Foxo Transcription Factors in T Cell Biology and Tumor Immunity

Abstract

The evolutionally conserved forkhead box O (Foxo) family of transcription factors is pivotal in the control of nutrient sensing and stress responses. Recent studies have revealed that the Foxo proteins have been rewired to regulate highly specialized T cell activities. Here, we review the latest advances in the understanding of how Foxo transcription factors control T cell biology, including T cell trafficking, naive T cell homeostasis, effector and memory responses, as well as the differentiation and function of regulatory T cells. We also discuss the emerging evidence on Foxo-mediated regulation in antitumor immunity. Future work will further explore how the Foxo-dependent programs in T cells can be exploited to cancer immunotherapy.

Introduction

The Foxo family of transcription factors governs a broad variety of fundamental cellular processes, including cell proliferation, apoptosis, energy metabolism, and stress responses to changes in the abundance of nutrients and growth factors [1]. Over the last decade, studies have shown that the evolutionarily ancient Foxo proteins have been co-opted to regulate specialized characteristic of T lymphocytes [2, 3]. Tremendous effort has been made so far to understand the functions of Foxo proteins in the settings of viral and bacterial infections as well as autoimmune diseases. With the resurgence of cancer immunotherapy over the past couple of years, studies started to uncover the roles of Foxo transcription factors in T cell-mediated antitumor immunity.

In this review, we first provide a general overview of T cell-mediated immunity, and how it impacts cancer. We then summarize the key observations made to date on the pleiotropic roles of Foxo transcription factors in T cell biology, including T cell migration, naive T cell survival, differentiation of effector subsets, memory T cell responses, and regulatory T (Treg) cell development and suppression. Next, we highlight the potential of targeting Akt-Foxo signaling in Treg cells to break immune tolerance in cancer, and discuss other aspects of Foxo-dependent T cell regulation that can be possibly manipulated to enhance tumor killing.

T cell-mediated immunity

The immune system has evolved to mount an effective defense against invading pathogens and to minimize deleterious reactions attacking healthy self-tissues and commensal microorganisms. T lymphocytes, a crucial component of the adaptive immune system, orchestrate antigen-specific immune responses and immunological memory. The αβ lineage of T cells are composed of two distinct functional subsets, CD4⁺ or CD8⁺ T cells [4]. Cytotoxic CD8⁺ T cells are crucial in mediating pathogen clearance as well as tumor eradication [5]. Full activation and differentiation of CD8⁺ T cells requires the help of CD4⁺ T cells, or helper T cells. CD4⁺ T cells, besides assisting CD8⁺ T cells, also provide help to B cells to produce antibodies and regulate innate immune populations such as dendritic cells and macrophages [6]. Regulatory T (Treg) cells, a subset of CD4⁺ T cells characterized by their expression of forkhead transcription factor Foxp3, suppress immune responses, and thus are pivotal in the maintenance of immune tolerance [7].

Immunity mediated by the CD4⁺ and CD8⁺ T cells includes a primary response by naive T cells, effector functions by activated T cells, and persistence and reactivation of antigen-specific memory T cells. Naive T cells, once developed in the thymus, enter the bloodstream and continually recirculate through secondary lymphoid organs, the lymph and the blood [8]. Naive T cells are characterized by low expression of CD44 (CD44^lo) and high expression of the lymph node-homing receptors L-selectin (CD62L) and CC-chemokine receptor 7 (CCR7) (CD62L^hiCCR7^hi) [9, 10]. They are relatively long-lived and can remain in interphase for several weeks [11, 12]. Under homeostatic conditions, naive T cells maintain a fairly stable cell number and diversity of T cell receptor (TCR) repertoire, which relies on signals from TCR-self-pMHC interaction and members of the common gamma chain (γ_C) family of cytokines, including interleukin 7 (IL-7) and to a lesser degree IL-15 [13–15].

Upon encountering of foreign antigens presented by antigen-presenting cells, naive T cells become activated and differentiate into effector T cells. This process is accompanied by robust proliferation, transcriptional, epigenetic and metabolic reprogramming, and the acquisition of cardinal features of effector T cells such as effector function and altered migratory pattern. Activated T cells downregulate CD62L and CCR7, and they disseminate broadly to sites of infection to exert effector functions [8]. Depending on the particular cytokine milieu, naive CD4⁺ T cells can differentiate into T helper 1 (Th1), Th2, Th17, T follicular helper (Tfh) and peripherally derived Treg (pTreg) cells [6, 16]. These subsets require the transcription factors T-bet, GATA-binding protein 3 (GATA3), retinoic acid receptor-related orphan receptor-γt (RORγt), B cell lymphoma 6 (Bcl-6) and forkhead box P3 (Foxp3), respectively, for their development [6]. They also play distinct roles in the regulation of immune responses: Th1 cells preferentially produce the cytokine IFN-γ, and are important for eradicating intracellular pathogens. Th2 cells produce large quantities of IL-4, IL-5 and IL-13, and are crucial in the regulation of humoral immune responses to extracellular pathogens, such as helminth worms. Th17 cells are known for their secretion of IL-17, and they are important in the elimination of extracellular bacteria and fungi [6]. Tfh cells are specialized providers of B cell help: they promote the germinal center (GC) reaction including B cell expansion, class switching, selection, and development of high-affinity antibody-forming cells [17]. pTreg cells, together with thymically-derived Treg (tTreg) cells, restrain effector T cell responses and keep autoimmunity in check [7]. CD8⁺ T cells have a less versatile repertoire of effector functions. They differentiate into cytotoxic effector T cells that recognize and kill cells infected by intercellular pathogens like virus and bacteria. CD8⁺ T cells accomplish killing through molecules including the granzymes and perforin, and Fas-mediated cell death mechanisms [18, 19]. Effector CD8⁺ T cells secrete inflammatory cytokines such as IFN-γ and tumor necrosis factor TNF as well [18, 19].

