Transcription factor Foxo1 represses T-bet mediated effector functions and promotes memory CD8⁺ T cell differentiation

SUMMARY

The evolutionary conserved Foxo transcription factors are important regulators of quiescence and longevity. Although, Foxo1 is known to be important in regulating CD8⁺ T cell trafficking and homeostasis, its role in functional differentiation of antigen stimulated CD8⁺ T cells is unclear. Herein, we demonstrate that inactivation of Foxo1 was essential for instructing T-bet transcription factor-mediated effector differentiation of CD8⁺ T cells. The Foxo1 inactivation was dependent on mTORC1 kinase, as blockade of mTORC1 abrogated mTORC2 mediated Akt (Ser473) kinase phosphorylation, resulting in Foxo1-dependent switch from T-bet to Eomesodermin transcription factor activation and increase in memory precursors. Silencing Foxo1 ablated interleukin-12 and rapamycin enhanced CD8⁺ T cell memory responses, and restored T-bet mediated effector functions. These results demonstrate an essential role of Foxo1 in actively repressing effector or terminal differentiation processes to promote memory CD8⁺ T cell development, and identify the functionally diverse mechanisms utilized by Foxo1 to promote quiescence and longevity.

INTRODUCTION

The availability and abundance of extracellular cues like antigen, co-stimulation and cytokines program differentiation of naïve CD8⁺ T cells for various effector and memory functions (Curtsinger et al., 2003). The role of these cell-extrinsic factors in regulating expression of master transcription factors that determine choice between short-lived effector and long-lived memory CD8⁺ T cells is being increasingly appreciated (Intlekofer et al., 2005; Joshi et al., 2007). Notably, the presence of pro-inflammatory cytokines like IL-12, enhance expression of transcription factors like T-bet to promote differentiation of CD8⁺ T cells into short-lived effectors, rather than long-lived memory cells (Joshi et al., 2007; Takemoto et al., 2006). In contrast, the transcription factor Eomesodermin (Eomes), which is required for important aspects of memory CD8⁺ T cell generation, is dampened by IL-12 (Intlekofer et al., 2005; Rao et al., 2010). Since inflammatory cytokines like IL-12 exert pluripotent effects on host immunity, rational strategies to regulate effector versus memory CD8⁺ T cell differentiation will require targeting cell-intrinsic mechanisms that regulate expression of transcriptional factors like T-bet and Eomes.

Several studies have identified a key role for the energy sensitive kinase mammalian target of rapamycin (mTOR), in regulating effector and memory differentiation of CD8⁺ T cells by regulating expression of master transcriptional factors T-bet and Eomes (Araki et al., 2009; Pearce et al., 2009; Rao et al., 2010). The mTOR protein is an evolutionarily conserved member of the phosphatidylinositol-3-OH kinase (PI3K)-related kinase family, whose ability to complex with various adaptors gives rise to two known variants, namely mTORC1 and mTORC2 (Wullschleger et al., 2006). The activation of mTORC1 can occur by signals generated via the PI3K and Akt pathways, and its ability to regulate activity of its downstream effectors like p70 S6 kinase (S6K) and translation repressor protein 4EBP1, imparts mTORC1 with multiple essential functions like protein translation, cell cycle regulation, differentiation, survival, etc (Wullschleger et al., 2006). In contrast, the upstream modulators of mTORC2 are yet to be fully determined, although the phosphorylation of Akt on Ser 473 (S473) has been used as a functional measure of mTORC2 activity (Guertin et al., 2006). Although TORC1 activity is essential for type I differentiation in both CD4⁺ and CD8⁺ T cells (Delgoffe et al., 2011; Rao et al., 2010), two independent studies using Rictor^−/− CD4⁺ T cells have shown divergent requirements for mTORC2 in T helper 1 (Th1) cell differentiation (Delgoffe et al., 2011; Lee et al., 2010). Furthermore, the role of mTORC2 in effector and memory CD8⁺ T cell differentiation and the interplay between the two mTOR complexes for the regulation of CD8⁺ T cell differentiation has not yet been reported. In addition, the mechanisms by which mTORC1 regulate transcriptional programs that impart terminal differentiation versus self-renewal potential remains enigmatic.

The forkhead box (Fox) family of transcription factors, which are mammalian orthologs of the Caenorhabditis elegans longevity gene DAF-16, have an evolutionarily conserved function in the regulation of organismal longevity, energy metabolism and tumor suppression (Lin et al., 1997; Paik et al., 2007). There are four members of the Foxo gene family in mammals: Foxo1, Foxo3, Foxo4 and Foxo6, which regulate key aspects of cell physiology, such as cell cycle progression, survival, differentiation, nutrient sensing, and response to stress (Burgering, 2008; Salih and Brunet, 2008). Many of these processes are also known targets of the mTOR protein complexes (Wullschleger et al., 2006). In addition to a variety of post-translational modifications, the activity of Foxo proteins is mainly regulated by the PI3K-Akt mediated phosphorylation on three conserved sites, leading to nuclear to cytoplasm export, degradation and decrease in transcriptional activity (Salih and Brunet, 2008). In the immune system, Foxo3a deficient T cells have been shown to undergo spontaneous proliferation and increased differentiation towards the Th1 cell phenotype (Lin et al., 2004). Recently, two of the Foxo proteins, Foxo1 and Foxo3a were shown to cooperatively enhance FoxP3 expression and dictate regulatory T (Treg) cell lineage commitment (Harada et al., 2010; Kerdiles et al., 2010). In addition, Foxo1 deficiency leads to impaired T cell trafficking and homeostasis partly due to its ability to regulate expression of the transcription factor Klf-2 and IL-7 receptor alpha (IL-7Rα) (Gubbels Bupp et al., 2009; Kerdiles et al., 2009; Ouyang et al., 2009); factors known to be enhanced upon rapamycin mediated mTORC1 inhibition (Araki et al., 2009; Sinclair et al., 2008). Despite the known ability of Foxo proteins to promote cellular and organismal longevity, its role in regulating CD8⁺ T cell differentiation for effector and memory functions has not been reported.

