Regulatory and Effector T Cell Activation Levels Are Prime Determinants of In Vivo Immune Regulation¹

Natural CD4⁺CD25⁺ regulatory T cells (Treg)⁶ play a major role in the prevention of autoimmune diseases. T_reg are also involved in the down-modulation of immune responses to tumors, allergens, and infectious agents (1) and regulate allogeneic immune responses (2). Because of the multiple functions of T_reg in the immune system and their therapeutic potentials, it is essential to have a better understanding of the conditions in which they exert their suppressive activity in vivo. Notably, it is important to determine these conditions at the initiation of potentially pathological autoimmune processes vs desirable inflammatory immune responses to pathogens.

In vitro experiments suggest that the level of T cell activation may be an important parameter affecting T_reg-mediated suppression. For instance, T_reg inhibited T cell proliferation in the presence of low but not high amounts of anti-CD3 mAb or Ag (3, 4). In addition, T_reg inhibited T cell proliferation induced by immature but not fully mature dendritic cells (DC) (4, 5, 6, 7). Also, T_reg lose their suppressive activity in a context of strong costimulation mediated by CD28 or glucocorticoid-induced TNFR family-related molecules (8, 9). These experiments suggest that highly activated T cells would be less sensitive to suppression by T_reg. However, the relevance of these in vitro findings has not been directly addressed in vivo.

In vivo properties of T_reg have been mostly analyzed in lymphopenic animals. However, T cell biology is significantly altered in lymphopenic animals and it has been observed that even naive T cells can have a regulatory function in this context (10). It is thus critical to evaluate the physiological activity of natural T_reg in nonlymphopenic animals. In this context, we and others have demonstrated that a fraction of autoreactive T_reg is continuously dividing at the steady state (11, 12). Because T_reg turn on their bystander suppressive activity upon activation (13, 14), these dividing T_reg likely exert basal and continuous immunosuppression. In this line, depleting endogenous T_reg led to inhibition of most but not all T-dependent immune responses in reported studies (15, 16, 17, 18, 19, 20). In addition, Ag-specific T_reg activated during an immune response also participate in the immunosuppression (21, 22, 23, 24). Thus, during an immune response to a given Ag, effective T_reg-mediated suppression could depend on the additive effect of basal and Ag-induced immunosuppression, the latter being dependent on the frequency of pre-existing Ag-specific T_reg.

In this study, we first designed an experimental strategy that allowed the evaluation of the in vivo suppressive activity of T_reg in nonlymphopenic mice. We compared the suppressive activity of endogenous T_reg vs transferred Ag-specific T_reg in different conditions of CD4⁺ T cell activation toward the same Ag. Our study suggests that relative activation of T_reg and effector T cells is critical in the intensity and nature of T_reg-mediated suppression in vivo. This conclusion is further supported by our findings showing that endogenous T_reg regulate type 1 diabetes (T1D) only in young NOD mice.

Six- to 8-wk-old BALB/cByJ (BALB/c) mice were obtained from Charles River Laboratories. The ins-HA-transgenic mice expressing hemagglutinin (HA) of influenza virus under the control of the insulin promoter in pancreatic islet β-cells (25) were backcrossed >10 generations onto the BALB/c genetic background. The TCR-HA mice that express a transgenic TCR recognizing the HA_126–138 epitope presented by I-A^d (26) were backcrossed >10 generations onto the BALB/c genetic background and then bred with Thy-1.1 BALB/c congenic mice to generate Thy-1.1 TCR-HA mice. Thy-1.1 BALB/c, TCR-HA, Thy-1.1 TCR-HA, ins-HA mice, and NOD mice were bred in our animal facility under specific pathogen-free conditions. They were manipulated according to European Union guidelines.

Mice were depleted of CD25⁺ cells by a single i.p. injection of 100 μg (purified mAb or ascites) of the depleting anti-CD25 mAb (PC61 hybridoma). Residual CD25⁺ cells were revealed by flow cytometry using the 7D4 anti-CD25 mAb. In our hands, this dose induced over 80% of CD4⁺CD25^high cell depletion in spleen and lymph nodes (LN) for 4 wk, which then returned to normal levels of functional T_reg within 2–3 wk (Ref. 27 and data not shown).

