Distinct roles of CTLA-4 and TGF-β in CD4+CD25+ regulatory T cell function



Both CTLA-4 and TGF-β have been implicated in suppression by CD4+CD25+ regulatory T cells (Treg). In this study, the relationship between CTLA-4 and TGF-β in Treg function was examined. Blocking CTLA-4 on wild-type Treg abrogated their suppressive activity in vitro, whereas neutralizing TGF-β had no effect, supporting a TGF-β-independent role for CTLA-4 in Treg-mediated suppression in vitro. In CTLA-4-deficient mice, Treg development and homeostasis was normal. Moreover, Treg from CTLA-4-deficient mice exhibited uncompromised suppressive activity in vitro. These CTLA-4-deficient Treg expressed increased levels of the suppressive cytokines IL-10 and TGF-β, and in vitro suppression mediated by CTLA-4–/– Treg was markedly reduced by neutralizing TGF-β, suggesting that CTLA-4-deficient Treg develop a compensatory suppressive mechanism through CTLA-4-independent production of TGF-β. Together, these data suggest that CTLA-4 regulates Treg function by two distinct mechanisms, one during functional development of Treg and the other during the effector phase, when the CTLA-4 signaling pathway is required for suppression. These results help explain contradictions in the literature and support the existence of functionally distinct Treg.


Regulatory T cells




5-(and-6)-carboxyfluorescein diacetate succinimidyl ester


Latency-associated peptide

1 Introduction

CD4+CD25+ regulatory T cells (Treg) are potent inhibitors of T cell activation in vitro and in vivo. The suppressive mechanism utilized by Treg is not well understood. It appears that in vitro the suppression is cell-contact dependent, likely mediated through direct T cell:T cell interactions 1. In most of these in vitro systems and some in vivo systems, suppression is not inhibited by antibodies specific for cytokines such as TGF-β and IL-10, and Treg from TGF-β- or IL-10-deficient mice exhibit uncompromised suppressive activity 28. However, in many other in vivo experimental models, Treg-mediated suppression requires TGF-β or IL-10, and in certain instances Treg themselves produce these cytokines 912, suggesting that the suppression is achieved through these soluble inhibitors.

CTLA-4-deficient mice develop a fatal lymphoproliferative disease 13, 14. Although the exact cause of the disease in these mice is not clear, the prevailing view is that the lymphoproliferation develops as a consequence of a lack of intrinsic control of T cell activation in the absence of CTLA-4. However, bone marrow chimera and co-transfer studies show that mice with both wild-type (WT) and CTLA-4–/– cells fail to develop lymphoproliferation. This suggests that the uncontrolled T cell activation in CTLA-4–/– mice may be due to a lack of extrinsic regulation of T cells 15, 16. Since Treg constitutively express CTLA-4, it has been postulated that CTLA-4 is critical for Treg function. In this regard, it has been shown that anti-CTLA-4 treatment blocks Treg function in vitro and that Treg from CTLA-4-deficient mice display impaired suppressive activity 17. Moreover, studies have demonstrated that CTLA-4 blockade in vivo abrogates Treg-mediated protection against autoimmune colitis and transplant rejection 10, 11. However, since T cells in CTLA-4-deficient mice undergo polyclonal activation (as evidenced in part by CD25 expression) and proliferation soon after birth, the reduced Treg function may be a result of contamination of Treg with activated T cells. In addition, increased autoimmunity after in vivo CTLA-4 blockade may be a consequence of enhanced pathogenicity of effector T cells instead of inhibition of Treg function. In this study, we investigated the roles of CTLA-4 and TGF-β in Treg-mediated suppression in vitro. We found that although CTLA-4 blockade abrogated WT Treg function, CTLA-4-deficient Treg exhibited uncompromised suppressive activity. However, suppression mediated by CTLA-4-deficient Treg is qualitatively different from that mediated by WT Treg in that it is at least partially TGF-β-dependent.