Herein, by employing a reductionist and in vivo approach, we identify an essential role for Foxo1 in regulating T-bet and Eomes expression for effector versus memory functional fate of CD8⁺ T cells. Our results demonstrate that instructions that program naïve CD8⁺ T cells for type I effector differentiation inactivate Foxo1 via mTORC1 and mTORC2 dependent Akt phosphorylation. Foxo1 actively represses type I effector maturation by blocking T-bet expression, and promotes Eomes gene transcription and memory precursor phenotype. Inhibition of mTORC1 enhances Foxo1 activity, which is essential for rapamycin mediated block in T-bet, and increase in Eomes expression. Importantly, experiments with shRNA knock-down implicate an essential role for Foxo1 in persistence and antigen-recall responses. These results have identified a critical role for Foxo1 in regulating transcriptional programs to determine functional differentiation of CD8⁺ T cells into effector and memory cell subsets.

RESULTS

Instructions that program naïve CD8⁺ T cells for type I effector differentiation inactivate transcription factor Foxo1

The differentiation of naïve CD8⁺ T cells into robust effector cells occurs at the expense of memory (Joshi et al., 2007; Williams and Bevan, 2007). Since Foxo1 promotes phenotype associated with memory precursor cells (CD62L, IL-7Rα and Bcl-2) (Kerdiles et al., 2009; Ouyang et al., 2009), we hypothesized that instructions, antigen (Ag), co-stimulation (B7.1) and pro-inflammatory cytokine (IL-12), that program type I effector functions in CD8⁺ T cells must inhibit Foxo1 expression and/or functions. To test this notion we first verified the ability of IL-12 to regulate Foxo1 phosphorylation in Ag and B7.1 (Ag+B7.1) stimulated OT-I cells. Stimulation of naïve OT-I cells with Ag+B7.1 induced rapid (2h) but transient phosphorylation of Foxo1, as the optimal phosphorylation observed at 12h was reduced to baseline levels by 48h (Figure 1A). Notably, addition of IL-12 produced enhanced and persistent phosphorylation of Foxo1, with maximum differences observed at later time-points (Figure 1A). These results demonstrate that instructions that program naïve CD8⁺ T cells for type I effector differentiation enhance and sustain the phosphorylation of transcription factor Foxo1.

Figure 1. Instructions that program naïve CD8⁺ T cell for type I effector maturation inactivate transcription factor Foxo1.

(A–E) Naïve OT-I cells stimulated with Ag (SIINFEKL, 10nM) plus B7.1 (Ag+B7.1) (+/−) IL-12 (2ng/ml) were evaluated for (A) phosphorylation of Foxo1 by intracytoplasmic staining (ICS) and flow cytometry at the indicated time points, (B) sub-cellular localization of Foxo1 by Image-stream based flow cytometry at 48h. Representative examples of bright-field (BF), total Foxo1-FITC, DAPI, and a merge image of the two stains is shown. The far right-top bar graph represents the similarity co-efficient between total Foxo1 and DAPI, obtained from pixel by pixel statistical analysis of each cell (n=10000) analyzed. The far right-bottom bar graph represents the intensity of Foxo1 protein in the cytoplasm; *p< 0.019 (C) total Foxo1 protein by ICS at 72h, (D) mRNA for Klf-2 at 48h by RT-PCR; *p< 0.029, and (E) cell cycle analysis by propidium iodide labeling at 48h; *p< 0.033. Experiments shown are representative of at least three (A–C) and two (D and E) independent experiments with similar outcomes. (Data are represented at mean +\− SEM)

Since phosphorylation of Foxo1 leads to its nuclear export, degradation and loss of transcriptional activity, we used Image Stream-based flow cytometry analysis to determine whether Ag+B7.1 and/or IL-12 stimulation promotes an increase in cytoplasmic amounts of Foxo1. As shown in (Figure 1B), Ag+B7.1 stimulation shifted the predominantly nuclear-Foxo1 in naïve OT-I cells to the cytoplasm, which was further enhanced in the presence of IL-12. In addition, the increased phosphorylation and cytoplasmic retention of Foxo1 in IL-12 conditioned OT-I cells was associated with a marked reduction in total Foxo1 protein content (Figure 1C). Notably, the reduction in Foxo1 total protein amounts occurred only at later time–points (72h). Furthermore, we observed a significant decrease in Klf-2 mRNA expression (a direct gene target of Foxo1 in T cells (Kerdiles et al., 2009)) (Figure 1D) and an increase in cell cycle progression (Figure 1E) (Foxo transcription factors inhibit cell cycle progression (Salih and Brunet, 2008)); thereby demonstrating a loss of Foxo1 activity in IL-12 conditioned OT-I cells. Collectively, these results identify Foxo1 as a target of instructions that program naïve CD8⁺ T cells for type I effector differentiation, and suggests a potential role for Foxo1 in type I effector differentiation of CD8⁺ T cells.

Instructional augmentation of mTORC1 is required for increased mTORC2 activity

The phosphorylation of Foxo1 is controlled, in part, by phosphorylation of Akt at Thr308 (T308) (downstream of PI3K) and S473 (hydrophobic motif, downstream of mTORC2). To characterize the pathways utilized by extracellular instructions to augment Foxo1 phosphorylation in CD8⁺ T cells, we evaluated the ability of Ag+B7.1 and/or IL-12 to induce Akt phosphorylation in OT-I cells. As shown in (Figure 2A), Ag+B7.1 stimulated OT-I cells rapidly (2h) enhanced phosphorylation of Akt at T308, which was further enhanced and sustained at late time-points by IL-12. In contrast, the induction of Akt phosphorylation at S473 was delayed, and was enhanced only by 12–24h (Figure 2B). However, similar to Akt T308 phosphorylation, the Ag+B7.1 induced Akt S473 phosphorylation was transient, and was maintained at higher levels only in the presence of IL-12 (Figure 2B). In addition, the increase in phosphorylation of Akt at T308 and S473 required PI3K activity (Figure S1A), which is consistent with an upstream role for PI3K in regulating PDK1 mediated Akt T308 phosphorylation and mTORC2 activity (Wullschleger et al., 2006). Because Akt phosphorylation on T308 preceded S473, it is reasonable to suggest that the PI3K mediated Akt activation precedes mTORC2 activation. Overall, these observations identify a role for instructions to enhance and sustain mTORC2 activity in CD8⁺ T cells and suggest a potential role for mTORC2 in IL-12 mediated type I effector differentiation.

