Transforming growth factor-β1 sustains the survival of Foxp3⁺ regulatory cells during late phase of oropharyngeal candidiasis infection

Abstract

As CD4⁺CD25⁺Foxp3⁺ regulatory T cells (T_regs) play crucial immunomodulatory roles during infections, one key question is how these cells are controlled during antimicrobial immune responses. Mechanisms controlling their homeostasis are central to ensure efficient protection against pathogens, as well as to control infection-associated immunopathology. Here we studied how their viability is regulated in the context of mouse oropharyngeal candidiasis (OPC) infection, and found that these cells show increased protection from apoptosis during late phase of infection and reinfection. T_regs underwent reduced cell death because they are refractory to T cell receptor restimulation-induced cell death (RICD). We confirmed their resistance to RICD, using mouse and human T_regs in vitro, and by inducing α-CD3 antibody-mediated apoptosis in vivo. The enhanced viability is dependent on increased transforming growth factor-β1 (TGF-β1) signaling that results in upregulation of cFLIP (cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein) in T_regs. Protection from cell death is abrogated in the absence of TGF-β1 signaling in T_regs during OPC infection. Taken together, our data unravel the previously unrecognized role of TGF-β1 in promoting T_reg viability, coinciding with the pronounced immunomodulatory role of these cells during later phase of OPC infection, and possibly other mucosal infections.

INTRODUCTION

CD4⁺CD25⁺Foxp3⁺ regulatory T cells (T_regs) are central for dominant tolerance to self-antigens and innocuous commensals.^{1, 2, 3} A majority of T_regs are natural or thymic T_regs, but a few also arise in vivo in the periphery from conventional naive CD4 T cells or can be generated in vitro. Reduced frequency of circulating T_regs and their defective functions are associated with autoimmunity and inflammation.⁴ In contrast, in patients with cancer, T_reg enrichment at the tumor sites is implicated in regulating tumor immune-surveillance.⁵ As the size of the T_reg population determines disease outcomes, manipulating it is an exciting strategy in human immune therapy. Therefore, basic mechanisms of T_reg viability and proliferation in different niches are being explored.^{6, 7} With respect to lymphocyte apoptosis, restimulation-induced cell death (RICD)^{8, 9} and cytokine withdrawal-induced death¹⁰ govern effector cell homeostasis in vivo.¹¹ RICD, also referred to as activation-induced cell death,^{12, 13} is responsible for the attrition and homeostasis of expanded T-cell population after antigen clearance.^{9, 14} RICD is an extrinsic apoptosis mechanism that is signaled through Fas and tumor necrosis factor-α receptors, leading to activation of caspases.¹⁵ Cytokine withdrawal-induced death is an intrinsic mitochondrial apoptosis that occurs because of cellular stressors such as withdrawal of survival cytokines,¹⁶ involving the activation of Bim and caspases.^{10, 17, 18} Interleukin-2 (IL-2) and other γ-chain cytokines control T_reg survival, without which T_regs are susceptible to Bim-dependent cytokine withdrawal-induced death.^{19, 20} It is well known that along with T_regs, RICD^{8, 9, 21} and cytokine withdrawal-induced death are also^{10, 16} required for immune homeostasis following antigen clearance during infections.^{22, 23} Although Bim and transforming growth factor-β1 (TGF-β1) have been implicated in regulating the size of T_regs,^{24, 25} their role in T_reg RICD is unknown. Two in vitro studies have examined RICD in T_regs, but the mechanistic details and the physiological relevance were not explored.^{26, 27} Although Foxp3⁺ T_regs are shown to accumulate at infection sites, little is known about their apoptosis mechanisms and homeostasis during infections.

In the context of microbial infections, T_regs perform important roles to limit inflammation, especially at late time points of infection. We and others have previously reported their immune-protective functions that vary depending on the context of infections.^{28, 29, 30, 31} Here we have systematically shown that T_regs resist RICD compared with T effector cells during oropharyngeal candidiasis (OPC) infection and chronic lymphocytic choriomeningitis virus (LCMV) infection. Increased survival is dependent on TGF-β1 signaling and upregulation of cFLIP (cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein) in T_regs. This novel mechanism that offers a survival advantage to these critical immunomodulatory T cells may be important for immune homeostasis and resolution of immunopathology after infection clearance, and possibly other inflammatory conditions.

RESULTS

T_regs undergo reduced apoptosis during later phases of oral Candida infection and reinfection in vivo

