Differential Control of iNKT Cell Effector Lineage Differentiation by the Forkhead Box Protein O1 (Foxo1) Transcription Factor

The invariant NKT (iNKT) cells recognize glycolipid antigens presented by the non-classical MHC like molecule CD1d. They represent an innate T-cell lineage with the ability to rapidly produce a variety of cytokines in response to agonist stimulation to bridge innate and adaptive immunity. In thymus, most iNKT cells complete their maturation and differentiate to multiple effector lineages such as iNKT-1, iNKT-2, and iNKT-17 cells that possess the capability to produce IFNγ, IL-4, and IL-17A, respectively, and play distinct roles in immune responses and diseases. Mechanisms that control iNKT lineage fate decisions are still not well understood. Evidence has revealed critical roles of Foxo1 of the forkhead box O1 subfamily of transcription factors in the immune system. However, its role in iNKT cells has been unknown. In this report, we demonstrate that deletion of Foxo1 causes severe decreases of iNKT cell total numbers due to impairment of late but not early iNKT cell development. Deficiency of Foxo1 results in decreases of iNKT-1 but increases of iNKT-17 cells. Our data reveal that Foxo1 controls iNKT effector lineage fate decision by promoting iNKT-1 but suppressing iNKT-17 lineages.

Foxo1 belongs to the Forkhead box "O" subfamily of transcription factors binding to a consensus DNA sequence 5 ′ -(A/C)AA(C/T)A to regulate transcription of many genes involved in cell survival, cell cycle, differentiation, and cancer (32). Many mechanisms have evolved to regulate Foxo1 at both transcriptional and post transcriptional levels. A key mechanism is its phosphorylation by Akt and serum/glucocorticoid regulated kinase 1 (SGK1) on different sites that leads to its export from the nucleus and sequestration in the cytosol due to increased association to 14-3-3 proteins (33). Many studies have revealed important roles of the Foxo subfamily in the immune system (32,34). Foxo1 promotes production of inflammatory cytokines in and regulates migration of dendritic cells (35,36), is essential for B cell development and Ig class-switch (37)(38)(39), and regulates T cell differentiation and function (34). It controls T cell survival and homing via increasing IL7 receptor expression, CCR7, and L-selectin expression (40), plays important roles for regulatory T cell development and function (41)(42)(43), negatively controls Th17 cell differentiation (44)(45)(46)(47) and Tfh differentiation (48)(49)(50), promotes Th9 responses (51)(52)(53), and regulates CD8 T cell effetor/memory differentiation (54)(55)(56). We report here that deficiency of Foxo1 does not grossly affect early iNKT cell development but severely impacts late iNKT cell development, manifested by severe reduction of iNKT1 cells but increases of iNKT17 cells. Our data suggest that Foxo1 is a critical regulator that controls iNKT1 and iNKT17 effector fate decision.

Mice
Foxo1 f /f and Cd2iCre mice were purchased from the Jackson Laboratory (57,58). Foxo1 f /f had further backcrossed to C57BL6/J background. Two to ten-week-old Foxo1 f /f -Cd2iCre and Foxo1 f /f or Cd2iCre littermate control mice were used for the experiments. For each experiment, one pair of test and control mice of the same sex littermates examined. All mouse experiments were performed according to protocols approved by the Duke University Institute Animal Care and Use Committee.

Generation of Chimeric Mice
CD45.1 + CD45.2 + WT mice in C57BL/6 background were irradiated with a single dose of 800 rad X-Ray and intravenously injected with 10-15 million of a mixture of BM cells from CD45.1 + WT mice and CD45.2 + Foxo1 f /f -Cd2iCre mice at 1:1 ratio. Recipient mice were euthanized and analyzed 8 weeks later.