2 Results

2.1 Normal Treg development and homeostasis in CTLA-4-deficient mice

In initial sets of experiments, we determined the role of CTLA-4 in CD4+CD25+ Treg development and peripheral homeostasis. One of the difficulties in analyzing Treg from CTLA-4-deficient mice is distinguishing Treg from activated T cells, which also express CD25. In this study, we used two approaches to minimize contamination of Treg with activated T cells. First, mice expressing a CTLA-4Ig transgene under the control of the keratin 14 promoter were used. The CTLA-4Ig fusion protein is constantly secreted by epithelial cells, resulting in an average serum concentration of approximately 20 ng CTLA-4Ig/ml. CTLA-4Ig binds to B7.1 and B7.2 and blocks their engagement with CD28, thereby delaying T cell activation and lymphoproliferation in CTLA-4-deficient mice 18. Generally, CTLA-4–/– mice die within 3–4 weeks of age. In contrast, 4-week-old CTLA-4Ig transgenic CTLA-4–/– mice are healthy, and the cellularity of their secondary lymphoid organs is normal (data not shown). Second, in several experiments CTLA-4–/– mice were treated with a mixture of anti-B7.1 and anti-B7.2 mAb beginning at 1 week of age to prevent lymphoproliferation. Finally, to ensure that we were working with true Treg, CD4+CD25+ T cells were further separated based on expression of CD62L. CD62L is expressed highly on resting T cells, and its expression is down-regulated on recently activated cells. Most CD25+ Treg in normal young unimmunized mice are CD62L+, and CD62L+ cells have been shown to exhibit regulatory activity 19. Therefore, using CD62L as an additional marker helps to exclude recently activated T cells from the Treg population. This is especially important when T cells from CTLA-4-deficient mice are analyzed. Even with the CTLA-4Ig transgene, CTLA-4–/– mice start to show signs of T cell activation and lymphadenopathy by 6 weeks of age. By using the additional CD62L marker, we were able to identify a distinct population of Treg that are present in young CTLA-4Ig transgenic CTLA-4–/– mice at a frequency similar to that seen in their WT littermates (Fig. 1A). It has recently been shown that Treg express high levels of the transcription factor FoxP3, which is critical for Treg development and function 2022. Because FoxP3 protein is not expressed in CD4+CD25 cells after activation, it is the only marker identified so far that is uniquely expressed at high levels in Treg. To determine if CD4+CD62L+CD25+ cells from CTLA-4–/– mice are of the same lineage as Treg from WT mice, we examined FoxP3 expression in CTLA-4–/– Treg by Western blot analysis. CTLA-4–/– Treg express 24-fold higher FoxP3 protein than CD4+CD62L+CD25 cells (Fig. 1B), similar to that observed in WT mice, suggesting that Treg developed in the absence of CTLA-4 are of the same lineage as those in WT mice.

Figure 1.

Normal development and homeostasis of regulatory T cells in CTLA-4–/– mice. (A) Percentages of Treg among CD4+ cells from LN of 5-week-old CTLA-4–/– CTLA-4 Ig transgenic and CTLA-4+/+ CTLA-4 Ig transgenic littermates were compared. The left two panels are representative dot plots of flow cytometry analysis of CD4+ cells. The graph to the right shows all mice examined in the experiment; each circle represents one mouse (n=3), and the bar indicates the mean of the group. (B) Western blot analysis of FoxP3 expression in CD4+CD62L+CD25 cells and Treg purified from CTLA-4+/+ CTLA-4 Ig transgenic mice and Treg purified from CTLA-4–/– CTLA-4Ig transgenic mice. The numbers below the bands indicate the relative band intensity of FoxP3. (C) CD4+ splenocytes from age-matched CTLA-4+/– untreated mice (left) or CTLA-4–/– mice treated with anti-B7.1 and anti-B7.2 on days 9, 12 and 15 after birth and harvested on day 10 (middle) or day 28 (right) after the start of treatment were analyzed by flow cytometry for expression of CD62L and CD25. Results are representative of five (A) and two (B and C) independent experiments.

We previously demonstrated that development and peripheral homeostasis of Treg is dependent on CD28 interaction with B7 molecules 23, 24. To determine if this effect is mediated indirectly through CTLA-4, we treated CTLA-4–/– mice on days 7, 11, and 13 after birth with a combination of anti-B7.1 and anti-B7.2 mAb and examined the percentage of Treg at various time points afterwards. The anti-B7 treatment of CTLA–/– mice led to a marked reduction of Treg (Fig. 1C middle panel) as observed previously in WT mice 23, 24. Moreover, the CTLA-4–/– Treg recovered with similar kinetics as their WT counterparts after the anti-B7 treatment was stopped, such that the percentage of Treg returned to normal levels in CTLA-4–/– mice by 28 days after the last anti-B7 treatment (Fig. 1C right panel). Thus, the homeostasis of the Treg is B7-dependent and CTLA-4-independent, and CTLA-4 deficiency does not affect the development or peripheral homeostasis of Treg.

