Journal of Clinical Immunology

, Volume 28, Issue 6, pp 640–646

Role of TGF-β in the Induction of Foxp3 Expression and T Regulatory Cell Function


    • Laboratory of ImmunologyNational Institute of Allergy and Infectious Diseases, National Institutes of Heath
  • Todd S. Davidson
    • Laboratory of ImmunologyNational Institute of Allergy and Infectious Diseases, National Institutes of Heath
  • Eva N. Huter
    • Laboratory of ImmunologyNational Institute of Allergy and Infectious Diseases, National Institutes of Heath
  • Richard A. DiPaolo
    • Laboratory of ImmunologyNational Institute of Allergy and Infectious Diseases, National Institutes of Heath
  • John Andersson
    • Laboratory of ImmunologyNational Institute of Allergy and Infectious Diseases, National Institutes of Heath

DOI: 10.1007/s10875-008-9240-1

Cite this article as:
Shevach, E.M., Davidson, T.S., Huter, E.N. et al. J Clin Immunol (2008) 28: 640. doi:10.1007/s10875-008-9240-1



A number of studies have suggested that transforming growth factor beta (TGF-β) plays a critical role in immune suppression mediated by Foxp3+ regulatory T cells. TGF-β in concert with interleukin 2 is a potent induction factor for the differentiation of Foxp3+ Treg from naive precursors. Polyclonal TGF-β-induced Treg (iTreg) are capable of preventing the autoimmune syndrome that develops in scurfy mice that lack Foxp3+ Treg. Antigen-specific iTreg can be used to both prevent and treat autoimmune gastritis that is induced by transfer of naive or primed autoantigen-specific T cells. TGF-β complexed with latency-associated peptide is expressed on the surface of activated thymus-derived Treg. Coculture of activated Treg with naive responder T cells results in the de novo generation of fully functional Foxp3+ T cells in a contact-dependent and TGF-β-dependent manner.

Conclusions and Speculations

Generation of functional Foxp3+ T cells via this pathway may represent a mechanism by which Treg maintain tolerance and expand their repertoire.


TGF-βgastritisinfectious toleranceimmune suppressionlatency-associated peptide


Transforming growth factor beta (TGF-β) plays an important role in maintaining immune homeostasis, in general, and in regulating T cells, in particular. Global disruption of the TGF-β gene in mice results in a spontaneous autoimmune syndrome that can be transferred to lethally irradiated wild-type recipients with bone marrow from TGF-β deficient (−/−) mice suggesting an important role for autocrine production of TGF-β in immune cells. T cell-specific disruption of TGF-β signaling by the generation of a T cell-specific dominant negative (DN) TGF-β-type RII transgene results a similar, but less aggressive, disease characterized by spontaneous activation of T cells, lymphocyte infiltration into multiple organs, and the production of autoantibodies. Specific intracellular intermediates that regulate TGF-β responses have also been globally disrupted in mice including genes encoding Smad2 and Smad3. All these models support the importance of TGF-β in maintaining immune homeostasis. Foxp3+ T regulatory cells (Treg) are the major population of Treg controlling all aspects of the immune response, and deficiency of Foxp3+ T cells results in the development of a rapidly lethal autoimmune syndrome in mice and man. Thus, it is appropriate to raise the possibility that TGF-β may play a critical role in immune suppression mediated by Foxp3+ Treg.

Does TGF-β Mediate the Suppressor Function of Foxp3+ T Cells In Vitro?

Considerable controversy exists on the role that TGF-β plays in the induction and execution of Foxp3+ Treg-mediated suppressor function in vitro. Most studies have concluded that suppression is not mediated by secreted TGF-β, as suppression requires cell contact between the suppressor and the responder and early studies claimed that suppression could not be abrogated with anti-TGF-β in murine, as well as human cocultures in vitro. However, Nakamura et al. [1] proposed a novel model in which Foxp3+ T cells express cell surface latent TGF-β and mediate suppression via a cell contact-dependent presentation of latent TGF-β to a TGF-β receptor on target Foxp3 T cells. This conclusion was based on the ability to detect expression of cell surface TGF-β by flow cytometry and to completely abrogate suppression with high concentrations of anti-TGF-β. Cell surface TGF-β was not expressed on resting Treg, but could be readily detected on the surface of activated Treg. These investigators also demonstrated that TGF-β in its inactive form complexed to latency-associated peptide (LAP) could also be detected on the surface of activated Treg and that activation only occurred upon cell contact of the Treg with the responder cells. Such a process of activation would be rapid and could only be blocked by much higher concentrations of anti-TGF-β than those used in the earlier studies. The biochemical mechanisms by which TGF-β was maintained on the cell surface and subsequently activated were not elucidated in these studies. One unique observation in this study was that Treg suppressed B cell immunoglobulin production and that anti-TGF-β could abolish this suppression.