Figure 2. mTORC1 activity is required for enhanced mTORC2 activity.

(A–B) Naive OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 and rapamycin (20ng/ml) were evaluated by ICS at the indicated time-points for (A) Akt phosphorylation at T308, and (B) Akt phosphorylation at S473. For mTOR inhibition, rapamycin was added 30 minutes prior to addition of Ag+B7.1+IL-12. (C) Naive OT-I cells stimulated with Ag+B7.1 (and/or) IL-12 were transduced with control or Raptor RNAi-GFP retroviral vector (RV), cultured for 4 days, re-stimulated and evaluated for Akt T308 and S473 phosphorylation. (D) Naive OT-I cells stimulated with Ag+B7.1 (and/or) IL-12 were transduced with control or DN-Akt GFP RV, cultured for 4 days, re-stimulated and evaluated for Foxo1 phosphorylation. Experiments shown are representative of at least three (A–B) and two (C–D) independent experiments with similar outcomes. (See also Figure S1)

Previously, we have reported that the presence of IL-12 during Ag+B7.1 stimulation induces PI3K-Akt dependent sustained mTORC1 activity in OT-I cells (Rao et al., 2010). Although, the upstream mechanisms underpinning mTORC2 activation remain elusive, recent studies have suggested that mTORC1 and mTORC2 may be linked via ribosomal proteins (Xie and Guan, 2011). To examine a possible role for Ag+B7.1+IL-12 enhanced mTORC1 in mTORC2 activity, we blocked mTORC1 activity by rapamycin and evaluated for Akt S473, as a functional measure of mTORC2 activity. Indeed, mTORC1 inhibition (rS6 phosphorylation (S6p); Figure S1B) blocked the ability of IL-12 to sustain phosphorylation of Akt at S473 at late time-points (24–72h) (Figure 2B). Notably, rapamycin treatment consistently induces an acute increase in Akt S473 phosphorylation in OT-I cells, as reported previously in other cell types (Sarbassov et al., 2006), which may be due in part to a cell stress response. The mTOR complex-2 is considered to be rapamycin-insensitive, and the inability of rapamycin to inhibit Akt S473 phosphorylation at early time-points confirms this notion. This is further substantiated by the fact that short-term rapamycin treatment (4h) that causes acute mTORC1 inhibition, failed to inhibit mTORC2 activity at later time points (48h) (Figure S1C), consistent with the requirement of prolonged rapamycin treatment to inhibit mTORC2 activity (Sarbassov et al., 2006). Importantly, mTORC1 inhibition causes no measurable changes in Akt phosphorylation on T308 (Figure 2A), indicating that Akt T308 is upstream of mTORC1, whereas mTORC2 and Akt S473 are downstream of mTORC1. These observations are further supported by the fact that inhibition of PI3K (PI3K acts upstream of mTORC1) affects phosphorylation on both Akt residues (Figure S1A), whereas mTORC1 inhibition affects only mTORC2 mediated Akt S473 phosphorylation, but not PI3K mediated phosphorylation on T308.

To confirm the requirement of mTORC1 in sustained mTORC2 activity observed in Ag+B7.1+IL-12 stimulated CD8⁺ T cells, we specifically disrupted mTORC1 activity by using retrovirus based RNA-interference (RNAi) to knock-down the adaptor gene Raptor. Retrovirus infected cells marked by GFP cells were re-stimulated and evaluated 48h post-stimulation for Akt S473 and T308 phosphorylation in Raptor or control retrovirus infected cells. Identical to our observations with rapamycin treatment (Figure 2A, B), specific knock down of mTORC1 (S6p; Figure S1D) abrogated IL-12 enhanced mTORC2 activity, with no detectable changes in Akt phosphorylation at T308 (Figure 2C). These results confirm the upstream role of mTORC1 in sustained mTORC2 mediated S473 phosphorylation of Akt in CD8⁺ T cells, and uncover the existence of cross-talk between the two TOR complexes. mTORC2 has the ability to regulate type I effector functions in CD4⁺ T cells (Lee et al., 2010), and identification of a cross-regulatory pathway between the two TOR complexes sheds new light into their role in regulating T cell functional differentiation.

To directly test the requirement for increased Akt activity in IL-12 enhanced Foxo1 phosphorylation, we ectopically expressed a dominant negative form of Akt (DN-Akt) in IL-12 conditioned OT-I cells and evaluated for Foxo1 phosphorylation. Indeed, DN-Akt expression which dampened Akt T308 and S473 phosphorylation (Figure S1E), abrogated IL-12 enhanced Foxo1 phosphorylation (Figure 2D), corresponding with an increase in expression of Foxo1 target genes; indicative of its transcriptional activity (Figure S1F). These results confirm that Ag+B7.1 and IL-12 mediated Akt phosphorylation is required for enhanced Foxo1 phosphorylation in CD8⁺ T cells.

mTORC1 inactivates Foxo1 via Akt dependent mechanisms

Based on our observations that IL-12 enhances Foxo1 phosphorylation via Akt dependent mechanisms, and IL-12-enhanced mTORC1 activity is required to sustain mTORC2 dependent Akt S473 phosphorylation, we envisaged that sustained mTORC1 activity was required for increased phosphorylation and inactivation of Foxo1. To test this notion, we stimulated naïve OT-I cells with Ag+B7.1+IL-12 in the presence of rapamycin, and performed kinetic analysis of Foxo1 phosphorylation. Similar to rapamycin induced block in Akt S473 phosphorylation (Figure 2B), the enhanced and sustained phosphorylation of Foxo1 in IL-12 conditioned OT-I cells was lost upon mTORC1 inhibition (Figure 3A). The decrease in phosphorylation was accompanied by a change in its sub-cellular localization, as mTORC1 inhibition led to a significant decrease in Foxo1 protein in the cytoplasm and a subsequent increase in its nuclear content (Figure 3B). Furthermore, inhibition of mTORC1 enhanced total Foxo1 protein levels (Figure 3C), along with corresponding increases Foxo1 target genes (Figure S2A–C). These results demonstrate an essential role for mTORC1 in regulating Foxo1 activity in CD8⁺ T cells, and implicate Foxo1 as a potential intermediary in mTORC1 regulated CD8⁺ T cell differentiation.

