Functional Reprogramming of Regulatory T cells in the absence of Foxp3

Summary

Regulatory T cells (T_reg cells) deficient in the transcription factor Foxp3 lack suppressor function and manifest a T effector (T_eff) cell-like phenotype. We demonstrate that Foxp3 deficiency dysregulates metabolic checkpoint kinase mTORC2 signaling and gives rise to augmented aerobic glycolysis and oxidative phosphorylation. Specific deletion of the mTORC2 adaptor gene Rictor in Foxp3-deficient T_reg cells ameliorated disease in a Foxo1 transcription factor-dependent manner. Rictor deficiency reestablished a subset of T_reg cell genetic circuits and suppressed the T_eff cell-like glycolytic and respiratory programs, which contributed to immune dysregulation. Treatment of T_reg cells from patients with FOXP3 deficiency with mTOR inhibitors similarly antagonized their T_eff cell-like program and restored suppressive function. Thus, regulatory function can be reestablished in Foxp3-deficient T_reg cells by targeting their metabolic pathways, providing opportunities to restore tolerance in T_reg cell disorders.

Introduction

Regulatory T cells (T_reg cells) enforce peripheral immunological tolerance by suppressing immune responses to self-antigens and innocuous environmental antigens ^{1, 2}. The Forkhead transcription factor Foxp3 is essential for T_reg cell differentiation and function ^{1, 3, 4}. While Foxp3 is not essential for thymic T_reg cell development, Foxp3-deficient T_reg cells (ΔT_reg cells) lack regulatory function ^{3, 4}. The core T_reg cell transcriptome and epigenome are largely preserved in ΔT_reg cells, although the expression of individual genes is frequently decreased ^{3, 4, 5}. Nevertheless, ΔT_reg cells acquire additional phenotypic and transcriptional attributes akin to those of effector T (T_eff) cell ^{3, 4, 6}. These findings led to the concept that Foxp3 stabilizes the transcriptome of T_reg cells and prevents their degeneration into T_eff cells ⁷.

T_eff cells undergo a profound change in their bioenergetic profile in favor of augmented aerobic glycolysis, oxidative phosphorylation (OXPHOS) and glutaminolysis as well as the de novo acquisition of biosynthetic pathways such as fatty acid synthesis ⁸. T_reg cell metabolism is biased towards fatty acid oxidation (FAO) and pyruvate-dependent OXPHOS ^{9, 10, 11, 12}, whereas glycolysis is kept under strict control. Enforced expression of Foxp3 promotes OXPHOS and suppresses glycolysis ¹³. Reciprocally, upregulation of the glucose transporter Glut1 by toll-like receptor signaling, acting via the mammalian Target of Rapamycin Complex 1 (mTORC1), or by its enforced expression promoted T_reg cell proliferation but dampened Foxp3 expression and T_reg cell suppressive function ¹⁴. We reasoned that interventions that deprive the T_eff cell-like programs of ΔT_reg cells of metabolic support may allow recovery of regulatory function by virtue of the preservation in those cells of a core regulatory transcriptome.

Results

Lineage-specific, Cre-mediated recombination in Foxp3-deficient T_reg cells

To enable genetic manipulations in ΔT_reg cells, we derived a bicistronic loss of function Foxp3 allele, Foxp3^ΔEGFPiCre, that contained a Frt element inserted between exons 9 and 10 that created an aberrant 3’ splice junction site, leading to a premature stop codon in the Foxp3 transcript. A ribosomal entry sequence inserted in the 3’ untranslated region of Foxp3 directed the expression of a humanized Cre recombinase (iCre) fused with the enhanced green fluorescent protein (EGFP) (Supplementary Fig. 1a–d). T_reg cells of Foxp3^ΔEGFPiCre mice expressed EGFP but not Foxp3 (Supplementary Fig. 1e, f). Analysis of Foxp3^ΔEGFPiCre mice harboring a Cre-inducible yellow fluorescent protein (YFP) transgene in the Rosa26 locus (R26^YFP) revealed that all EGFP⁺ cells were YFP⁺ and virtually all within the CD4⁺ T cell population (Supplementary Figure 1e). Comparison with wild-type mice carrying a bacterial artificial chromosome (BAC) harboring a Foxp3 promoter-driven EGFP and Cre recombinase and crossed with R26^YFP (Foxp3^EGFPCreR26^YFP) showed that the EGFP⁺ cells of Foxp3^ΔEGFPiCre mice were expanded, in agreement with previous studies on Foxp3-deficient mice (Supplementary Fig. 1e) ^{3, 4}. Within the YFP⁺ cell population, the frequency of cells that have downregulated their Foxp3 locus activity (EGFP^–YFP⁺) was modestly increased in the Foxp3^ΔEGFPiCreR26^YFP males, pointing to incremental ΔT_reg cell instability.

Analysis of Foxp3^EGFPCre and Foxp3^ΔEGFPiCre EGFP^– and EGFP⁺ CD4⁺ T cells revealed that Foxp3 transcripts were highly enriched in EGFP⁺ cells (Supplementary Fig. 1g). Foxp3 transcripts were 20 fold higher in Foxp3^EGFPCre EGFP⁺ compared to Foxp3^ΔEGFPiCre EGFP⁺ cells, reflecting a positive feed-forward regulation by Foxp3 of its own gene expression (Supplementary Fig. 1g) ⁴. EGFP⁺ ΔT_reg cells exhibited similar demethylation of the Foxp3 CNS2 promoter region as EGFP⁺ control T_reg cells from Foxp3^EGFP knockin reporter allele mice, indicating that the ΔT_reg cells were epigenetically of T_reg cell lineage (Supplementary Fig. 1h).

Analysis of Foxp3^ΔEGFPiCre/+ heterozygous females revealed that similar to the indigenous wild-type Foxp3 allele, expression of Foxp3^ΔEGFPiCre in thymocytes was overwhelmingly restricted to the CD4⁺ lineage, starting at the CD4 single positive thymocyte stage (Supplementary Fig. 1i). However, the ΔT_reg cells were dramatically decreased in the periphery compared to wild-type T_reg cells, indicative of their reduced fitness (Supplementary Fig. 1i) ^{3, 4}. Overall, these results established that Foxp3^ΔEGFPiCre allele was specifically expressed in the T_reg cell lineage, and rendered these cells totally deficient in Foxp3 while enabling EGFP and Cre recombinase expression.

Inactivation of mTORC2 in ΔT_reg cells ameliorates disease

The mTOR pathways play a key role in promoting T_eff cell function ^{15, 16}. The mTOR kinase exists as part of two complexes: mTORC1 and mTORC2, that differentially contribute to T_H cell effector functions, while mTOR-deficient T cells differentiate in vitro into iT_reg cells ^{17, 18, 19}. mTORC1 acts as a finely-tuned T_reg cell metabolic checkpoint essential for T_reg cell homeostasis and function ²⁰. Activated mTORC1 phosphorylates the downstream substrates S6 and 4EBP, while activated mTORC2 phosphorylates the serine/threonine kinase AKT at serine residue 473 (p_S473AKT) ^{20, 21}. ΔT_reg cells from Foxp3^ΔEGFPiCre hemizygous mutant males, but not from heterozygous Foxp3^ΔEGFPiCre/+ females, had increased pS6 and p4EBP at steady state compared to Foxp3-sufficient T_reg cells from Foxp3^EGFPCre mice (Fig. 1a,b and Supplementary Fig. 2a,b). However, both ΔT_reg cell populations upregulated their pS6 phosphorylation in response to anti-CD3 mAb stimulation to the same extent as Foxp3-sufficient T_reg cells, indicating that increased basal mTORC1 activity in male ΔT_reg cells was cell-extrinsic. Induction of p_T308AKT, a target of upstream Phosphoinositide-dependent kinase ²², was also increased to a similar extent in T_reg and ΔT_reg cells of Foxp3^EGFPCre and Foxp3^ΔEGFPiCre mice, respectively, upon anti-CD3 mAb stimulation (Supplementary Fig. 2a,b).

Figure 1. Inactivation of mTORC2 but not mTORC1 in ΔT_reg cells mitigates Foxp3 deficiency.

(a,b) Representative flow cytometric analysis (a) and mean fluorescence intensity (MFI) (b) of phosphorylated S6 (pS6) and phosphorylated AKT at Ser473 residu (p_S473AKT) expression of unstimulated (US) or anti-CD3 stimulated (α-CD3) T_reg cells from Foxp3^EGFPCre, Foxp3^ΔEGFPiCre/+ and Foxp3^ΔEGFPiCre mice (n=5 per group). Results represent 1 of 3 independent experiments. (c,d) Body weight at 25–28 days of age (c) and survival (d) of Foxp3^K276X, Foxp3^ΔEGFPiCre, Foxp3^ΔEGFPiCreRictor^Δ/WT, Foxp3^ΔEGFPiCreRictor^Δ/Δ, Foxp3^ΔEGFPiCreRptor^Δ/WT, Foxp3^ΔEGFPiCreRptor^Δ/Δ, Foxp3^ΔEGFPiCreRictor^Δ/ΔRptor^Δ/Δ mice and control littermates. Results represent a pool of 5 independent experiments. (e) Gross appearance of Foxp3^ΔEGFPiCre, Foxp3^ΔEGFPiCreRictor^Δ/Δ and control littermate and their respective spleens and peripheral lymph nodes. (f,g) Representative microscopic pictures of H&E staining (original magnification ×200) (f) and histological scores (g) of skin, lung, colon and liver of Foxp3^ΔEGFPiCre (n=14), Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=7) and control littermates (n=5). Results represent a pool of 3 independent experiments. Statistical significance was determined by a one-way ANOVA with Tukey’s multiple comparisons (c, g), two-way ANOVA with Sidak’s multiple comparisons (b) and log rank test (d) (P values as indicated).

