Protein phosphatase 2A is requisite for the function of regulatory T cells

Abstract

Immune homeostasis depends on the proper function of regulatory T (T_reg) cells. Compromised T_reg cell suppressive activity leads to autoimmune disease, graft rejection and promotes anti-tumor immunity. Here we report the previously unrecognized requirement of the serine/threonine phosphatase Protein Phosphatase 2A (PP2A) for the function of T_reg cells. T_reg cells exhibited high PP2A activity and T_reg cell-specific ablation of the PP2A complex resulted in a severe, multi-organ, lymphoproliferative autoimmune disorder. Mass spectrometric analysis revealed that PP2A associates with components of the mTOR pathway and suppresses mTORC1 activity. In the absence of PP2A, T_reg cells altered their metabolic and cytokine profile and were unable to suppress effector immune responses. Therefore, PP2A is requisite for the function of T_reg cells and the prevention of autoimmunity.

Introduction

Immunological tolerance is achieved through the elimination of self-antigen specific T cell clones generated in the thymus and through the active suppression of autoreactive T cell thymic escapees in the periphery by regulatory T cells (T_reg cells)¹. T_reg cells express the signature transcription factor Foxp3 and have a distinct metabolic, proliferation and cytokine profile^2,3. These characteristics are inherent in their ability to suppress allowing them to maintain immune homeostasis and loss of T_reg cell function leads invariably to autoimmunity in mice⁴ and humans⁵.

Protein phosphatase 2A (PP2A) is a highly conserved serine/threonine phosphatase that is the assembly product of three distinct subunits - termed scaffold A, regulatory B and catalytic C - into a trimolecular complex^6,7. The heterodimer of the scaffold A and the catalytic C subunit (PP2A_A/PP2A_C) forms the PP2A core enzyme that associates with one of the regulatory B subunits. The PP2A holocomplex regulates key cellular processes, such as cell cycle progression, apoptosis, cellular metabolism and migration⁷. PP2A is involved in the development of cancer⁸, neurodegenerative diseases⁹ and systemic lupus erythematosus (SLE)¹⁰. In SLE, PP2A has been implicated in the regulation of the production of interleukin 2 (IL-2) and IL-17 by CD4⁺ T cells and in the control of T cell apoptosis induced upon IL-2 deprivation^10,11. Furthermore, PP2A plays a central role in MyD88-dependent endotoxin tolerance¹², T cell-mediated anti-tumor responses¹³ and in the termination of IRF3-dependent type I interferon signaling after viral infection¹⁴.

T_reg cells depend on several activating signals including the T cell antigen receptor (TCR), CD28 and IL-2 signaling pathways for their survival and function. Specifically, T_reg cells are agonist-selected by high-affinity TCR ligands in the thymus¹⁵ and continuous TCR engagement is required for their maintenance in the periphery¹⁶. Loss of CD28 (ref. 17) or the IL-2–IL-2 receptor^18,19 signaling results in profound T_reg cell impairment and autoimmunity. Paradoxically, while T_reg cell function needs the constant presence of these activating signals, T_reg cells display diminished activity of several key downstream signaling pathways including the mechanistic target of rapamycin (mTOR)^3,20 and the phosphatidylinositol-3-OH kinase (PI(3)K)-AKT^21,22 pathway compared to other antigen-experienced T cells. Therefore, T_reg cells utilize additional negative regulators compared to conventional T (T_conv) cells to rewire these downstream signaling relays. Previous reports have established that negative regulation of the PI(3)K-AKT pathway by the Nrp1-SEMA4a axis²³ and of the mTORC2 pathway by PTEN²² in T_reg cells is indispensable for the maintenance of their suppressive function. However, very little is known about how T_reg cells control the mTORC1 complex in a cell-intrinsic manner and whether this regulation is integral for their function.

In this report, we demonstrate that the serine-threonine phosphatase PP2A controls the activity of the mTORC1 complex in T_reg cells allowing them to maintain a metabolic and cytokine profile that is essential for their suppressive function. T_reg cell-specific loss of PP2A causes a severe lymphoproliferative and autoimmune disorder with spontaneous immune system activation and autoantibody production.

Results

Ablation of PP2A in T_reg cells leads to autoimmunity

The PP2A holoenzyme structurally consists of three different proteins: the catalytic C subunit (PP2A_C), the scaffold A subunit (PP2A_A) and the regulatory B subunit (PP2A_B)^6,7. When we compared the catalytic activity of the PP2A complex in T_reg and T_conv cells, T_reg cells displayed increased PP2A activity (Supplementary Fig. 1a). The nascent catalytic PP2A_C subunit -encoded by two different isoforms Cα and Cβ̃ is produced in an inactive state and undergoes an activation process that is coupled to its incorporation with the scaffold PP2A_A subunit into the heterodimeric PP2A_A-PP2A_C core^24–26. The absence of PP2A_A prevents the maturation of the catalytic subunit into its active state and the PP2A catalytic activity is impaired²⁴. The scaffold PP2A_A subunit is also encoded by two isoforms, Aα and Aβ with gene names Ppp2r1a and Ppp2r1b respectively, with the former being the dominant in primary and secondary lymphoid organs²⁷ as well as in isolated CD4⁺ T cells (Supplementary Fig. 1b). Accordingly, to study the role of PP2A in T_reg cell function, we deleted the dominant α isoform (Ppp2r1a) of the scaffold PP2A_A subunit in a T_reg cell-specific manner by crossing Foxp3^YFP-cre mice with Ppp2r1a^flox/flox mice to generate Foxp3^YFP-crePpp2r1a^flox/flox (termed here PP2A^flox) and Foxp3^YFP-crePpp2r1a^flox/+ or Foxp3^YFP-crePpp2r1a^+/+ (both termed here PP2A^wt because we did not appreciate any haploinsufficiency of Ppp2r1a in the T_reg cell population).