Figure 3. mTORC1 activity is required for inactivation of Foxo1.

(A–C) Naive OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 and rapamycin were evaluated for (A) Foxo1 phosphorylation by ICS at the indicated time-points, (B) sub-cellular localization of Foxo1 by Image-stream based flow Cytometry at 48h. Representative examples of bright-field (BF), total Foxo1-FITC, DAPI, and a merge image of the two stains is shown. The far right-top bar graph represents the similarity co-efficient between total Foxo1 and DAPI, obtained from pixel by pixel statistical analysis of each cell (n=10000) analyzed. The far right-bottom bar graph represents the intensity of Foxo1 protein in the cytoplasm; *p<0.011 and *p<0.0086 and (C) total Foxo1 protein expression by ICS at 72h. (D) Naive OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 and rapamycin were transduced with control or Myr-Akt GFP RV, cultured for 4 days, re-stimulated and evaluated for Foxo1 phosphorylation by ICS. Experiments shown are representative of at least three (A–C) and two (D) independent experiments with similar outcomes. (See also Figure S2)

We next evaluated whether the loss of Akt phosphorylation upon rapamycin treatment is responsible for decrease in phosphorylation of Foxo1. To test this, we transduced rapamycin treated, IL-12 conditioned OT-I cells with myristoylated-Akt (constitutively active form, myr-Akt) and evaluated for Foxo1 phosphorylation. Indeed, active Akt restored the loss of Foxo1 phosphorylation observed upon rapamycin treatment (Figure 3D) and led to a decrease in Foxo1 transcriptional activity (Klf-2 mRNA expression) (Figure S2D). These results demonstrate the ability of mTORC1 to inactivate Foxo1, via Akt dependent mechanisms, and indicate a potential role for Foxo1 in mTORC1 regulated T-bet expression in CD8⁺ T cells.

Inactivation of Foxo1 is required for enhanced T-bet expression in CD8⁺ T cells

Since mTORC1 inhibition increases Foxo1 activity (Figure 3), inhibits T-bet expression and blocks type I effector maturation (Rao et al., 2010), we hypothesized that the ability of IL-12 to enhance Foxo1 phosphorylation, and reduce its activity is essential for enhanced and sustained T-bet expression in CD8⁺ T cells. To test this notion, we transduced Ag+B7.1 (with or without) IL-12 conditioned OT-I cells with Foxo1-ER-Thy1.1 retroviral vector, wherein the expression of Foxo1 is regulated by Tamoxifen (Tm.). Addition of Tm led to an increase in total Foxo1 protein amounts in IL-12 conditioned-Thy1.1⁺ OT-I cells (Figure 4A), along with an increase in Klf-2 gene expression (Figure S3A). Thus, ectopic Foxo1 expression restores IL-12-mediated loss of Foxo1 protein and its activity. The over-expression of Foxo1 inhibited IL-12 enhanced T-bet mRNA (Figure 4B) and protein expression (Figure 4C), leading to a decrease in CD122 expression (IL-15Rβ), a direct gene target of T-bet in CD8⁺ T cells. Notably, Ag+B7.1 induced T-bet protein expression was also abrogated upon ectopic Foxo1 expression (Figure S3B). To further implicate the role of Akt-Foxo1 in regulating T-bet, we induced ectopic expression of DN-Akt (inhibits Foxo1 phosphorylation (Figure 2D), and increases its activity (Figure S3C)) in IL-12 conditioned OT-I cells. Indeed, DN-Akt abrogated the IL-12 mediated increase in T-bet expression (Figure 4D). These results demonstrate that transcription factor Foxo1 can act to repress T-bet expression, while promoting expression of target genes like Klf-2 and CD62L.

Figure 4. Foxo1 represses T-bet expression in CD8⁺ T cells.

(A) Naive OT-I cells stimulated with Ag+B7.1+IL-12 were transduced with Foxo1-ER-Thy1.1 RV, cultured for 4 days (+/−) Tm (10nM), and evaluated for total Foxo1 protein expression on Thy1.1⁺ cells by ICS. (B–C) Transduced OT-I cells from (A) were re-stimulated (+/−) Tm, and evaluated for (B) T-bet mRNA expression by RT-PCR on sorted Thy1.1⁺ OT-I cells; **p<0.0023; and (C) T-bet protein and CD122 expression on Thy1.1⁺ cells (D) Naive OT-I cells stimulated with Ag+B7.1+IL-12 were transduced with control or DN-Akt GFP RV, cultured for 4 days, re-stimulated and evaluated for Foxo1 phosphorylation. (E) Purified CD8 SP thymocytes from WT and Foxo1^−/− mice were stimulated with anti-CD3 and anti-CD28 (+/−) IL-12 and evaluated 72h post-stimulation for T-bet protein expression. Experiments shown are representative of three (A–C) and two (D,E) independent experiments with similar outcomes. (Data are represented at mean +\− SEM). (See also Figure S3)

To confirm the role of Foxo1 as a repressor of T-bet expression, we next evaluated for T-bet protein expression in CD8⁺ T cells derived from mice with a T-cell specific deficiency of Foxo1 (Foxo1^−/−). Since the T cells from Foxo1^−/− mice have a defect in homeostatic maintenance (Gubbels Bupp et al., 2009; Kerdiles et al., 2009), we purified CD8 SP mature T cells from the thymus of Foxo1^−/− mice and confirmed the loss of Foxo1 expression and its target genes (Figure S3D, E). Remarkably, Foxo1^−/− CD8⁺ T cells showed increased basal T-bet expression (data not shown), which was enhanced by stimulation with Ag+B7.1 alone, and with IL-12 (Figure 4E), indicating that loss of Foxo1 itself is sufficient to promote T-bet expression in naïve and antigen stimulated CD8⁺ T cells. These observations identify a role for Foxo1 in regulating T-bet expression, and demonstrate that the ability of IL-12 to inactivate Foxo1 is essential to promote T-bet expression in CD8⁺ T cells.