p_S473AKT was selectively increased in ΔT_reg cells of both hemizygous males and heterozygous females following anti-CD3 mAb stimulation, indicating that Foxp3 deficiency dysregulated mTORC2 activation in a cell intrinsic manner irrespective of the inflammatory environment (Fig. 1a,b). The increased mTORC2 activity in ΔT_reg cells was associated with decreased protein expression of phosphatidylinositol-3,4,5-trisphosphate 3-phosphatase phosphatase and tensin homolog (PTEN) and PH-domain leucine-rich-repeat protein phosphatase 1 (PHLPP1), two phosphatases implicated in mTORC2 regulation ^{23, 24} (Supplementary Fig. 2c). We also evaluated proximal TCR signaling in ΔT_reg cells of Foxp3^ΔEGFPiCre mice by measuring Zeta-chain-associated protein kinase 70 (ZAP70) phosphorylation at tyrosine 319 residue upon anti-CD3 mAb stimulation, an event critical to downstream calcium mobilization, Ras activation and NFAT-dependent transcription ²⁵. ΔT_reg cells exhibited attenuated ZAP70 phosphorylation in response to anti-CD3 mAb stimulation compared to Foxp3-sufficient T_reg cells (Supplementary Fig. 2d). These results suggested that the enhanced mTORC2 activation in ΔT_reg cells of Foxp3^ΔEGFPiCre mice reflected impaired regulation by phosphatases.

Male mice hemizygous for the Foxp3^ΔEGFPiCre allele were runted and died early due to autoimmune lymphoproliferation, similar to another Foxp3-deficient mouse strain, Foxp3^K276X, that we previously generated ²⁶ (Fig. 1c–f). We examined the role of dysregulated mTOR signaling in ΔT_reg cells of Foxp3^ΔEGFPiCre mice in their disease by cell-specific deletion of Rptor and Rictor, encoding the mTORC1 and mTORC2 regulators Raptor and Rictor, respectively. Expression of Rptor and Rictor transcripts was specifically abrogated in EGFP⁺ cells of Foxp3^ΔEGFPiCreRptor^Δ/Δ and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice, respectively (Supplementary Fig. 2e,f). Rictor protein expression was increased in EGFP⁺ cells of Foxp3^ΔEGFPiCre mice, mirroring the increase in its transcripts, and was also abrogated in the ΔT_reg cells of Foxp3^ΔEGFPiCreRictor^Δ/Δ mice, as confirmed by immunoblotting (Supplementary Fig. 2g). ΔT_reg cell-specific Rictor deletion markedly increased the body weight and survival of Foxp3^ΔEGFPiCre mice in a gene dose-dependent manner, whereas Rptor deletion did not (Fig. 1c,d). Furthermore, combined deletion of Rictor and Rptor in ΔT_reg cells of Foxp3^ΔEGFPiCreRictor^Δ/ΔRptor^Δ/Δ mice did not increase survival or body weight compared to Foxp3^ΔEGFPiCre or Foxp3^ΔEGFPiCreRptor^Δ/Δ mice (Fig. 1c,d). Foxp3^ΔEGFPiCreRictor^Δ/Δ mice had decreased tissue inflammation compared to Foxp3^ΔEGFPiCre mice but persistent lymphoproliferation (Fig. 1e–g).

The marked increase in ΔT_reg cells in Foxp3^ΔEGFPiCre mice was reversed in Foxp3^ΔEGFPiCreRptor^Δ/Δ but not Foxp3^ΔEGFPiCreRictor^Δ/Δ mice, indicating its dependence on mTORC1 activation (Supplementary Fig. 2h). The phenotype of EGFP⁺ ΔT_reg cells of Foxp3^ΔEGFPiCre mice closely resembled that of ΔT_reg cells analyzed in earlier models ^{3, 4}. In addition to absent Foxp3 expression, ΔT_reg cells had lower expression of CD25 and CD73 and higher expression of ICOS, Helios, GITR, CD127, IL-2 and Granzyme B compared to WT T_reg cells (Supplementary Fig. 2i,j and L-M.C., data not shown). ΔT_reg cells of Foxp3^ΔEGFPiCreRictor^Δ/Δ mice had upregulated expression of ICOS, Nrp1, GITR and CD73 compared to ΔT_reg cells from Foxp3^ΔEGFPiCre mice (Supplementary Fig. 2i,j). These results indicated that mTORC2 was specifically dysregulated in ΔT_reg cells in a cell-intrinsic manner, and promoted disease in Foxp3-deficient mice.

Rictor deficiency restores ΔT_reg cell suppressor function

Analysis of Foxp3^ΔEGFPiCreRictor^Δ/Δ mice revealed that the frequencies of CD4 and CD8 effector memory cells were decreased by about 50% compared to Foxp3^ΔEGFPiCre mice, whereas those of naive and central memory T cells were increased (Fig. 2a,b and Supplementary Fig. 3a,b). Expression of the T_H1-associated transcription factor T-bet and IFN-γ was markedly increased in ΔT_reg and T_eff cells of Foxp3^ΔEGFPiCre compared to control Foxp3^EGFPCre mice, but was down-regulated by Rictor deficiency (Foxp3^ΔEGFPiCreRictor^Δ/Δ) By contrast, expression of the T_H2-associated transcription factor Gata-3 and IL-4, which was also increased in Foxp3^ΔEGFPiCre ΔT_reg and T_eff cells, was further upregulated in ΔT_reg cells by Rictor deficiency, thus confirming a critical role for mTORC2 in T_H1 but not T_H2 programming of ΔT_reg cells (Fig. 2a, b and Supplementary Fig. 3a,b). In contrast, expression of ROR-γt and IL-17 was unchanged (Supplementary Fig. 3a,b).

Figure 2. mTORC2 blockade endows ΔT_reg cells with regulatory function.

(a,b) Representative flow cytometric analysis (a) and frequencies (scatter plots with mean) (b) of CD62L^hiCD44^lo and CD62L^loCD44^hi CD4⁺ T cells or IFN-γ and IL-4 expression by CD4⁺ T_reg (YFP⁺) and T_eff (YFP^–) cells from spleen of Foxp3^EGFPCreR26^YFP (n=5 for CD62L^hiCD44^lo and CD62L^loCD44^hi CD4⁺ T cells, n=4 for IFN-γ and IL-4 expression), Foxp3^ΔEGFPiCreR26^YFP (n=4) and Foxp3^ΔEGFPiCreRictor^Δ/ΔR26^YFP mice (n=5 for CD62L^hiCD44^lo and CD62L^loCD44^hi CD4⁺ T cells, n=4 for IFN-γ and IL-4 expression). Results represent 1 of 4 independent experiments (c) In vitro suppression of the proliferation of WT CD4⁺ T_eff cells (denoted as T responder or T_resp) by Foxp3^EGFPCre T_eff cells pre-treated for 1h with DMSO (Veh WT T_eff cells) or 1μM Rapamycin (Rapa WT T_eff cells), Foxp3^EGFPCre T_reg cells pre-treated for 1h with DMSO (Veh WT T_reg cells) or 1μM Rapamycin (Rapa WT T_reg cells) or Foxp3^ΔEGFPiCre T_reg cells pre-treated for 1h with DMSO (Veh ΔT_reg cells) or 1μM Rapamycin (Rapa ΔT_reg cells) (denoted as T suppressor or T_supp) (n=3 per group). Results represent 1 of 3 independent experiments (d, e) In vitro suppression of T_resp cells by Foxp3^EGFPCre T_reg cells (Foxp3^EGFPCre), Foxp3^ΔEGFPiCre T_reg cells pre-treated for 1h with DMSO (Veh Foxp3^ΔEGFPiCre) or 1μM Rapamycin (Rapa Foxp3^ΔEGFPiCre), Foxp3^ΔEGFPiCreRictor^Δ/Δ T_reg cells pre-treated for 1h with DMSO (Veh Foxp3^ΔEGFPiCreRictor^Δ/Δ) or 1μM Rapamycin (Rapa Foxp3^ΔEGFPiCreRictor^Δ/Δ) (n=3 per group) (Results represent 1 of 3 independent experiments) (d), and Foxp3^ΔEGFPiCreRptor^Δ/Δ T_reg cells pre-treated for 1h with DMSO (Veh Foxp3^ΔEGFPiCreRptor^Δ/Δ) or 1μM Rapamycin (Rapa Foxp3^ΔEGFPiCreRptor^Δ/Δ) (n=3 per group) (Results represent 1 of 3 independent experiments) (e). (f) In vitro suppression of T_resp cells by Rictor-sufficient or -deficient T_eff (CD4⁺CD25⁻ cells from CD4^cre and CD4^creRictor^Δ/Δ) and Foxp3-sufficient T_reg (Foxp3^YFPcre and Foxp3^YFPcreRictor^Δ/Δ) cells (n=3 per group) (Results represent 1 of 2 independent experiments). Results are expressed as mean ± SEM in panels c-f. Statistical significance was determined by a one-way ANOVA (b) or two-way ANOVA with Tukey’s multiple comparisons (c-f) (P values as indicated).

We next analyzed the impact of mTOR inhibition on T_reg, ΔT_reg and T_eff cell in vitro suppressive capacities. Pre-treatment of T_eff cells with the mTOR inhibitor Rapamycin did not confer them suppressive capacity. Foxp3^ΔEGFPiCre ΔT_reg cells had a small but detectable suppressive activity as compared to T_eff cells, which was augmented by Rapamycin pretreatment. Rapamycin pretreatment also enhanced the suppressive function of wild-type T_reg from Foxp3^EGFPCre (Fig. 2c).