2.2 In vitro suppression by WT Treg is CTLA-4-dependent and TGF-β-independent

Although Treg are present in CTLA-4–/– mice in normal numbers, the disease in CTLA-4-deficient mice could result from a functional defect in the resident Treg. It has been proposed that Treg-mediated suppression is dependent on costimulation through CTLA-4 17, which may induce TGF-β production by T cells 25 and Treg in particular 26, although these later findings remain controversial 27. We therefore examined the effect of blocking CTLA-4 or TGF-β in our in vitro suppression assays. The mAb we used have been shown to function as antagonists in a variety of systems both in vitro and in vivo. CD4+CD62L+CD25 cells from CTLA-4Ig transgenic CTLA-4–/– mice were used as responder cells to distinguish effects of the anti-CTLA-4 mAb treatment on the Treg versus responder populations. Responder cells were labeled with 5-(and-6)-carboxyfluorescein diacetate succinimidyl ester (CFSE), and their proliferation was assessed by dilution of the intracellular dye. Proliferation of the CTLA-4-deficient responder cells was markedly inhibited by WT Treg (Fig. 2A, left panel). Moreover, the addition of anti-CTLA-4 Fab fragments completely abolished the suppression by WT Treg (Fig. 2A, right panel). The effect of the anti-CTLA-4 antibody was specific, as it had no effect on the Treg activity of CTLA-4-deficient T cells (see below). In contrast, suppression by WT Treg was not affected in the presence of a high concentration of TGF-β neutralizing antibody (Fig. 2B). Thus, our results are in agreement with previous reports 3, 17 that suggested that CTLA-4 expression on Treg is critical for their function. However, this function of CTLA-4 is not mediated through TGF-β, since the in vitro function of Treg was TGF-β independent.

Figure 2.

In vitro suppression by WT Treg is CTLA-4-dependent and TGF-β-independent. (A) CFSE-labeled CD4+CD62L+CD25 cells from CTLA-4–/– CTLA-4Ig transgenic mice were stimulated with WT T-depleted splenocytes and anti-CD3 alone (filled) or in the presence of an equal number of unlabeled WT Treg (solid line) and in the absence (left panel) or presence (right panel) of anti-CTLA-4 Fab fragments (100 μg/ml) added at the initiation of the culture. Cells were harvested at 72 h and analyzed by flow cytometry for CFSE intensity. (B) WT Treg were co-cultured with CD4+CD62L+CD25 cells from CTLA-4+/– CTLA-4Ig transgenic mice at the indicated ratio with (filled) or without (open) 50 μg/ml anti-TGF-β mAb. These data are representative of two (A) and six (B) independent experiments.

2.3 CTLA-4–/– Treg exhibit normal suppressive activity in vitro

In an attempt to substantiate the results obtained using anti-CTLA-4 antibodies, we next compared the suppressive activity of Treg from CTLA-4–/– and WT mice in vitro. Surprisingly, CTLA-4-deficient Treg were able to suppress proliferation of WT responders (Fig. 3A, left panel). Addition of anti-CTLA-4 Fab to the suppression assay enhanced proliferation of the WT responders, as demonstrated by the increased fraction of cells diluting CFSE in responder alone (CD25) cultures that received the antibody (Fig. 3A, right panel, shaded histograms). The same treatment did not affect proliferation of the CTLA-4–/– effectors as illustrated in Fig. 2A (similarity of CFSE dilution of the shaded histograms in the right and left panels). Moreover, the anti-CTLA-4 Fab did not change the suppression of the CTLA-4–/– Treg (Fig. 3A, compare CFSE dilution between lined histograms in the right and left panels). These results demonstrate that at the concentration used in the experiments presented in Fig. 2A and 3A, the anti-CTLA-4 Fab was efficacious and its activity was specific. A more quantitative comparison of the suppressor activity of WT and CTLA-4–/– Treg at various Treg to responders ratios showed that suppression by the two populations of Treg was indistinguishable (Fig. 3B), demonstrating that loss of CTLA-4 expression does not affect the suppressive activity of Treg in these knockout mice. In addition, CTLA-4–/– Treg were equally efficient at suppressing proliferation of CTLA-4–/– responders (data not shown). These findings were different from previous reports showing that CTLA-4–/– Treg have a 50% reduction in regulatory activity 17; it is likely that this previously observed reduction was due to contamination of activated cells with Treg. By using young CTLA-4Ig transgenic CTLA-4–/– mice combined with sorting for CD62Lhigh Treg, we were able to assess the activity of CTLA-4-deficient Treg more accurately than previously reported. Taken together, our results demonstrate that although anti-CTLA-4 mAb affects suppressor function in WT Treg, Treg can develop and function normally in CTLA-4–/– mice.