Statistical Analysis
Data were presented as mean ± SEM and analyzed for statistical differences using the Prism 5/GraphPad software. Comparisons were made using the two-tailed paired or unpaired Student t-test. Each pair represents age-and sex-matched littermates that were examined in a single experiment and is indicated by a connecting line between test and control mice. P-values less than 0.05 were considered significant ( * p < 0.05, * * p < 0.01, * * * p < 0.001).

Impairment of iNKT Cell Development in
To investigate the role of Foxo1 in iNKT cell development, we bred Foxo1 f /f mice (57) with hCD2-iCre (Cd2iCre) mice (58) to generate Foxo1 f /f -Cd2iCre (Foxo1KO) mice. CD2iCre induces gene ablation of floxed genes in both T cells and B cells (58) and in CD4 + CD8 + double positive (DP) thymocytes ( Figure 1A). We used TCRβ and PBS-57 loaded CD1d tetramer (CD1dTet) to detect iNKT cells. In Foxo1 f /f -Cd2iCre mice, TCRβ + CD1dTet + iNKT cell percentages and numbers were decreased 64.41% and 68.56%, respectively in the thymus compared with Foxo1 f /f controls (Figures 1B-D). Moreover, iNKT cell percentages and numbers were also reduced in the spleen and liver in Foxo1 f /f -Cd2iCre mice with the exception of splenic iNKT cells in one Foxo1 f /f -Cd2iCre mice due severe splenomegaly likely caused by defective function of regulatory T cells. In contrast, CD4 + CD8 − single positive (SP) TCRβ + and CD4 − CD8 + SP TCRβ + thymocyte numbers were similar between control and Foxo1KO mice ( Figure 1E). Thus, Foxo1 deficiency resulted in severe impairment of iNKT cell but not conventional αβT cell development.

Intrinsic Control of iNKT Development by Foxo1
Because Foxo1 was ablated in both T and B cell lineages and Foxo1 deficiency causes severe B cell developmental defects, impairment of regulatory T cell function, and Th lineage and CD8 effector/memory differentiation (32,59), the severe decreases of iNKT cells could result from change of environment in these mice. To determine if abnormal iNKT cell development in Foxo1 f /f -Cd2iCre mice was autonomous, we generated mixed bone marrow (BM) irradiation chimeric mice by injecting a mixture of CD45.1 + WT and CD45.2 + Foxo1 f /f -Cd2iCre BM cells at a 1:1 ratio into sublethally irradiated CD45.1 + CD45.2 + recipient mice. Two months after reconstitution, recipient mice contained about 1:1 ratio of CD45.2 + and CD45.1 + CD11b + Ly6G + Ly6C − neutrophils in the spleen (Figures 2A,B) suggesting equal reconstitution of these two types of hematopoietic stem cells (HSCs). Additionally, the ratios of CD45.2 + to CD45.1 + CD4 + CD8 + double positive (DP) thymocytes, the immediate precursors of iNKT cells, and non-iNKT TCRβ + conventional T cells were also 1:1 (Figures 2C,D,F). However, the CD45.2 + Foxo1KO to CD45.1 WT ratios for iNKT cells were drastically decreased to 0.2:1 (Figures 2D,F). In contrast, the ratio for CD1dTet − TCRβ + conventional αβT cells in the thymus were not decreased (Figures 2D,F). Consistent with severe defects in iNKT cell generation, very few iNKT cells in the spleen and liver in the recipients were derived from Foxo1KO HSCs (Figures 2E,F). Thus, Foxo1 deficiency intrinsically inhibited iNKT but not conventional αβT cell development.  3 (13, 14). In Foxo1 f /f -Cd2iCre mice, stage 0 iNKT cell percentages were increased but numbers were not altered compared with control mice (Figures 3A,B). Stages 1 and 2 iNKT cell percentages displayed increased trend, although not statistically significant. However, their numbers were not obviously altered (Figures 3A,C). Importantly, stage 3 iNKT cell percentages and numbers were both decreased in Foxo1 f /f -Cd2iCre thymus (Figures 3A,C). In Foxo1KO spleen, stage 2 iNKT cell numbers were similar to WT control but stage 3 showed decreased except one mouse (Figure 3D), which contained increased stage 3 iNKT cells due to considerable enlargement of the spleen that was likely caused defects in regulatory T cells and lymphoproliferative disorders. In Foxo1KO liver, stage 2 iNKT cell numbers were also similar to WT control but stage 3 iNKT cells were obviously decreased ( Figure 3E). Thus, Foxo1 deficiency appears to selectively affect late stage iNKT cell development.