Given that Foxo1 can act both as a transcriptional activator and repressor (Salih and Brunet, 2008), we sought to determine whether Foxo1 inhibits T-bet expression by direct DNA binding. The Foxo family of transcription factors binds to a consensus DNA sequence of (T/C/G)(T/C/G)T(G/A)TTTT(A/G/T) (Paik et al., 2007). Detailed bioinformatics analysis identified three putative forkhead-binding sites on the T-bet promoter region (Figure S3F). To determine whether Foxo1 can directly bind within these sites, we performed chromatin immunoprecipitation (ChIP) experiments using primer sets designed to amplify regions located in each of these sites. We were unable to detect direct binding of Foxo1 on the T-bet promoter at all three sites (Figure S3G). However, binding of Foxo1 to the IL-7R locus from the same ChIP lysates was observed (Figure S3G), suggesting that Foxo1 inhibits T-bet expression via mechanisms that are independent of direct DNA binding, and may mediate the repressive action via indirect mechanisms.

Foxo1 regulates type I effector maturation in T-bet dependent manner

PDK1 can control TCR signal-induced IFN-γ production via Foxo dependent mechanisms in CD8⁺ T cells (Macintyre et al., 2011). However, the role of Foxo1 in IL-12 enhanced and sustained type I effector differentiation of CD8⁺ T cells has not been reported. Given the importance of IL-12 in inactivating Foxo1 to maintain sustained levels of T-bet expression, we predicted that Foxo1 regulates expression of key effector molecules in CD8⁺ T cells in a T-bet dependent manner. To test this notion, we over-expressed Foxo1 in IL-12 conditioned OT-I cells and evaluated their ability to express effector molecules such as IFN-γ and Granzyme-B. Over-expression of Foxo1 blocked IL-12 enhanced IFN-γ production and granzyme B expression in OT-I cells (Figure 5A, B). In agreement, we also observed increased IFN-γ production and granzyme B expression from Foxo1^−/− cells in the presence or absence of IL-12 (Figure 5C, D). Because type I IFNs, like IFN-α can also enhance T-bet expression and IFN-γ production in CD8⁺ T cells, we determined whether IFN-α also regulates Foxo1 to promote type I effector differentiation in OT-I cells. As shown in (Figure S4A and B), IFN-α enhances Foxo1 phosphorylation along with an increase in IFN-γ production in Ag+B7.1 activated CD8⁺ T cells. These results clearly demonstrate that inactivation of Foxo1 is sufficient to promote type I effector maturation in CD8⁺ T cells.

Figure 5. Foxo1 inhibits type I effector differentiation in a T-bet dependent manner.

(A–B) Naive OT-I cells stimulated with Ag+B7.1/+IL-12 were transduced with Foxo1-ER-Thy1.1 RV, cultured for 4 days (+/−) Tm (10nM), re-stimulated and evaluated on gated Thy1.1⁺ cells for (A) IFN-γ production, and (B) Gzmb expression. (C–D) Purified CD8 SP thymocytes from WT and Foxo1^−/− mice were stimulated with anti-CD3 and anti-CD28 (+/−) IL-12 and evaluated 72h post-stimulation for (C) IFN-γ production, and (D) Gzmb expression. (E) Naive WT or Tbx21^−/− OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 were transduced with control or Foxo1 shRNA-hCD2 RV, cultured for 4 days, re-stimulated and evaluated for IFN-γ production on gated hCD2⁺ cells; *p<0.0275, **p<0.0061 and n.s.-not significant. (F) Naïve OT-I cells stimulated with Ag+B7.1/+IL-12 (+/−) rapamycin were transduced with T-bet-ER RV, cultured for 4 days (+/−) 4-HT (10nM), re-stimulated and evaluated for IFN-γ production; ***p< 0.0005 and **p<0.0025. Experiments shown are representative of three (A, B and F) and two (C–E) independent experiments with similar outcomes. (Data are represented at mean +\− SEM) (See also Figure S4)

The master regulator T-bet imparts type I effector fate in both CD4⁺ and CD8⁺ T cells. To determine whether Foxo1 regulates expression of IFN-γ in a T-bet dependent manner, we silenced Foxo1 in Ag+B7.1 stimulated Tbx21^−/− OT-I cells and evaluated their ability to produce IFN-γ. Decrease in total Foxo1 protein expression, upon retroviral transduction with vector carrying Foxo1 shRNA (Figure S4C), led to a significant increase in IFN-γ production from Ag+B7.1 treated WT OT-I cells (Figure 5E). Knockdown of Foxo1 in Tbx21^−/− OT-I cells failed to increase IFN-γ production, thereby indicating that Foxo1 mediated regulation of IFN-γ production in CD8⁺ T cells is T-bet dependent. In addition, knock-down of Foxo1 in IL-12 conditioned Tbx21^−/− OT-I cells led to modest, but insignificant increase in IFN-γ production (data not shown). Importantly, ectopic re-expression of T-bet in rapamycin treated OT-I cells restored IFN-γ production (Figure 5F), although the direct gene targets of Foxo1 like CD62L remained unchanged (data not shown). These results identify the master transcriptional factor T-bet as a downstream target of Foxo1, and demonstrate that the ability of Foxo1 to dampen type I effector maturation in CD8⁺ T cells is mediated via its actions on T-bet.

Inhibition of mTORC1 leads to loss of T-bet expression via Foxo1 dependent mechanisms

Since mTORC1 inhibition and over-expression of Foxo1 inhibits T-bet expression, we sought to determine whether increase in Foxo1 activity upon mTORC1 inhibition was responsible for loss of T-bet expression and type I effector maturation in CD8⁺ T cells. We stimulated WT or Foxo1^−/− CD8⁺ T cells with Ag+B7.1+IL-12 in the presence or absence of rapamycin and evaluated for T-bet protein expression and IFN-γ production. As shown in (Figure 6A), inhibition of mTORC1 blocked IL-12 enhanced T-bet expression in WT CD8⁺ T cells, whereas rapamycin treatment failed to block T-bet expression in Foxo1^−/− CD8⁺ T cells. Similarly, in agreement with its inability to curtail T-bet expression, rapamycin treatment failed to inhibit IFN-γ production from Foxo1^−/− CD8⁺ T cells (Figure 6B). Since Foxo1 enhances CD62L expression, we next evaluated whether rapamycin mediated increases in CD62L expression requires Foxo1. As shown in (Figure 6C), rapamycin treatment failed to enhance CD62L expression in IL-12 conditioned Foxo1^−/− CD8⁺ T cells. Taken together, these results identify Foxo1 as an essential target of mTORC1 that regulates expression of master transcription factor T-bet for type I effector maturation in CD8⁺ T cells.