Following the peak of effector expansion, and the resolution of inflammation and pathogen eradication, the majority of effector T cells die, leaving behind a heterogeneous pool of memory T cells [20, 21]. On the basis of their function, proliferative capacity, anatomical location and migration pattern, memory T cells can be divided into discrete subsets, including two circulating populations – effector memory T (T_EM) cells and central memory T (T_CM) cells – as well as non-circulating tissue-resident memory T cells [22]. T_CM cells express high levels of lymph node homing molecules, CCR7 and CD62L, and home to secondary lymphoid organs and bone marrow. T_EM cells lack the expression of these lymph homing molecules, and are most commonly found in non-lymphoid tissues [23, 24]. Functionally, T_CM cells can proliferate extensively and produce IL-2, whereas T_EM cells possess heightened effector functions such as cytolytic activity among CD8⁺ T cells, and they are commonly the immediate responders. Both T_CM and T_EM populations continuously circulate through blood vessels, and they might interconvert as they pass through lymphoid and non-lymphoid tissues [23]. Memory T cells downregulate much of the activation program of effector T cells, yet they maintain the ability to rapidly reactivate effector functions upon restimulation [11].

T lymphocytes in cancer

Extensive work over the past two decades has demonstrated the dual host-protective and tumor-promoting roles of immune system in cancer [25]. The immune system can not only suppress tumor growth by recognizing and destroying cancer cells (cancer immunosurveillance), but can also facilitate tumor progression either by selecting for tumor cells that are more fit to survive in an immunocompetent host (possibly through shaping the immunogenicity), or by establishing conditions within the tumor microenvironment that are beneficial for tumor growth [26]. T lymphocytes have central functions in the many facets of immune-tumor interactions.

The host-protective role of T cells is supported by the observations that mice lacking T cells, such as Tcrb^−/− Tcrd^−/− and Nude strains, develop more carcinogen-induced tumors and spontaneous cancer than wild-type mice [27, 28]. CD8⁺ T cells are at the core of adaptive response. They recognize and destroy the tumor cells expressing peptide-MHC class I complexes on the surface. Activated effector CD8⁺ T cells release IFN-γ that can mediate anti-tumor effects by inhibiting tumor cell proliferation and angiogenesis, or by activating macrophages [29]. Additionally, CD8⁺ T cells can induce tumor cell apoptosis by interacting with Fas and TNF-related apoptosis-inducing ligand (TRAIL) receptors on tumor cells, or through secreting perforin and granzymes. Depleting effector molecules including IFN-γ, perforin, Fas Ligand, TRAIL in mice causes increased susceptibility to carcinogen-induced or spontaneous tumors [30–32].

On the other hand, because most tumors do not express MHC class II, the potential antitumor protective role of CD4⁺ T helper cells has been less obvious. Contribution of CD4⁺ T cells to host protection against tumors has been typically attributed to Th1 cells, which produce large amounts of IFN-γ [33]. Th1 cells enhance the priming and expansion of CD8⁺ T cells, and help recruit natural killer (NK) cells and type I macrophages to tumor sites, which can act in concert towards tumor eradication [34]. The contribution of Th2 and Th17 cells to antitumor immunity has been somewhat contradictory. While Th2-drived IL-4 and IL-13 have been associated with antitumor immunity via recruitment of eosinophils, IL-5 produced by Th2 cells may have tumor-promoting effects [35, 36]. Additionally, despite that low-level chronic exposure to Th17-associated inflammation may facilitate tumor progression, Th17 cells have been shown to induce antitumor inflammatory activity and have stem cell-like properties, therefore making them a prominent candidate for adoptive T cell therapy [37–39]. Furthermore, specific host condition can induce cytotoxic activity in a subpopulation of CD4⁺ T cells, which exert antitumor activity via secretion of granzyme B and IFN-γ [40, 41].