Intrinsic Promotion of Late Stage iNKT Maturation by Foxo1
We further examined whether Foxo1 intrinsically promotes late stage iNKT cell maturation using the mixed BM chimeric mice described in Figure 2. Within CD45.2 + Foxo1KO iNKT cells, the percentages of stages 0, 1, and 2 were increased 11.08, 2.54, and 1.87 folds compared with CD45.1 + WT iNKT cells, respectively; but stage 3 were drastically decreased (Figures 4A,B), indicating an intrinsic role of Foxo1 for iNKT cell terminal maturation to stage 3. Because iNKT cells were predominantly stage 3 and its dramatic decreases could affect the percentages of stage 0-2 iNKT cells, we further calculated CD45.2 Foxo1KO/CD45.1 WT ratios of each stage of iNKT cells in individual recipient mice. As shown in Figures 4C,D, the ratios of CD45.2/CD45.1 of stage 0 iNKT cell were increased to 3.4:1 compared with the ratios of DP thymocytes. But the ratios of stages 1, 2, and 3 were progressively decreased from 0.60:1, to 0.35:1, and finally to 0.073:1. These results suggested that Foxo1 deficiency exerted weak inhibition on stage 1 and stage 2 iNKT cell maturation but much strong impact on stage 3 iNKT cell maturation. The increases of stage 0 ratios could be caused by enhanced early iNKT cell generation and/or blockade of stage 0 to stage 1 iNKT cell transition in the absence of Foxo1. Thus, Foxo1 intrinsically promoted iNKT cell maturation, particularly late stage maturation.

Differential Effects of Foxo1 Deficiency on iNKT Effector Lineage Differentiation
iNKT cells differentiate to multiple effector lineages in the thymus that can be defined based expression transcriptional factors/repressors PLZF, GATA3, T-bet, and RORγt (15). Within iNKT cells from Foxo1 f /f -Cd2iCre thymus, the percentages of PLZF + GATA3 + iNKT2 and RORγt + T-bet − iNKT17 cells were increased but RORγt − T-bet + iNKT1 cells were decreased (Figures 5A,B). Thus, Foxo1 deficiency greatly inhibited iNKT1 development, which was consistent with severe decreases of stage 3 iNKT cells that are mostly iNKT1 cells.