The positive impact of rapamycin on ΔT_reg cell suppressive capacity mapped to mTORC2. Rictor deletion substantially upregulated ΔT_reg cell-mediated suppression (Fig. 2c), an effect that was not related to decreased ΔT_reg cell proliferation, as evidenced by Ki-67 staining (Supplementary Fig. 3a,b). Furthermore, Rictor but not Raptor deficiency abrogated the enhanced ΔT_reg cell suppressive activity mediated by Rapamycin, indicating that Rapamycin promoted ΔT_reg cell suppression by inhibiting mTORC2 (Fig. 2d,e).

We further compared the suppressive capacity of CD4⁺CD25^– T_eff cells isolated from Cd4^Cre versus Cd4^CreRictor^Δ/Δ mice. Results showed that Rictor deficiency did not confer any suppressive capacity to T_eff cells. In contrast, Rictor deficiency in otherwise Foxp3-sufficient T_reg cells (Foxp3^YFPCreRictor^Δ/Δ) resulted in a small but significant improvement in their suppressive capacity (Fig. 2f), consistent with the results of treating wild-type T_reg cells with Rapamycin (Fig. 2c).

The significance of improved wild-type T_reg cell function upon Rictor deficiency was evaluated in vivo using the lymphopenia colitis model ²⁷. Foxp3^YFPCreRictor^Δ/Δ T_reg cells proved superior in their capacity to control the intestinal inflammation in Rag1-deficient mice reconstituted with naive CD4⁺CD45RB^high T_eff cells. Under limiting T_reg cell availability (1:10 T_reg:T_eff cell ratio), the Foxp3^YFPCreRictor^Δ/Δ T_reg cells prevented weight loss, improved the intestinal inflammation and the inflammation-induced colon shortening, and suppressed the infiltration of the intestines by T_eff cells, including IFN-γ⁺, IL-17⁺ and IFN-γ⁺IL-17⁺ cells (Supplementary Fig. 4a–g).

We next analyzed the impact of Rictor deletion in ΔT_reg cells on their stability. Whereas ex-T_reg cells, corresponding to the EGFP^– fraction of total YFP⁺ CD4 T cells, represented approximately 7% in Foxp3-sufficent control mice (Foxp3^EGFPcreR26^YFP) and 15% in Foxp3-deficient Foxp3^ΔEGPiCreR26^YFP mice, their frequency was significantly decreased to 9–10% in Foxp3^ΔEGFPiCreRictor^Δ/ΔR26^YFP mice (Supplementary Fig. 3a,b), indicating that mTORC2 destabilized ΔT_reg cells, possibly by affecting their epigenetic demethylation ²⁸.

We employed heterozygous Foxp3^ΔEGFPiCre/+ female mice, which are phenotypically normal, to examine the impact of Rictor deficiency on ΔT_reg cells in the absence of systemic inflammation. ΔT_reg cells isolated from Foxp3^ΔEGFPiCre/+ females had more suppressive activity at high ΔT_reg/T_eff cell ratios than those of Foxp3^ΔEGFPiCre males, indicative of a detrimental role for inflammation in the loss of ΔT_reg suppressive function (Fig. 3a). To assess T_reg cell fitness, we examined the frequencies of wild-type and ΔT_reg cells in Foxp3^ΔEGFPiCre/+R26^YFP versus Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP heterozygous female mice. Whereas ΔT_reg cells were profoundly under-represented in Foxp3^ΔEGFPiCre/+ mice compared to T_reg cells carrying the wild-type Foxp3 allele, their frequency increased in Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP mice, indicative of their improved fitness (Fig. 3b,c). ΔT_reg cells from Foxp3^ΔEGFPiCre/+R26^YFP and Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP heterozygous females minimally expressed T-bet/IFN-γ and Gata-3/IL-4 compared to ΔT_reg cells of hemizygous males Fig. 3d,e). These results indicate that the acquisition by ΔT_reg cells of T_H cell-like phenotypes was augmented by the intense inflammatory environment in Foxp3-deficient males. Further characterization of ΔT_reg cells of Foxp3^ΔEGFPiCre/+R26^YFP and Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP heterozygous females revealed increased expression of CD25 and Nrp1 in Rictor-deficient ΔT_reg cells (Fig. 3f). Overall, the results in Fig. 2 and Fig. 3 show that mTORC2 was cell intrinsically dysregulated in ΔT_reg cells, and that its inhibition substantially restored ΔT_reg cell function.

Figure 3: Cell intrinsic and extrinsic determinants of the T_eff cell-like phenotype of ΔT_reg cells.

(a) In vitro suppression of CD4⁺ T_eff cell (T_resp) proliferation by Foxp3-sufficient T_reg cells from Foxp3^EGFPcre mice and ΔT_reg cells from Foxp3^ΔEGFPiCre/+ and Foxp3^ΔEGFPiCre mice (T_supp) (n=3 per group). Results are expressed as mean ± SEM and represent 1 of 2 independent experiments. (b, c) Representative flow cytometric analysis of Foxp3 and YFP among CD4⁺ T cells (b) and frequencies of ΔT_reg cells (c) among total T_reg cells from Foxp3^ΔEGFPiCre/+R26^YFP (n=12) and Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP (n=10) mice. Results are expressed as scatter plot and mean and represent a pool of 3 independent experiments. (d, e) Representative flow cytometric analysis of IL-4 and IFN-γ among YFP⁺ T_reg cells (d) and frequencies (Scatter plot and mean) of T-Bet⁺, Gata-3⁺, IFN-γ⁺ and IL-4⁺ YFP⁺ T_reg cells (e) from Foxp3^EGFPCreR26^YFP, Foxp3^ΔEGFPiCre/+R26^YFP, Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP, Foxp3^ΔEGFPiCreR26^YFP and Foxp3^ΔEGFPiCreRictor^Δ/ΔR26^YFP mice (n=4 per group). Results represent 1 of 2 independent experiments. (f) CD25, CTLA4, Nrp1 and Helios MFI in WT T_reg and ΔT_reg cells from Foxp3^EGFPCreR26^YFP, Foxp3^ΔEGFPiCre/+R26^YFP, Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP female mice (n=4 per group). ND, Not Determined. Results represent 1 of 2 independent experiments. Statistical significance was determined by unpaired t-test (c), one-way ANOVA (e) or two-way ANOVA (a, f) with Tukey’s multiple comparisons (P values as indicated).

mTORC2 dependent and independent transcriptional programs in ΔT_reg cells

We compared the transcriptomes of ΔT_reg and T_reg isolated from female mice that carried one mutant Foxp3^ΔEGFPiCre allele and a second competent Foxp3 allele (Foxp3^RFP) that also directed the expression of the red fluorescent protein (RFP), thus allowing for color sorting of the respective populations ²⁹. We extended these studies to examine the transcriptome of ΔT_reg cells of Foxp3^ΔEGFPiCre/+Rictor^Δ/Δ females to identify transcriptional pathways altered by Rictor. Results revealed global changes in the transcriptional landscape induced by Foxp3 deficiency. Concurrent Rictor deficiency normalized some of these changes, while also inducing de novo Foxp3-independent transcriptional alterations of its own Fig. 4a,b and Data Set 1). ΔT_reg cells exhibited decreased expression of some genes associated with the T_reg cell transcriptome (e.g. Itgae, Fgl2, Nrp1, Ebi3a, Il2ra, Ikzf2, Tigit) and the acquisition of several others associated with T_eff cells, including T_H17 cells (e.g. Rorc, Tgfb3, Il17a, Nr1d1, Sgk1), cytokines and cellular activators (Il2, Il21, Cd40lg).

Figure 4. mTORC2-dependent and -independent gene expression profiles in ΔT_reg cells.

Gene expression profiles of Foxp3^RFP (WT), Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ T_reg cells (n=4 per group) isolated from spleen of Foxp3^{RFP/ΔEGFPiCre}R26^YFP and Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP female mice. (a), Gene expression profiles represented as Fold change (Foxp3^RFP vs Foxp3^ΔEGFPiCre) versus Fold change (Foxp3^RFP vs Foxp3^ΔEGFPiCreRictor^Δ/Δ). (b), Heatmap representation, (c), Log₂ expression and (d) volcano plot of differential gene expression in Foxp3^ΔEGFPiCre versus Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells. FDR, false discovery rate; log2FC, log2(fold change). (e), Flow cytometric analysis of IL-10 and Blimp1 expression in T_reg cells of Foxp3^EGFPCre (n=4 and n=8 respectively) and ΔT_reg cells of Foxp3^ΔEGFPiCre (n=4 and n=10 respectively) and Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=4 and n=6 respectively) mice. Results are expressed as scatter plot and mean and represent 1 of 3 independent experiments. (f), Chromatin immunoprecipitation (ChIP) assay for the binding of Blimp1 and control isotype to the Il10 conserved non-coding sequence located 9kb in proximal of il10 promoter (il10 CNS–9) in Foxp3^ΔEGFPiCre (n=5) and Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=13) ΔT_reg cells. Results are expressed as scatter plot and mean and represent a pool of 2 independent experiments. (g), In vitro suppression of T_resp cells by Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells carried out in the presence of either an isotype control or anti-IL-10 mAb (n=3 per group). Results are expressed as mean ± SEM and represent 1 of 2 independent experiments. (h) ChIP assays for the binding of Foxo1 and control isotype to Tbx21 and Ifng promoters in Foxp3^ΔEGFPiCre (n=5) and Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=13) ΔT_reg cells. Results are expressed as scatter plot and mean and represent a pool of 2 independent experiments. Statistical significance was determined one-way ANOVA (e) or two-way ANOVA (f-h) with Tukey’s or Sidak’s multiple comparisons (P values as indicated).