By the age of 10–14 weeks, the PP2A^flox mice developed spontaneously severe, progressive, multi-organ autoimmunity characterized by wasting, dermatitis, scaly tails and ears, eyelid crusting and in some occasions overt skin rash and ulcerations (Fig. 1a and Supplementary Fig. 1c–f). The clinical picture displayed similarities to the scurfy (sf) phenotype, where the sf mice harbor a spontaneous mutation mapped to the Foxp3 gene²⁸ and develop early-onset, multi-organ autoimmunity. Upon macroscopic examination of the organs, it was evident that the PP2A^flox mice suffered a lymphoproliferative syndrome with secondary lymphoid organ enlargement (Fig. 1b and Supplementary Fig. 1g and h). Histologic examination of the PP2A^flox mice revealed extensive inflammatory infiltrates in the lungs, stomach, pancreas, salivary glands and the skin (Fig. 1c and Supplementary Fig. 1d). In the lungs, extensive lymphocytic infiltrates were appreciated, especially in the perivascular and peribronchial areas. This cellular infiltration was accompanied by parenchymal consolidation and thickening of the alveolar walls. In the skin, dermal thickening and epidermal hyperplasia with spongiosis, hyperkeratosis and parakeratosis were noted. Lymphoid and polymorphonuclear aggregates in the dermis were abundant and occasionally extended into the epidermis. Focal micro-abscess formation in the stratum corneum and ulceration were also observed. Periductal and perivascular lymphocytic infiltrates were evident in the pancreas and salivary glands and around the pancreatic islets of the PP2A^flox mice. In the stomach, inflammatory aggregates, composed of lymphocytes, polymorphonuclear leukocytes and eosinophils, were observed in the submucosa protruding into the adjacent mucosal layer.

Figure 1. PP2A^flox mice develop multi-organ autoimmunity.

(a, b) The body weight (a, n=17 for PP2A^wt and n=9 for PP2A^flox) and the spleen weight (b, n=5 per group) of 10–14 week-old PP2A^wt and PP2A^flox mice were quantified. (c) H&E staining of the lungs, the skin, the pancreas, the salivary glands and the stomach of PP2A^wt and PP2A^flox mice (scale bar represents 200 μm). Images are from one experiment representative of three independent experiments with similar results (n=3 mice per group per experiment). Mean ± s.e.m. is shown, *P<0.001 (unpaired, two-tailed t-test).

The spontaneous autoimmune phenotype of the PP2A^flox mice prompted us to examine the status of their immune system compared to their wild-type littermates. We observed increased T cell activation in both CD4⁺ and CD8⁺ T cells, as evidenced by the presence of increased percentages of CD44⁺CD62L^–CD4⁺, CD44⁺CD62L^–CD8⁺ and CD44⁺CD62L⁺CD8⁺ T cells in the spleen, peripheral and mesenteric lymph nodes (Fig. 2a). When stimulated ex vivo, CD4⁺ T cells of the PP2A^flox mice produced significantly higher amounts of IL-17 and IL-2 and the CD8⁺ T cells of the PP2A^flox mice produced increased amounts of interferon-γ (IFN-γ) and tumor necrosis factor (TNF) (Fig. 2b). EdU incorporation by CD4⁺ T cells was accentuated in PP2A^flox mice demonstrating their higher proliferative state (Supplementary Fig. 2). PP2A^flox mice demonstrated higher concentrations of all immunoglobulin classes (IgM, IgG, IgE and IgA) (Fig. 2c). Congruent with these findings was the increased percentage of T follicular helper (CD3⁺CD4⁺PD1⁺CXCR5⁺) cells and germinal center B (CD19⁺FAS⁺GL7⁺) cells in the spleens of the PP2A^flox mice (Fig. 2d). To determine whether PP2A^flox mice produced autoantibodies, we used an autoantigen array that interrogates 128 known autoantigens. We detected the presence of IgG autoantibodies for 76 autoantigens that included both tissue-restricted and lupus-associated nuclear autoantigens (Fig. 2e). These results indicated that the dominant tolerance exerted by the T_reg cells was lost in the PP2A^flox mice. Additionally, the absence of any signs of autoimmunity or spontaneous immune system activation in the female Foxp3^YFP-cre/+Ppp2r1a^flox/flox mice (data not shown), where approximately half of the T_reg cell population is PP2A sufficient, further suggested that the loss of T_reg cell-mediated suppression is responsible for the phenotype recorded in the PP2A^flox mice.

Figure 2. Spontaneous T cell and B cell activation in PP2A^flox mice.

(a, b) CD4⁺ and CD8⁺ T cells of 10–14 week-old PP2A^wt and PP2A^flox mice were stained for CD62L and CD44 (a, spleen, peripheral and mesenteric lymph nodes) and for the production of the indicated cytokines (b, spleen). Data shown are from one experiment representative of three independent experiments (n=3 mice per group) (c) The levels of serum IgM, IgG, IgA and IgE in PP2A^wt and PP2A^flox mice were quantified (n=8 for PP2A^wt and n=5 for PP2A^flox mice). (d) The percentages of T follicular helper (T_fh) cells and germinal center (GC) B cells in the spleen from PP2A^wt and PP2A^flox mice are shown. Data shown are from one experiment representative of two independent experiments (n=3 mice per group per experiment). (e) Sera from PP2A^wt and PP2A^flox mice were analyzed for the presence of autoantibodies. Each individual column of the heatmap represents an individual mouse (total 4 mice per group). Mean ± s.e.m is shown. *P<0.01 (unpaired, two-tailed t-test).