Figure 3.

CTLA-4–/– Treg exhibit uncompromised suppressive activity in vitro. (A) CFSE-labeled CD4+CD62L+CD25 cells from WT CTLA-4Ig transgenic mice were stimulated with WT T-depleted splenocytes and anti-CD3 alone (filled) or in the presence of equal number of unlabeled CTLA-4–/– Treg from CTLA-4Ig transgenic mice (solid line) in the absence (left panel) or presence (right panel) of anti-CTLA-4 Fab fragments (100 μg/ml) added at the initiation of the culture. Cells were harvested at 72 h and analyzed by FACS for CFSE intensity. (B) Treg from WT CTLA-4Ig transgenic (filled) or CTLA-4–/– CTLA-4 transgenic mice (open) were co-cultured with CD4+CD62L+CD25 cells from WT CTLA-4Ig transgenic mice at the indicated ratios. These data are representative of seven independent experiments.

2.4 Lack of a contribution of IL-10 to the functions of CTLA-4–/– Treg

The conflicting results presented thus far suggested that there may be compensatory changes in CTLA-4–/– Treg that enable them to suppress in a CTLA-4-independent manner. We therefore compared expression of the immunosuppressive cytokine IL-10 in Treg from WT and CTLA-4–/– mice. The mRNA levels for IL-10 were 5-fold higher in CTLA-4–/– Treg than in WT Treg as assessed by quantitative real-time PCR analysis (Fig. 4A). CTLA-4–/– Treg consistently secreted 10-fold more IL-10 upon in vitro stimulation with anti-CD3 and splenic APC (Fig. 4B). However, anti-IL-10 antibody had no effect on the in vitro suppression mediated by either WT (data not shown) or CTLA-4–/– Treg (Fig. 4C). Thus, our results show that although CTLA-4–/– Treg express significantly higher levels of IL-10, IL-10 does not contribute to the suppression function of CTLA-4–/– Treg.

Figure 4.

Function of CTLA-4–/– Treg is not dependent on their enhanced IL-10 expression. (A) Real-time PCR analysis of IL-10 mRNA levels in freshly isolated Treg from CTLA-4–/– CTLA-4Ig transgenic or CTLA-4+/+or+/– CTLA-4Ig transgenic mice. (B) Treg purified from CTLA-4–/– CTLA-4Ig transgenic or CTLA-4+/+or+/– CTLA-4Ig transgenic mice were stimulated in vitro for 48 h, and the culture supernatants were analyzed for IL-10 secretion by ELISA. (C) CTLA-4–/– Treg were co-cultured with CD4+CD62L+CD25 cells from CTLA-4+/– CTLA-4Ig transgenic mice at the indicated ratios with (filled) or without (open) 20 μg/ml anti-IL-10 mAb. Results are representative of two independent experiments.

2.5 In vitro suppression by CTLA-4–/– Treg is partially TGF-β-dependent

We next examined the expression of TGF-β by CTLA-4–/– Treg. No increase in TGF-β message (Fig. 5A) or protein (Fig. 5B) was observed when compared to that expressed by WT Treg. TGF-β and TGF-β receptors are widely expressed by normal tissues, and the biological activity of TGF-β is mainly controlled after the secretion of the cytokine (reviewed in 28). Most of the TGF-β is secreted in its latent inactive form due to its linkage to the latency-associated peptide (LAP), which has to be cleaved to release the active TGF-β. LAP has two mannose-6-phosphate (M6P)-containing carbohydrate modifications that mediate its binding to the cell surface through M6P/insulin-like growth factor II receptor. Moreover, the cell-surface anchoring of latent TGF-β has been shown to be important for its conversion to the biologically active form 2931. The ELISA we used to quantify secreted TGF-β involved chemical activation of TGF-β by denaturing LAP with acid and could therefore not distinguish between latent and active TGF-β. In addition, since culture supernatant was analyzed, the level of membrane-bound TGF-β was not assessed by the assay. Thus, we examined the levels of membrane-bound TGF-β on various cells using flow cytometry. Treg from WT mice did not show any TGF-β or LAP staining even after in vitro activation, whereas LAP was readily detectable on Treg freshly isolated from CTLA-4–/– mice, and TGF-β expression was detected after in vitro activation (Fig. 5C). Thus, Treg from CTLA-4–/– mice express more TGF-β, especially after activation.