CD25⁺ and CD25⁻ cells were prepared as previously described (28). Briefly, brachial, axillary, cervical, and inguinal LN and spleen were mechanically dissociated. Cells, incubated with biotin-labeled anti-CD25 mAb (7D4; BD Biosciences), were coated with anti-biotin microbeads (Miltenyi Biotec). The CD25⁺ cell fraction (T_reg) was obtained after two consecutive runs on magnetic cell separation LS columns (Miltenyi Biotec), reaching 85% of CD25⁺ cells, >90% of them being Foxp3⁺. The CD25-depleted cells were harvested from the flow-through. The CD25⁻ fraction contained 30% of CD4⁺ T cells and 0.5% of residual CD25⁺ cells. To follow cell division, CD25⁻ cells and T_reg were then labeled with CFSE for 5 min in serum-free PBS at room temperature and were washed twice in PBS before injection. BALB/c or ins-HA mice were injected i.v. with 1.5 × 10⁶ CD25⁻ cells prepared from Thy-1.1 TCR-HA mice with or without cotransfer of 1.5 × 10⁶ (low dose) or 4.5 × 10⁶ (high dose) of T_reg purified from TCR-HA or BALB/c mice. Alternatively, BALB/c mice were injected i.v. with 1 × 10⁶ T_reg prepared from Thy-1.1 TCR-HA mice (see Fig. 5 B). When needed, mice were injected with the anti-CD25-depleting mAb 10 days before cell transfer.

For DC preparation, spleens from BALB/c mice were digested for 30 min with liberase (0.42 mg/ml; Boehringer Mannheim) diluted in a RPMI 1640 solution (Invitrogen Life Technologies) containing DNase I (1 μg/ml; Boehringer Mannheim). Splenocytes were then incubated with anti-CD11c-coated microbeads, and DC were purified after two consecutive runs on magnetic cell separation LS columns (Miltenyi Biotec), giving 85% of CD11c⁺ cells. DCs were then incubated overnight for maturation in RPMI 1640/10% FCS (PAA Laboratories) medium containing 10 ng/ml GM-CSF (R&D Systems) and 2 or 20 μg/ml HA_126–138 peptide and were washed before injection.

The day after adoptive T cell transfer, BALB/c mice were immunized with various protocols: s.c. injection in footpad of 0.3 × 10⁶ DC pulsed in vitro with 2 μg/ml (DC2 condition) or 20 μg/ml HA_126–138 peptide (DC20 condition), s.c. injection of 0.3 × 10⁶ DC pulsed in vitro with 20 μg/ml HA_126–138 peptide associated with i.v. injection of 33 μg of agonistic anti-CD40 mAb (FGK45; Alexis Biochemicals), and 30 μg of LPS (strain 055:B5; Sigma-Aldrich) (DC20/CD40/LPS condition). Alternatively, BALB/c mice were immunized by s.c. injection of 2 μg of HA_126–138 peptide emulsified in CFA (CFA condition).

Nondraining (mix of brachial and axillary) and draining (popliteal for BALB/c mice, pancreatic for ins-HA mice) LNs were mechanically dissociated. For cell surface staining, cells were preincubated with the 2.4G2 mAb to block FcR binding, and then stained in PBS 3% FCS buffer with saturating concentrations of the following mAbs: allophycocyanin-labeled anti-CD4, PE-labeled or CyChrome-labeled anti-Thy-1.1, biotinylated-labeled anti-CD25, revealed with streptavidin-PerCP or streptavidin-CyChrome (all from BD Biosciences). For intracellular cytokine staining, 6 × 10⁶ LN cells were stimulated by 6 × 10⁶ BALB/c splenocytes pulsed with 4 μg/ml HA_126–138 peptide for 6 h at 37°C in 6 ml in the presence of 1 μl/ml GolgiPlug (BD Biosciences). Then, cell surface staining was performed as described above, followed by fixation and permeabilization of cells with CytoFix/CytoPerm buffer (BD Biosciences) for 20 min at 4°C. Cells were then stained for intracellular cytokines for 30 min at 4°C with the following mAb diluted in Perm/Wash buffer (BD Biosciences): PE-labeled anti-IFN-γ and anti-IL-2 (BD Biosciences). PE-conjugated isotypic controls mAb were used to substrate the background staining. Cells were acquired on a FACSCalibur (BD Biosciences) and analyzed with CellQuest (BD Biosciences) or FlowJo (Tree Star) software.