Ceramide accumulation activates PP2A in T_reg cells

Our results suggested that PP2A is necessary for T_reg cells to maintain their suppressive function. Thus, we investigated the mechanisms that regulate the activity of the PP2A complex in T_reg cells. Whereas the abundance of the catalytic PP2A_C subunit is remarkably stable in a particular cell state²⁹, its activity is controlled through post-translational modifications at its carboxy-terminal tail³⁰. Specifically, phosphorylation of the Y307 residue at the C-terminal end of the PP2A_C protein results in PP2A inactivation^30,31. When we activated CD4⁺ T_conv cells with anti-CD3 plus anti-CD28, we detected PP2A_C (Y307) phosphorylation within 24 h of activation (Fig. 3a). However, we did not detect the same behavior in T_reg cells, which retained low amounts of Y307 phosphorylation after CD3 plus CD28 stimulation (Fig. 3b and Supplementary Fig. 3a) consistent with their increased PP2A activity (Supplementary Fig. 1a). Notably, the total amount of the PP2A_C subunit in T_reg cells was similar to T_conv cells both before and after CD3 plus CD28 stimulation (Fig. 3c).

Figure 3. TCR activation induces SET-mediated phosphorylation of PP2A_C at Y307 in T_conv cells but not T_reg cells.

(a, b) Naïve CD4⁺ T cells (a) or T_conv and T_reg cells (b) were stimulated with CD3 plus CD28 antibodies for the indicated time periods. Immunoblotting for p-PP2A_C (Y307) was then performed. Data are from one experiment representative of two independent experiments with similar results (n=3 mice per experiment). (c, f) Splenocytes isolated from Foxp3^IRES-GFP mice were stimulated with CD3 plus CD28 antibodies for 24 hours or left unstimulated. Intracellular staining was then performed for PP2A_C (c) or SET (f). Data are from one experiment representative of three independent experiments with similar results (n=3 mice per experiment). (d) Naïve CD4⁺ T cells were stimulated with CD3 plus CD28 for 0, 0.5, 1, 3, 6, 18 and 24 hours. Intracellular staining was then performed for SET and p-PP2A_C (Y307). A representative histogram (left) and quantification of the results (right) is shown from one of two independent experiments with similar results (n=3 per group). (e) Intracellular staining for p-PP2A_C (Y307) of naïve CD4⁺ T cells activated with CD3 plus CD28 antibodies for 24 hours and then spin-infected with an empty or a Set-expressing mCherry lentiviral vector. The analysis was done on mCherry⁺ T cells (n=5 per group, one of two experiments with similar results is shown). Mean ± s.e.m. is shown, MFI: Mean fluorescent intensity, *P<0.01 (unpaired, two-tailed t-test).

The PP2A_C (Y307) phosphorylation is under the control of the ceramide-SET pathway. The SET protein (originally named I²PP2A for inhibitor 2 of PP2A) has been identified as an inducer of Y307 phosphorylation and thus an inhibitor of PP2A activity^32–34. In CD4⁺ T cells, we found that Set is a highly TCR-responsive gene with kinetics of induction that paralleled those of the Y307 phosphorylation of PP2A_C (p-PP2A_C) following CD3 plus CD28 stimulation of CD4⁺ T cells (Fig. 3d). Lentiviral infection of CD4⁺ T cells with a Set-expressing vector and knockdown experiments with a Set-specific shRNA promoted and decreased p-PP2A_C (Y307) respectively after anti-CD3 plus anti-CD28 stimulation (Fig. 3e and Supplementary Fig. 3b), indicating that SET is responsible for the Y307 phosphorylation of PP2A_C upon activation of CD4⁺ T cells. However, T_reg and T_conv cells exhibited similar induction of SET upon stimulation with anti-CD3 plus anti-CD28 (Fig. 3f). Thus, to investigate the cause of differential Y307 phosphorylation of PP2A_C in T_reg and T_conv cells, we turned our focus to ceramide, the upstream regulator of the ceramide-SET pathway. Specifically, the SET-mediated induction of p-PP2A_C (Y307) can be abolished by an increase in the intracellular abundance of ceramide, because ceramide interacts with SET and constraints its inhibitory action on the PP2A complex, thus representing an important endogenous PP2A activator^35–37. Indeed, treatment of T cells with sphingomyelinase (SMase), which increases endogenous ceramide content (Fig. 4a), reduced the association of the SET protein with PP2A_C (Supplementary Fig. 3c). Accordingly, SMase treatment of activated CD4⁺ T cells decreased the Y307 phosphorylation of PP2A_C (Fig. 4b and Supplementary Fig. 3d). To quantify the abundance of the ceramide species present in T_reg and T_conv CD4⁺ cells, we used high performance liquid chromatography coupled with electrospray ionization tandem mass spectrometry (ESI-MS/MS) and found that T_reg cells display higher amounts of several ceramide species compared to T_conv cells (Fig. 4c). Flow cytometry for total ceramide content corroborated these results at the single-cell level (Fig. 4d).

Figure 4. T_reg cells display higher ceramide content through Foxp3-mediated inhibition of Sgms1.

(a, b) Naïve CD4⁺ T cells were stimulated with anti-CD3 plus anti-CD28 for 24h and then treated with sphingomyelinase (SMase, 0.5 units/mL) or vehicle (50% glycerol in PBS) for 1h. The cells were then stained for ceramide (a) or p-PP2A_C (Y307) (b). A representative histogram (left) and quantification of the results (right) is shown (n=3, one of two experiments with similar results is shown). (c) T_conv and T_reg cells were stimulated for 24h and then subjected to ESI-MS/MS for the quantification of ceramide species. (d) Ceramide content of CD3⁺CD4⁺Foxp3⁻ and CD3⁺CD4⁺Foxp3⁺ was quantified using flow cytometry (n=3 mice). Data are from one of three experiments with similar results. (e) T_conv and T_reg cells were subjected to chromatin immunoprecipitation using a Foxp3-specific antibody or an IgG control antibody. Shown is binding enrichment at the mouse Sgms1 gene, normalized to the input (n=3 per group, one representative of two independent experiments with similar results is shown). (f) Naïve CD4⁺ T cells were stimulated for 24h and then spin-infected with an mCherry-expressing lentivirus that harbored or not the murine Sgms1 coding sequence. CD4⁺mCherry⁺ cells were analyzed for ceramide content using flow cytometry. A representative histogram (left) and quantification of the results (right) is shown. Data are from one experiment representative of two independent experiments with similar results. Mean ± s.e.m. is shown, MFI: Mean fluorescent intensity, *P<0.05, **P<0.01, ***P<0.001 (unpaired, two-tailed t-test).