Figure 5.

In vitro suppression by CTLA-4–/– Treg is partially dependent on TGF-β. (A) Real-time PCR analysis of TGF-β mRNA levels in freshly isolated Treg from CTLA-4–/– CTLA-4Ig transgenic or CTLA-4+/+or+/– CTLA-4Ig transgenic mice. (B) Treg purified from CTLA-4–/– CTLA-4Ig transgenic or CTLA-4+/+or+/– CTLA-4Ig transgenic mice were stimulated in vitro for 48 h, and the culture supernatants were analyzed for TGF-β secretion by ELISA. (C) LN cells (left four panels) from CTLA-4–/– CTLA-4Ig transgenic or CTLA-4+/+or+/– CTLA-4Ig transgenic mice were stained with antibodies specific for CD4, CD62L, CD25, and cell-surface LAP or TGF-β and analyzed by flow cytometry. Histograms of LAP or TGF-β (shaded) and background (bold line) staining on Treg is shown. Purified Treg cells (right two panels) were stimulated as described in (B) for 24 h, and cell-surface TGF-β expression was determined by flow cytometry. Results are representative of three independent experiments. (D) CTLA-4–/– Treg were co-cultured with CD4+CD62L+CD25 cells from CTLA-4+/– CTLA-4Ig transgenic mice at the indicated ratios with (filled) or without (open) 50 μg/ml anti-TGF-β mAb. These data are representative of six independent experiments.

To determine if the higher level of TGF-β expression by CTLA-4–/– Treg contributes to their suppressive activity in vitro, we examined the suppression by CTLA-4–/– Treg in the presence of TGF-β neutralizing antibodies in vitro. Anti-TGF-β treatment led to a significant reduction of suppression by these cells, especially at lower Treg to responder ratios (Fig. 5D). Addition of anti-IL-10 mAb did not have any effect on the partial reversion of suppression by anti-TGF-β antibodies (data not shown). Similar results were observed using WT T effector cells in the suppression assays (data not shown). Taken together, these results suggest that membrane-bound TGF-β on the surface of CTLA-4–/– Treg contributes to their suppressive activity.

3 Discussion

In this study, we have shown that suppression by WT Treg is CTLA-4-dependent and TGF-β-independent. However, CTLA-4–/– Treg exhibited uncompromised suppressive activity and expressed higher levels of the suppressive cytokines IL-10 and TGF-β. Moreover, the in vitro suppression by CTLA-4–/– Treg was partially dependent on TGF-β.

The role of CTLA-4 in Treg-mediated suppression has been controversial. It has been reported that the suppressive activity of T cell clones derived from CD4+CD25+ cells in human PBL strongly correlates with a high level of CTLA-4 expression 8, consistent with a role of CTLA-4 in Treg function. Similarly, we have observed by comparing gene expression in Treg and CD4+CD25 cells that CTLA-4 is one of the few genes that is consistently highly overexpressed in mouse and human Treg under various activation conditions (data not shown). However, direct demonstration of a functional role of CTLA-4 in vitro has been controversial. Several findings in the present study help to explain these discrepancies. First, we observed that CTLA-4 blockade abrogated WT Treg suppressor activity but found that a relatively high concentration of anti-CTLA-4 Fab fragments (100 μg/ml) was required to completely abrogate suppression by WT Treg in vitro. Previous studies used much less antibody in their suppression assays and often used whole antibodies 2, which might have agonist effects in the culture 32. To avoid the confounding issues of the agonist effect of whole anti-CTLA-4 antibodies, we exclusively used Fab fragments in all the experiments presented in this study. Because we had previously shown 33 that Fab fragments of anti-CTLA-4 are ten times less efficient in binding to CTLA-4 than whole anti-CTLA-4 mAb, we used relatively high concentrations of Fab in our cultures. In addition, it is important to distinguish the effects of CTLA-4 blockade on the CD25 responders from those on the Treg. In the current study, we took advantage of T cells isolated from CTLA4Ig transgenic CTLA-4–/– mice to ensure that the antibody effects were limited to the Treg populations. Interestingly, even at these high concentrations, the anti-CTLA-4 antibody did not break the anergic phenotype of the Treg (unlike engagement of GITR 34, 35) or induce death of Treg (data not shown). Thus, the basis for CTLA-4-mediated regulation of Treg is most likely due to a direct role on their suppressive function. CTLA-4 engagement of B7 may result in inhibition of responder proliferation. This may occur directly or via the production of a secondary inhibitor. In accordance with this notion, it has been demonstrated that CTLA-4 can engage B7 on antigen-presenting cells and induce indeolamine 2,3-dioxygenase expression, leading to the degradation of tryptophan and resultant inhibition of T cell proliferation 36, 37. Alternatively, it is possible that CTLA-4 sends a signal to Treg to activate a suppressor mechanism. Treg must be activated through their T cell receptors to induce their suppressive activity. Thus, CTLA-4 may contribute to the induction of suppression by sending an independent signal or by altering TCR signaling.