Figure 6. Rapamycin inhibits T-bet expression via Foxo1 dependent mechanisms.

(A–C) Purified CD8 SP thymocytes from WT and Foxo1^−/− mice were stimulated with anti-CD3 and anti-CD28 (+/−) IL-12 and rapamycin, and evaluated 72h post-stimulation for (A) T-bet protein expression, (B) IFN-γ production, and (C) CD62L protein expression. (D) Naive OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 and rapamycin were transduced with control or Myr-Akt GFP RV, cultured for 4 days, re-stimulated and evaluated for T-bet protein expression. Experiments shown are representative of two independent experiments with similar outcomes.

To confirm that the loss of Akt-dependent-Foxo1 phosphorylation due to mTORC1 inhibition is responsible for the decrease in T-bet expression, we ectopically expressed myr-Akt in rapamycin treated, IL-12 conditioned OT-I cells, and tested their ability curtail T-bet expression. Indeed, active Akt restored the loss of T-bet expression in rapamycin treated IL-12 conditioned OT-I cells (Figure 6D), along with an increase in IFN-γ production (data not shown). These results indicate that mTORC1, via Akt dependent mechanisms reduce Foxo1 activity and/or expression to enhance T-bet expression and type I effector maturation in CD8⁺ T cells.

Foxo1 regulates Eomes gene transcription and is essential for CD8⁺ T cell persistence and memory functions

Previously, we have shown that inhibition of mTORC1 skews T-bet for Eomes expression, and transitions type I effector cells to memory precursor cells (CD62L^hi, Bcl-2^hi, KLRG1^lo) (Rao et al., 2010). Based on our observations with the ability of Foxo1 to repress T-bet, we hypothesized that increase in Foxo1 activity upon mTORC1 inhibition is responsible for an increase in Eomes expression and memory precursor phenotype in OT-I cells. To test this notion, we first evaluated the ability of ectopic Foxo1 expression to enhance Eomes expression in Ag+B7.1 and/or IL-12 conditioned OT-I cells. In agreement with the inverse expression patterns for T-bet and Eomes (Takemoto et al., 2006), addition of IL-12 decreased Eomes mRNA expression in antigen activated OT-I cells, although we did not observe any changes at the protein level (Figure 7A, B). Over-expression of Foxo1, led to significant increases in Eomes mRNA and protein expression in Ag+B7.1 and/or IL-12 conditioned OT-I cells (Figure 7A, B). These results demonstrate the novel ability of Foxo1 to regulate the relative expression of transcription factors that promote effector (T-bet) versus memory (Eomes) differentiation of CD8⁺ T cells. To confirm whether increase in Foxo1 activity upon mTORC1 inhibition is responsible for increased Eomes expression, we stimulated purified SP CD8⁺ T cells from WT or Foxo1^−/− mice with Ag+B7.1+IL-12 in the presence or absence of rapamycin and evaluated for Eomes protein expression. The expression of Eomes was lower in IL-12 conditioned Foxo1^−/− cells in comparison to their WT counterparts (Figure 7C), confirming the requirement of Foxo1 to induce Eomes expression in CD8⁺ T cells. Moreover, the mTOR inhibition-induced increase in Eomes expression was abrogated in Foxo1^−/− CD8⁺ T cells, thereby demonstrating that the ability of rapamycin to increase Foxo1 activity is essential for enhanced Eomes expression in CD8⁺ T cells. Collectively, these results suggest the possibility that Foxo1 might directly regulate Eomes transcription, and as such, we inspected Eomes promoter for putative Foxo1 binding sites and identified one proximal (158 bp upstream of transcription start site) and two distal Foxo1 binding sites. As shown in (Figure 7D), anti-Foxo1 immunoprecipitated the DNA fragments corresponding to the proximal region of Eomes promoter, but not the distal region. These results demonstrate that Foxo1 induces Eomes gene expression by directly binding to Eomes promoter.

Figure 7. Foxo1 regulates Eomes gene transcription and is essential for memory CD8⁺ T cell functions.

(A–E) Naive OT-I cells stimulated with Ag+B7.1 (+/−) IL-12 were transduced with Foxo1-ER-Thy1.1 RV, cultured for 4 days (+/−) Tm (10nM), and evaluated on Thy1.1⁺ cells for (A) Eomes mRNA expression by RT-PCR; *p<0.021,**p<0.0091, and (B) Eomes protein expression. (C) Purified CD8⁺ T cells from WT and Foxo1^−/− mice were stimulated with anti-CD3 and anti-CD28 (+/−) IL-12 and rapamycin, and evaluated 72h post-stimulation for Eomes protein expression. (D) Schematic structure of the Eomes promoter with putative Forkhead binding sites, and ChIP analysis of Foxo1 binding to the Eomes promoter in OT-I cells. Results analyzed by RT-PCR are presented as fold of template enrichment in immunoprecipitates of Foxo1 antibody relative to control immunoprecipitation (Ig). (E–F) Naive OT-I cells (Thy1.1⁺) stimulated with Ag+B7.1 (+/−) IL-12 and rapamycin were transduced with control or Foxo1 shRNA-hCD2 RV, sorted for hCD2-Thy1.1⁺ cells, adoptively transferred (2 × 10⁶ cells) into BL/6 recipients, and (E) evaluated for the frequency of adoptively transferred cells in the spleen at day 40 post adoptive transfer. The circles indicate OT-I/Thy1.1 population and the numbers indicate percent frequency. (F) The recipient mice were immunized with IFA-OVA on day 40 post transfer and secondary CD8⁺ T cell responses were measured 3 days later. The absolute numbers of adoptively transferred cells before (pre-rechallenge) and after (post-rechallenge) immunization in the spleen, is shown. The numbers in parenthesis indicate fold expansion of CD8α⁺Thy1.1⁺ from day 40 to day 43; *p< 0.0125 and **p< 0.0087. Experiments shown are representative of three (A,B) and two (C–F) independent experiments with similar outcomes. (See also Figure S5)