Relevant to the mTOR pathway, ΔT_reg cells had increased expression of Prr5l, encoding an mTORC2 regulator that directs its substrate specificity ³⁰, and decreased expression of Phlpp1, encoding the phosphatase PHLPP1 ²³. Concurrent Rictor deficiency upregulated a subset of the core T_reg transcriptome genes, including Fgl2, Il10, Lag3, Tnfrsf18 (Gitr), Ebi3, Nrp1, Il2ra, Ctla4 and Nt5e. It also down-regulated the expression of genes associated with the T_H17 (Rorc, Tgfb3) and T follicular helper cell (Cxcr5 and Bcl6) lineages, while upregulating the Bcl6 antagonist Prdm1 Fig. 4c,d and Data Set 2). These results suggested that Rictor deficiency altered the ΔT_reg cell transcriptome in favor of improved regulatory function.

Of particular interest was the role of increased IL-10 and Blimp1 expression in Rictor-deficient ΔT_reg cells (Fig. 4e). Blimp1 binds to the regulatory CNS –9 element located proximal to the Il10 gene to promote Il0 transcription ³¹. Chromatin immunoprecipitation (ChIP) studies demonstrated increased binding of Blimp1 to the Il10 CNS–9 element of Rictor-deficient ΔT_reg cells (Fig. 4f). IL-10 neutralization largely abrogated the improved in vitro suppressive capacity of Rictor deficient ΔT_reg cells, indicating a mechanistic role for this pathway in their augmented regulator function (Fig. 4g).

mTORC2 signaling in ΔT_reg cells activates the AKT-FOXO1 axis

We examined mTORC2 activity, as monitored by p_S473AKT staining, in wild-type T_reg cells from Foxp3^EGFP mice, and in ΔT_reg cells from Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice. The increase in p_S473AKT staining in Foxp3^ΔEGFPiCre ΔT_reg cells following their stimulation with anti-CD3+anti-CD28 mAbs was totally reversed in the Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells Fig. 5a,b). AKT phosphorylates the transcription factor Foxo1, resulting in its retention in the cytosol and its ubiquitination and degradation ³². Foxo1 in turn negatively regulates T_H1 differentiation of T_reg cells by suppressing the transcription of Ifng and other T_H1 genes ^{33, 34}. Expression of several Foxo1-regulated genes, including Il7r, Cd55, Cd83, Emb, Etv5 and Mafg, was affected in Foxp3^ΔEGFPiCre ΔT_reg cells and normalized in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells (Data Set 2). Foxo1 was decreased in the nucleus in anti-CD3 mAb-stimulated ΔT_reg cells of Foxp3^ΔEGFPiCre compared to control Foxp3^EGFP T_reg cells, a deficit that was reversed in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells Fig. 5c,d). Also, Rictor deletion increased Foxo1 binding at the Ifng and Tbx21 promoters (Fig. 4g).

Figure 5. Contribution of AKT/Foxo1 axis to the Rictor-dependent ΔT_reg cell phenotype.

(a) Representative flow cytometric analysis and (b) MFI of p₄₇₃-AKT expression in unstimulated (US) and anti-CD3/CD28 mAb-stimulated (α-CD3+α-CD28) T_reg cells from Foxp3^EGFPCre, Foxp3^ΔEGFPiCre, Foxp3^ΔEGFPiCreRictor^Δ/Δ mice (n=3 per group). Results are expressed as scatter plot and mean and represent 1 of 3 independent experiments. (c,d) Representative confocal microscopic merge image of Foxo1 (Red) and DAPI (Blue) (c), and percent of nuclear Foxo1 (d) in unstimulated (US) or 0.1μg/mL anti-CD3 (α-CD3) stimulated T_reg cells from Foxp3^EGFPCre (n=53 for US, n=71 for α-CD3), Foxp3^ΔEGFPiCre (n=74 for US, n=95 for α-CD3), Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=78 for US, n=46 for α-CD3) mice. Results are expressed as scatter plot and mean and represent a pool of 2 independent experiments. (e) Representative flow cytometric analysis and (f) frequencies (scatter plot and mean) of CD62L^loCD44^hi CD4⁺ T_eff cells, and IFN-γ and IL-4 expression by CD4⁺ T_reg and T_eff cells in spleens of Foxp3^EGFPCre (n=10), Foxp3^ΔEGFPiCre (n=7), Foxp3^ΔEGFPiCreRictor^Δ/Δ (n=6), Foxp3^ΔEGFPiCreFoxo1^Δ/Δ (n=3), Foxp3^ΔEGFPiCreRictor^Δ/ΔFoxo1^Δ/Δ (n=4) and Foxp3^ΔEGFPiCreR26^Foxo1AAA (n=5) mice. Results represent a pool of 3 independent experiments. Statistical significance was determined one-way ANOVA with Tukey’s multiple comparisons (f) or two-way ANOVA with Sidak’s multiple comparisons (b, d) (P values as indicated).

We examined the consequences of deleting a floxed Foxo1 gene in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells. The triple mutant Foxp3^ΔEGFPiCreRictor^Δ/ΔFoxo1^Δ/Δ mice were similar to single mutant Foxp3^ΔEGFPiCre and the double mutant Foxp3^ΔEGFPiCreFoxo1^Δ/Δ mice in terms of growth and survival, dysregulated T_eff cells and IFN-γ expression in both ΔT_reg and T_eff cells, whereas the IL-4 response was unaffected Fig. 5e,f and Supplementary Fig. 5a–e). Foxo1 deletion reversed the upregulation of Il10 expression in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg (Supplementary Fig. 5f). Reciprocally, ΔT_reg cell-specific expression of Cre-regulated AKT-insensitive Foxo1 transgene (R26^Foxo1AAA) partially recapitulated the effects of Rictor deficiency in improving weight gain, prolonging survival and augmenting ΔTreg cell suppressive capacity (Supplementary Fig. 5a–d) ³³. It also down-regulated T_eff cell activation and the T_H1 programming of ΔT_reg cells Fig. 5e,f and Supplementary Fig. 5e,g). These results indicated that the effects of Rictor deletion in promoting ΔT_reg cell function largely proceeded by Foxo1-dependent mechanisms.

Rictor deficiency resets the metabolism of ΔT_reg cells

Dysregulation of the mTORC2-Foxo1 axis upregulates glycolysis and OXPHOS ^{35, 36}. T_reg cells and naive T cells exhibit low levels of glycolysis in favor of increased OXPHOS, whereas T_eff cells manifest increased aerobic glycolysis, products of which feed into DNA and protein synthesis ³⁷. Foxp3^ΔEGFPiCre ΔT_reg cells had increased expression of several enzymes in the glycolytic and pentose phosphate shunt pathways, several of which were reset back to baseline upon concurrent Rictor deficiency (e.g. Gpi, Aldoa, Eno1, Pkm2 and the pentose phosphate shunt enzymes) (Supplementary Fig. 6a). Some glycolytic enzymes were down-regulated by Rictor deficiency but remained increased above their levels in wild-type T_reg cells (e.g. Hk2, Gapdh), while others such as Tpi and the glycolytic regulator Pfkfb3, which were upregulated in Foxp3-deficient T_reg cells, were either unaffected or further upregulated by Rictor deficiency.

Unlike wild-type T_reg cells, Foxp3^ΔEGFPiCre ΔT_reg cells stimulated with anti-CD3 mAb and IL-2 exhibited exaggerated lactate production similar to activated T_eff cells, which was markedly reduced in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells (Supplementary Fig. 6b). Glycolysis and OXPHOS were further evaluated in Foxp3^EGFP (wild-type) T_reg cells and Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells isolated from hemizygous males using an extracellular metabolic flux analyzer. Of the three T_reg cell populations, Foxp3^ΔEGFPiCre ΔT_reg cells exhibited the largest increase in extracellular acidification rate (ECAR), a measure of glycolysis, upon D-glucose supplementation, while Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg had an intermediate phenotype Fig. 6a,b). All three T_reg cells populations exhibited a modest drop in ECAR upon Oligomycin supplementation, indicating that glycolysis had peaked following glucose addition ³⁸.

Figure 6. mTORC2 promotes aerobic glycolysis and OXPHOS in ΔT_reg cells.

(a, b) Extracellular acidification rate (ECAR) under glycolysis stress test conditions (n=4 per group) and Oxygen consumption rate (OCR) under mitochondrial stress test conditions (n=5 per group) of T_reg/ΔT_reg cells isolated from Foxp3^EGFP, Foxp3^ΔEGFPiCre or Foxp3^ΔEGFPiCreRictor^Δ/Δ males (n=4 per group). Results are expressed as mean ± SEM and represent a pool of 2 independent experiments. (c) OCR under mitochondrial fuel test conditions (n=4 per group). Results are expressed as mean ± SEM and represent a pool of 2 independent experiments (d) Pie chart representation of the contribution of glucose, fatty acids and glutamine to the OXPHOS capacities in Foxp3^EGFP, Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells. (e) Quantification of metabolites (expressed as scaled index) of glucose metabolism (Phosphoenolpyruvate and lactate), tricarboxylic acid (TCA) cycle (Citrate, fumarate and Malate) and lipid metabolism (Acetylcarnitine (C2), Carnitine, 1-palmitoyl-2-decosahexaenoyl and Phospho-ethanolamine) isolated from Foxp3^EGFP, Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells (n=5 per group). Results are expressed as mean ± SEM and represent 1 experiment. Statistical significance was determined one-way ANOVA with Tukey’s multiple comparisons (e) or two-way ANOVA with Sidak’s multiple comparisons (b) (P values as indicated).