As we have previously shown that T_regs play important roles in oral candidiasis model,³⁰ we first sought examine the T_reg frequency during the course of OPC infection. We infected Foxp3^GFP reporter mice with Candida, isolated the cells from draining cervical lymph nodes (CLNs), and estimated the frequency of Foxp3^GFP+ cells using flow cytometric analyses. As expected, sham-infected mice showed ∼10% of Foxp3^GFP+ cells among CD4⁺ T cells (Figure 1a,b). Although on days 1 and 2 there was a drop in the frequency of GFP⁺ T_regs, the frequency of T_regs kept increasing even 5 days after infection in the CLNs (Figure 1a,b). We have previously shown that mice clear the infection in ∼3–4 days, and in the absence of T_regs they succumb to severe infection and immunopathology.^{30, 31} Although tumor necrosis factor-α has protective functions during acute infection,³² T_regs reduce tumor necrosis factor-α levels in CD4 T cells, thereby limiting tongue inflammation at later time points. Therefore, we hypothesized that T_regs may proliferate or survive better over effector cells at later time points, coinciding with immunopathology control. Although Foxp3+T_regs did not proliferate better than Foxp3^GFP− T effector cells, they showed increased viability, as measured by propidium iodide (PI) staining (Supplementary Figure S1A online). Fas- and Bim-dependent apoptosis of effector cells,^{23, 33} along with immunomodulation by T_regs, are known to contribute to immune response shutdown and control of immunopathology after infections.^{14, 28} Because T_regs are already known to be susceptible to Bim-dependent apoptosis induced by cytokine withdrawal,^{7, 19} we hypothesized that their increased viability is attributed to their reduced ability to undergo RICD during infection. Therefore, we induced RICD of T cells in vivo by reinfecting Foxp3^GFP mice at late time points of primary Candida infection. We assessed the viability of the cells on day 1 after reinfection. We harvested the cells from axillary lymph nodes and CLN, the draining lymph nodes, as well as spleen and inguinal lymph nodes to assess CD4⁺ cell viability. We refer to the non-T_reg (Foxp3−) cells activated by the infection in vivo as effector cells (T_effs). We gated on the control CD4⁺Foxp3^GFP− T_effs and CD4⁺ CD25⁺Foxp3^GFP+ T_regs (Supplementary Figure S1B, C), and measured the viability by PI staining. We found that the frequency of PI⁺ dead cells among CD4+Foxp3^GFP+ T_regs was 10–12%, and was significantly lower than in CD4+ Foxp3^GFP− effector cells (∼24%) in draining lymph nodes (Figure 1c and Supplementary Figure S1C). In addition, by examining the absolute cell numbers ex vivo at various time points after primary infection, we found that although the effector cells undergo an expansion followed by contraction at late time points, T_regs did not show reduction in cell counts (Supplementary Figure S1D) coinciding to increased survival at later time points. In spleen and inguinal lymph nodes, although T_regs had slightly increased viability than effector cells, the differences were smaller than in draining lymph nodes (Figure 1c). Next, we adoptively transferred fluorescence-activated cell sorting (FACS)-sorted naive CD4+CD25−GFP− cells (conventional or control CD4+ cells; T_cons) or CD4+CD25+GFP+ T_regs into Rag1^−/− mice, 4 days before primary OPC infection, and assessed their viability 1 day after reinfection. We refer to the non-T_reg (Foxp3− T_cons) cells activated by the infection in vivo as T_effs. We found that the frequency of PI⁺ cells was greater among T_effs than T_regs (Figure 1d), showing that T_regs survive better than conventional CD4 T cells during RICD at late phase of infection. To confirm the role of Fas in the contraction of CD4⁺ T cells, we infected Fas mutant lympho-proliferation (lpr) mice, comparing with the wild-type (WT) mice. Although there was an increase in the frequency of T_regs in WT mice, there was no increase in lpr mice at late time points (Supplementary Figure S2). These results demonstrate that Fas is largely contributing to contraction of effector cells, without which the apparent increase in proportion of T_regs is not observed at late time points.

Figure 1

Regulatory T cells (T_regs) show increased viability during Candida reinfection and in vivo. (a, b) CD4⁺Foxp3⁺ T_regs show increased enrichment during the course of oropharyngeal candidiasis (OPC) in mice. Foxp3^GFP reporter mice (n=5/group) were infected with Candida albicans (Candida) as described in Methods. Flow cytometric plots (a) and statistical analyses (b) showing the frequencies of cells expressing CD4 and intracellular Foxp3 (gated on CD4 cells) in cervical lymph nodes (CLNs) at various days (d) after C. albicans infection. Percentages of Foxp3-expressing cells are shown. P-value is determined by Mann–Whitney test. (c) Mice were infected and reinfected as in a. Axillary lymph nodes (ALNs), CLNs, spleen (SPLN), and inguinal lymph nodes (ILNs) were harvested on day 1 after reinfection to examine cell death of Foxp3^GFP− T_eff (left) and Foxp3^GFP+T_regs (right) ex vivo by green fluorescent protein (GFP) and propidium iodide (PI) staining. Flow cytometric contour plots (left panel) and statistical data (right panel) show the frequencies of PI⁺ cells (gated on CD4 cells). Statistical significance was determined using Mann–Whitney test. (d) Rag1^−/− CD45.1 mice (n=5/group) were reconstituted with CD4⁺Foxp3^GFP− T_eff or CD4⁺Foxp3^GFP+T_reg cells obtained from congenic CD45.2 mice. Recipient mice in each group were reinfected with Candida. ALNs and CLNs were harvested on day 1 after reinfection for examining cell death ex vivo by PI staining (gated on CD45.2⁺ cells). (e, f) Foxp3^GFP reporter mice were injected with phosphate-buffered saline (PBS) or α-CD3 antibody. Flow contour plots of yellow fluorescent protein (YFP) and PI staining histograms (e) of Foxp3^YFP− T_eff (top) and Foxp3^YFP+ T_regs (bottom) and quantification of % PI⁺ cells (f) of Foxp3^GFP− T_eff (blue circles) and Foxp3^YFP+ T_regs (red squares) 24 h after α-CD3 antibody injection in mice (n=4/group) (gated on CD4⁺ cells). Statistical significance was determined using Mann–Whitney test. Data from one of three to five independent experiments are shown.

PowerPoint slide

To further validate that T_regs have decreased susceptibility to T cell receptor (TCR)-mediated RICD in vivo, we directly examined whether T_regs resist α-CD3 antibody-mediated deletion in vivo. We injected 100 μg of α-CD3 antibody in Foxp3^YFP mice and measured the viability of T_regs and conventional CD4⁺ cells.³⁴ Correlating to the results from OPC infection experiments, we found that the percentage of PI⁺ cells was significantly lower among Foxp3^YFP+ T_regs compared with Foxp3^YFP− cells in vivo in CLN (Figure 1e), spleen, and other lymph nodes upon α-CD3 antibody injection (Figure 1e,f). To confirm whether T_regs resist apoptosis during a chronic infection, we infected the mice with LCMV clone 13, and assessed apoptosis of CD4⁺ T cells at different time points after infection. We found that at all time points, T_regs showed significantly reduced apoptosis than effector cells after infection in vivo (Supplementary Figure S3).