We have shown in this study that CTLA-4–/– Treg have normal suppressive activity, whereas it was previously reported that CTLA-4–/– Treg are less effective suppressors 17. This was likely due to contamination of Treg with activated T cells in previous studies. T cells from CTLA-4–/– mice express activation markers such as CD25 as early as 7 days after birth. Even in the presence of a CTLA-4Ig transgene, a concomitant increase in CD25 and decrease in CD62L markers became apparent in CTLA-4–/– mice at 6 weeks of age (data not shown). In mice that exhibited overt signs of T cell activation, the separation between CD62Lhigh and CD62Llow Treg became less distinctive, and even CD62Lhigh Treg from these mice showed less regulatory activity when compared to Treg from young CTLA-4–/– mice with minimal in vivo T cell activation (data not shown).

Our observation that CTLA-4–/– Treg express higher levels of the suppressive cytokines IL-10 and TGF-β than WT Treg suggests that in addition to its role in Treg effector function, CTLA-4 also controls the functional development of Treg. Two functionally distinct populations of Treg have been described, the natural and adaptive Treg (reviewed in 38). Natural Treg develop in the thymus, express CD4+CD25+ markers, are naturally suppressive, and suppress in a cell-cell contact-dependent manner. In contrast, adaptive Treg are not naturally suppressive but acquire the function through IL-10 and/or TGF-β induction and depend on these cytokines for their suppressive activity. Interestingly, most in vivo evidence supports a requirement for suppressive cytokines in the function of both populations of Treg. While the adaptive Treg can be generated in vitro with various treatments from either CD25+ or CD25 cells (reviewed in 3941), how these cells arise in vivo is not clear. Our data suggest that adaptive Treg can derive from natural Treg in the absence of a CTLA-4 signal. Consistent with this idea, we have observed that chronic in vitro stimulation of WT Treg with anti-CD3 and anti-CD28 in the absence of APC (therefore no B7 to engage CTLA-4) leads to high levels of IL-10 and increased TGF-β expression (data not shown). These results suggest that CTLA-4 expression in Treg controls the functional differentiation of Treg in vivo. We and others have recently reported that Treg proliferate vigorously in vivo in normal hosts, which is likely due to their higher sensitivity to self antigens 24, 42. It is possible that CTLA-4 controls the functional differentiation of Treg during this homeostatic process to prevent the emergence of adaptive Treg and nonspecific immunosuppression. In the setting of autoimmunity and transplant rejection, chronic and strong activation of Treg may override CTLA-4 function, thus permitting their differentiation into adaptive Treg. Although we did not detect LAP or TGF-β expression on CD4+CD62L+CD25+ WT Treg even after activation, the expression of LAP was readily seen on the CD62LCD25+CD4+ cells in WT mice (data not shown). It has been suggested that Treg lose CD62L expression after extensive in vivo activation and proliferation 42. Therefore, it is possible that these cells represent chronically activated adaptive Treg. Our observations may also help to reconcile the difference between our finding that CD4+CD62L+CD25+ Treg do not express TGF-β and previous reports that showed TGF-β expression by CD4+CD25+ Treg 26, 43. It is conceivable that the TGF-β-expressing cells detected in WT mice in previous studies were mostly CD62L Treg, and the prevalence of this population may vary greatly depending on the age, strain, and antigen experience of the mice used in the studies.