We subsequently explored the mechanism that leads to T_reg-specific accumulation of intracellular ceramide. Sphingomyelin synthase 1 (gene name Sgms1) is an enzyme of the sphingolipid metabolism pathway that has been previously shown to be part of the T_reg cell signature³⁸ and underexpressed in T_reg cells. Sgms1 catalyzes the conversion of ceramide and phosphatidylcholine to diacylglycerol and sphingomyelin. Low amounts of Sgms1 result in accumulation of ceramide intracellularly^39,40. Chromatin immunoprecipitation (ChIP) experiments demonstrated that the Sgms1 gene is a direct FoxP3 target (Fig. 4e) and retroviral overexpression of FOXP3 in Jurkat T cells decreased the expression of SGMS1 at the mRNA and protein level (Supplementary Fig. 4a and b) in agreement with previous data⁴¹. Lentiviral infection of murine CD4⁺ T cells with an Sgms1-expressing vector reduced, accordingly, the intracellular abundance of ceramide (Fig. 4f). Therefore, T_reg cells have a cell-intrinsic mechanism to maintain increased PP2A activity. Foxp3-mediated suppression of Sgms1 results in accumulation of ceramide in T_reg cells that leads to activation of the PP2A complex.

PP2A inhibits mTORC1 in T_reg cells

To delineate the specific contribution of ceramide-mediated PP2A activation in T_reg cell function, we activated the PP2A complex in Jurkat T cells using SMase. PP2A_C immunoprecipitation was performed and bands that represented increased association with PP2A_C after SMase treatment were subjected to mass spectrometry (Supplementary Fig. 5a). The results were controlled for false discovery rate and any hits that were identified in the bands of the same height of the IgG fraction were subtracted. We identified 269 unique proteins associated with PP2A_C after ceramide activation (Supplementary Table 1). Ingenuity Pathway Analysis (IPA) revealed that the major associated pathway was the mTOR pathway (Supplementary Fig. 5b). We verified the association of PP2A_C after SMase treatment with the key component of the mTORC1 complex Raptor (Supplementary Fig. 5c) that represented one of the identified proteins. To determine the specific effect that PP2A activation has on the mTOR signaling pathway, we examined the activity of the mTORC1 complex by checking the phosphorylation of the ribosomal S6 protein (p-S6), of the mTORC2 complex by checking the S473 phosphorylation of Akt (p-Akt^S473) and of the PI(3)K by checking the T308 phosphorylation of Akt (p-Akt^T308). Ceramide activation had a unique effect on mTORC1 complex activity and a negligible effect on mTORC2 and PI(3)K (Fig. 5a). Similar results were obtained when we used isolated primary murine CD4⁺ T cells (Fig. 5b). Specifically, SMase treatment of the CD4⁺ T cells reduced significantly the phosphorylation of S6 and pharmacologic inhibition of PP2A with okadaic acid increased p-S6. Importantly, the SMase-induced decrease of p-S6 was completely abrogated when PP2A was inhibited with okadaic acid. We observed a slight decrease in p-Akt^S473 upon treatment of murine CD4⁺ T cells with SMase and no change in p-Akt^T308. Flow cytometric analysis of p-S6 directly ex vivo in PP2A^wt and PP2A^flox mice showed that PP2A^flox T_reg cells have significantly increased phosphorylation of S6 (Fig. 5c).

Figure 5. PP2A inhibits the mTORC1 pathway in T_reg cells.

(a) Jurkat T cells were treated with SMase (0.5 units/mL) or vehicle (50% glycerol in PBS) for 1 hour and then stained for p-AKT^T308, p-AKT^S473 and p-S6. Data are from one experiment representative of two independent experiments with similar results. (b) Naïve CD4⁺ T cells were stimulated with CD3 plus CD28 antibodies (2 μg/mL) for 24h and then incubated with okadaic acid or DMSO for 3 hours. During the last hour the cells were treated with SMase or vehicle and then stained for p-AKT^T308, p-AKT^S473 and p-S6 (n=3 per treatment group). Data are from one experiment representative of two independent experiments with similar results. (c) Foxp3⁺ T_reg cells were isolated from PP2A^wt and PP2A^flox mice and stained for p-S6 (n=3 mice per group). A representative histogram (left) and quantification of the results (right) is shown. Data are from one experiment representative of three independent experiments with similar results. Mean ± s.e.m., MFI: Mean fluorescent intensity, ns P>0.05, *P<0.05, **P<0.01, ***P<0.001 (unpaired, two-tailed t-test).

mTOR is centrally involved in the regulation of several key cellular processes, including nutrient sensing, cell proliferation and metabolism⁴². In T cells, mTOR additionally exerts control over cytokine production and T cell differentiation³. The increased activity of the mTORC1 complex in PP2A^flox T_reg cells (Fig. 5c), coupled with the loss of T_reg cell-mediated tolerance in the PP2A^flox mice, prompted us to examine the phenotypic characteristics of PP2A^flox and PP2A^wt T_reg cells. Specifically, we quantified the glycolytic and mitochondrial respiratory capacity using a Seahorse XF24 instrument. PP2A^flox T_reg cells had significantly elevated baseline and maximum glycolytic rate (Fig. 6a). Similarly, PP2A^flox T_reg cells displayed increased oxidative phosphorylation rate both at baseline and at maximum capacity (Fig. 6b). The level of augmented glycolytic and oxidative phosphorylation rate of the PP2A^flox T_reg cells was comparable or higher than the corresponding level of glycolytic and oxidative phosphorylation rate of wild-type T_conv cells before and after CD3 plus CD28 stimulation (Supplementary Fig. 6a). In vivo determination of the proliferation rate of T_reg cells was done with EdU injected i.p. in PP2A^wt and PP2A^flox mice. PP2A^flox T_reg cells displayed increased EdU incorporation indicating that they proliferate more than the PP2A^wt T_reg cells (Fig. 6c). These results were congruent with our observation of an increased percentage of T_reg cells in the spleens, peripheral and mesenteric lymph nodes of the PP2A^flox mice (Fig. 6d). In addition, ex vivo stimulation resulted in increased production of IL-2 and IL-17 in PP2A^flox T_reg cells compared to PP2A^wt T_reg cells (Fig. 6e). PP2A^flox T_reg cells exhibited similar expression of the surface markers CD25, CTLA-4, PD-1, LAP, CD73, GITR and FR-4 (Fig. 6f and Supplementary Fig. 6b) to PP2A^wt T_reg cells, but increased expression of the large neutral amino acid transporter (LAT1) CD98 (Fig. 6g) that depends on mTORC1 activity⁴³. PP2A^flox T_reg cells did not suppress to the same extent as the PP2A^wt T_reg cells the proliferation of effector T cells in an in vitro suppression assay (Fig. 6h).