Statistical significances were calculated using the two-tailed unpaired Student t test with 95% confidence intervals.

Using an adoptive transfer model of TCR-transgenic T cells specific for a peptide of influenza virus HA, we designed different immunization conditions leading to different levels of T cell activation (Table I). The lowest level of HA-specific T cell activation was obtained after transfer of CFSE-labeled HA-specific T cells in ins-HA-transgenic mice expressing HA under the control of the insulin promoter (ins-HA condition). In this noninflammatory environment, donor T cells were weakly activated in draining pancreatic LN as shown by low expansion and proliferation and undetectable cytokine production (Table I). The three other conditions were based on a transfer of CFSE-labeled HA-specific T cells in BALB/c mice, subsequently immunized by s.c. injection of HA-pulsed splenic mature DC. To get a gradually stronger immune response, DC were pulsed with increasing amounts of HA peptide (DC2 and DC20 conditions) with or without the administration of the proinflammatory anti-CD40 agonist mAb and LPS reagents (DC20/CD40/LPS condition). In these three different immunization conditions, high expansion of donor HA-specific T cells (Thy-1.1⁺CD4⁺) were observed compared with nonimmunized BALB/c mice. However, gradual T cell activation was attested by progressive acquisition of the CD25 activation marker on responder T cells and by the fact that compared with the DC20 condition, the addition of the proinflammatory reagents led to an increased proportion of IFN-γ- and IL-2-producing cells by a factor of 2–3 (Table I). Interestingly, lower expansion of donor T cells was observed in the DC20/CD40/LPS condition compared with the DC20 condition despite higher levels of cytokine production and CD25 expression. This suggests an increase in activation-induced cell death possibly due to exhaustion or IFN-γ-mediated cell death (29).

We then analyzed the effects of depletion of endogenous T_reg by an anti-CD25 mAb on proliferation, expansion, and cytokine production of HA-specific T cells in the four different immunization conditions described above. In the ins-HA noninflammatory condition, T_reg depletion induced a significant increase of the few divided HA-specific T cells in draining pancreatic LN, whereas IL-2 and IFN-γ cytokine production was not detected (Fig. 1,A). In contrast, in BALB/c mice immunized with HA-pulsed DC, T_reg depletion did not induce expansion of transferred HA-specific T cells, irrespective of the condition (Fig. 1,A). However, in the DC2 condition, T_reg depletion induced a significant augmentation of the numbers of IFN-γ- and IL-2-producing T cells, which was increased by a factor of 2–3 compared with controls, as shown in graphs depicting all performed experiments (Fig. 1,A) and in a representative experiment (Fig. 1,B). In the DC20 condition, T_reg depletion induced a significant increase of the numbers of IFN-γ-producing cells that was less pronounced than in the DC2 condition. Depleting endogenous T_reg in the DC20/CD40/LPS high inflammatory condition did not have any effect on expansion or cytokine production of HA-specific T cells (Fig. 1). As an additional immunization protocol, BALB/c mice were immunized with HA peptide emulsified in CFA (CFA condition). Similarly, in this highly inflammatory condition, T_reg depletion did not significantly affect expansion or IL-2 production of HA-specific T cells (Fig. 1). In this condition, IFN-γ production was barely detectable as previously reported (30). Overall, these experiments show that endogenous T_reg efficiently regulated T cell expansion only in the ins-HA condition. In the DC2 or DC20 conditions, T_reg exerted selective suppressive activity affecting only T cell differentiation, i.e., cytokine production. In the DC20/CD40/LPS and CFA conditions, T_reg suppressed neither expansion nor cytokine production.

