjugated secondary antibody (Thermo Fisher, Rockford IL) Bands were visualized using chemiluminescence reagents (Thermo Fisher, Rockford IL) and presented by use of an HP Deskscan system (Hewlett-Packard, Palo Alto CA). Blots initially probed for FOXO3 were subsequently stripped and re-probed with b-Actin (Sigma-Aldrich, St. Louis MO) to serve as a loading control. Figure S1 Early investigation of proliferation in +/+ and FOXO32/2 mice during LCMV infection. After 5 days of infection with LCMV, splenocytes were isolated from +/+ and FOXO32/2 mice and stained with anti-CD8, MHC-I tetramer and anti-Ki67. The percentage of Ki67 positive cells amongst tetramer binding CD8 T cells was determined by flow cytometry. Data is the mean from at least 2 individual experiments with 4-6 mice/group/experiment. (TIF) Figure S2 Representative plots of BIM and Bcl-2 MFIs from day 6 PI in +/+ and FOXO32/2 mice after 6 hours of culture with and without IL-15. Splenocytes from +/+ and FOXO32/2 mice were isolated after 6 days PI. Splenocytes were cultured for 6 hours with and without IL-15 (10 ng/ml). After 6 hours, splenocytes were stained with anti-CD8, MHC-I tetramers and anti-BIM, anti-Bcl-2 or the respective isotype control. Representative tracings of BIM (top) or the isotype control from pooled samples without (left) or with 10 ng/ml IL-15 (right) are shown. Representative Bcl-2 or isotype control MFIs (bottom) in +/+ and FOXO32/2 mice without (left) or with 10 ng/ml IL-15 (right) are also illustrated. Data are from pooled samples from at least two experiments. (TIF) Figure S3 Characterization of FOXO32/2 and FOXO3L mice. (A) Mononuclear cells from spleens of +/+, FOXO3+/2 and FOXO32/2 mice were isolated and stained with anti-CD8, MHC-I tetramer, anti-FOXO3 antibody or FOXO3 antibody pre-incubated with its specific antigenic peptide. Histograms represent the MFI of FOXO3 signal for each FOXO3 variant. Numbers represent MFI of FOXO3 for each group. (B) Mononuclear cells from spleens of +/+ and FOXO3L mice were stained for anti-CD8, anti-CD4, anti-B220, anti CD11c and anti-FOXO3. Histograms represent the MFI of the FOXO3 signal gated on either total positive CD8s, CD4s, B220 or CD11c as indicated. Numbers represent MFI of FOXO3 for each group. (C) Splenocytes from naïve +/+ and FOXO3L mice were collected and T cell and non-T cell fractions were purified using the MACS system (Miltenyi Biotec, Auburn CA) by positive selection for CD90.2. SDS-PAGE followed by immunoblotting with anti-FOXO3 shows preferential loss of FOXO3 in the T cell compartment of FOXO3L mice but not +/+ mice. b-Actin probing was used to ensure equal loading. (D) After 8 days of LCMV infection, mononuclear cells from spleens of +/+, FOXO32/2 and FOXO3L mice were isolated and stained with anti-CD8, MHC-I tetramer, and anti-FOXO3 antibody. Histograms represent the FOXO3 signal from GP33 +ve CD8 T cells for each group. Numbers are the MFI of FOXO3 for each respective group. (TIF) Figure S4 Investigation of proliferation during early infection in +/+ and FOXO3L mice. At 5 days PI, lymphocytes from spleens of +/+ and FOXO3L mice were collected and stained with anti-CD8, MHC-I tetramers and anti-Ki67. The percentage of Ki67 +ve cells amongst tetramer binding CD8 T cells was determined by flow cytometry. Data is the mean from at least 2 individual experiments with 3-6 mice/group/ experiment. (TIF) Figure S5 Total levels of BIM and Bcl-2, measured directly ex-vivo at 6 days PI. Total levels of immuno-reactive BIM (left) or Bcl-2 (right) in tetramer positive CD8 T cells from spleen, liver and lymph nodes were assessed directly ex-vivo on day 6 PI. Each solid bar represents the observed MFI for BIM or Bcl-2, and each checkered bar represents the MFI for the respective isotype control. (TIF)