Figure 6. PP2A^flox T_regs exhibit metabolic, proliferation and cytokine production abnormalities.

(a, b) The extracellular acidification rate (a, ECAR) and the oxygen consumption rate (b, OCR) of PP2A^wt and PP2A^flox T_reg cells are shown. Left: Representative stress tests (a: glycostress test, b: mitostress test). Right: quantification of the results is shown (n=4). Data are from one experiment representative of three independent experiments with similar results. (c) The proliferation rate of PP2A^wt and PP2A^flox T_reg cells was quantified in vivo after i.p. injection of EdU (n=3, one experiment is shown representative of two independent experiments). (d) The percentage of CD25⁺Foxp3-YFP⁺ T cells in the spleens, peripheral and mesenteric lymph nodes of 10–14 week-old PP2A^wt and PP2A^flox mice is shown (n=4 mice per group, one experiment is shown representative of two independent experiments). (e) T_reg cells from the spleens of 10–14 week-old PP2A^wt and PP2A^flox mice were stimulated with PMA/Ionomycin for 6h and then stained intracellularly for the indicated cytokines (n=3 mice per group). Data are from one experiment representative of three independent experiments with similar results. (f, g) PP2A^wt and PP2A^flox T_reg cells were stained for CD25, CTLA-4, PD-1 and LAP (f) or CD98 (g, representative histogram and quantification of the results is shown). Data are from one experiment representative of two independent experiments with similar results (n=3 mice per group) (h) In vitro suppression assay using effector CD45.1⁺CD4⁺ (T_eff) T cells and PP2A^wt or PP2A^flox T_reg cells at the indicated ratios (T_eff:T_reg cells). Mean ± s.e.m., MFI: Mean fluorescent intensity, *P<0.001 (unpaired, two-tailed t-test).

To determine the functional importance of the deregulation of the mTORC1 pathway in PP2A-deficient T_reg cells, we assessed the ability of the mTORC1 inhibitor rapamycin to reverse the abnormal phenotype of the PP2A^flox T_reg cells. Rapamycin treatment in vitro of PP2A^flox T_reg cells reduced the production of IL-2 and IL17 (Fig. 7a). In addition, rapamycin normalized their increased glycolytic and oxidative phosphorylation rate to the levels of the PP2A^wt T_reg cells (Fig. 7b). In vivo administration of rapamycin (3 mg/kg i.p. every other day⁴⁴) to PP2A^flox mice reduced the leukocyte infiltration in the affected peripheral tissues (Fig. 7c). Rapamycin administration also normalized the proportion and numbers of effector CD4⁺ and CD8⁺ T cells in PP2A^flox mice such that those were comparable to PP2A^wt mice (Supplementary Fig. 7a and b). T_reg cells from treated mice displayed decreased p-S6 (Fig. 7d), indicating decreased mTOR signaling. This inhibition was also evident by a reduction of the surface expression of the mTOR-dependent amino acid transporter CD98 (Fig. 7e). Importantly, T_reg cells from rapamycin-treated PP2A^flox mice suppressed effectively the proliferation of in vitro activated wild-type naïve CD4⁺ T cells (Fig. 7f). In addition, rapamycin treatment decreased the proportion of follicular helper T cells and germinal center B cells in PP2A^flox mice (Supplementary Fig. 7c and d). We conclude that PP2A-mediated inhibition of mTORC1 is necessary for the suppressive function of T_reg cells and treatment of PP2A^flox T_reg cells with rapamycin in vitro and in vivo re-establishes their suppressive function, metabolic and cytokine profile.

Figure 7. Rapamycin reverses the abnormal profile of the PP2A^flox T_reg cells.

(a, b) FACS-sorted PP2A^wt and PP2A^flox T_reg cells were stimulated with CD3 plus CD28 and then expanded for 4 days in the presence of 100 IU/mL of IL-2. During the final 24 hours, they were treated with 100nM of Rapamycin or DMSO. (a) The T_reg cells (gated on FoxP3-YFP⁺CD4⁺ T cells) were re-stimulated with PMA/Ionomycin for 6 hours and stained for IL-17 and IL-2. (b) The ECAR and OCR of PP2A^wt and PP2A^flox T_reg cells (sorted again as Foxp3-YFP⁺CD4⁺ T cells at the end of the culture) were measured (n=3 mice per group, unpaired, two-tailed t-test). Data are from one experiment representative of two independent experiments with similar results. (c) Left: H&E staining of the lungs and the salivary glands of PP2A^wt and PP2A^flox mice treated or not with rapamycin (scale bar represents 100 μm). Images are from one experiment representative of two independent experiments with similar results. Right: combined clinical score of inflammation in the liver, skin, stomach, salivary glands, lungs and pancreas of PP2A^wt and PP2A^flox mice treated or not with rapamycin (one-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test). (d, e) Ex vivo amounts of p-S6 (d) and CD98 expression (e) in T_regs (FoxP3-YFP⁺) from mice in (c). (f) In vitro suppression assay using effector CD45.1⁺CD4⁺ (T_eff) T cells and T_reg cells from PP2A^wt or PP2A^flox rapamycin-treated and vehicle-treated mice. Representative histograms showing CFSE dilution and percentage of divided cells (ratio T_reg/T_eff: 1/1). Data are from one experiment representative of two independent experiments with similar results. Mean ± s.e.m., *P<0.05, **P<0.01, ***P<0.001.