Mouse and human T_regs are resistant to RICD and show concomitant reduction in active caspase-3 levels

To investigate the mechanism by which T_regs resist RICD, we examined their viability during RICD apoptosis in vitro. We isolated CD4⁺ CD25⁺ Foxp3^GFP+ cells (T_reg) and conventional or control CD4⁺ CD25⁻ Foxp3^GFP− naive T cells (T_con) from Foxp3^GFP reporter mice (Supplementary Figure S4) and stimulated and restimulated them with α-CD3 antibody (1 μg ml⁻¹) and IL-2. We measured the cell viability by their forward scatter shift and PI staining 12–18 h after RICD induction using flow cytometry. Whereas 80–85% of the T_cons were PI⁺, only 30–40% of T_regs showed PI staining (Figure 2a). Then, we induced RICD with different concentrations of α-CD3 antibody. We measured apoptosis by annexin-V and PI staining (Figure 2b) and found that T_regs underwent significantly reduced apoptosis. Next, we isolated CD4⁺CD25⁺CD127^lo T_regs and CD4⁺ CD25⁻ T_cons from human peripheral blood mononuclear cells and induced RICD. Human T_regs also showed resistance to RICD compared with T_cons (Figure 2c,d). Upon induction of RICD, both mouse and human T_regs showed substantially lower levels of cleaved caspase-3 compared with T_cons (Figure 2e,f). Taken together, these results reveal that although conventional CD4 cells are highly susceptible to RICD, T_regs show reduced RICD with concomitantly lower levels of active caspase-3.

Figure 2

Mouse and human regulatory T cells (T_regs) are resistant to restimulation-induced cell death (RICD) and show decreased activated caspase levels. (a) Flow cytometric histogram plots of propidium iodide (PI) staining of mouse T_cons (upper panel) and mouse T_regs (lower panel) derived and pooled from spleen/lymph nodes (SPLN/LN), stimulated under RICD conditions for 20 h with 0 μg ml⁻¹ (left) or 1 μg ml⁻¹ (right) of α-CD3 antibody. As controls, we stimulated the cells with interleukin-2 (IL-2) only, but no α-CD3 antibody (0 μg ml⁻¹). (b) Quantification of % apoptosis of the T_cons (gray bars) or T_regs (black bars) restimulated with indicated concentrations of α-CD3 antibody for 18 h, based on evaluating the annexin-V and PI staining by flow cytometry. (c) Flow cytometric analyses of PI staining of human T_cons (upper panel) and human T_regs (lower panel) stimulated under RICD conditions for 21 h with 0 μg ml⁻¹ (left), 0.5 μg ml⁻¹ (middle), or 1 μg ml⁻¹ (right) of α-CD3 antibody. (d) % Apoptosis quantification of human cells stimulated as in c. (e) Mouse cells or (f) human cells were stimulated for 15 h as in a and c, respectively, and cleaved caspase-3 levels were determined by intracellular staining and flow cytometry. P-values are determined by Mann–Whitney test in b and d. Data in a and b represent at least 10 independent experiments and in c–f are from one of three independent experiments showing similar results.

PowerPoint slide

RICD resistance is not because of retarded proliferation in T_regs

Previous studies show that T cells producing high levels of IL-2 and proliferating at higher rates have an increased propensity to die of RICD.³⁵ T_regs produce little or no IL-2 and proliferate slower that might contribute to their reduced apoptosis. IL-2 was ruled out because we added excess of IL-2 during RICD. However, during the initial stimulation for 4 days, T_regs displayed reduced proliferation compared with T_cons as determined by CPD670 labeling (Figure 3a). We assessed cell death, gating on proliferating and nonproliferating cells, upon RICD. Nonproliferating cells (in the prior stimulation) in both the populations showed little or no apoptosis (Figure 3b,d). However, proliferating T_regs were still exquisitely resistant to RICD compared with proliferating T_cons (Figure 3c,d). Moreover, 16 h after RICD induction, the time point at which we measured apoptosis, there was no difference in proliferation between T_cons and T_regs (data not shown). Moreover, to examine whether T_regs and T_cons are activated at different levels, we tested their nuclear factor (NF)-κB activation by its nuclear translocation. Confocal microscopy revealed that nuclear translocation of the p65 subunit of NF-κB was comparable in T_regs and T_cons after RICD stimulation (Figure 3e). These results showed that RICD resistance of T_regs is not due to their reduced proliferation or activation.

Figure 3

Reduced restimulation-induced cell death (RICD) in regulatory T cells (T_regs) is not dependent on retarded proliferation or reduced activation. (a) Flow cytometric histogram plots showing CPD670 dilution of mouse T_cons (upper panel) and mouse T_regs (lower panel) derived and pooled from spleen/lymph nodes (SPLN/LN), sorted as described in Methods, and stimulated for 4 days with α- CD3, α-CD28, and interleukin-2 (IL-2) without RICD induction. CPD670 dilution shows the frequency of proliferating (prol.) and nonproliferating cells (n.prol.). (b, c) T_con or T_reg cells labeled and stimulated as in a were restimulated with (right) or without (left) 1 μg ml⁻¹ of α-CD3 antibody. Flow cytometric contour plots show CPD670 dilution and propidium iodide (PI) staining of T_con (top) or T_reg (bottom) cells gated on (b) nonproliferating cells and (c) proliferating cells. Plots show the frequency of PI-negative viable cells. (d) Statistical representation of % apoptosis of nonproliferating cells (upper panel) and proliferating cells (lower panel). P-value is determined by Mann–Whitney test. Data represent one of two independent experiments showing similar results. (e) T_cons (upper panel) or T_regs (lower panel) were restimulated with 1 μg ml⁻¹ α-CD3 or 0 μg ml⁻¹ α-CD3 under RICD conditions for 30 min. Cells were fixed, stained using α-p65 antibody (red) and 6-diamidino-2-phenylindole (DAPI) (nucleus; blue), and assessed by confocal microscopy. Data from one of three independent experiments are shown.