In contrast to the experiments of Nakamura et al., we could not identify a role for TGF-β in Foxp3+ Treg-mediated suppression in vitro [2]. We made use of several genetic model systems in which Foxp3+ T cell suppressor function could be assessed in vitro in the complete absence of Foxp3+ T cell-derived TGF-β production and target cell responsiveness to TGF-β (Fig. 1). We did not observe any effects of even high concentrations of anti-TGF-β or a soluble TGF-βRII-Fc on the immunosuppressive effects of activated CD4+CD25+ T cells. We have also been unable to consistently observe significant differences in the expression of cell surface TGF-β1 on activated CD25+ cells when compared to similarly activated CD25- T cells. Both cell populations stained very weakly. We also used CD25+ T cells that could not produce TGF-β1 and CD25 T cells that could not respond to TGF-β-induced growth arrest. Activation of CD25 T cells from the Smad3−/− and DNRIITg mice was completely resistant to the immunosuppressive effects of TGF-β, but were readily suppressed by CD4+CD25+ T cells from wild-type mice. CD4+CD25+ T cells from TGF-β −/− mice were as suppressive as CD25+ T cells from wild-type mice when mixed with CD25 T cell from wild-type mice. Collectively, these studies rule out the possibility that production of or responsiveness to TGF-β was absolutely required for Foxp3+ T cell-mediated suppression in vitro.
Fig. 1

Treg from TGF-β1 −/− mice are fully competent to suppress the proliferative responses of wild-type CD4+CD25 T cells. Wild-type CD4+CD25 T cells were stimulated for 72 h with soluble anti-CD3 in the presence of T-depleted spleen cells. CD4+CD25 or CD4+CD25+ T cells from wild type or TGF-β1 −/− mice were added as indicated

Does TGF-β Mediate the Suppressive Functions of Foxp3+ T Cells In Vivo?

Interpretation of the effects of anti-cytokine reagents on the reversal of Treg-mediated suppression in vivo is difficult. As an alternative, several groups have evaluated the suppressive capacity of Treg from TGF-β −/− mice to function in vivo. Mamura et al. [3] have reported that CD4+CD25+Treg cells were present in normal numbers in TGF-β −/− mice and expressed all the characteristic Treg cell-specific marker antigens (CTLA-4, GITR, Foxp3). Transfer of splenocytes from TGF-β −/− mice to RAG −/− recipients resulted in a global autoimmune disease similar to that observed in TGF-β −/− mice. Cotransfer of Treg cells from TGF-β −/− mice attenuated disease in RAG −/− recipients of CD4+CD25+-depleted spleen and lymph nodes cells from TGF-β −/− mice. Taken together, these studies demonstrate that Treg cells develop normally in TGF-β −/− mice and that Treg production of TGF-β is not required for these cells to suppress inflammation in this model in vivo.

It was first reported that suppression of inflammatory bowel disease (IBD) induced by transfer of CD4+CD45RBhigh cells into severe combined immunodeficiency recipients by cotransfer of CD4+CD45RBlow T cells could be reversed by treatment of the recipients with anti-TGF-β [4]. Although CD45RBlow cells are highly enriched in Foxp3+Treg, they also contain Foxp3 T cells derived from the memory/effector cell pool. The cellular source of the TGF-β that mediates control of IBD in this model is unknown. TGF-β could be produced by the Treg, the T effector cells, or even by host lymphoid or non-lymphoid cells, such as healing gut epithelium. We failed to reverse suppression of autoimmune gastritis (AIG) induced by transfer of CD4+CD25 T cells by cotransfer of CD4+CD25 T cells into nu/nu recipients by an anti-TGF-β regimen similar to the one used to reverse Treg suppression of IBD [2]. However, IBD and AIG differ in their pathogenesis, as intestinal bacteria are essential for disease induction of the former, but play no role in the latter. CD4+CD25+ Treg cell-mediated control of CD8+ T cell-mediated tumor immunity also appears to require the action of TGF-β, but the cellular source of the activated TGF-β was not determined [5]. Similar results were observed in a model of type I diabetes induced by CD8+ T cells [6].