Discussion

With the exception of PTEN, the role of intracellular phosphatases in the control of T_reg cell function is poorly understood. In this communication, we report that T_reg cells require PP2A, an evolutionary conserved serine/threonine phosphatase, to suppress effector immune cell function. Mass spectrometric, pathway enrichment and intracellular signaling analysis revealed that PP2A operates though inhibition of the mTORC1 pathway. Previous studies have shown that increased activity of the mTOR pathway negatively impacts T_reg cell generation and function^20,22,45,46. Interestingly, abrogation of the mTORC1 complex in T_reg cells in Foxp3^YFP-creRaptor^flox/flox mice⁴³ also disrupts T_reg cell function, indicating that fine regulation of the activity of this pathway rather than its complete absence is the determining factor for T_reg cell operational fitness. Ceramide-mediated activation of PP2A provides T_reg cells with required phosphatase activity to control mTORC1 and establish their tolerogenic metabolic and cytokine profile.

Foxp3, the main transcriptional regulator of T_reg cells, utilizes the pre-existent transcriptional landscape of the activated CD4⁺ T cell to establish T_reg cell function⁴⁷ acting largely as a repressor⁴¹. However, our understanding of the translation of the Foxp3-dependent epigenetic regulation to intracellular signaling events is limited. Ceramides have recently been recognized as important intracellular bioactive metabolites and second messengers⁴⁸. A major effect of ceramide is the activation of the PP2A complex^35–37. SGMS1, an enzyme of the sphingolipid metabolism, regulates the ceramide amount of the cell and has consistently been recognized in microarray studies as one of the main genes repressed in T_reg cells³⁸. Here we show that Sgms1 is a direct Foxp3 target and, thus, we provide a link between Foxp3 function and PP2A-mediated signaling effects.

PP2A is implicated in the regulation of key intracellular signaling pathways, including but not limited to the AKT, MAPK and JAK-STAT pathways^7,8. The relative importance of each of these associations depends on the cellular type and the interrogated biologic response. Thus, it was crucial in our system to define the main downstream signaling events following PP2A activation. Our experiments highlight the importance of the interaction between PP2A and the mTOR pathway in T_reg cells. Given the well-established association of PP2A with AKT⁴⁹ and AKT with T_reg cells^21–23,50, we investigated independently the effect of PP2A activation on AKT phosphorylation. We found that ceramide-mediated activation of PP2A has minimal effect on the mTORC2-dependent Akt^S473 phosphorylation and no effect on the PI(3)K-mediated Akt^T308 phosphorylation. We concluded that activation of PP2A in T_reg cells in a ceramide-dependent manner targets mainly the mTORC1 signaling.

In this report, we have provided evidence that T_reg cell function depends on the presence of PP2A and in its absence mice develop profound lymphoproliferation and autoimmunity. We propose that therapeutic enhancement of PP2A activity in T_reg cells should mitigate autoimmunity and transplant rejection and inhibition of the PP2A complex in T_reg cells should compromise their function and promote the fight of the immune system against cancer.

Online Methods

Mice

C57BL/6J (strain: 000664), Foxp3-IRES-eGFP (strain: 018628) and CD45.1 (strain: 002014) mice were purchased from the Jackson Laboratory. To generate PP2A^flox, Ppp2r1a^flox/flox mice⁵¹ (strain: 017441, Jackson Laboratory, FVB background) were back-crossed for at least 7 generations into C57BL/6J mice. Subsequently they were bred with Foxp3^YFP-cre mice⁵² (strain: 016959, Jackson Laboratory) to generate Foxp3^YFP-crePpp2r1a^flox/flox mice. Both age- and sex-matched male and female mice at the age of 10–14 weeks (unless indicated otherwise in the figure legend) were used for experiments. For the in vivo treatment experiments with rapamycin, 3mg/kg of rapamycin was administrated by i.p. injection every other day as described previously⁴⁴ for 4–5 weeks. Blood was collected before the first administration and weekly afterwards to assess lymphocyte activation. All mice were bred and housed in a specific pathogen-free environment in a barrier facility in accordance to the Beth Israel Deaconess Medical Center Institutional Animal Care and Use Committee (IACUC).