OXPHOS, as assessed by the oxygen consumption rate (OCR), was also markedly increased in Foxp3^ΔEGFPiCre ΔT_reg cells. Both basal respiration, ATP-coupled respiration and respiratory reserve were increased in Foxp3^ΔEGFPiCre ΔT_reg compared to control T_reg cells whereas Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg had an intermediate phenotype.. Fig. 6a,b).

The ECAR and OCR of ΔT_reg cells of Foxp3^ΔEGFPiCre/+ females were also highly increased, indicative of cell-intrinsic metabolic changes (Supplementary Fig. 6c,d). Rictor deficiency completely reversed the increase in ECAR in ΔT_reg cells of Foxp3^ΔEGFPiCre/+ females, but only partially reversed the increase in OCR, indicating that additional pathways contributed to the latter’s dysregulation (Supplementary Fig. 6c,d).

The contribution of inputs from glycolysis, glutaminolysis and FAO to the OCR was measured upon the sequential addition of specific metabolic inhibitors: UK5099 for glycolysis, Etomoxir for FAO and BTPES for glutaminolysis (Fig. 6c). Whereas the OCR in wild-type T_reg cells is fatty acid-dependent, it becomes glucose- and to a lesser extent glutamate-dependent in Foxp3^ΔEGFPiCre ΔT_reg cells, pointing to a shift from FAO to aerobic glycolysis and, secondarily, glutaminolysis (Fig. 6c). Rictor deficiency reduced the OCR and decreased aerobic glycolysis Fig. 6c). Transcripts of key factors associated with aerobic glycolysis, including Myc and Hif1a ^{12, 39}, were markedly increased in Foxp3^ΔEGFPiCre ΔT_reg cells, and partially corrected in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells (Supplementary Fig. 6e).

Metabolomic analysis on Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells, and control T_reg and T_eff cells from Foxp3^EGFP mice revealed increased glycolytic (phosphoenolpyruvate, lactate) and citric acid cycle (fumarate, malate) metabolites in Foxp3^ΔEGFPiCre ΔT_reg cells, which were reversed in Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells Fig. 6d and Supplementary Fig. 6f). Foxp3^ΔEGFPiCre ΔT_reg cells were deficient in carnitine, necessary for the transport of fatty acids to the mitochondria for beta oxidation, which was not corrected by Rictor deficiency Fig. 6d and Supplementary Fig. 6f) ⁴⁰. These results established that Foxp3^ΔEGFPiCre ΔT_reg cells exhibited robust mTORC2-dependent aerobic glycolysis and OXPHOS.

Metabolic blockade inhibits the T_eff-like phenotype of ΔT_reg cells

The role of dysregulated glycolysis in ΔT_reg cell dysfunction was probed by cell-specific deletion of Pfkfb3, encoding 6-Phosphofructo-2-Kinase/Fructose-2,6-Biphosphatase 3. This enzyme synthesizes fructose-2,6-bisphosphate, an allosteric activator of phosphofructokinase-1 and a powerful stimulator of glycolysis ⁴¹. ΔT_reg cell-specific deletion of Pfkfb3 (Foxp3^ΔEGFPiCrePfkfb3^Δ/Δ) normalized their ECAR, while minimally impacting the OCR (Fig. 7a). Foxp3^ΔEGFPiCrePfkfb3^Δ/Δ mice showed decreased memory T_eff cells, at values intermediate between Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice (Fig. 7b). ΔT_reg cell-specific PFKFB3 deficiency suppressed IFN-γ production by ΔT_reg and T_eff cells, whereas IL-4 production was not affected. Concurrent deletion of Pfkfb3 and Rictor were not additive, indicating that effects of PFKFB3 deficiency on glycolysis were subsumed under mTORC2 deficiency (Fig. 7b). PFKFB3 deficiency marginally improved in vitro ΔT_reg cells suppression of T_eff cell proliferation (Fig. 7c), Thus, the inhibition of dysregulated glycolysis by Rictor or Pfkfb3 deletion restored ΔT_reg cells regulation of T_H1 cell responses.

Figure 7. Blockade of glycolysis improves the scurfy phenotype of Foxp3^ΔEGFPiCre mice.

(a) ECAR under glycolysis stress test conditions and OCR under mitochondrial stress test conditions of ΔT_reg cells isolated from Foxp3^ΔEGFPiCre or Foxp3^ΔEGFPiCrePfkfb3^Δ/Δ males (n=3 per group). Results are expressed as mean ± SEM and represent 1 of 2 independent experiments. (b) Frequencies of CD62L^loCD44^hi CD4⁺ T_eff cells, and IFN-γ and IL-4 expression by CD4⁺ T_reg and T_eff cells in spleens of Foxp3^EGFPCreR26^YFP, Foxp3^ΔEGFPiCreR26^YFP, Foxp3^ΔEGFPiCrePfkfb3^Δ/ΔR26^YFP, Foxp3^ΔEGFPiCreRictor^Δ/ΔR26^YFP, Foxp3^ΔEGFPiCreRictor^Δ/ΔPfkfb3^Δ/ΔR26^YFP mice. Results are expressed as scatter plot and mean and represent a pool of 3 independent experiments. (c) In vitro suppression of CD4⁺ T_eff cell (T_resp) proliferation by Foxp3^EGFPCre, Foxp3^ΔEGFPiCre or Foxp3^ΔEGFPiCrePfkfb3^Δ/Δ T_reg cells (T_supp) (n=3 per group). Results are expressed as mean ± SEM and represent 1 of 2 independent experiments. (d) Frequencies of IFN-γ⁺ and IL-4⁺ Foxp3^ΔEGFPiCreR26^YFP and Foxp3^ΔEGFPiCreRictor^Δ/ΔR26^YFP sorted T_reg cells that were either sham treated or treated ex-vivo with 2-deoxyglucose (2DG) (n=3 per group). Results are expressed as scatter plot and mean and represent 1 of 2 independent experiments. (e) In vitro suppression of CD4⁺ T_eff cell (T_resp) proliferation by Foxp3^ΔEGFPiCre ΔT_reg (T_supp) cells that were pre-treated for 12h with either PBS (Veh ΔT_reg cells) or 2DG (2DG ΔT_reg cells) (n=3 per group). Results are expressed as mean ± SEM and represent 1 of 2 independent experiments. (f, g) Representative hematoxylin and eosin-stained tissue histological sections (f) and tissue histological scores (g) of ears, livers and lungs of Foxp3^ΔEGFPiCreR26^YFP mice treated with PBS (n=) or 2 μg/g 2DG (n=) every other day from day 14 to day 28. Results are expressed as scatter plot and mean and represent a pool of 2 independent experiments. (h) Frequencies of CD62L^loCD44^hi CD4⁺ and CD8⁺ T_eff cells, IFN-γ⁺ and IL-4⁺ T_reg and CD4⁺ T_eff cells and T-Bet⁺ and Gata-3⁺ T_reg and CD4⁺ T_eff cells in Foxp3^ΔEGFPiCreR26^YFP mice treated with PBS (n=6) or 2DG (n=7). Results are expressed as scatter plot and mean and represent a pool of 2 independent experiments. Statistical significance was determined unpaired t-test (g, h), one-way ANOVA with Tukey’s multiple comparisons (b) or two-way ANOVA with Sidak’s multiple comparisons (a, d) or with Tukey’s multiple comparisons (c, e) (P values as indicated).

We further investigated the dysregulated metabolism of ΔT_reg cell using 2-deoxyglucose (2DG), a competitive inhibitor of glycolysis. While Pfkfb3 deletion attenuated the ECAR in ΔT_reg cells, 2DG more profoundly suppressed it (Fig. 6a). 2DG strongly inhibited the production by Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ ΔT_reg cells of both of IFN-γ and IL-4 (Fig. 7d), and at high concentrations partially restored Foxp3^ΔEGFPiCre ΔT_reg cell suppression, suggesting a role for aerobic glycolysis-dependent T_H cytokine production in poor ΔT_reg cell suppression (Fig. 7e).

In vivo, treatment of Foxp3^ΔEGFiCreR26^YFP mice with 2DG resulted in decreased tissue inflammation Fig. 7f–h). 2DG reduced the frequencies of memory CD62L^loCD44^hi CD4⁺ and CD8⁺ T cells, and the production of IL-4 and IFN-γ and the expression of Gata-3 and T-bet expression by both CD4⁺ T_eff and T_reg cells. 2DG treatment of Foxp3^ΔEGFPiCreRictor^Δ/Δ mice also reduced the memory frequencies and cytokine production by CD4⁺ and CD8⁺ T cells (Supplementary Fig. 7).

mTOR inhibition upregulates human mutant FOXP3 T_reg cell suppressive function

FOXP3 deficiency causes a human autoimmune lymphoproliferative disease, Immune dysregulation, Enteropathy, Polyendocrinopathy X-Linked or IPEX ⁴². Similar to ΔT_reg cells, IPEX T_reg cells are present in the periphery but are deficient in suppressive functions ^{43, 44}. We analyzed T_reg cells of five IPEX patients with different FOXP3 mutations, including N-terminal (C169Y), linker region (R309N) and forkhead domain (A384T and N388S) missense mutations and an inactivating mutation in the 3’ polyadenylation signal (Fig. 8a). The C169Y, R309N and N388S are novel mutations, while the A384T and the 3’ polyadenylation signal mutation have been reported ⁴⁵. All five IPEX patients exhibited detectable but lower expression of FOXP3 in their mutant T_reg cells, defined as CD4⁺CD25⁺CD127^{lo 46, 47}, compared to control subject T_reg cells (Supplementary Figure 8). There was a similar increase in pS6 phosphorylation in control and IPEX T_reg cells in response to cell activation, which was abrogated by treatment with the dual mTOR inhibitor Ku-0063794 (Fig 8b). By contrast, increased p_S473AKT phosphorylation was exclusively observed in stimulated IPEX T_reg cells, which was also abrogated by the inhibitor (Fig 8c). Thus, mTORC2 activity was also dysregulated in IPEX T_reg cells.