PowerPoint slide

T_regs and in vitro T_regs resist TCR-mediated RICD and direct Fas-mediated apoptosis

Next, we explored whether TCR-induced RICD was Fas–Fas ligand (FasL) dependent, and therefore induced RICD in the presence or absence of α-FasL antibody. α-FasL antibody abrogated TCR-mediated RICD (Figure 4a), indicating the RICD we observed was Fas–FasL apoptosis. We then analyzed the expression of Fas and FasL at 0 and 6 h after RICD induction. In vitro induced T_regs are referred to as in vitro T_regs in our experiments. Although Fas levels were comparable (Figure 4b), T_regs and in vitro T_regs showed reduced FasL expression compared with T_cons (Figure 4c). Next, we determined whether reduced apoptosis in T_regs was completely attributed to decreased FasL expression, or whether these cells also have intrinsic reduced sensitivity to Fas-mediated apoptosis. We directly crosslinked the Fas molecule using plate-bound α-Fas antibody to induce Fas-mediated apoptosis. Both T_regs and in vitro T_regs showed significantly reduced apoptosis upon direct Fas crosslinking, independent of reduced FasL expression (Figure 4d).

Figure 4

Regulatory T cells (T_regs) and in vitro T_regs show decreased Fas ligand (FasL)/FAS mediated death. (a) % Apoptosis of T_cons (white bars), T_regs (black bars), or in vitro T_regs (gray bars) derived and pooled from spleen/lymph nodes (SPLN/LN) that were stimulated under restimulation-induced cell death (RICD) conditions with or without 1 μg ml⁻¹ of α-CD3 antibody for 12 h, as in Figure 2b. Isotype control antibody or α-FasL antibody (10 μg ml⁻¹) was added 30 min before RICD induction. (b, c) Flow cytometric contour plots of Fas (b) or FasL (c) staining of T_con (top row), T_reg (middle row), or iT_reg (bottom row) cells, stimulated as in Figure 2b. Left columns show isotype control staining on T_con cells. Gates show the frequencies of Fas- or FasL-expressing cells. Data represent one of three independent experiments showing similar results. (d) % Apoptosis of T_cons (white bars), T_regs (black bars), or in vitro T_regs (gray bars) that were restimulated with 1 or 10 μg ml⁻¹ of α-Fas antibody and human interleukin-2 (IL-2) for 8 h. As controls, some cells were restimulated with 0 or 1 μg ml⁻¹ of α-CD3 antibody. P-values are determined by Mann–Whitney test.

PowerPoint slide

RICD resistance is partially dependent on soluble TGF-β1 in T_regs

Next we determined whether a soluble factor might influence RICD resistance in T_cons and T_regs. Therefore, we stimulated T_cons and T_regs, and collected the supernatants from 4-day cultures. We then induced RICD in the presence of T_con or T_reg supernatants. T_reg supernatants partially, but significantly, reduced T_con apoptosis (Figure 5a). We reasoned that IL-10,³⁶ IL-35,³⁷ or TGF-β,³⁸ the most prominent soluble factors produced by T_regs, may underlie apoptosis resistance, and sought to test this hypothesis. However, Il10^−/− T_regs were as resistant to RICD as WT T_regs (Figure 5b). Excluding the role of IL-35 in T_reg survival, the frequency of Foxp3⁺ T_regs was normal in Ebi3^−/− mice (Supplementary Figure S5). Although previous studies reported the role of TGF-β in Foxp3 expression and the development of T_regs,³⁹ its effect on T_reg RICD is undetermined. Based on the antiapoptotic functions of TGF-β in general,⁴⁰ we examined the role of TGF-β during RICD. We either added TGF-β only during induction of RICD in T_cons or preconditioned them with TGF-β both during the initial stimulation and again during RICD induction. TGF-β reduced RICD moderately but significantly when added only during RICD induction (Figure 5c). However, preconditioning with TGF-β completely abrogated RICD (Figure 5c), and also induced Foxp3 expression (in vitro T_regs) in naive T cells. To confirm the role of TGF-β in conferring resistance to RICD in T_regs, we preblocked TGF-β (1d11 antibody) in T_cons and T_regs during primary stimulation and RICD induction. Blocking TGF-β abrogated RICD resistance partially but significantly in T_regs (Supplementary Figure S6). Similar to T_regs and consistent to a previous publication, in vitro T_regs also showed decreased apoptosis (Figure 5d).²⁴ Although TGF-β did not inhibit proliferation in in vitro T_regs, these cells also showed reduced active caspase levels compared with T_cons upon RICD induction (Supplementary Figure S7A,B). Taken together, these results show that TGF-β plays a crucial role in reducing TCR-induced RICD.