To assess activation of endogenous T_reg, we analyzed their frequency and their size in the different immunization protocols. Except for the CFA condition, no significant increased proportions of CD4⁺CD25^high cells were observed in draining LN compared with nondraining LN (Fig. 2,A), suggesting that endogenous T_reg were poorly or not activated by these various treatments. This was further supported by their small size, compared with transferred HA-specific T cells, except for the ins-HA condition (Fig. 2,B). To assess cell division of polyclonal T_reg, we adoptively transferred CFSE-labeled polyclonal T_reg from BALB/c mice in the DC20/CD40/LPS or CFA inflammatory conditions. These transferred T_reg, which likely behave as endogenous T_reg in nonlymphopenic mice (12), were not activated, as assessed by their size and their proliferation profile (Fig. 2,C). In addition, as observed with endogenous T_reg, transferred polyclonal T_reg did not suppress expansion and cytokine production of transferred HA-specific T cells (Fig. 2 D). Altogether, our results show that polyclonal T_reg (endogenous or transferred) were poorly activated in the various immunization protocols.

We then investigated whether cotransfer of T_reg from TCR-HA-transgenic mice could suppress expansion and cytokine production of CFSE-labeled HA-specific T cells in the DC (DC2, DC20, DC20/CD40/LPS) and CFA conditions. In all conditions, cotransfer of HA-specific T_reg inhibited expansion of HA-specific T cells in a dose-dependent manner. Transferred divided CD4⁺ T cells represented 1.8–3% of draining LN cells, which dropped to 0.2–0.6% after cotransfer of a high dose of HA-specific T_reg (Fig. 3,A). Inhibition of the expansion of HA-specific T cells was likely due to reduced proliferation because highly divided cells (>4 divisions) represented 80% in the absence of transferred T_reg vs 30–50% when high numbers of HA-specific T_reg were cotransferred. Consequently, the proportions of undivided cells and of cells that had divided less than five times were increased in the presence of HA-specific T_reg (Fig. 3, B and C). Regarding cytokine production, a high dose of T_reg induced a significant decrease of cells producing IFN-γ and IL-2 as shown in a representative experiment (Fig. 4,A) and graphs showing all performed experiments (Fig. 4,B). In the DC2 and DC20 conditions, IFN-γ and IL-2 production decreased from 7 to 13% of divided donor CD4⁺ T cells in controls to 1–3% of the cells in mice injected with a high dose of HA-specific T_reg. In the DC20/CD40/LPS condition, IFN-γ- and IL-2-secreting cells decreased from 23 and 15% in controls to, respectively, 4 and 7% in T_reg-injected mice. Finally, in the CFA condition, HA-specific T_reg also strongly inhibited IL-2 production of HA-specific CD4⁺ T cells (Fig. 4). Cotransfer of a lower number of HA-specific T_reg resulted in an intermediate inhibition of cytokine production of donor CD4⁺ T cells (Fig. 4 B).

To assess activation of transferred HA-specific T_reg, we analyzed their proliferation and size. Compared with nonimmunized control mice, division (CFSE loss) and increase in cell size of cotransferred T_reg was readily observed in all conditions of immunization (Fig. 5,A). However, highly divided CFSE⁻ T_reg could not be discriminated from the pool of endogenous CD4⁺ cells because they shared the same Thy-1.2 allele. We thus transferred HA-specific Thy-1.1⁺ T_reg alone and confirmed that they divided in all conditions except in nonimmunized mice (Fig. 5 B). Overall, our results show that contrary to polyclonal T_reg, HA-specific T_reg were able to efficiently suppress expansion and cytokine production in strong T cell activation contexts. Importantly, this was correlated with their strong activation.