Conversely, T cells can promote tumor evasion. One such mechanism is through immunoediting that selects outgrowth of tumor cells lacking strong rejection antigens [42]. Another mechanism is mediated by Treg cells, which dampen effector responses and prevent immune-mediated rejection of cancer [43]. In murine tumor models, transient ablation of Treg cells results in activation of CD4⁺ or CD8⁺ effector T cells and rejection of solid tumors [44]. In human patients, solid organ tumors are often characterized by abundant Treg cell infiltration compared to their counterparts in normal individuals. This high frequency of Treg cells is associated with poor prognosis in various types of cancers, including breast, lung, esophageal, renal, hepatocellular, ovarian, melanoma and colorectal cancers [43, 45–48]. However, in some studies of colorectal and head and neck tumors, Foxp3⁺ T cell infiltration indicates better patient outcome [49, 50], which can be explained by the heterogeneity of Foxp3⁺ populations in human and the infiltration of Foxp3⁺ non-regulatory T subsets [51, 52].

PI3K-Akt-Foxo signaling pathway

The phosphoinositide 3-kinases (PI3Ks) are a family of kinases that regulate diverse biological process, including cell growth, differentiation, proliferation, survival, metabolism and migration, through the generation of lipid second messengers. On the basis of the structural similarities, the PI3K family can be divided into four classes, among which class IA and class IB PI3Ks have been most extensively studied in immune cells [53]. Class I PI3Ks phosphorylate PIP₂ phosphatidylinositol (4,5)-bisphosphate (PIP2) into phosphatidylinositol (3,4,5)-trisphosphate PIP₃, which mediates the recruitment and activation of numerous signaling molecules. Class IA PI3Ks are activated by receptor tyrosine kinases such as the TCR, costimulatory receptors and cytokine receptors, whereas class IB PI3Ks are primarily stimulated by G-protein-coupled receptors such as chemokine receptors [53]. Each PI3K comprises a regulatory subunit and a catalytic subunit. Most studies in the immune system focus on the p110δ class IA and p110γ class IB catalytic subunits due to their high expression [54].

PIP₃ recruits PH domains of 3-phosphoinositide-dependent protein kinase 1 (PDK1) and its target Akt to the plasma membrane. PDK1 phosphorylates Akt at Thr308, and the full activation of Akt requires a second phosphorylation by mechanistic target of rapamycin complex 2 (mTORC2) or DNA-dependent protein kinase (DNA-PK) at Ser473 [55, 56]. In the nucleus, activated Akt phosphorylates Foxo transcription factors at three conserved sites, leading to diminished DNA-binding activity of Foxo and translocation from nucleus to cytoplasm in a complex with the 14-3-3 scaffolding protein [2]. Phosphorylated Akt also activates mTORC1 via Rheb-GTPase [57]. Several phosphatases negatively regulate the PI3K pathway, including the lipid phosphatases Pten and SHIP that dephosphorylate PIP₃, and the PH-domain leucine-rich-repeat protein phosphatase (PHLPP) that dephosphorylates Akt [58, 59].

Foxo family of transcription factors

Foxo transcription factors belong to the family of forkhead proteins, characterized by the forkhead (FKH) DNA-binding domain. They are key players in an evolutionary conserved pathway downstream of insulin and insulin-like growth factors [1]. Foxo transcription factors regulate a variety of processes including cellular metabolism, organ development, cell cycle progression or apoptosis [1]. In mammals, the Foxo subclass is comprised of four members, Foxo1, Foxo3, Foxo4 and Foxo6 [60]. Foxo6 expression is confined to specific region of the brain, whereas Foxo1, 3 and 4 are ubiquitously expressed, but among different cell types and organs, a heterogeneous pattern of expression has been described [61]. Foxo1 is highly expressed in B cells, T cells and ovaries. Constitutive deletion of Foxo1 gene causes embryonic lethality in mice at day 10.5 due to impaired vascular development [62–64]. Lymphocytes and myeloid cells express high levels of Foxo3. Foxo3-mutant mice exhibit minimally noticeable phenotype with the exception of early ovarian follicle depletion in female mice [63, 65]. Foxo4 is expressed at a lower level, and no apparent phenotype has been reported in the Foxo4 knockout animal [64].

One mechanism by which Foxo proteins regulate gene transcription is through their binding as monomers to cognate DNA-binding sequence (5′-TTGTTAC-3′) [66]. Besides, they can associate with many transcriptional cofactors, including STATs, Smad3, p300 and β-catenin, to regulate context-dependent transcription programs [67]. The activity of Foxo proteins is tightly regulated by post-translational modifications, primarily phosphorylation and acetylation, which alter the subcellular localization and protein abundance of Foxo [66]. In response to growth factors, insulin or cytokines stimulation, kinases downstream of PI3K, such as Akt and serum glucocorticoid kinase (SGK1) phosphorylate Foxo proteins, resulting in their nuclear export into the cytoplasm and potentially proteasomal degradation [1, 68]. Ubiquitination and degradation of Foxo1 can be mediated by Skp1/Cul1/F-box protein ubiquitin complex [69]. On the contrary, oxidative stress activated JNK or Mst1 kinases activates Foxo and triggers the relocalization of Foxo members from the cytoplasm to the nucleus [1].