Although CTLA-4–/– Treg expressed both IL-10 and TGF-β upon in vitro reactivation, neutralizing TGF-β alone abrogated the suppression, whereas IL-10 neutralization had no effect. It was suggested in a chemical-induced colitis model that IL-10 facilitates the differentiation of TGF-β-producing cells but has little effect itself during the effector phase of the suppression 44. Indeed, it has been shown that Th1 cytokines suppress TGF-β expression by naive T cells 45. We recently demonstrated that naive CTLA-4–/– T cells skew to a Th2 type upon activation, which is likely due to a higher TCR signaling strength in the absence of CTLA-4 46. CTLA-4-deficient cells produce very little IFNγ and high level of the Th2 cytokines IL-4 and IL-5, which may indirectly favor the generation of IL-10- and TGF-β-expressing cells. This result also suggests that TGF-β expression can be induced independently of CTLA-4 and that the lymphoproliferation in CTLA-4–/– mice is not due to the lack of TGF-β expression in the absence of CTLA-4 signaling as previously proposed 25.

In summary, the results presented in this study support a role of CTLA-4 in both the function and the development of Treg. CTLA-4 is functional during the effector phase of Treg activity and controls the development IL-10- and TGF-β-producing adaptive Treg. We propose that CTLA-4 and TGF-β maintain tolerance by controlling T cell activation at two distinct stages. Future experiments will be focused on identifying the contributions of CTLA-4 and TGF-β in in vivo suppression by Treg. Understanding the actions of these molecules may help us to design more effective therapeutics for autoimmune disease and transplantation.

4 Materials and methods

4.1 Mice

WT (CTLA-4+/+), CTLA-4–/–, and CTLA-4+/– mice on the C57BL/6 background were generated by breeding CTLA-4+/– mice. Genotypes of the progeny were determined by PCR within the first week after birth. CTLA-4–/– mice were treated with a mixture of anti-B7.1 and anti-B7.2 mAb (100 μg each) by i.p. injection on days 8, 11, and 14 after birth to delay the generalized lymphoproliferation. In some experiments, WT, CTLA-4+/–, or CTLA-4–/– mice carrying a CTLA-4Ig transgene under the control of the keratin-14 promoter were used 18, as the expression of CTLA-4Ig in CTLA-4–/– mice delays the onset of lymphoproliferation until 6 to 8 weeks of age. These mice are on a mix of C57BL/6 and 129 backgrounds and were used for experiments between 4 and 6 weeks of age. All the mice were housed in the specific pathogen-free animal facilities at UCSF.

4.2 Antibodies and other reagents

Anti-B7.1 (16–10A1), anti-B7.2 (GL-1), anti-CD3 (145–2C11), anti-CD28 (PV-1), and anti-CTLA-4 (4F10) mAb were purified from hybridoma culture supernatants in our laboratory. Purified anti-TGF-β (2G7) mAb was a kind gift from Dr. Chatenoud (INSERM, France). Purified anti-IL-10 mAb (JES5–16E3) was purchased from BD-PharMingen (San Diego, CA). FITC-, RPE-, and SpectroRed-conjugated anti-CD25, anti-CD4, and anti-CD62L mAb were purchased from Southern Biotechnology Associates (Birmingham, AL). Allophycocyanin (APC)-conjugated anti-CD62L, anti-CD4, and streptavidin were purchased from BD-PharMingen. Biotinylated rabbit anti-human TGF-β and LAP antibodies were purchased from R&D Systems (Minneapolis, MN). Anti-FcR (2.4G2) mAb was used as a culture supernatant to block FcR-mediated binding before the staining of cells for analysis by flow cytometry or FACS.

4.3 Purification of Treg using FACS

Splenocytes were enriched for CD4+ cells by negative selection on AutoMACS (Miltenyi, Germany). The CD4-enriched splenocytes were then combined with LN cells, incubated with anti-FcR culture supernatant for 5 min, and stained with fluorochrome-conjugated anti-CD25, anti-CD4, and anti-CD62L mAb. CD4+CD25+CD62L+ Treg were separated on a Moflo cell sorter (Cytomation, Fort Collins, CO). The sorted Treg were between 92% and 98% pure.