Intrinsic Control of iNKT Effector Lineage Differentiation by Foxo1
Because iNKT effector lineages may compete with each other during development and are influenced by thymic environment,  we further examined whether Foxo1 intrinsically controls iNKT cell effector lineage differentiation using the mixed BM chimeric mice described in Figure 2. Similar to data shown in Figure 5, the percentages of iNKT1 were decreased but iNKT17 and, to certain extent, iNKT2 were increased within CD45.2 + Foxo1KO iNKT cells (Figures 6A,B). Additionally, the Foxo1KO to WT ratios of iNKT1, iNKT2, and iNKT17 cells were 0.077:1, 0.36:1, and 1.1:1, respectively ( Figure 6C). Because the Foxo1KO/WT iNKT17 ratios (1.1:1) were greater than stage 2 iNKT ratios (0.35:1, Figure 4D), it suggested that Foxo1KO iNKT17 differentiation was enhanced and that Foxo1 deficiency intrinsically inhibited iNKT1 but promoted iNKT17 differentiation. In Foxo1KO mice, iNKT1 cells but not iNKT2 or iNKT17 cells were prone to death compared with WT iNKT cells ( Figure 6D). ROS levels were weakly increased in stage 3 iNKT cells (Figure 6E), which might contribute increased death and decreases of iNKT1 cells as most stage 3 iNKT cells are iNKT1 cells. Foxo1 inhibits T-bet expression in NK cells (60). Within Foxo1KO iNKT1 cells, T-bet expression was increased compared with WT controls, suggesting that Foxo1 also inhibited Tbet expression in iNKT cells (Figure 6F). Additionally, CD122, the IL2/15Rβ chain that mediates IL15 signal to positively regulate T-bet expression and to promote iNKT1 differentiation and survival (61,62), was expressed at similar levels between WT and Foxo1KO iNKT1 cells (Figure 6F), suggesting that the reduction of Foxo1KO iNKT1 cells was unlikely due to altered T-bet or CD122 expression. RORγt is crucial iNKT17 differentiation (18). RORγt levels were not obviously changed in Foxo1KO iNKT17 cells or in DP thymocytes (Figure 6F). IL7 receptor (IL7R) signal and ICOS costimulatory signal promote iNKT17 lineage differentiation/homeostasis (8,61,63). However, both IL7Rα and ICOS levels were reduced in iNKT cells at different stages (Figure 6G), suggesting that Foxo1 deficiency alleviated the requirement of these signals for iNKT17 cells and might promote iNKT17 lineage via other mechanisms. Connecting lines indicate individual pairs of sex-and age-matched test and control mice in each experiment. One pair of mice was examined in each experiment. *p < 0.05; **p < 0.01 determined by two-tail pair-wised Student t-test.