Figure 8. mTOR inhibition augments the suppressive function of human FOXP3 mutant T_reg cells.

(a) Schematic representation of FOXP3 illustrating the exons, the protein domains and mapped mutations of five patients. Amino acid changes are referred to by their single letter code. The N-terminal proline rich repressor domain (Repressor), zinc finger (ZF) motif, leucine zipper domain (LZ) and the forkhead DNA-binding domain (FKH DBD) are indicated. (b,c) Mean Fluorescence Intensity (MFI) of pS6 and p_S473AKT in T_reg cells of healthy control subjects (HC) (n=3) and IPEX patients (n=3; P1, P2, P3) stimulated with anti-CD2/CD3/CD28 mAbs (α-CD2/3/28) in the absence or presence of a competitive dual mTOR inhibitor (mTOR inh). Results are expressed as scatter plot and mean and represent 1 of 2 independent experiments. (c) Representative ECAR tracings (HC1 and P1), expressed as mean ± SEM. (d) Evaluation of glycolysis and glycolytic reserve (n=5 each for HC and IPEX groups). Results are expressed as mean ± SEM and represent a pool of 5 independent experiments. (e) Representative OCR tracings (HC2 and P2). Basal respiration, ATP production and maximal respiration in DMSO (Vehicle) or mTOR inhibitor (mTOR inh) treated HC and IPEX T_reg cells (HC n=3 and IPEX: P1, P2, P4). (f) In vitro suppression of third party CD4⁺ T_eff cell (T_resp) proliferation by T_reg (T_supp) cells of a HC or an IPEX subject (P3) that were pre-treated with vehicle or mTOR inhibitor. (g) Compilation of in vitro suppression assay results for HC and IPEX subjects at the ratio of 1:1 T_reg:T_eff cells without or with T_reg cell mTOR inhibitor pretreatment. (h) IFN-γ and IL-4 secretion by vehicle or mTOR inhibitor pre-treated IPEX T_reg cells. Statistical significance was one-way ANOVA with Tukey’s multiple comparisons (h) or two-way ANOVA with Sidak’s multiple comparisons (b-d) or with Tukey’s multiple comparisons (e, f, g) (P values as indicated).

Examination of metabolic activities revealed that IPEX T_reg cells exhibited increased ECAR in response to D-glucose, indicative of heightened glycolytic function, while showing equivalent glycolytic reserves following oligomycin addition (Fig. 8d). Both glycolytic components were inhibited by treatment of T_reg cells with the dual mTOR inhibitor. IPEX T_reg cells also exhibited exaggerated OCR responses, which were also inhibited by treatment with the dual mTOR inhibitor (Fig. 8e).

While control T_reg cells effectively suppressed T_eff cell proliferation, IPEX T_reg cells did not Fig. 8f,g). Pretreatment of the latter with the dual mTOR inhibitor uniformly improved their suppressor function irrespective of the underlying mutation involved Fig. 8f,g). mTOR inhibition also upregulated the suppressive function of T_reg cells of healthy subjects Fig. 8g), similar to Foxp3-sufficient but Rictor-deficient mouse T_reg cells (Fig. 2f and Supplementary Fig. 4). IPEX T_reg cells expressed both IFN-γ and IL-4, indicative of their T_eff-like phenotype. mTOR inhibition suppressed the T_H1 but not the T_H2 response, reflecting the role of mTORC2 in driving the T_H1 response in ΔT_reg cells (Fig. 8h). These results support the targeting of mTOR pathways to restore T_reg cell function in IPEX, as well as augmenting Foxp3-sufficient T_reg cell function in immune dysregulatory diseases.

Discussion

We demonstrate that the skewing of T_reg cells towards a T_eff cell-like phenotype upon Foxp3 deficiency is critically dependent on a limited set of molecular pathways, including mTORC2 and glycolysis. Furthermore, inhibition of these pathways specifically in Foxp3-deficient T_reg cells partially restored regulatory function and attenuated disease. Targeting the same pathways may also improve the stability and function of Foxp3-sufficient T_reg cells in inflammatory and autoimmune diseases.

Our studies implicated impaired regulation by phosphatases, including PTEN and PHLPP1, in mTORC2 dysregulation in ΔT_reg cells. Pten deletion dysregulates mTORC2 and results in the loss of T_reg cell regulatory function ³⁸. PTEN and AKT phosphatase PHLPP1 were markedly decreased in ΔT_reg cells, while transcripts encoding both Rictor and the Rictor-associated protein PRR5L, which regulates mTORC2 activity in a substrate-specific manner ³⁰, were increased. Collectively, these abnormalities may favor heightened mTORC2 activity.

The mechanisms by which mTORC2 dysregulation impair ΔT_reg cell regulatory functions included disruption of Blimp1 and Foxo1-dependent induction of IL-10 ^{31, 48}. The salutary effects of Rictor deficiency on ΔT_reg cells were were reversed by Foxo1 deficiency. Reciprocally expression of a Foxo1 gain-of-function mutant transgene recapitulated the effects of Rictor deficiency, rendering ΔT_reg cells less T_H1 cell-like, consistent with the role of Foxo1 in suppressing T_reg cells T_H1 reprogramming ³³.

The relationship of the metabolic resetting induced by Rictor deletion with enhanced ΔT_reg cell function was further established using pharmacological and genetic approaches, including inhibition of aerobic glycolysis with 2-DG and ΔT_reg cell-specific deletion of Pfkfb3. Rictor deficiency did not attenuate the T_H2 program in ΔT_reg cells and in fact exacerbated it, an effect not observed upon glycolytic pathway inhibition with 2DG treatment or Pfkfb3 deletion. These results suggest that the T_H2 cell program in ΔT_reg cells is regulated by mTORC2 by distinct mechanisms(s)

Unlike ΔT_reg cells of Foxp3^ΔEGFPiCre hemizygous male mice, those of Foxp3^ΔEGFPiCre/+ heterozygous females manifested residual suppressive activity and did not express T_H cell cytokines, indicating that the intense inflammatory environment of mutant male mice further contributed to the T_eff cell-like phenotype of ΔT_reg cells. These findings suggested that suppression of systemic immune activation may act in an adjunct manner with mTORC2 inhibition to further antagonize the T_eff cell-like phenotype of ΔT_reg cells and to promote their regulatory function.

IPEX-causing FOXP3 mutations may completely inactivate FOXP3 or more selectively impair its functions ⁴⁵. Studies on mice harboring IPEX-causing FOXP3 mutations, including I363V, A384T and R397W, have identified divergent mechanisms by which these mutations impair Foxp3 function ^{49, 50}. Common to these mutations is the acquisition by the T_reg cells of T_eff cell-like attributes, with the impaired regulatory function ⁴⁹. IPEX T_reg cells also responded favorably to mTOR inhibitors, with decreased cytokine expression and improved suppression, indicating that targeting mTORC2 is a viable therapeutic strategy in this disorder. More broadly, our studies open up the possibility of combinatorial interventions that target distinct metabolic pathways in T_reg cells to restore immune tolerance in a variety of immune dysregulatory diseases.

Methods

Human subjects

Male subjects with IPEX-causing FOXP3 mutations and healthy controls were recruited under protocols approved by the local Institutional Review Boards at the respective referring centers. Patient P1, P2, P3 and P4 each harbored a missense mutation at c.505G>A (p.C169Y), c.926G>A (p.R309N), c.1150G>A (p.A384T), c.1163A>G (p.N388S) respectively (reference sequence NM_014009.3). Patient P5 suffered from a previously described inactivating mutation in the FOXP3 3’ polyadenylation signal ⁵¹.

Generation of Foxp3^ΔEGFPiCre mice

Foxp3 genomic DNA was isolated from a bacterial artificial chromosome clone (Genome Systems) and subcloned into the plasmid vector pKO (Lexicon). A DNA cassette encoding for an improved Cre recombinase (iCre) was isolated from the paavCAG-iCre plasmid (Addgene) and inserted at the BSGRI site of a pIRES2-EGFP (Addgene) vector containing a DNA cassette composed of an internal ribosomal entry sequence (IRES) linked to downstream EGFP and SV40 poly-A sequences. The IRES-EGFP-iCre sequence was inserted by blunt-end ligation at the SSPI restriction site immediately downstream of the Foxp3 translational stop codon and upstream of the endogenous polyadenylation signal. A PGK-neo cassette was also inserted at the EcoRI site in intron 9 of Foxp3 in the opposite orientation of Foxp3 and was flanked by two FRT sites to allow excision by FLP-mediated recombination. The targeting construct also included a diphtheria toxin gene (DT) for negative selection against randomly inserted targeting constructs.