We previously observed that NOD mice genetically deficient in T_reg exhibited exacerbated diabetes, showing that T_reg are major players in diabetes regulation in this model (31, 32). We performed transient T_reg depletion in NOD mice at various ages to reveal their role at different phases of a chronic progressive inflammatory disease. Indeed, the autoimmune process in NOD mice is progressive with the first signs of autoimmunity starting at 3 wk of age, with few APCs and lymphocytes invading the periphery of pancreatic islets. This is followed by a chronic and progressive increase of the level of inflammation and infiltrating T cells around and then inside the islets (33, 34, 35), which is associated with a decrease in the proportion of T_reg in islets and draining pancreatic LN and thus a dysbalance of the T_reg-T cell ratio (36). Eventually, mice develop clinical diabetes in 75% of females and 25% of males at 25 wk of age. Strikingly, when T_reg depletion was induced at 3 wk of age in NOD mice, we observed early onset and higher incidence of diabetes. Eighty percent of treated males and females were diabetic as early as 15 wk of age and all mice were diabetic by 21 wk of age (Fig. 6). Exacerbation of T1D was also observed when T_reg depletion was induced at 4 wk of age in males although with a lesser magnitude compared with the treatment induced in 3-wk-old mice. In contrast, no significant increase in diabetes incidence was observed when T_reg depletion was performed after 6 wk of age in both sexes (Fig. 6), indicating that T_reg do not control the disease after this age. Importantly, the fact that the treatment did not reduce diabetes incidence suggests that it did not significantly affect CD25⁺ diabetogenic T cells. Indeed, anti-CD25 mAb treatment mostly depleted cells expressing high levels of CD25 (data not shown) whereas diabetogenic T cells express a low level of CD25 (37). These results show that endogenous T_reg control T1D only before 6 wk of age, during the early phases of the disease in NOD mice.

Three major conclusions can be drawn from our experiments. 1) The efficiency of T_reg-mediated basal immunosuppression is related to the level of responder T cell activation. 2) Only Ag-specific T_reg can block activation of CD25⁻ T cells in a context of strong T cell activation. 3) Depending on the experimental approach, T_reg inhibit either T cell expansion or cytokine production or both.

It has been shown that T_reg can efficiently suppress the effector response in some inflammatory situations. For instance, depleting T_reg in vivo led to an increased immune response to various pathogens (17, 18, 38). This could be explained by an increased activity of T_reg during an infectious disease (17) due to various mechanisms (39). Inflammation associated with infection could activate DC, which in turn would induce expansion of T_reg (7, 40, 41). In addition, danger signals released during infection by tissue damage may also increase T_reg-mediated suppression (39). Finally, T_reg express multiple TLR, and their ligands expressed by pathogens can also directly induce expansion of highly suppressive T_reg (42). Our study suggests that basal immunosuppression exerted by endogenous T_reg on CD4⁺ T cells in nonlymphopenic mice is overwhelmed in a context of strong T cell activation, in apparent contrast with the above observations. This discrepancy could be explained by the poor activation of polyclonal T_reg, as observed in our model, due to several factors. Our immunization protocols induce limited tissue damage and thus minimal danger signals would be released to activate endogenous T_reg. Also, endogenous T_reg would not be activated by TLR signaling in our immunization conditions.

The importance of T_reg activation for efficient immunosuppression is further suggested in our experiments with transferred HA-specific T_reg. In the DC and CFA conditions, these T_reg strongly divided and exerted effective immunosuppression. In that line, Ag-dependent T_reg immunosuppression has been correlated with proliferation in response to immunization with Ag emulsified in Freund adjuvant (22, 24). The correlation between T_reg-mediated suppression and T_reg division confirms that T_reg activation is a key requirement for effective immunosuppression.