Foxo-dependent regulation in conventional T cells

T cell trafficking

Over the last decade, accumulating evidence has shown that the evolutionarily ancient Akt-Foxo signaling has been co-opted to play a highly specialized role in the immune system [2, 3]. Foxo1 is abundantly expressed in lymphoid cells, and has been shown to control T cell homing to secondary lymphoid organs [64, 70, 71]. Overexpression of a constitutively active mutant of Foxo1 in which the three Akt phosphorylation sites (Tyr24, Ser256 and Ser319) are mutated to alanine, elevates the expression of CD62L in Jurkat cells and human primary T cells [72]. In line with these gain-of-function studies, CD4⁺ and CD8⁺ T cells in mice harboring T-cell specific Foxo1-deficiency express diminished levels of trafficking molecules, including CD62L (encoded by Sell gene), CCR7 (Ccr7), and Sphingosine-1-phosphate receptor 1 (S1P1, encoded by Edg1 gene) [64, 70, 71]. One way in which Foxo1 regulates T cell trafficking is through direct transcriptional regulation on the Ccr7 gene [73]. Additionally, Foxo1 induces the expression of Krupple-like factor 2 (Klf2), a transcription factor that in turn promotes the transcription of Sell and Edg1 [72, 74, 75] (Figure 1a).

Figure 1. Foxo family transcription factors-mediated regulation in conventional T cells and regulatory T cells.

(a) Foxo1 controls T cell trafficking by promoting the expression of the transcription factor Klf2, and trafficking molecules CD62L, CCR7, and S1P1. Foxo1 is essential for naive T cell homeostasis via its regulation of IL-7Rα. Foxo proteins regulate many aspects of effector T cell differentiation and memory responses. Foxo1 and Foxo3 directly activate the expression of Eomes, thereby control effector versus memory functional fate by regulating T-bet and Eomes expression. Foxo1 also promotes memory T cell differentiation via its induction of Tcf-7. The positive feedback between Foxo1 and PD-1 supports the homeostasis of cytotoxic lymphocytes during persistent antigen-stimulation. In CD4⁺ T cells, Foxo3 drives Eomes-dependent differentiation of pathogenic Th1 cells. Moreover, Foxo1 inhibits the accumulation of Tfh cells via its transcriptional repression on Bcl6. (b) In immature thymocytes, the Foxo family proteins Foxo1 and Foxo3 are both highly expressed. They bind to the Foxp3 locus and cooperatively induce Foxp3 gene transcription. Foxo1 and Foxo3 regulates TGF-β-induced conversion from conventional CD4⁺ T cells to Treg cells as well. In mature Treg cells, Foxo1 is highly expressed, and it plays non-redundant role in the regulation of Treg cell function and trafficking. Foxo1 represses the transcription of the pro-inflammatory cytokine IFN-γ, and promotes the expression of CTLA-4. Foxo1 also directly regulates the expression of lymphoid organ homing molecules CCR7 and CD62L. Notably, Foxo1 activity is subject to modulation of PI3K-Akt signaling. Activated Akt phosphorylates Foxo1, leading to its nuclear expulsion and loss of transcriptional activity. Failure to downregulate Foxo1 activity results in sustained high expression of CCR7 and CD62L, and altered migration pattern of Treg cells.

Naive T cell homeostasis

Extensive studies in non-lymphoid cell lines have revealed that Foxo transcription factors are negative regulators of cell cycle progression, as they promote the expression of several cyclin-dependent kinases inhibitors such as p27^kip1 [1]. Moreover, Foxo proteins are positively connected with apoptosis due to their transactivation of the pro-apoptotic molecules Bim and Fas ligand [1]. In naive T cells, Foxo1 and Foxo3 mainly reside in the nucleus, and TCR engagement triggers phosphorylation of Foxo proteins and their extrusion from the nucleus to the cytoplasm [76]. Interestingly, Foxo1- and Foxo3-deficient naive T cells do not spontaneously enter the cell cycle progression or are more resistant to apoptosis, implying a different type of regulation from non-lymphoid cell lines [77]. Foxo1 has instead been shown indispensable in sustaining naive T cell survival [2, 3]. Adult mice with T-cell specific deletion of Foxo1 have a significantly reduced population of naive T cells, which is caused by the severe defect in the IL-7 receptor α chain (IL-7Rα, also known as CD127, encoded by Il7r gene) expression [64, 70]. At the molecular level, Foxo1 binds to an evolutionarily conserved noncoding region of the Il7r locus, promoting its transcription [64, 70] (Figure 1a). Compromised IL-7Rα expression then leads to diminished expression of the anti-apoptotic molecule Bcl-2 in Foxo1-deficient T cells [70].