Figure 5

Restimulation-induced cell death (RICD) in T_cons is partially reduced by a soluble factor secreted by regulatory T cells (T_regs) and the reduced apoptosis is dependent on transforming growth factor-β1 (TGF-β). (a) T_cons and T_regs derived and pooled from spleen/lymph nodes (SPLN/LN) were stimulated under RICD conditions for 20 h with indicated concentrations of α-CD3 antibody as in Figure 2b, in the absence (white bars) or the presence of supernatants collected from 4-day T_con cultures (T_con supe.) (hatched bars) or those collected from T_reg cultures (T_reg supe.) (black bars). % Apoptosis, as assessed by annexin-V and propidium iodide (PI) staining. (b) % Apoptosis of wild-type (WT) or Il10 ^− − T_cons (white bars) or T_regs (black bars) that were stimulated under RICD conditions for 10 h as in Figure 2b. (c) % Apoptosis of WT T_cons restimulated as in Figure 2b in the presence of T_reg supe., or increasing concentrations of human TGF-β1 added during RICD induction only. Some T_cons were cultured with 4 ng ml⁻¹ of TGF-β1 during initial stimulation for 4 days (preconditioning (PC)) and during RICD. (d) In vitro T_regs are resistant to RICD. % Apoptosis of T_cons (gray bars) or in vitro T_regs (black bars) that were restimulated with indicated concentrations of α- CD3 antibody for 17 h as in Figure 2b. P-values are determined by Mann–Whitney test.

PowerPoint slide

Absence of TGF-β signaling sensitizes T_regs to TCR-mediated RICD

Based on the antiapoptotic effects of TGF-β1 that we observed above, we hypothesized that T_regs resist RICD because they produce TGF-β1 and are exposed to self-stimulatory TGF-β1 signaling. To determine whether T_regs and in vitro T_regs exhibit increased TGF-β1 signaling, we measured the mothers against decapentaplegic homolog (SMAD) proteins.⁴¹ Both T_regs and in vitro T_regs exhibited increased levels of phosphorylated SMAD2 (pSMAD2) (Figure 6a, upper panel) and pSMAD3 (Figure 6a, lower panel), whereas only a fraction of T_cons showed very low levels of these molecules (Figure 6a). To conclusively show that TGF-β1 signaling is required for RICD resistance of T_regs and in vitro T_regs, we utilized the transgenic mice that express a dominant-negative form of the TGF-β receptor II under the direction of the mouse CD4 antigen promoter (Cd4-tgfβr2dn^tg), in which CD4 cells lack TGF-β signaling.⁴² We flow cytometrically sorted CD4⁺CD25⁻CD44^lo naive cells and CD4⁺CD25⁺T_reg cells from WT or Cd4-tgfβr2dn^tg to induce RICD. T_regs from WT and Cd4-tgfβr2dn^tg mice were respectively ∼92 and 80% Foxp3-positive before RICD induction (Supplementary Figure S8A). We found that Cd4-tgfβr2dn^tg T_cons underwent RICD similar to WT T_cons, and both WT T_regs and WT in vitro T_regs showed reduced apoptosis compared with T_cons (Figure 6b). In striking contrast, Cd4-tgfβr2dn^tg T_regs and Cd4-tgfβr2dn^tg naive cells stimulated under iT_reg conditions (that did not upregulate Foxp3) showed increased RICD comparable to T_cons (Figure 6b and Supplementary Figure S8B). These data demonstrate that T_regs show increased TGF-β signaling, and that is central to their RICD resistance.

Figure 6

Resistance to restimulation-induced cell death (RICD) is dependent on increased transforming growth factor-β1 (TGF-β1) signaling in regulatory T cells (T_regs) and in vitro T_regs. (a) T_cons, T_regs, and in vitro T_regs derived and pooled from spleen/lymph nodes (SPLN/LN) were stimulated for 4 days with α-CD3, α-CD28, and IL-2, and restimulated in the presence of TGF-β1 for 30 min. Flow cytometric contour plots show Foxp3 and pSmad2 (top) or pSMAD3 (bottom) staining of T_con (left), T_reg (middle), or iT_reg (right) cells. (b) % Apoptosis of T_cons, T_regs, or in vitro T_regs from wild-type (WT) or Cd4-tgfβr2dn^tg mice were restimulated as in Figure 2b without (white bars) or with 1 μg ml⁻¹ (black bars) of α-CD3 antibody for 23 h. Data represent one of three independent experiments showing similar results. P-value is determined by Mann–Whitney test.

PowerPoint slide

Expression profiles of apoptosis-related molecules in T_regs

Next we profiled a panel of apoptosis-related genes using quantitative PCR array analysis of messenger RNA (mRNA) isolated from T_cons and T_regs 3 h after induction of RICD. Il10 was used as a positive control for T_regs. As expected, whereas mRNA level of Il10 was upregulated, Fasl level was downregulated in T_regs (Figure 7a). Some of the Bcl-2 members such as Bcl-2 and Bcl-xl (Bcl2l1) were unchanged or moderately changed compared with T_cons. However, we found that antiapoptotic Cflip (Cflar) mRNA level was upregulated, and proapoptotic Bax and Bok were downregulated in T_regs compared with T_cons. cFLIP is a cytoplasmic protein that is capable of binding to FADD (Fas-associated death domain) and preventing the initiation of the Fas death pathway.⁴³ More importantly, TGF-β has previously been shown to upregulate cFLIP in other cell types.⁴⁰ Therefore, we validated the expression profiles of cFLIP and other apoptotic genes by assessing their protein levels. Although T_regs and in vitro T_regs upregulated Bcl-2, Bcl-xl, and Bim, their levels were slightly lower, or unchanged, compared with T_cons (Figure 7b). However, supporting our quantitative PCR array data, T and in vitro T_regs showed increased frequency of cFLIP^high cells, whereas most of the T_cons were cFLIP^low (Figure 7b). These results suggest that TGF-β1-mediated upregulation of cFLIP may also be involved in T_regs and in vitro T_regs in conferring resistance to RICD.