Antibodies and reagents

Antibodies against PP2A_C (clone: 1D6, Millipore), phospho-PP2A_C (Y307, clone: E155, Epitomics), SET (clone: EPR12972(B), Abcam), Raptor (clone: 24C12, Cell Signaling), Foxp3 (clone: 150D, BioLegend and clone: D6O8R, Cell Signaling), Ceramide (clone MID 15B4, Sigma-Aldrich), phospho-S6 (Pacific Blue conjugate, clone: D57.2.2E, Cell Signaling), phospho-Akt (Ser473, Alexa Fluor 647 conjugate, clone: D9E, Cell Signaling), phospho-Akt (T308, Alexa Fluor 488 conjugate, clone: C31E5E, Cell Signaling) and SGMS1 (Thermo Scientific, cat# PA1-12761) were used for immunoblot, immunoprecipitation, ChIP and flow cytometry experiments. Staining for phospho-flow experiments was performed using the Beckman Coulter PerFix Expose kit (cat#: B26976). All flow cytometry antibodies for T cell and B cell sorting and cytokine production (CD3, clone: 145-2C11; CD4, clone: GK1.5; CD8, clone: 53-6.7; CD44, clone: IM7; CD62L, clone: MEL-14; CD45.1, clone: A20; CXCR5, clone: 2G8; PD1, clone: 29F.1A12; GL7, cat # 114612; FAS, clone: 15A7; CD25, clones: 3C7 and PC61; CTLA4, clone: UC10-4B9; CD73, clone: TY/11.8 LAP, clone: TW7-16B4; GITR, clone: DTA-1; FR4, clones: TH6 and 12A5; CD98, clone: RL388; IL-17, clone: TC11-18H10.1; IL-2, clone: JES6-5H4; IFN-γ, clone: XMG1.2; IL-4, clone: 11B11; TNF, clone: MP6-XT22) were from BioLegend. All antibodies have been validated by their suppliers and references can be found on their website or on the online validation databases Antibodypedia and 1DegreeBio. Antibody dilutions were 1:1000 for immunoblot, immunoprecipitation and ChIP experiments and 1:100 for flow cytometry experiments. Sphingomyelinase (SMase, used at 0.5 units/ml) from Staphylococcus aureus (cat#: S8633), Okadaic Acid (cat#: O9381, used at 10 nM) and Rapamycin (cat#: R0395) were from Sigma-Aldrich. ELISAs for murine IgM, IgG, IgE and IgA were from Affymetrix-eBiosciences. For in vivo cell proliferation experiments, the Click-iT Plus EdU Alexa Fluor 647 Flow Cytometry Assay Kit (cat#: C10634, Life Technologies) was used per the manufacturer’s instructions.

Glycolytic and mitochondrial respiration rate measurement

For metabolic experiments, a Seahorse XF24 instrument was used. FACS sorted FoxP3-eGFP⁺ or Foxp3-YFP⁺CD4⁺ T cells (Fig. 6a and b) and CD4⁺CD25⁺ T cells (isolated as described below; Supplementary Fig. 6a) were seeded at a density of 5 × 10⁵ per well. The extracellular acidification rate (ECAR) and the oxygen consumption rate (OCR) for each well were calculated while the cells were subjected to the XF Cell Mito or the XF Glycolytic stress test protocols. The XF Cell Mito and the XF Glycolytic stress test kits were purchased from Seahorse Biosciences.

T cell isolation

T_reg cell isolation for immunoblotting, mass-spectrometric and ChIP experiments was performed using the magnetic separation Dynabeads FlowComp Mouse CD4⁺CD25⁺ T_reg Cells kit (cat#: 11463D) from Life Technologies. For all other experiments, T_reg cells were sorted on a FACS-Aria II (Becton, Dickinson) as CD3⁺CD4⁺CD25⁺Foxp3-eGFP⁺ T cells. Naïve CD4⁺ T cell isolation was performed with the magnetic separation Naive CD4⁺ T Cell Isolation kit (cat#: 130-104-453) from Miltenyi Biotech. In all instances, cells were sorted to more than 95% purity.

Mass-spectrometry

Ceramide species were quantified using high performance liquid chromatography – electrospray ionization tandem mass spectrometry (ESI-MS/MS) at the Lipidomics Analytical Core Facility of the Medical University of South Carolina (MUSC), as previously described⁵³. 3 × 10⁶ T_conv and 3 × 10⁶ T_reg cells were analyzed, after lipid extraction, together with internal sphingolipid controls on a Thermo Fisher Scientific triple quadrupole mass spectrometer operating in a multiple reaction monitoring (MRM) positive ionization mode for the quantification of ceramide species.

For PP2A_C–interacting proteins, Jurkat T cells were treated with sphingomyelinase (SMase, 0.5 units/ml) or vehicle (50% glycerol in PBS) for 1 h and then lysed for protein immunoprecipitation with a PP2A_C-specific antibody (clone: 1D6, Millipore) or a mouse IgG control antibody. Immunoblot was performed under denaturating conditions with the immunoprecipitates separated on two different gels: a 4–12% Bis-Tris gel (NuPage Novex Life Technologies, SDS-PAGE, pH 7.0) with effective range of separation of 2–200 kDa (Supplementary Fig. 5a) and a 3–8% Tris-Acetate gel (NuPage Novex Life Technologies, pH 8.1) with effective range of separation of 36–400 kDa (data not shown). Bands that were differentially represented between the SMase and the vehicle treated groups were subjected to mass spectrometry for identification of PP2A_C-interacting proteins after ceramide-mediated PP2A activation. The bands with corresponding height at the IgG control lane were also excised and any identified proteins were excluded. The results were controlled for false discovery rate and the identified proteins were analyzed with Qiagen’s Ingenuity Pathway Analysis (IPA) software for the identification of pathway enrichment in the dataset.

Autoantigen microarray

Serum from 10–14 week-old PP2A^wt and PP2A^flox mice was analyzed for the presence of autoantibodies using the Autoantigen Microarray Super Panel (128 antigen panel) at the UTSW Genomics and Microarray Core facility. Autoantibodies that were statistically different between the two groups were used to create a row-normalized heatmap analysis with R software.

Code availability

The pathway analysis for the mass spectrometric data in Supplementary Fig. 5b was done with the use of QIAGEN’s Ingenuity Pathway Analysis (QIAGEN Redwood City, www.qiagen.com/ingenuity, IPA Winter Release 2014).

R software was used to analyze and generate the results of the autoantigen array, shown in Fig. 2e. “R Development Core Team (2008). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-project.org”.

PP2A_C enzymatic activity

PP2A_C enzymatic activity was assessed using the PP2A Immunoprecipitation Phosphatase Assay Kit (cat#: 17-313, Millipore) per the manufacturer’s instructions. Briefly, protein extracts were immunoprecipitated with a PP2A_C-specific antibody (clone: 1D6, Millipore). An appropriate phosphopeptide (amino acid sequence: K-R-pT-I-R-R) was added to the immunoprecipitated immune complexes as a substrate for PP2Ac and samples were incubated at 30 °C in a shaking incubator for 10 min. Supernatants (25 μl) were transferred in a 96-well plate and released phosphate was measured by adding 100 μl malachite green phosphate detection solution in a 15-min colorimetric reaction. Phosphate concentrations were calculated from a standard curve created using serial dilutions of a standard phosphate solution.