Targeting plasmids were introduced by electroporation into SCC10 embryonic stem cells and subjected to G418 selection. Resistant clones were screened by Southern blot. Successfully targeted clones were injected into C57BL/6 blastocysts, and chimeric males were mated with Wild-type (WT) C57/BL6/J females to derive N1 females that have transmitted the bicistronic allele together with the PGK-neo cassette insert (Foxp3^EGFPiCre-neo) in the germline. The PGK-neo cassette was removed by mating founder males with Flp-deleter female mice that harbor a constitutive, ubiquitously expressed FLP recombinase allele ⁵². Male offspring hemizygous for Foxp3^ΔEGFPiCre allele (minus the PGK-neo cassette) presented Scurfy phenotype. The PCR products of Exon9/Exon10 obtained using complementary and genomic DNA (cDNA and gDNA, respectively) were amplified by the following primer sequences: Foxp3 Exon9 Forward 5’-CTTCCACAACATGGACTACTTCAA-3’ and Foxp3 Exon10 Reverse 5’-AAGTAGGCGAACATGCGAGT-3’. The PCR product obtained from gDNA was purified and sequenced by Sanger sequencing using Foxp3 Exon9 Forward primer.

Mice

Foxp3^EGFPCre, Foxp3^YFPCre, CD4^Cre, Rptor^fl/fl, Foxo1^fl/fl, R26^YFP and Rag₁^–/– mice were obtained from the Jackson Laboratory. Rictor^fl/fl mice were obtained from the Mutant Mouse Regional Resource Center. R26^Foxo1AAA and Pfkfb3^fl/fl mice were generated as described ^{33, 41}. Foxp3^K276X, Foxp3^EGFP, CD45.1 Foxp3^EGFP, Foxp3^ΔEGFPiCre and their respective crosses were backcrossed 8–10 generations on C57/BL6/J, excepted Foxp3^ΔEGFPiCreFoxo1^Δ/Δ and Foxp3^ΔEGFPiCreRictor^Δ/ΔFoxo1^Δ/Δ strain. A list of all mouse strains used is reported in Supplementary Table 1. Excepted when it was specified, 25–28 days old mice were used in this study. The mice were housed under specific pathogen-free conditions and used according to the guidelines of the institutional Animal Research Committees at the Boston Children’s Hospital.

Real-time PCR

Total RNA was isolated from sorted cells with RNeasy kit (Qiagen). Reverse transcription was performed with the SuperScript II RT-PCR system (Invitrogen) and quantitative real-time reverse transcription (RT)-PCR with Taqman^® Fast Universal PCR master mix, internal house-keeping gene mouse (Actin VIC-MGB dye) and specific target gene primers (FAM Dye) (Applied Biosystems) on Step-One-Plus machine. Relative expression was normalized to Actin for genes encoding for the enzymes of the glycolysis cascade (Scl2a3, Hk1, Hk2, Gpi, Pfkfb3, Aldoa, Gapdh, Tpi, Pdk1, Pgam1, Pgam5, Eno1, Pkm2, Ldha, Scl16a1), the pentose phosphate cycle (G6pdx, Pgd, Rpia, Rpe), Foxp3, Rictor, Rptor, HIF1a and Myc and calculated as fold change compared to WT CD4⁺GFP^– T_eff cells or WT CD4⁺GFP^– T_reg cells isolated from Foxp3^EGFP mice.

Flow cytometry

Viability dye and antibodies against mouse CD4, CD8, CD16/CD32, CD90.2, CTLA4, ICOS, CD73, Blimp-1 (biolegend), CD25, CD44, CD45.1, CD45.2, CD62L, Foxp3, T-Bet, Gata-3, Helios, IFN-γ, IL-4, IL-17A, IL-2, IL-10, OX40, Nrp1, GITR, pS6, p4EBP (eBioscience), ROR-γt, p_T308AKT (BD biosciences), p_S473AKT (Cell signaling) were used. Cell suspensions were incubated for 10 min with CD16/CD32 then stained for surface markers and viability dye for 20 min on PBS/0.5%FCS. Foxp3, Helios, T-Bet, Gata-3, ROR-γt and CTLA4 staining was performed overnight using the BD Cytofix/Cytoperm™ kit. For cytokine detection, cell suspensions were pre-incubated with 50 ng/mL PMA, 500 ng/mL ionomycin and 10 μg/mL brefeldin A for 4h in complete medium. Following CD16/32 blocking with specific mAbs, the cells were surface stained for the indicated markers then permeabilized and stained intracellularly overnight with mAbs against IL-2, IL-4, IL-10 IFN-γ, or IL-17 using the BD Cytofix/Cytoperm™ kit. For p_S473AKT, p_T308AKT, pS6 and p4EBP staining, spleen cells were stimulated for 30 min with soluble anti-CD3 mAb (1μg/mL), then fixed with PBS/2% paraformaldehyde for 20min, permeabilized in 90% methanol for 30 min on ice and stained for CD4, p_S473AKT and p_T308AKT or CD4, pS6 and p4EBP in PBS. For ex-vivo T_reg cell stimulation, isolated T_reg cells were cultured for 2 days with plate-bound anti-CD3 (1μg/mL) and IL-2 (100U/mL) in presence or absence of 2DG (Sigma; 2mg/mL) and PMA/ionomycin/BrefeldinA the last 4 hours before staining for IFN-γ and IL-4. All flow cytometry acquisitions were performed on a BD Fortessa cytometer using DIVA software (BD Biosystems) and analyzed using FlowJo Version 10 (Tree Star). All mouse and human antibodies used are listed in Supplementary Table 2 and 3, respectively.

Suppression assays

For mouse studies, CD4⁺YFP^– T_eff cells from Foxp3^EGFPcreR26^YFP mice were isolated by cell sorting on FACSAria (Becton Dickinson), labeled with CellTrace™ Violet Cell Proliferation dye (Life Technologies) according to the manufacturer’s instructions and used as responder cells. T_reg (CD4⁺YFP⁺) cells were similarly isolated by cell sorting and used as suppressor cells. Responder cells were cultured in triplicates at a fixed number of 10⁵ cells/well in 96-well round-bottom plates and stimulated for 3 days with 1 μg/mL of soluble anti-CD3 mAb in the presence of 4 × 10⁵ feeder spleen cells from Rag1^−/− mice. In some experiments, T_eff and T_reg cells were pretreated for 1h with Rapamycin (Sigma; 1μM) or vehicle (DMSO), or 2DG (40mg/mL) or vehicle (PBS) overnight before being extensively washed and used as suppressor cells. For human studies, CD4⁺CD127^hiCD25^lo T cells from control subjects were by cell sorter, labeled with CellTrace™ Violet Cell Proliferation dye (Life technologies) according to the manufacturer’s instructions and used as responder cells. T_reg (CD4⁺CD127^loCD25^hi) cells were isolated on FACSAria from control or IPEX subjects and used as suppressor cells. Responder cells were cultured in triplicates at a fixed number of 5 × 10⁴ cells per well and stimulated for 3 days with T cell activation and expansion beads (Miltenyi) in 96-well, round-bottom plates.

Adoptive transfer induced Colitis

Naïve (CD45.1 CD4⁺CD45RB^highGFP^–) and T_reg (CD4⁺YFP⁺) cells are respectively isolated from the spleen of CD45.1 Foxp3^EGFP and CD45.2 Foxp3^YFPCre or Foxp3^YFPCreRictor^Δ/Δ mice. Colitis was induced in Rag₁^–/– males by i.p. injection of 5.10⁵ CD45.1 naïve ± 5.10⁴ T_reg cells. Mice were weighed and monitored for signs of disease twice weekly. Large intestines were dissected from the mice and the fecal contents were flushed out using PBS containing 2% FCS. The intestines were cut into 1cm pieces and treated with PBS containing 2% FCS, 1.5 mM dithiothreitol, and 10mM EDTA at 37 °C for 30 min with constant stirring to remove mucous and epithelial cells. The tissues were then minced and the cells were dissociated in RPMI containing collagenase (2 mg/mL collagenase II; Worthington), DNase I (100 μg/mL; Sigma), 5mM MgCl₂, 5mM CaCl₂, 5mM HEPES, and 10% FBS with constant stirring at 37 °C for 45 min. Leukocytes were collected at the interface of a 40%/70% Percoll gradient (GE Healthcare). The cells were washed with PBS containing 2% FCS and used for experiments.

Immunoblotting

Cell extracts were prepared by using RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100) supplemented with a complete protease inhibitor cocktail (Roche), a Phos STOP phosphatase inhibitor cocktail (Roche). The lysates were mixed with 4x loading buffer (Biorad) and denatured by heating for 5 minutes in 100°C. Samples were subjected to SDS-PAGE. The resolved proteins were then electrically transferred to a PVDF membrane (Millipore). Immunoblotting was probed with indicated antibodies followed by anti-rabbit IgG, HRP-linked antibody. The protein bands were visualized by using a SuperSignal West Pico chemiluminescence ECL kit (Pierce). Signal intensities of immunoblot bands were quantified by Image J software. For p_Y319-ZAP70 and ZAP70 immunoblotting, cell-sorted T_reg cells were stimulated with plate-bound anti-CD3 (145–2C11, 1 μg/mL) for 0, 2, 5 or 10 min prior protein extraction. A list of all antibodies used for immunoblotting is reported in Supplementary Table 4.