Figure 7

Expression of apoptosis-related genes and proteins in regulatory T cells (T_regs). (a) T_cons and T_regs derived and pooled from spleen/lymph nodes (SPLN/LN) were restimulated as in Figure 2b. Total RNA was prepared 3 h after stimulation to assess the mRNA levels by quantitative PCR (qPCR) array using RT² Profiler PCR array (indicated by y axes). These data represent one of three independent experiments. (b) Flow cytometric histograms of T_cons (blue), T_regs (red), and in vitro T_regs (green) that were stimulated as above for 12 h, fixed, and stained for intracellular Bcl-xl, Bcl-2, Bim, and cFLIP (cellular FLICE (FADD-like IL-1β-converting enzyme)-inhibitory protein).

PowerPoint slide

WT but not Cd4-tgfβr2dn^tg T_regs undergo reduced apoptosis during oral Candida reinfection in vivo

To determine whether increased viability of T_regs during the OPC infection was dependent on TGF-β, we adoptively transferred WT or Cd4-tgfβr2dn^tg CD4⁺ cells into Rag1^−/− mice and assessed their viability as in Figure 1 in draining lymph nodes. We also examined the CD4 viability in mouse oral intraepithelial and lamina propria leukocytes isolated from oral tissues, the site of infection. The proportion of VD-ef660⁺ dead cells was significantly lower among WT Foxp3⁺T_regs than in WT Foxp3⁻effector cells (Figure 8a,b). However, in mice that received Cd4-tgfβr2dn^tg CD4⁺ cells, cell death of the Foxp3⁺ T_regs was higher than WT T_regs (Figure 8a,b). We then determined the frequency of CD25⁺Foxp3⁺ T_regs 2 days after reinfection and found that Cd4-tgfβr2dn^tg recipient mice showed at least 50% reduced frequency of Foxp3⁺T_regs than WT recipients (Figure 8c,d). These results demonstrate that T_regs resist apoptosis during later phases of infection and reinfection, and require TGF-β1 signaling for their increased viability.

Figure 8

Increased viability in regulatory T cells (T_regs) during Candida reinfection is transforming growth factor-β1 (TGF-β1) dependent in vivo. (a) Rag1^−/− mice (n=11) were reconstituted with CD4⁺ cells from C57BL/6 or Cd4-tgfβr2dn^tg congenic mice. Recipient mice in each group were infected and reinfected with phosphate-buffered saline (PBS) sham controls or Candida. Axillary lymph nodes (ALNs), cervical lymph nodes (CLNs), and mouse oral intraepithelial and lamina propria leukocytes (MOIL) were harvested on day 1 after reinfection for examining cell death of Foxp3⁻ T_eff in wild-type (WT) cell recipients (left), Foxp3^GFP+T_regs in WT cell recipients (middle), and Foxp3^GFP+T_regs in Cd4-tgfβr2dn^tg cell recipients (right) ex vivo by intracellular Foxp3 and VD-ef660 staining (gated on CD4⁺ cells). (b) Quantification of the frequency of VD-ef660⁺ dead cells among Foxp3⁻ T_effs (blue circles) and Foxp3⁺ T_regs (red squares) from ALNs, CLNs, MOIL, and inguinal lymph nodes (ILNs) as in a (gated on CD4⁺ cells). Flow cytometric dot plots (c) and quantification (d) showing CD25- and Foxp3-expressing cells (T_regs) in WT cell recipients (c, left) and Cd4-tgfβr2dn^tg cell recipients (c, right) ex vivo (gated on CD4⁺ cells) 2 days after reinfection as in a. Data represent three independent experiments showing similar results. Statistical significance was determined using Mann–Whitney test.

PowerPoint slide

DISCUSSION

Our data reveal that during late points of OPC infection (Figure 1) and chronic LCMV infection (Supplementary Figure S2), T_regs are exquisitely resistant to apoptosis as compared with conventional CD4⁺ T cells. During initial phases of acute OPC infection, there was a slight drop in T_reg frequencies. This is consistent with a previous observation that demonstrated an infection-induced partial loss of T_regs because of insufficiency in IL-2.⁴⁴ However, at late time points, whereas most of the effector cells undergo attrition, T_regs survive RICD. We and others have shown that during infections, Toll-like receptor-2 signaling can promote proliferation and accumulation of T_regs in the infected tissues and the draining lymph nodes, and is required to limit immunopathology.^{31, 45} Although T_regs underwent proliferation, their expansion is less than that of the effector cells, and could not have contributed to increased viability compared with effector cells (Supplementary Figure S1A). It is conceivable that although most effector cells die after infection clearance, T_regs have adopted a survival mechanism to be retained longer in the tissues and limit inflammation and tissue damage. This is consistent to the previous data showing that memory T_regs stay in the system, and serve as more potent suppressors mediating resolution of organ-specific autoimmunity in mice.⁴⁶