In vitro suppression assay

CD45.1⁺CD4⁺CD62L⁺CD25^–CD44^– T cells (1 × 10⁵, T_eff) were stained with CFSE (Life Technologies) and cultured for 4 days with PP2A^wt or PP2A^flox T_reg cells in serially decreasing ratios (1:1, 1:2, 1:4, 1:8, only T_eff cells) in the presence of 1 × 10⁵ irradiated splenocytes and plate-bound anti-CD3 (2 μg/ml, clone 145-2C11; BioXcell). T_eff cell proliferation was assessed with the dilution of the CFSE dye using FACS at the end of the 4-day culture.

Plasmid, retroviral and lentiviral infections

For SET-silencing experiments, the psi-mU6 mCherry-expressing vector was used harboring a SET-specific shRNA or a scrambled control (purchased from Genecopeia). For lentiviral overexpression of SGMS1 and SET, the full coding sequences of murine Sgms1 and Set were cloned from cDNA generated from murine CD4⁺ T cells using the following primers:

• Sgms1-cloning-for 5′-ATCTGGAATTCGCAAGCTGGGGGTACTGAAT-3′

• Sgms1-cloning-rev 5′-ATTCGGGATCCCATTCCCTAGTCGGCAGAGC-3′

• SET-cloning-for 5′-ATCTGGAATTCCTGTCTCCCGGTCATCTCCC-3′

• SET-cloning-rev 5′-ATTCGGGATCCAGGGAGGAAAGGACTGCAAC-3′5′-

They were then incorporated using EcoRI and BamHI restriction enzymes into the pLVX-EF1α-IRES-mCherry Lentiviral Vector (cat#: 631987, Clontech Laboratories) and viral particles were generated in HEK293FT cells using the ViraPower Lentiviral Packaging Mix (cat#: K4975-00, Life Technologies). Viral supernatants were concentrated 100-fold with Lenti-X™ Concentrator (cat#: 631231, Clontech Laboratories). Lentiviral infection of naïve CD4⁺ T cells was performed 24 h after activation with anti-CD3 plus anti-CD28 (2 μg/ml) using spin-infection during which cells were centrifuged at 900g for 90 min at 32 °C in the presence of 6 μg/ml of polybrene. Downstream flow cytometry analysis of the infected cells was performed by gating on mCherry⁺ T cells.

The MSCV-FOXP3-IRES-Thy1.1 used in the FOXP3 overexpression experiments was a gift from Anjana Rao⁵⁴ (Addgene plasmid # 17443). Viral particles were generated as explained previously, using pCG-GagPol and pVSVg helper plasmids and Jurkats T cells were infected two consecutive times as described above. 24 h after the last infection Thy1.1⁺ cells were sorted and lysed for RNA and protein extraction. For detection of transcripts encoding for FOXP3 and SGMS1 the following primers were used:

• hFOXP3-forward 5′-GAGAAGCTGAGTGCCATGC-3′

• hFOXP3-reverse 5′-AGCCCTTGTCGGATGATG-3′

• hSGMS1-forward 5′-GCGATTGGGGTGCGGAG-3′

• hSGMS1-reverse 5′-GCCCATTCAGGGATCGTACA-3′

Chromatin Immunoprecipitation

Chromatin immunoprecipitation experiments were performed with the MAGnify Chromatin Immunoprecipitation System (cat#: 49-2024, Life Technologies) following the manufacturer’s instructions. Briefly, 2 × 10⁶ cells were cross-linked with 1% formaldehyde for 10 min at 37 °C. The reaction was stopped with glycine for 5 min, and the samples were lysed and sonicated to obtain 200- to 500-bp fragments. Immunoprecipitation was performed with a Foxp3-specific antibody or an IgG control. Cross-linking was reversed, and DNA was eluted and purified using DNA purification magnetic beads. Enrichment of the murine Sgms1 genomic sequence in the samples was quantified by real-time quantitative PCR and normalized with the input samples. The primers used were Sgms1-ChIP-forward: 5′-TGCTGTGAGGTTTTAAGGGGAA and Sgms1-ChIP-reverse: 5′-CTTCATAAGCACAAGACGGGC.

T cell stimulation

T cells were stimulated in complete RPMI 1640 with plate-bound anti-CD3 (2 μg/ml, clone 145-2C11; BioXcell) and anti-CD28 (2 μg/ml, clone 37.51; BioLegend) in flat-bottom 96-well plates. For staining of intracellular cytokines, T cells were stimulated with PMA (50 ng/ml) and Ionomycin (250 ng/ml) for 6 h in the presence of GolgiPlug (BD Biosciences) and stained using the Cytofix/Cytoperm kit from BD Biosciences (cat#: 555028).

Cell lines

The human acute T cell leukemia cell line Jurkat (clone E6-1) was purchased from ATCC (cat#: ATCC TIB-152) and has been tested for mycoplasma by ATCC.

Histology

Formalin-fixed tissues were processed, stained with haematoxylin and eosin (H&E) and evaluated blindly.

The clinical score depicted in Fig. 7c represents the combined inflammation score of the following organs: liver, skin, stomach, salivary glands, lungs and pancreas. The inflammation was scored blindly. No infiltration was scored as 0, some cellular infiltration was scored as 0.5 and clear abundant infiltrates were scored as 1. All the scores were added for each organ to produce the combined clinical score.

Statistical Analysis

Student’s two-tailed t-test, two-way analysis of variance (ANOVA) followed by Bonferroni’s test and one-way ANOVA followed by Tukey’s multiple comparison test were used to calculate statistical significance among groups as indicated in the figure legends. A P value of less than 0.05 was considered statistically significant. Results were expressed as the mean ± s.e.m.