Effector T cell differentiation

Foxo proteins are also involved in the differentiation of diverse effector T cell subsets. T-cell specific deletion of Foxo1 results in a significant increase in the production of IFN-γ and IL-17 in peripheral T cells [78, 79]. Compound deletion of Foxo1 and Foxo3 further heighten the levels of pro-inflammatory cytokines [78–80]. Foxo1 and Foxo3 have been shown cooperatively regulate the generation of Foxp3⁺ Treg cells from conventional T cells as well, which will be discussed further in the section below [77, 78]. In the absence of Foxo1, TCR engagement, costimulation, and concurrent TGF-β stimulation fail to efficiently induce iTreg differentiation from naive CD4⁺ T cells [77]. Instead, the Foxo1-deficient CD4⁺ T cells are misdirected to become IFN-γ-producing Th1 cells, suggesting a role of Foxo1 in Th1 cell fate specification [77]. Additionally, recent work by Stienne et al revealed the function of Foxo3 in modulating pathogenic Th1 cell differentiation as well [81]. The expression of Foxo3 is increased in CD4⁺ T cells upon TCR stimulation [81]. Foxo3 deficiency blunts the differentiation of IFN-γ⁺GM-CSF⁺ pathogenic Th1 cells, thereby rendering total Foxo3-deficient mice and mice with a T cell-specific deletion of Foxo3 less susceptible to the development of central nervous system inflammation [81]. Mechanistically, Foxo3 controls IFN-γ and GM-CSF production in CD4⁺ T cells through its direct transcriptional regulation of Eomesodermin (Eomes) expression [81] (Figure 1a).

Besides the involvement in Th1 cell differentiation, Foxo transcription factors have been shown negatively regulate the generation of Th17 cells. One way in which Foxo1 suppresses Th17 program is its interaction with RORγt, the key lineage specifying transcription factor of Th17 cells [79, 82]. Foxo1 binds to RORγt via its DNA binding domain, hereby blocking RORγt from its target genes such as Il17a and Il23r (encoding IL-23 receptor, IL-23R) [79, 82]. Such Foxo1-mediated RORγt repression is further impacted by environmental factors. For example, a modest increase in salt concentration induces activation of the salt-sensing kinase SGK1, which phosphorylates and deactivates Foxo1, thus promoting IL-23R expression and Th17 cell differentiation in vitro and in vivo [82]. Foxo1 has also been recently reported as a negative regulator of Th17 cell pathogenicity [83]. The pro-inflammatory cytokine IL-6 activates Stat3, which then induces the expression of the microRNA-183-96-182 cluster in Th17 cells [83]. This microRNA cluster directly represses Foxo1 expression, relieving RORγt from Foxo1-mediated suppression, and resulting in elevated IL-1R1 expression and Th17 pathogenicity [83].

Emerging evidence has also pointed a role of Foxo transcription factors in Tfh cell development. Tfh cells are specialized providers of T cell help to B cells. Expression of Foxo1 protein is significantly lower in CXCR5⁺PD-1^lo Tfh cells and CXCR5^hiPD-1^hi GC-Tfh cells than in naive T cells or non-Tfh cells [84]. Mice with T-cell specific Foxo1-deletion accumulate a large population of Tfh cells and develop B cell autoimmunity, which is manifested by GC formation and production of circulating, anti-DNA antibodies [77, 84]. Conversely, enforced nuclear localization of Foxo1 results in a reduction of Tfh cells, indicating a negative function of Foxo1 in Tfh cell differentiation [85]. At the molecular level, Foxo1 binds to the Bcl6 gene and mediates its transcriptional repression [85] (Figure 1a). The negative regulation of Foxo1 on Bcl6 can be relieved by ICOS signaling-triggered inactivation of Foxo1 or by the E3 ubiquitin ligase Itch-mediated Foxo1 degradation [84, 85]. However, the role of Foxo transcription factors in Tfh cell differentiation is further complicated by several other observations. Despite an amplified total pool of CXCR5⁺ Tfh cells in T-cell specific Foxo1-deficient mice, the CXCR5^hiPD-1^hi GC-Tfh cells subset is notably reduced, suggesting that Foxo1 may promote the differentiation of the particular GC-Tfh cell subpopulation, yet the exact molecular mechanism remains unknown [85]. Additionally, in polarized Th1 cells, Foxo1 and Foxo3a are positive correlated with Bcl6 expression and the induction of a subset of Tfh genes, when IL-2 is limiting [86]. It is possible that distinct cytokine milieu modulates Foxo signaling differently, generating diverging outcomes of Tfh cell development.

Collectively, these observations underline the key importance of Foxo transcription factors in the fate determination of effector T cells. Future investigation on how Foxo proteins integrate various signals under diverse contexts is needed to fully unveil their multifaceted regulation in these processes.

Memory responses

In addition to effector T cell fate specification, a number of studies have elucidated the roles of Foxo transcription factors, in particular, Foxo1, in memory T cell differentiation and function. Using an in vitro activation system, Rao et al revealed that Foxo1 is negatively correlated with T-bet expression and effector T cell responses [87]. The T-box transcription factors T-bet and Eomes are required for the formation and function of effector and memory CD8⁺ T cells, respectively [20]. Knockdown of Foxo1 reduced the persistence of these in-vitro-activated T cells when transferred into mice, which was associated with diminished expression of Eomes [87]. These observations imply that Foxo1 promotes the differentiation of memory T cells while counterbalancing effector responses [87].