4.4 Suppression assay

Assays were set up in 96-well U-bottom plates with 2×104 to 5×104 CD4+CD62L+CD25 cells, equal numbers of T-depleted irradiated splenocytes, graded numbers of Treg, and anti-CD3 mAb (3 μg/ml). In some experiments, Fab fragments of anti-CTLA-4 mAb (100 μg/ml), anti-TGF-β mAb (50 μg/ml), or anti-IL-10 (20 μg/ml) were added to the wells at the beginning of the culture. Proliferation was quantified by [3H]-thymidine incorporation. The cultures were incubated at 37°C for 64–72 h and pulsed with [3H]-thymidine (Perkin Elmer Biosciences, Shelton, CT) during the last 6–14 h.

4.5 CFSE labeling and proliferation assay

Sorter-purified T cells were washed twice in PBS and then labeled with 2.5 μM CFSE for 5 min and quenched with heat-inactivated FCS. Cells were washed twice in culture medium before they were plated for the suppression assay as described above. Proliferation of the labeled cells was determined at 72 h after stimulation using flow cytometry.

4.6 Analysis of membrane TGF-β expression by flow cytometry

The level of membrane-bound TGF-β or LAP expression on freshly isolated T cells was determined using unfractionated whole LN cells that were incubated with anti-FcR mAb and then biotinylated anti-TGF-β or anti-LAP antibodies on ice for 30 min. The cells were washed before labeling with fluorochrome-conjugated anti-CD4, anti-CD25, anti-CD62L, and streptavidin. Expression of these markers was determined using a FACSCalibur (Becton Dickenson, San Jose, CA). To assess the expression of TGF-β and LAP on activated Treg and CD4+CD62L+CD25 cells, the cells were purified first on a Moflo and then activated with plate-bound anti-CD3 (10 μg/ml), soluble anti-CD28 (2 μg/ml), and 100 U/ml IL-2 as described 26. After 24 h, the cells were harvested and stained as described above for fresh cells.


The levels of IL-10 and TGF-β in culture supernatants were determined by ELISA using antibody pairs purchased from BD-PharMingen. For the TGF-β ELISA, the culture supernatant was first treated with acid to lower the pH to 2.0, which denatures LAP and allows the detection of active TGF-β. The supernatant was then brought back to neutral pH before the ELISA.

4.8 Real-time RT-PCR

Total RNA was extracted from cell sorter-purified Treg using RNAeasy columns (Qiagen, Valencia, CA). The cDNA was synthesized from 0.5 μg total RNA using Superscript II RNase H-reverse transcriptase and oligo dT as primers (Invitrogen, Carlsbad, CA). Primer and probe pairs for IL-2, IL-10, TGF-β1, and HPRT were purchased from Applied Biosystems (Foster City, CA). The real-time PCR was performed in duplicate on an ABI 7700 using TaqMan Universal PCR master mix (Applied Biosystems), and the average threshold cycles (Ct) of the duplicates were used to compare the relative abundance of the mRNA. Ct of HPRT was used to normalize all samples.

4.9 Western blot

Treg and CD4+CD62L+CD25 cells were purified using a Moflo cell sorter. Cells (2×105) of each type were lysed in sample buffer (62.5 mM Tris pH 6.8, 12.5% glycerol, 2% SDS, 30 ng/ml bromophenoblue), sonicated, and passed through 28-gauge needles. The lysates were clarified by centrifugation and boiled for 5 min before separation on a 10% SDS-PAGE gel. The samples were transferred to a PVDF membrane after electrophoresis and incubated with rabbit anti-FoxP3 antiserum (gift from Dr. Steven Ziegler) followed by HRP-conjugated anti-rabbit Ig. The blot was developed with SuperSignal® Chemiluminescent Substrate (Pierce, Rockford, IL), visualized on a Kodak Image Station 440CF (Eastman Kodak), and quantified using Kodak Digital Science 1D Image Analysis software 3.0.


The authors would like to thank Shuwei Jiang and Cliff McArthur for technical help with cell sorting, Paul Wegfahrt for mouse handling, and Derek Park and Dr. Jianqin Ye for technical assistance. This work is supported by NIH grants AI466430 (J. A. B.) and F32 AI10360 (Q. T.). E. K .B. is a Howard Hughes Medical Institute medical student fellow.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5