Metabolic profiling

Mouse splenocyte suspensions were enriched for CD4⁺ T cells by positive selection with magnetic beads (Miltenyi). T_reg cells were cell sorted based on GFP expression and stimulated with plate-bound anti-CD3 (145–2C11, 1 μg/mL) and 100U recombinant mouse IL-2 (Peprotech) overnight. Sorted human T_reg (CD4⁺CD127^loCD25^hi) cells were isolated from blood after PBMCs enrichment by Ficoll-Paque (Sigma-Aldrich) density gradient centrifugation and treated with the competitive dual mTOR inhibitor KU 0063794 (Tocris) at 8 μg/mL or with vehicle (DMSO) in presence of T cell activation and expansion beads (Miltenyi) and 100U/mL of recombinant human IL-2 (Peprotech). Cells were washed and seeded in Seahorse 8 wells plate at 10⁵ cells per well. ECAR and OCR were measured for both mouse and human T_reg cells using an XFp Extracellular Flux Analyzer respectively under glycolysis, mitochondrial stress and mitochondrial fuel test conditions (Seahorse Bioscience-Agilent). For glycolysis stress test, assay buffer was made of non-buffered DMEM medium supplemented with 2 mM glutamine andD-glucose, Oligomycin and 2-DG were sequentially injected at a final concentration of 10mM, 1μM and 50mM, respectively. For Mitochondrial stress test, assay buffer was made of non-buffered DMEM medium supplemented with 2.5 mM D-glucose, 2 mM glutamine and 1 mM sodium pyruvate and Oligomycin, Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) and Rotenone/Antimycin A were sequentially injected at a final concentration of 1μM, 1μM and 500nM, respectively. For Mitochondrial fuel test, assay buffer was made of non-buffered DMEM medium supplemented with 2.5 mM D-glucose, 2 mM glutamine and 1 mM sodium pyruvate and UK5099, Etomoxir and Bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide (BPTES) were sequentially injected at a final concentration of 2μM, 4μM and 3μM, respectively. Baseline ECAR (for glycolysis stress test) and OCR (for mitochondrial stress and mitochondrial fuel tests) values were averaged between technical replicates for these first three successive time intervals. For lactate production by ELISA, 10⁵ purified mouse naïve T_eff cells (CD4⁺CD62L^hiCD44^lo) from Foxp3^EGFP mice or T_reg cells from Foxp3^EGFP, Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice were seeded in flat bottom 96 well plate in complete RPMI medium in absence or presence of plate-bound anti-CD3 (1μg/mL) and 100U/mL of recombinant IL-2. Supernatants were collected after 48h and extracellular lactate production was measured using L-Lactate Assay Kit (Abcam). For metabolomics profiling, 2.5×10⁶ purified T_reg cells from Foxp3^EGFP, Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice were seeded in flat bottom 48 well plate in complete RPMI medium in presence of plate-bound anti-CD3 (1μg/mL) and 100U/mL of recombinant IL-2. Supernatants and cell pellets were collected after 48h and analyzed for metabolomics studies (Metabolom ®, Morrisville, North Carolina, USA).

Transcriptome profiling

Splenic T_reg (CD4⁺RFP+ or CD4⁺YFP⁺) cells were double-sorted from 4 weeks old female Foxp3^{ΔEGFPiCre/RFP}R26^YFP and Foxp3^ΔEGFPiCre/+Rictor^Δ/ΔR26^YFP mice. (n=4 per group). Cells were collected directly into TRIzol (Invitrogen). Total RNA was extracted and converted into double-stranded DNA (dsDNA), using SMART-Seq v4 Ultra Low Input RNA kit (Clontech). dsDNA was then fragmented to 200- to 300-bp-sized fragments, using M220 Focused-ultrasonicator (Covaris), and these were used for the construction of libraries for Illumina sequencing, using the KAPA Hyper Prep Kit (Kapa Biosystems). Libraries were then quantified using Qubit dsDNA HS (high-sensitivity) Assay Kit on Agilent High Sensitivity DNA Bioanalyzer. RNA sequencing data was demultiplexed by using perfect matches to indices and was quality-inspected using FastQC. The sequencing data was aligned to the mm10 build (Gencode annotation) of the mouse genome using STAR62, and counts were quantified using HTSeq63. Raw counts were filtered for non-mitochondrial protein-coding genes with at least three counts in one sample, and were normalized using the DESeq2 package in R64. Pairwise comparisons of differential gene expression were computed using DESeq2.

Histology

Large intestine, lung, Liver and ear sections were stained with hematoxylin and eosin. Histopathological scoring of tissue was done by a blinded observer, and the final scores reflected averages of scores from 5 different ×200 fields per tissue per mouse. Large intestinal sections were scored as follows ⁵³: 0, no inflammation; 1, mild, scattered infiltrates; 2, moderate infiltrates without loss of epithelium integrity; 3, moderate and diffuse or severe inflammation; 4, Severe inflammation associated with loss of the epithelial barrier integrity. Lung inflammation was scored separately for cellular infiltration around blood vessels and airways, as follows: 0, no infiltrates; 1, few inflammatory cells; 2, a ring of inflammatory cells 1 cell layer deep; 3, a ring of inflammatory cells 2–4 cells deep; 4, a ring of inflammatory cells >4 cells deep. A composite score was determined by adding the inflammatory scores for both vessels and airways. Liver inflammation was scored at portal areas, as follows: 0, no inflammatory cells; 1, mild, scattered infiltrates; 2, moderate infiltrates occupying less than 50% of the portal areas; 3, extensive infiltrates in the portal areas; 4, severe, with infiltrates completely packing the portal area and spilling over into the parenchyma. Ear inflammation was scored as followed: 0, no inflammation, no infiltration; Mild inflammation associated with few cells infiltration; 2, moderately severe inflammation associated with mild infiltration; 3, severe inflammation associated with large infiltration of cells and mild skin dryness; 4, very severe inflammation associated with skin dryness and cartilage erosion.

Methylation analysis

The methylation status of Foxp3 T_reg-specific demethylation region (TSDR or CNS2) in splenic T_reg cells of 25 d old male mice was assessed by bisulfite sequence analysis, as described ⁵⁴. The TSDR of converted DNA was amplified by methylation-specific primer sequences: Foxp3 CNS2 Forward 5’-TATTTTTTTGGGTTTTGGGATATTA-3’ (forward) and Foxp3 CNS2 Reverse 5’-AACCAACCAACTTCCTACACTATCTAT-3’. The PCR product was subcloned and sequenced ⁵⁴. Blast analyses were done by comparing the resulting sequences with converted Foxp3 gene sequences.

Confocal microscopy

Confocal microscopic analysis of Foxo1 nuclear and cytoplasmic distribution was carried out as described ²⁸. T_reg cells were purified and incubated on pre-coated coverslip (poly-L-lysin 50 μg/mL, ± anti-CD3 0.1 μg/mL) at 37°C for 30 min in RPMI/10%FCS. After fixation with PBS/4% paraformaldehyde, cells were permeabilized with PBS/0.1% saponin and blocked on PBS/4% bovine serum albumin (BSA). Cells were incubated with 1:100 diluted rabbit anti-Foxo1 (C29H4, Cell Signaling) followed by 1:500 diluted Alexa fluor 555-anti rabbit secondary antibody (Life technologies) in PBS/1%BSA. Slides were mounted with gold anti-fade reagent with DAPI (Invitrogen). Images were acquired with a Zeiss LSM700 confocal microscopy and ZEN imaging software. Five to 10 fields were selected randomly and total cells in the field were analyzed for percentage of Foxo1 nuclear localization using ImageJ software. Percentage of nuclear Foxo1 localization was obtained by the formula: 100 X corrected nuclear fluorescence/corrected total cell fluorescence and corrected fluorescence was obtained by the formula: Integrated Density – (Area of selected cell or nucleus X Mean fluorescence of background).

Chromatin Immunoprecipitation

ΔT_reg cells were isolated from spleen of Foxp3^ΔEGFPiCre and Foxp3^ΔEGFPiCreRictor^Δ/Δ mice. Cells were first cross-linked with 10% paraformaldehyde (PFA) for 8 minutes at room temperature (RT). Chromatin was centrifuged and the pellet was resuspended in lysis buffer I for 20 minutes at RT. The chromatin was pelleted again at 8000 rpm for 5 min at 4°C. The chromatin was then resuspended in 75 μl lysis buffer II containing 4% SDS. The chromatin was incubated for 5 min at RT to ensure the disruption of the nucleus and the release of the chromatin. The supernatant was then diluted to 1% SDS content using lysis buffer II without any SDS. Afterwards, the chromatin was sonicated using the bioruptor (Diagenode, Denville, NJ, USA) toward a proper size of 200–400 bp per DNA fragment for further immunoprecipitation. For the chromatin immunoprecipitation (ChIP) 3 antibodies were used to pull down the DNA fragments. Blimp1 (clone 3H2-E8, Invitrogen, USA), Foxo1 (clone 3B6, Invitrogen, USA) and mock IgG1 isotype control (Invitrogen, USA). Quantitative RT-PCR with the precipitated chromatin was performed to calculate the percentage of input and a fold change Blimp1/Isotype control or Foxo1/Isotype control was performed for each sample. All amplifications were performed in triplicate with SYBR Green PCR Master Mix (Qiagen) and specific primers for the different locations (5’-CTTGAGGAAAAGCCAGCATC-3’ Forward and 5’-TTTGCGTGTTCACCTGTGTT-3’ Reverse primers for il10 CNS–9, 5’-AGGCGAGATCTGAAGTGCAT-3’ Forward and 5’-CCTGCCGGTTGTAATCAACT-3’ Reverse primers for Tbx21 Promoter and 5’-TACTTCCTGCTCAGACCTGC-3’ Forward and 5’-TTCCCATCTCCTTCCTGTGG-3’ Reverse primers for Ifng Promoter). Control ChIP was performed with a respective isotype control antibody to ensure specificity. After normalization of the data according to the isotype control, the specific pulldown (percentage of input chromatin) was calculated.

Statistical analysis

Data were analyzed by paired and unpaired two-tailed Student’s t-test, one- and two-way ANOVA with post-test analyses and log-rank test, as indicated. Differences in mean values were considered significant at a P < 0.05.

Data Availability

Data presented in the manuscript, including de-identified patient results, will be made available to investigators following request to the corresponding author. Any data and materials to be shared will be released via a Material Transfer Agreement. RNAseq datasets have been deposited in the GEO with the accession code GSE129472.