In order to probe the precise function of Foxo1 in antigen-experienced T cell responses without confounding effects on naive T cell homeostasis as observed in T cell-specific Foxo1-deficient mice, the Cre recombinase under the control of human granzyme B promoter (GzmB-Cre) was used to deplete Foxo1 in later studies [73, 88]. In the GzmB-CreFoxo1^f/f mice, the generation of memory CD8⁺ T cells upon acute bacterial or viral infections was substantially diminished, and the recall response against secondary antigen challenge was severely impaired as well [73, 88]. Nevertheless, Foxo1 deficiency had minimal impact on effector T cell responses during the early stages of acute infection [73, 88]. Later on, after pathogen clearance, Foxo1-depleted CD8⁺ T cells retained high expression of effector molecules such as GzmB, suggesting that although superfluous for early effector responses, Foxo1 is critically required for the effector-to-memory transition [88]. Furthermore, recent studies led by Hedrick showed that the memory phenotype of CD8⁺ T cells in both acute and chronic viral infections must be actively maintained by the presence of Foxo1 [89, 90]. Acute inactivation of Foxo1 disrupts T cell memory and can lead to reversion to a short-lived effector phenotype [89, 90].

Consistent with the notion that Foxo1 plays a more important role in the precursors to memory cells than in effector T cells, KLRG1^lo memory precursor effector cells (MPECs) expressed higher levels of Foxo1 compared with KLRG1^hi short lived effector T cells (SLECs) [73, 88]. Mechanistically, Foxo1 regulates memory T cell differentiation through transactivating the Tcf7 gene, which encodes T cell factor 7 (Tcf-7, also known as Tcf-1), an essential transcription factor for memory formation [73, 88] (Figure 1a). Intriguingly, Delpoux et al reveled that naive CD8⁺ T cell Tcf7 expression is independent of Foxo1, yet Foxo1 and Tcf7 oppose effector programming and enable early memory prosperities in post-infection CD8⁺ T cells [91]. Moreover, Foxo1 has been shown involved in the epigenetic regulation of effector versus memory T cell fate. A recent study has indicated that SLECs have more repressive H3K27me3 deposition at numerous pro-memory and pro-survival genes relative to MPECs, and the H3K27me3 deposition is dependent on Foxo1 [92]. Reduced Foxo1 expression in SLECs may allow for increased activity of polycomb repressive complex 2 (PRC2) and stable epigenetic repression of the pro-memory genes, whereas increased Foxo1 expression in MPECs might shield the pro-memory genes from H3K27me3 deposition [92]. Nevertheless, the precise mechanism by which Foxo1 controls such chromatin remodeling awaits future investigation.

Staron et al compared the regulation of the PI3K-Akt-Foxo pathway in antigen-specific cytotoxic T lymphocytes (CTLs) during acute and chronic infections [93]. Antigen persistence during chronic infection triggers the differentiation of dysfunctional or exhausted CTLs that are characterized by their elevated expression of the inhibitory receptor PD-1 [94]. Exhausted CTLs show impaired PI3K-Akt-mTOR activation compared with the antigen-specific CTLs from acute infection, which is contributed in part by the PD-1 signaling [93]. Dampened Akt-mTOR activation then leads to enhanced nuclear retention and protein expression of Foxo1 in the exhausted CTLs [93]. Interestingly, the high Foxo1 activity is in turn needed to sustain PD-1 expression (via transactivation of the Pdcd1 gene), and the acquisition of PD-1^hiEomes^hi terminally exhausted state [93] (Figure 1a). Depletion of Foxo1 in mice resulted in markedly reduced population of viral-specific CD8⁺ T cells and poor control of viral load during chronic infection, even though the Foxo1-deficient T cells expressed higher level of GzmB [93]. Therefore, Foxo1 is critically required to sustain the expression of PD-1 and cell survival of CTLs during chronic infection. An intriguing positive feedback between PD-1 and Foxo1 is exploited by cytotoxic lymphocytes to support their homeostasis in response to persistent infection.

Foxo3 is the main isoform expressed in the myeloid compartment, and is expressed at a lower level in T cells compared with Foxo1. Nevertheless, numerous studies have demonstrated a role of Foxo3 in the regulation of memory T cell differentiation. Early work using global inactivation of Foxo3 revealed that Foxo3 restrains the expansion of T cells upon viral infection through limiting the production of inflammatory cytokines by dendritic cells and macrophages [80]. With T cell-specific Foxo3-deficient mice, Sullivan et al then showed that Foxo3 regulates memory responses in a T cell-intrinsic manner [95, 96]. In both acute and chronic viral infection models, loss of Foxo3 results in enhanced accumulation of antigen-specific T cells, which is associated with a reduction in cell death [95, 96]. In contrast to Foxo1, which promotes CD8⁺ central memory formation, loss of Foxo3 did not alter the composition of MPECs versus SLECs among antigen-specific T cells [97]. These findings imply that Foxo1 and Foxo3 regulate divergent aspects of effector and memory T cell responses. The exact molecular programs regulated by Foxo1 and Foxo3, and how they vary in different settings remain to be determined.