Our data unequivocally show that TGF-β1 is required for RICD resistance in T_regs. Addition of exogenous T_reg supernatant or TGF-β1 partially relieved even the T_con cells from RICD (Figure 5a), demonstrating a partial role of soluble TGF-β1 in decreasing apoptosis during RICD. Moreover, Cd4-tgfβr2dn^tg T_regs showed complete susceptibility to RICD similar to T_con levels, substantiating a critical role of TGF-β1 signaling in T_reg RICD resistance (Figure 6b). We have shown for the first time that TGF-β1 confers increased viability in T_regs at the site of infection (mouse oral intraepithelial and lamina propria leukocyte tissue) during an oral infection in vivo (Figure 8a), phenotypically and mechanistically recapitulating the in vitro results. It is possible that both soluble TGF-β1 and membrane-bound TGF-β1 (latency-associated peptide) are jointly involved in apoptosis resistance in T_regs. Thus, in the absence of TGF-βRII, unable to respond to both the forms of TGF-β, T_regs showed high susceptibility to RICD. Validating the role of TGF-β further in CD4 cells, TGF-β1-induced in vitro T_regs exhibited RICD resistance as well. Because in vitro T_regs express Foxp3, it can be argued that Foxp3 imparts increased viability in T_regs.²⁷ However, our results showing high RICD susceptibility of FACS-sorted T_regs from Cd4-tgfβr2dn^tg mice despite Foxp3 expression (Supplementary Figure S8B) rules out this possibility. A potential caveat to the experiment where we adoptively transferred Cd4-tgfβr2dn^tg T_regs could be the loss of Foxp3 expression in some of the T_regs. Moreover, as we purified these cells using CD4 and CD25 only, there is likely a contamination of CD25⁺ T_eff cells. However, in our viability analyses we gated only on Foxp3⁺ T_regs (Figure 8a). Therefore, although the starting populations may have been different in terms of Foxp3 expression, even gating exclusively on Foxp3+ T_regs, Cd4-tgfβr2dn^tg T_regs had poor viability. These results show that despite Foxp3 expression in Cd4-tgfβr2dn^tg cells, they have poor survival, thus substantiating the role of TGF-β in maintaining T_reg survival. Moreover, the overall viable CD4 count of the cells recovered from CLN showed that Cd4-tgfβr2dn^tg CD4 cell recipient mice had lower CD4 cell numbers than the WT CD4 cell recipients on day 2 after reinfection (Supplementary Figure S9). These results show that Cd4-tgfβr2dn^tg CD4+ cells did not expand better than WT CD4 T cells at this time point. Therefore, decrease in Foxp3⁺ cells in Figure 8c,d could not be because of preferential expansion of the activated cell population (Foxp3 negative) on day 2 at which we measured the frequency of Foxp3⁺ cells.

Thus, we believe that TGF-β1 can confer RICD resistance independently of Foxp3 expression. As shown previously,²⁷ although we observed lower FasL expression in T_regs and in vitro T_regs compared with T_cons (Figures 4b, 7a), we do not attribute the RICD resistance of T_regs solely to reduced FasL expression. This is because human T_regs show little or no reduction in FasL expression compared with T_cons (data not shown),⁴⁷ yet are resistant to RICD (Figure 2e,f). Moreover, in contrast to the previous findings in naive T_regs,⁴⁸ activated T_regs and in vitro T_regs showed increased apoptosis resistance even to intrinsic Fas apoptosis that was induced directly using α-FAS antibody (Figure 4d).

We believe that TGF-β1 is involved in cFLIP upregulation in T_regs and in vitro T_regs (Figure 7b), based on a previous study that demonstrated cFLIP induction by TGF-β.⁴⁰ When we stimulated naive cells using α-CD3 antibody in the presence of IL-2 and TGF-β, the naive T cells not only expressed Foxp3 and became in vitro T_regs, but they also expressed higher levels of cFLIP compared with cells without TGF-β (Figure 7b). These results show that TGF-β is involved in inducing cFLIP along with Foxp3 expression in naive T cells. cFLIP may inhibit intrinsic Fas signaling, contributing to FasL-independent component of RICD resistance in T_regs. Future studies are warranted to investigate cFLIP-dependent survival in T_regs during infections. Thus, our data not only support the previously demonstrated role of TGF-β1 in inducing T_regs and controlling inflammation,³⁹ but also reveal its role in promoting T_reg resistance to Fas-dependent RICD in vitro and during infections. Taken together, we show that TGF-β1 contributes to apoptosis resistance mechanism to T_regs and in vitro T_regs during RICD that has implications in controlling immunopathology during microbial infections.

METHODS

Antibodies, reagents, and mice. Purified and fluorochrome-conjugated CD3 (145-2C11), CD28, CD25 (3C7), CD4, Foxp3, and p-Akt antibodies were purchased from eBiosciences (San Diego, CA). Fluorochrome-conjugated annexin-V, Bcl-2, Fas, FasL, Bcl-xl, and Bim antibodies were all purchased from BD Biosciences (San Diego, CA). cFLIP, cleaved caspase-3, caspase-9, and pAkt antibodies were purchased from Cell Signaling Technologies (Danvers, MA). FasL blocking antibody was purchased from R&D Systems (Minneapolis, MN). Fas crosslinking antibody was purchased from Alexis Biochemicals (Farmingdale, NY). Antibody to NF-κB (p65) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse CD4 isolation kits were purchased from Miltenyi Biotec (Auburn, CA) or from Stem Cell Technologies (Vancouver, BC, Canada). Mouse IL-2, IL-10, and human TGF-β were purchased from R&D Systems and BioBasic (Amherst, NY). Complete RPMI-1640 (Hyclone, GE Healthcare, Pittsburgh, PA) supplemented with 10% fetal bovine serum, 100 U ml⁻¹ penicillin (BioBasic), 100 μg ml⁻¹ streptomycin (BioBasic), 1 × Glutamax (Gibco, Thermo Fisher Scientific, Grand Island, NY), and 50 μM β-mercaptoethanol was used for cell cultures. Foxp3^GFP reporter mice on Balb/C or C57BL/6 background, and Foxp3^YFPcre, Rag1^−/−, Il10^−/− Ebi3^−/−, and Cd4-tgfβr2dn^tg mice and the appropriate control mice were purchased from Jackson Laboratories (Bar Harbor, ME). Experiments were performed at Case Western Reserve University (CWRU) in compliance with the CWRU School of Medicine Institutional Animal Care and Use Committee, and adhered to national guidelines published in Guide for the Care and Use of Laboratory Animals, 8th Edn, National Academies Press, 2001(protocol no. 2012-0140). Human cells were obtained from peripheral blood mononuclear cells of the healthy donors obtained under an approved protocol, reviewed, and approved by the University Hospitals Case Medical Center institutional review board (protocol no. 03-13-15). All subjects provided written informed consent, and participants <18 years of age were not enrolled in the study.