DISCUSSION
iNKT cells differentiate to multiple effector lineages that play distinct roles in immune responses and diseases. Evidence suggests that iNKT1 and iNKT17 are competing lineages during iNKT effector differentiation. Both intrinsic program and environmental factors are both involved in control the balance between iNKT1 and iNKT17 cells (8,15,(63)(64)(65)(66)(67). In this report, we demonstrated that deficiency of Foxo1 intrinsically inhibited iNKT1 but enhanced iNKT17 differentiation without obviously affecting early iNKT cell development, suggesting that Foxo1 controls iNKT effector lineage fate decision.
In Foxo1 f /f -Cd2iCre mice, stage 0, 1, and 2 iNKT cell numbers are not obviously different from WT control mice and only stage 3 iNKT cell numbers and percentages were decreased. In mixed BM chimeric mice reconstituted with Foxo1 f /f -Cd2iCre and congenic WT BM cells, Foxo1KO to WT ratios of stage 0 iNKT cells are slightly increased compared with the ratios of CD4 + CD8 + thymocytes. Thus, Foxo1 is dispensable for early iNKT cell development. However, other members of the Foxo subfamily are also expressed in developing thymocytes, our data do not rule out the possibility that those factors may function redundantly with Foxo1 and compensate for its deficiency during early iNKT cell development. Further studies with compound deficiency of Foxo1 and other family members are needed to fully rule out a role of Foxo1 in early iNKT cell development.
Foxo1 deficiency leads to considerable decreases of CD44 + NK1.1 + T-bet + iNKT1 cells but increases of CD44 + NK1.1 − RORγt + iNKT17 cells. iNKT1 cell percentages and numbers are both decreased, revealed a positive role of Foxo1 in promoting iNKT1 differentiation/ maintenance. The increases of iNKT17 cells in the absence of Foxo1 are not solely resulted from decreases of iNKT1 cells and consequent increases of iNKT17 ratios. The Foxo1KO to WT ratios of iNKT17 cells are significantly overrepresented compared with those of stage 1 and 2 iNKT cells in mixed BM chimeric mice, indicating that Foxo1 negatively controls iNKT17 differentiation. Thus, Foxo1 plays an intrinsic role in controlling iNKT1/17 effector fate decision and may rheostat the balance between these two iNKT effector lineages.
Numerous mechanisms have been found to control iNKT1/17 lineage differentiation. Many transcription factors and regulators such as cMAF, Runx1, MED2/3, and NKAP are important for iNKT17 differentiation (68)(69)(70)(71)(72)(73)(74), while the epigenetic modifiers of the TET-family dioxygenases inhibit iNKT17 but promote iNKT1 differentiation (67). Foxo1 deficient iNKT1 but not iNKT2/17 cells were prone to death and contained increased ROS, suggesting that Foxo1 might regulate ROS production to promote iNKT1 cell survival. IL15R signal is critical for iNKT1 cell differentiation and homeostasis in part by increasing T-bet expression (61,62). We have found IL15Rβ chain expression is not reduced in Foxo1KO iNKT1 cells. Thus, Foxo1 deficiency may cause iNKT1 defect via mechanisms other than altered expression of T-bet or CD122. IL7R signal and ICOS costimulatory signal promote iNKT17 cell homeostasis (8,61,63). Both IL7Rα and ICOS levels were decreased in Foxo1KO iNKT cells. In conventional T cells, Foxo1 binds to Il7ra and Icos loci to promote their transcription (48,75). Foxo1 may function similarly to promote expression of these molecules in iNKT cells. However, it is unlikely that the decreased expression of IL7R and ICOS causes enhanced iNKT17 differentiation in Foxo1KO mice. Although ICOS and IL7R are not crucial for iNKT1 differentiation, our data do not rule out the possibility that decreased expression of these molecules may impair iNKT1 lineage differentiation/homeostasis in the context of Foxo1 deficiency.
Foxo1 inhibits Th17 differentiation and IL-17A expression by directly suppressing transcription of RORγt and IL-23R expression and by interacting with RORγt (44,47,76). Although such mechanisms could operate similarly iNKT cells, we do not observe increased RORγt protein levels in Foxo1KO iNKT17 cells. It is interesting to note that Foxo1 inhibits T-bet mediated effector differentiation of CD8 T cells and Th1 differentiation (55,77) and binds to the Tbx21 promoter to inhibit T-bet expression in NK cells (60). In Foxo1 deficient iNKT1 cells, Tbet levels were increased, suggesting that Foxo1 may also directly suppress T-bet expression in iNKT cells. However, increased Tbet expression should not be the reason for deficiency of iNKT1 cells in Foxo1KO mice.
Foxo1 is regulated by multiple mechanisms. Akt, SGK1, and the serine/threonine kinase CK2 phosphorylate Foxo1 to inhibit its nuclear localization and activity. mTOR regulates iNKT1 and iNKT17 differentiation via mTOR complex 1 (mTORC1) and mTORC2 and their tight regulation by the tumor suppressor TSC1 ensures proper iNKT1/17 balance (8,66,78,79). mTORC2 phosphorylates Akt and SGK1 to enhance their enzyme activities (80)(81)(82). Deficiency of either Akt2 or mTORC2 causes decreases of iNKT17 cells, which correlates with decreased Foxo1 phosphorylation or increased Foxo1 nuclear localization (66,83). The data we report here provide direct evidence that Foxo1 plays critical roles in iNKT cell development and effector lineage differentiation. Although the importance of SGK1 and CK2 in iNKT cells has been unclear, both SGK1 and CK2 promote Th17 differentiation at least partially through inhibition of Foxo1 (45,84,85). Our data together with these observations indicate that Foxo1 may integrate signals from mTORC2/Akt/SGK1 and other enzymes to control iNKT cell effector lineage fate decision.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The protocol was approved by the Duke University Institute Animal Care and Use Committee.

AUTHOR CONTRIBUTIONS
HT and LL designed and performed experiments, analyzed data, and were involved in manuscript preparation. YG and ZW were involved in data analyses and manuscript preparation. X-PZ conceived the project, designed experiments, analyzed data, and wrote the manuscript.

FUNDING
This study was supported by the National Institutes of Health (AI079088 and AI101206) to X-PZ.