Foxo proteins control regulatory T cell development, function and migration

Besides their central roles in the control of conventional T cell responses, Foxo transcription factors are pivotal regulators of T cell tolerance as well. Mice with T cell-specific deficiency of Foxo1 have impaired thymic Treg cell development, and deletion of Foxo3 further exaggerate the phenotype [77, 78]. Foxo1 and Foxo3 also modulate TGF-β-mediated iTreg cell induction in vitro [77, 78, 98]. Mechanistic studies reveal that Foxo1 and Foxo3 regulate Foxp3 expression by binding to the promoter and a conserved intronic enhancer region (conserved noncoding sequence 2: CNS2) of the Foxp3 locus [77, 78] (Figure 1b).

In mature Treg cells, Foxo1 rather than Foxo3, is highly expressed [99]. Disruption of Foxo1 in developed Treg cells using the Cre recombinase driven by the Foxp3 promotor (Foxp3^Cre) triggers an early fatal lymphoproliferative disease akin to that developed in Foxp3-deficient mice [99]. Nevertheless, Foxp3⁺ Treg cells are unaffected or even expanded in these mice, implying that the inflammatory disease is caused by a loss of Treg cell function rather than a loss of Treg cell population [99]. At the molecular level, Foxo1 directly regulates a genetic program consist of about 300 genes, including the pro-inflammatory cytokine Ifng [99]. Deficiency of Foxo1 leads to relieved repression of IFN-γ production in Treg cells, thereby abrogating their suppressive activity [99]. Moreover, Foxo1-mediated transactivation of the Ctla4 gene also contributes to potent Treg suppressive capacity [77] (Figure 1b). Taken together, these findings identify Foxo proteins as central regulators of Treg cell differentiation and suppressive function.

Another key aspect of Foxo signaling is its dynamic regulation. Foxo1 is phosphorylated downstream of the PI3K-Akt-mTOR signaling pathway, resulting in its translocation from the nucleus to the cytoplasm and subsequent inactivation of its transcriptional activity [1]. In contrast to conventional T cells, in which low dose TCR stimulation is sufficient to induce robust Akt phosphorylation and Foxo1 nuclear clearance, Treg cells are more resistant to TCR induced nuclear expulsion of Foxo1 [99]. To explore how the dynamic regulation of Foxo1 impacts Treg cell biology, we performed a close examination of Foxo1 subcellular localization in different Treg cell subsets. CD62L^loCD44^hi activated-phenotype Treg (aTreg) cells are the major Treg subset in non-lymphoid tissues, and they have a slower turnover rate relative to CD62L^hiCD44^lo resting Treg (rTreg) cells, which primarily populate secondary lymphoid organs [100, 101]. Foxo1 predominantly resides in the nucleus of rTreg cells, whereas it is translocated to the cytoplasm of aTreg cells [100]. Moreover, we found that the differentiation of aTreg cells is associated with repression of Foxo1-mediated transcriptional program [100].

To further dissect the role of proper downregulation of Foxo1 in Treg cells, we utilized a mouse model in which an Akt-insensitive Foxo1 mutant allele (also known as Foxo1 constitutively active form, Foxo1CA) preceded by a “STOP-flox” cassette was knocked into the Rosa26 locus [100]. The Foxo1CA mice were crossed to the Foxp3^Cre mice to drive Treg cell-specific hyperactivation of Foxo1 [100]. Expression of Foxo1CA from one allele (Foxp3^CreFoxo1CA/+) or two alleles (Foxp3^CreFoxo1CA/Foxo1CA) triggered a dose-dependent increase of Foxo1 target gene expression, as well as a gradual reduction of aTreg cells [100]. Mechanistically, hyperactivation of Foxo1 in Treg cells prevented the downregulation of lymphoid organ homing molecules, including CCR7 and CD62L, and impeded Treg cell trafficking to non-lymphoid organs [100] (Figure 1b). Mice containing two alleles of Foxo1CA in Treg cells developed a lethal autoimmune disorder that was associated with augmented effector CD8⁺ T cell response, whereas mice expressing one allele of Foxo1CA were largely spared from the immunopathology [100]. Collectively, these findings reveal that in addition to its indispensable role in the control of Treg cell development and suppressive function, proper downregulation of Foxo1 is also critically required for Treg cell trafficking to non-lymphoid tissues. Furthermore, these observations highlight how the Akt-Foxo signaling in Treg cells needs to be tightly controlled such that too high, too low, or the inability to dynamically regulate the Akt-Foxo signaling can lead to devastating immune dysfunction.