RICD induction in mouse and human CD4 cells in vitro. CD4⁺CD25⁺Foxp3^GFP+ T_regs or CD4⁺CD44^low CD25⁻ Foxp3^GFP− naive T cells (T_con) were isolated from Foxp3^GFP reporter mice and cultured in flat-bottom 96-well plates in the presence of soluble 1 μg ml⁻¹ α-CD3 and 2 μg ml⁻¹ α-CD28 antibodies and human IL-2 (10 ng ml⁻¹). For induction of T_regs in vitro human TGF-β1 (4 ng ml⁻¹) was added to naive cells during the primary stimulation. After 4–5 days, dead cells were removed using Ficoll-Paque centrifugation, and restimulated with plate-bound α-CD3 antibody and IL-2 (10 ng ml⁻¹) for RICD induction. Where indicated, cells were labeled with cell proliferation dye-670 (CPD670) at the beginning of stimulation. For some experiments, CD4⁺CD25⁺ T_regs and CD4⁺CD44^low CD25⁻ T_cons were isolated. For human cell experiments, CD4⁺CD25^hi CD127^lo T_regs and CD4⁺CD44^lowCD25⁻ T_cons were used for initial stimulation with 1 μg ml⁻¹ α-CD3 (OKT3) and 2 μg ml⁻¹ α-human CD28 antibodies. After 5 days, RICD was induced using plate-bound OKT3 antibody and IL-2.

Measurement of cell viability by flow cytometry. Cell viability was examined on the basis of forward scatter (FSC) or annexin-V and PI staining of the cells from triplicate culture wells. The percentage of annexin-V⁺PI⁺ or FSC^lo PI⁺ in flow cytometric plots was used as frequency of dead cells to calculate percentage apoptosis (% Apoptosis). % Apoptosis plotted in all analyses is normalized to the cultures that were stimulated with 0 μg ml⁻¹ α-CD3 antibody, whose cell death was set at 0%. In some experiments, where intracellular Foxp3 expression was also determined, we measured the viability of CD4⁺ cells using viability dye-efluor660 (VD-ef660, eBiosciences, San Diego, CA). Data were acquired using BD FACS Calibur or BD LSR Fortessa (Franklin Lakes, NJ) analyser cytometers and were analyzed using FlowJo 9.8 software (Tree Star, Ashland, OR).

NF-κB nuclear translocation. After 30-min restimulation, cells were collected for fixation and permeabilization. To determine the nuclear translocation of the NF-κB, cells were cytospun down on the slides, stained using α-p65 antibody, and were assessed by confocal microscopy.

Apoptosis induction in vivo by administration of α-CD3 antibody in mice. Foxp3^GFP reporter mice were injected with phosphate-buffered saline or 100 μg of α-CD3 antibody (eBiosciences) by intraperitoneal injection. At 1 day after injection, mice were killed, the spleen and lymph nodes were isolated in single-cell suspensions, and the viability of the cells was assayed using green fluorescent protein (GFP) and PI staining and flow cytometry.

Quantitative reverse transcriptase-PCR array analysis. RT² Profiler PCR array plates (PAMM-012Z) from Qiagen/SABiosciences (Valencia, CA) were used for profiling apoptosis-related mRNA expression according to the manufacturer’s instructions. Arrays contained 84 apoptosis-related genes and 12 housekeeping genes. T_regs or T_cons were harvested 1 h after RICD stimulation, and RNA was recovered using an EZ-10 RNA isolation kit (BioBasic). Genomic DNA was removed by DNA Away (Ambion; Life Technologies (Grand Island, NY; AM1906)) and complementary DNA was synthesized from total RNA using MuLV reverse transcriptase enzyme (BioBasic) with OligodT primers. Then, 1 μl of complementary DNA was used per array well in one reaction using the SYBR Green quantitative PCR reaction in a real-time PCR machine ABI7900HT (Applied Biosystems, Grand Island, NY).

Oral Candida infection in mice. Mice were infected as previously described.^{30, 49} Briefly, they were sublingually infected under anesthesia by placing a 3 mm diameter cotton ball saturated with 1 × 10⁸ Candida albicans (CAF2-1) blastospores for 90 min. For experiments involving adoptive transfer of cells, Rag1^−/− mice were reconstituted with indicated congenic cells 3–4 days before infection. Mice were reinfected on day 5 after primary infection for assessing the viability during RICD in vivo.

LCMV infection in mice. Foxp3^YFP reporter mice were obtained from Jackson Laboratory and were infected intravenously with 2 × 10⁶ PFUs of the LCMV clone 13. Virus was diluted in 200 μl phosphate-buffered saline and injected via retro-orbital route in mice anesthetized with isoflurane. Viral stocks were prepared and viral titers were measured as previously described.⁵⁰

Statistical analyses. Statistical significance and P-values were determined by Mann–Whitney tests in Prism 6.0 (GraphPad Software, La Jolla, CA). The s.d. values are shown in the data unless otherwise specified.