Given the ability of Foxo1 to regulate expression of memory-associated transcription factor Eomes, we envisaged a possible role for Foxo1 in memory CD8⁺ T cell differentiation. To test this notion, we first evaluated the role of Foxo1 in regulating CD8⁺ T cell memory precursor phenotype. shRNA-mediated knockdown of Foxo1 (Figure S5A) led to considerable decrease in CD62L, CD127, and increase in KLRG1 expression (Figure S5B), indicating the requirement of Foxo1 to promote phenotype associated with enhanced homing to secondary lymphoid compartments, homeostatic renewal and long term survival. To directly test the role of Foxo1 in CD8⁺ T cell memory, we evaluated the persistence (day 40) and antigen recall response (day 43) of adoptively transferred Ag+B7.1 and/or IL-12 and rapamycin treated OT-I/Thy1.1⁺ cells, that were knocked down for Foxo1. As previously reported, the OT-I cells treated with rapamycin demonstrate remarkably greater persistence than IL-12 conditioned OT-I cells (Figure 7E). Knockdown of Foxo1 significantly reduced the ability of Ag+B7.1, IL-12 conditioned and/or rapamycin treated cells to persist, as demonstrated by the decreased frequency OT-I⁺Thy1.1⁺ cells detected on day 40 (Figure 7E and S5C). Moreover, the ability of OT-I cells to undergo vigorous antigen-recall clonal expansion (Figure 7F) and effector function (IFN-γ) (Figure S5D) was significantly impaired in Foxo1 shRNA transduced cells (day 43). These results demonstrate an essential role for transcription factor Foxo1 in programming antigen, IL-12 and rapamycin induced CD8⁺ T cell persistence and functional memory responses.

DISCUSSION

Extracellular instructions direct differentiation of naïve CD8⁺ T cells into effector and memory cell subsets, but these subsets occur at variance to each other, as the factors that promote terminal differentiation into short-lived effector cells will curtail differentiation into long-lived memory cells, and vice-versa. We and others have shown a central role for energy sensitive kinase mTOR in effector and memory differentiation process, as this kinase has the unique ability to integrate extracellular instructions and regulate gene programs associated with effector and memory functions (Araki et al., 2009; Li et al., 2011; Pearce et al., 2009; Rao et al., 2010). However, the precise downstream mechanisms by which mTORC1 regulate expression of these master transcriptional factor genes is not well understood. In this study, we provide direct evidence for a unique role of Foxo1, downstream of mTORC1, in regulating gene programs to actively suppress effector maturation and/or terminal differentiation and promote longevity and functional memory responses in CD8⁺ T cells.

To date studies have demonstrated the ability of Foxo1 to regulate properties like quiescence, homing and survival of naïve CD8⁺ T cells that are required for memory responses, but its ability to regulate transcriptional programs that determine effector and memory functions of CD8⁺ T cells has not been reported. Resting T cells express Foxo1, which is inactivated upon antigen stimulation to enable proliferation and clonal expansion. However, the regulation of Foxo1 by cytokines that program functional differentiation of CD8⁺ T cells has not been characterized. Our results are the first to show that pro-inflammatory cytokine IL-12, augments the phosphorylation and inactivation of Foxo1 in antigen-activated cells, which is essential for T-bet mediated type I effector functions. In addition to increased phosphorylation, IL-12 causes a loss in total Foxo1 protein levels, which may be essential in its ability to impart T-bet-dependent “heritable” type I effector differentiation of CD8⁺ T cells. At this juncture, it is not fully clear as to how IL-12 causes a loss in Foxo1 protein levels. Based on our results and a recent report (Stittrich et al., 2010), it is reasonable to suggest that IL-12 inhibits Foxo1 at the post-transcriptional level, given the fact that we observe loss of Foxo1 protein only at late time points. Since inactivation of Foxo1 by IL-12 leads to a loss of Klf-2 and CD62L expression, the observations provide a mechanism for the exclusion of effector CD8⁺ T cells from the secondary lymphoid organs. The dual role for Foxo1 in augmenting Eomes, Klf-2, CD62L, IL-7Rα, and Bcl-2 expression, and dampening T-bet-mediated type I effector differentiation affords new opportunities to regulate CD8⁺ T cell differentiation for desirable health benefits.

The mTOR complexes (mTORC1 and mTORC2) have been shown to regulate differentiation in a variety of tissues. In CD8⁺ T cells, signals emanating from the TCR, CD28 co-receptor, and cytokines (IL-12, IFN-γ, IL-7 and IL-2) have been shown to modulate mTORC1 activity (Li et al., 2011; Rao et al., 2010; Sinclair et al., 2008); however, relatively little is known about the factors that modulate mTORC2 activity in T cells. Our results have identified PI3K-induced mTORC1 as an upstream regulator of mTORC2 activity, such that increased mTORC1 activity promoted by IL-12 is required to sustain mTORC2 activity in CD8⁺ T cells. A recent study demonstrated that mTORC2 can be activated by direct association with the ribosomes in yeast and mammalian cells (Zinzalla et al., 2011). Since, mTORC1 is a direct regulator of S6 and ribosome biogenesis, it can be suggested that Ag+B7.1 and/or IL-12 induced mTORC1 increases mTORC2 activity by increasing ribosomal levels within CD8⁺ T cells. Although compelling, these concepts need further experimentation. Nevertheless, by identifying a causal link between mTORC1 and mTORC2 in regulating T cell differentiation is significant, as these mTOR protein complexes are known to exert distinct functions in other cell types, and by delineating their interplay we may be able to achieve finer regulation of CD8⁺ T cell functional outcomes for immunity.

The ability of mTORC1 to inhibit Foxo1 transcriptional activity is consistent with an inverse co-relation of activity between mTORC1 and Foxo1 (Chen et al., 2010). Collectively, our observations have unraveled pathways by which mTORC1, depending upon physiological state of a CD8⁺ T cell, may modulate Foxo1 to regulate proliferation, survival, migration and functional differentiation. Since Foxo1 is a direct transcriptional activator of Klf-2 and IL-7Rα, our results have identified mechanisms by which inhibition of mTORC1 by rapamycin leads to an increase in their trafficking into secondary lymphoid compartments and promote their survival, and memory differentiation. Importantly, the ability of Foxo1 to regulate master transcription factor T-bet unravels a new function for Foxo1 in CD8⁺ T cell differentiation, and further defines the mechanisms by which mTORC1 controls functional differentiation of CD8⁺ T cells. Although, we demonstrate the ability of Foxo1 to inhibit T-bet mRNA expression in CD8⁺ T cells, we did not detect direct binding of Foxo1 on T-bet promoter. Given the inverse co-relation of expression between T-bet and Eomes, and the identification of Eomes being a direct gene target of Foxo1, it can be envisaged that Foxo1 may indirectly regulate T-bet expression by controlling expression of Eomes. In any case, further studies are clearly warranted, as better understanding of mechanisms underpinning Foxo1 mediated T-bet expression in CD8⁺ T cells could shed light on the potential role for Foxo proteins in immunity and/or life-span extension.