Foxo-mediated control in tumor-infiltrating T cells

While substantial efforts have been made to elucidate how Foxo transcription factors regulate T cell responses in the settings of infection and autoimmunity, the roles of Foxo proteins in antitumor immunity have been less clear. Our study on Treg cells revealed that tumor-infiltrating aTreg cells show more substantial reduction of Foxo1-target genes compared with those residing in healthy tissues, for instance, the liver and intestine [100]. Given that appropriate downregulation of Foxo1 activity is required for the differentiation and trafficking of aTreg cells [100], it is possible that tumor-infiltrating aTreg cells are more susceptible to the depletion triggered by Foxo1 gain-of-function, such as the expression of Foxo1CA mutant. Indeed, expression of Foxo1CA in Treg cells at a dose that does not cause overt autoimmune diseases (Foxp3^CreFoxo1CA/+) is sufficient to deplete tumor-associated aTreg cells, and to evoke effector CD8⁺ T cell responses, resulting in delayed tumor growth in both spontaneous and transplantable tumor models [100]. Such selective depletion of tumor-associated Treg cells suggest that the Foxo1 signaling can be titrated to preferentially break tumor immune tolerance without inflicting autoimmunity, a common advert event associated with antitumor immunotherapy.

Interestingly, attenuated tumor growth has also been observed in mice containing Treg cell specific-deletion of the leukocyte-specific PI3K isoform p110δ [102]. It remains to be determined whether p110δ blockade impairs the generation of tumor-associated aTreg cells in a similar manner as Foxo1CA expression. Nevertheless, these findings imply that the PI3K-Akt-Foxo signaling pathway in Treg cells would be manipulated to unleash antitumor immunity.

Whether and how Foxo-dependent program in conventional T cells can be harnessed for cancer immunotherapy is still enigmatic. Given that Foxo1 regulates T cell migration by promoting the expression of lymphoid organ homing molecules [64, 70, 71], we speculate that dampening Foxo1 activity would facilitate T cell trafficking into tumor sites. Additionally, Foxo1 promotes PD-1 expression and the survival of antiviral CD8⁺ T cells during chronic infection [93]. Persistent antigen exposure in cancer also triggers the differentiation of PD-1⁺ dysfunctional or exhausted T cells [103]. Therefore, manipulation of Foxo1, and possibly other members of the PI3K-Akt pathway, can modulate the PD-1 axis in the tumor-infiltrating T cells, which could lead to novel therapeutic options to fight cancer.

Concluding remarks

Significant progress has been made in our understanding of how Foxo transcription factors regulate T-cell mediated immune responses. It is now clear that Foxo plays an indispensable role for the proper trafficking of both conventional and regulatory T cells. Foxo proteins have also been shown critically required to sustain naive T cell survival, and to coordinate effector and memory T cell differentiation. Foxo proteins are instrumental in the differentiation and function of Treg cells as well. Nevertheless, how Foxo transcription factors integrate various extracellular cues, and possibly intracellular signals, to exert transcriptional and epigenetic regulation remains incompletely understood.

An exciting development is the potential use of PI3K-Akt-Foxo signaling modulators in cancer immunotherapy. Using a genetically modified mouse model, we showed that tumor-infiltrating Treg cells are more susceptible to depletion triggered by Foxo1 hyperactivation, compared with their counterparts in healthy noncancerous tissues. There is accumulating evidence that removal of Treg cells is capable of boosting antitumor immune responses, yet systemic Treg depletion may concurrently elicit devastating autoimmunity. One strategy to circumvent the autoimmune problem is to identify cell surface molecules specifically expressed by tumor-infiltrating Treg cells. The chemokine receptor CCR4 has been reported highly expressed in effector-type Treg cells in tumor, and cell-depleting antibodies against CCR4 can be promising candidates for cancer immunotherapy [104]. Insights on Foxo pathway thus provide another approach -- agents targeting the PI3K-Akt signaling may impinge on the Foxo pathway to selectively break Treg cell-mediated tumor immune tolerance. Finally, beyond Treg cells, future work should focus on how Foxo-dependent regulation in conventional T cells can be exploited to evoke antitumor immune responses.

Table 1.

Foxo-dependent regulation	Molecular Mechanism	Reference
T cell trafficking	Foxo1 promotes the expression of CD62L, CCR7, S1P1, Klf2	64, 70, 71, 72, 73
Naive T cell homeostasis	Foxo1 positively regulates IL7Rα expression	64, 70
Effector T cell differentiation	Foxo3 regulates Eomoes expression Foxo1 blocks RORγt from binding its target genes Foxo1 inhibits Bcl6 expression	81 79, 82 85
Memory T cell responses	Foxo1 promotes Tcf7 expression	73, 88
Regulatory T cell differentiation	Foxo1 and Foxo3 facilitates Foxp3 transcription	77, 78
Regulatory T cell function	Foxo1 inhibits IFN-g expression Foxo1 promotes CTLA-4 expression	77, 96