Altogether, our cognitive model of various immune responses to the same Ag suggests that T_reg-mediated suppression is dependent on the relative activation of both T_reg and effector T cells. Indeed, T_reg efficiently suppressed expansion when effector T cells and T_reg were both poorly activated (i.e., ins-HA condition, Figs. 1 and 2). In contrast, poorly activated T_reg did not suppress if effector T cells were strongly activated (i.e., polyclonal T_reg in DC20/CD40/LPS or CFA conditions, Figs. 1 and 2). Finally, strongly activated T_reg efficiently suppressed strongly activated effector T cells (i.e., HA-specific T_reg in DC20/CD40/LPS or CFA conditions, Figs. 3–5).

Analyzing in vivo expansion and cytokine production by Ag-specific CD4⁺ T cells following immunization has been performed in various models of adoptive transfer of Ag-specific T cells from TCR-transgenic mice into nonlymphopenic mice. Using this experimental approach, it has been shown that T_reg selectively inhibit CD25⁻ T cells at the level of cytokine production but not of their proliferation or expansion (43, 44), whereas other investigators reported an effect of T_reg on CD25⁻ T cell expansion (22, 24). The reasons for these conflicting data could be due to several experimental differences: 1) the type of immunization (Ag-pulsed DC (43, 44) vs Ag in Freund adjuvant (22, 24)); 2) the type of approach (deficit of endogenous T_reg (43, 44) vs cotransfer of Ag-specific T_reg (22, 24)); 3) the Ag specificity of the transgenic TCR. Our study compares different immunization protocols (pulsed DC vs Freund adjuvant), different approaches (T_reg depletion vs T_reg cotransfer) with T cells specific for the same HA Ag. Our findings emphasize the importance of the experimental approach on the nature of T_reg-mediated suppression. Endogenous T_reg affect cytokine production by CD25⁻ T cells without inhibiting their proliferation, showing that cytokine production is not directly linked to cell division (this study and others (43, 44)). In contrast, cotransfer of Ag-specific T_reg also affects proliferation of CD25⁻ T cells (this study and others (22, 23, 24)). This may rely on the fact that self-reactive endogenous T_reg and TCR-transgenic T_reg could function in an entirely different way. Whatever the mechanism, we speculate that T_reg may preferentially affect expansion or cytokine production of effector T cells depending on their activation level.

Previous publications showed that blocking CTLA-4 or ICOS pathways induced diabetes exacerbation in NOD mice expressing an islet Ag-specific transgenic TCR. This was observed only when performed in young animals exhibiting no or mild insulitis. Blocking these pathways may have altered T_reg function, suggesting that endogenous T_reg play a major role at the beginning of the autoimmune process (37, 45). Here, we directly tested whether T_reg would play a regulatory function predominantly in young NOD mice. Depleting endogenous T_reg led to T1D exacerbation only if performed in young (3–6 wk old) NOD mice, when inflammation in islets is still mild (33, 34, 35). At this stage, depletion of endogenous T_reg would remove basal immunosuppression leading to increased activation of islet-specific CD25⁻ T cells. As the disease progresses, T_reg-mediated basal immunosuppression would be overwhelmed in a context of a rising number of islet-specific effector T cells (46, 47) and a proportional decrease of T_reg (36). The efficiency of this basal immunosuppression would thus depend on the balance between activated effector T cells and activated T_reg locally in islets or draining LN, as suggested in our cognitive model using HA-specific T cells. This hypothesis is further supported by the findings that transfer of islet-specific, but not polyclonal, T_reg can control advanced diabetes (48, 49, 50). We cannot rule out, however, that CD25⁺ T cells of aging NOD mice have lesser suppressive activity than the ones of young animals, or that CD25⁻ T cells of aging NOD mice become progressively refractory to suppression by T_reg, as reported in vitro (51, 52). Altogether, T_reg specific for target Ag could be envisioned for the treatment of autoimmune diseases either acute or chronic. It is now critical to design conditions for generation of high numbers of human Ag-specific T_reg.

We are grateful to Olivier Boyer for critical comments on the manuscript. We thank Frédéric Jacob, Simon Blanchard, and Douglas Vasey for their contribution to some of the experiments.

The authors have no financial conflict of interest.