The Foxo transcription factors play a crucial role in promoting organismal longevity, and also impart quiescence and survival to T cells. The results obtained with knockdown of Foxo1 leading to a loss of survival and memory functions in CD8⁺ T cells provides further support for the role of Foxo1 as a longevity factor in CD8⁺ T cells. In addition, the inability of rapamycin to enhance Eomes expression and memory precursor phenotype in Foxo1^−/− cells, provide a mechanistic basis on how inhibition of mTORC1 promotes memory CD8⁺ T cell differentiation. Given the role of rapamycin and Foxo1 in life-span extension, identification of the novel role of Foxo1 repressing terminal effector differentiation of CD8⁺ T cells may be harbinger of its additional role in other tissues, and its potential contribution to longevity.

In summary, we have identified Foxo1 as an important downstream mediator of mTORC1 that regulates key transcriptional programs to determine effector versus memory differentiation of CD8⁺ T cells. Moreover, it indicates an as yet unappreciated role for Foxo1 in blocking T-bet mediated type I effector differentiation of CD8⁺ T cells and demonstrates its importance in memory generation. Furthermore, these studies unravel key molecular mechanisms by which mTORC1 matches physiologic and metabolic state of CD8⁺ T cell to control growth, proliferation and differentiation. The information offers new opportunities to target CD8⁺ T cell differentiation for immunity against infectious diseases and cancers.

EXPERIMENTAL PROCEDURES

Mice and Reagents

The C57BL/6, CD8⁺ TCR transgenic Rag2^−/− (OT-I, WT), (OT-I Thy1.1⁺), and Tbx21^−/− OT-I mice were bred, housed and used according to IACUC guidelines at RPCI. Conditional deletion of Foxo1 in T cells was achieved in Foxo1^fl/fl-Lck-Cre+/− mice by breeding Foxo1^fl/fl mice as described (Gubbels Bupp et al., 2009) with Lck.cre transgenic mice (C57BL/6NTac-TgN(Lck-Cre), Taconic Farms). The Foxo1^fl/fl WT and Foxo1^fl/fl-Lck-Cre mice used in experiments were 3 month old sex-matched littermates and were bred, housed, and used according to IACUC guidelines at Randolph-Macon College. The rmIL-12 was a gift from Wyeth, Inc. (Cambridge, MA). IFN-α was a gift from T. Tomasi (RPCI) rmIL-7 was purchased from Peprotech (Rocky Hill, NJ). 4-HT and rapamycin were purchased from Sigma Aldrich (St. Louis, MO). LY290042 was purchased from Calbiochem.

Antibodies and Flow cytometry

The antibodies for flow cytometric analysis, anti-CD62L (MEL14), anti-Thy1.1 (OX-7), and isotype matched controls were purchased from BD PharMingen. Anti-TCR Vα2 (B20.1) was purchased from Biolegend. Anti-CD8 (53-6.7), anti-CD122 (5H.4), anti-IFN-γ (XGM1.2), anti-T-bet (eBio4B10), anti-Eomesodermin (21Mags8), anti-KLRG1 (2F1), anti granzyme-B (16G6), anti-hCD2 (RPA-2.10), goat anti-rabbit IgG, and anti-IL-7Rα (AR734) were purchased from eBioscience. The following anti-bodies were from Cell Signaling: antibody to S6 ribosomal protein phosphorylated at Ser 235/236 (D57.2.2E), antibody to Akt phosphorylated at Thr 308 (C31E5E), antibody to Akt phosphorylated at Ser473 (D9E), isotype matched control antibody (D1AE), antibody to FoxO1 phosphorylated at Ser256 and total FoxO1 antibody (L27). Immuno-stained samples were run on BD FACS Calibur and analyzed using FCS Express (Version 3).

Stimulation of OT-I cells

Unless otherwise stated, naïve OT-I cells were stimulated with latex microspheres expressing H-2K^b-SIINFEKL and B7.1. In some experiments, the cell line derived from embryonic fibroblasts namely; BOK (MEC.B7.SigOVA: expressing H-2K^b, OVAp (SIINFEKL) and B7.1), were used as antigen-presenting cells to stimulate naïve OT-I cells. The cells obtained from the thymus of Foxo1-WT and Foxo1^−/− cells, were purified by negative selection (Cedarlane Labs), and were stimulated with anti-CD3 and anti-CD28 coated latex microspheres.

Adoptive transfer and evaluation of Ag-specific CD8⁺ T cells in vivo

OT-I T cells (CD8α⁺/Thy1.1⁺) transduced with control or FoxO1 shRNA were harvested 4 days post-transduction and adoptively transferred (2 × 10⁶) (i.v.) into intact Thy1.2⁺ congenic BL/6 recipients. Recipients were re-challenged with 5 μg of chicken OVA protein and mixed with an equal volume of IFA (Sigma-Aldrich) via subcutaneous injection on day 40 post-transfer. The numbers of OT-I/Thy1.1⁺ cells were determined after staining and flow cytometry evaluation. The total number of adoptively transferred cells was calculated by multiplying the total lymphocyte count by the ratio of CD8α⁺/Thy1.1⁺ T cells gate. For ex vivo IFN-γ expression analysis, spleen cells were re-stimulated in vitro with 0.2μM SIINFEKL peptide. After 1h of incubation at 37°C, Brefeldin A, 10μg/ml, was added to block protein transport. After 5h at 37°C, the cells were surface stained with antibody to CD8α and Thy1.1, fixed, permeabilized, and intracellularly stained with antibody to IFN-γ and fixed again.

Statistical analysis

For statistical analysis, the unpaired Student’s t test was applied. Significance was set at p < 0.05.