The capacity of Treg from TGF-β −/− mice to suppress IBD has also been evaluated. Nakamura et al. [7] were reported that the weight loss induced by transfer of CD4+CD45RBhigh T cells to RAG −/− mice could be prevented by cotransfer of CD4+CD45RBlow cells or CD4+CD25+ T cells from wild-type mice, but not by CD4+CD25+ T cells from TGF-β −/− mice. These results were confirmed in histologic studies. Although these results support the view that the Treg are the cell type that secretes the TGF-β, a major difficulty with the interpretation of these experiments was that the CD4+CD25+ T cells alone were not tested for their colitogenic potential and it remains possible that they were actually activated effector cells rather than Treg. CD4+LAP+ (CD25+ or CD25), but not CD4+LAP T cells from normal mice have also been shown to prevent colitis in the transfer model [7].

We have been unable to confirm these in vivo studies using a similar, but more extensive, genetic approach. CD4+CD25+ Treg purified from neonatal TGF-β −/− mice were almost as efficient as wild-type Treg in their capacity to suppress IBD as measured by clinical disease, histology score, and production of interferon gamma (IFN-γ) in the colon [8]. Anti-TGF-β enhanced the colitogenic potential of effector cells in the absence of CD4+CD25+ Treg suggesting a role for non-Treg-derived TGF-β in the control of IBD. To rule out the possibility that the TGF-β was acting on the Treg cells themselves, we demonstrated that Treg from Smad3 −/− mice were as suppressive as Treg from wild-type mice. In addition, TGF-β did not appear to act on the effector cells, as T effectors from Smad3 −/− mice were as colitogenic as T effectors from wild-type mice and both were susceptible to inhibition by Treg from wild type or Smad3−/− mice. Thus, Treg can mediate suppression of IBD in the complete absence of Treg-produced TGF-β and functional Smad3 responsiveness in either the Treg or the effector cells. We did observe a very slight increase in inflammation in groups receiving Treg alone from the TGF-β −/− mice. Although this low level of disease may have been because of contaminating effector T cells, it remains possible that autocrine production of TGF-β and its action on Treg themselves may be needed for their optimal survival and function in vivo [9].

Although these studies raise the possibility that TGF-β is not involved in protection from IBD, Fahlen et al. [10] demonstrated that CD4+CD45RBhigh cells that express the DN TGF-βRII and cannot respond to TGF-β and cannot be controlled by Treg cells. Treg cells from the thymus of these mice can completely inhibit the development of colitis indicating that T cell responsiveness to TGF-β is not required for the function of Treg. In agreement with the studies of Kulberg et al. [8], Treg cells from TGF-β −/− mice were able to prevent colitis with an efficiency equal to that of Treg cells from wild-type mice. Most importantly, anti-TGF-β treatment of recipients that received either wild type or TGF-β −/− Treg abrogated suppression of disease. These studies clearly demonstrate that TGF-β plays a critical role in Treg-mediated suppression of colitis, but it need not be produced by the Treg themselves. It remains possible that the Treg cells induce production of TGF-β by other lymphoid or non-lymphoid cell types.

Does TGF-β Act on Treg and/or on Conventional T Cells?

The studies described above focus on the effects of TGF-β as a Treg-produced suppressor effector molecule. A number of studies have suggested that TGF-β may also have effects on the Foxp3+ Treg themselves. Constitutive expression of TGF-β in pancreatic islets prevents the development of diabetes in non-obese diabetic mice, but leads to fibrosis of the islets. When mice with an inducible TGF-β gene were studied, a short pulse of TGF-β in the islets either during the priming or the effector phase of the disease protected against the development of disease [11]. It appeared that TGF-β inhibited the development of anti-islet specific effector T cells, but had no effect on antigen presenting cell function or on Th1/Th2 differentiation. Exposure to TGF-β for the first 8 weeks of life dramatically enhanced the frequency of Treg among the total islet cell CD4+ T cells. The expanded Treg population could prevent the adoptive transfer of diabetes. If TGF-β was turned on after 8 weeks of age, the accumulation of CD4+CD25+ T cells was not observed, but protection from diabetes was still seen. It appears that TGF-β is a positive regulator of the Treg pool in vivo. One likely possibility is that TGF-β converts peripheral non-Treg to Foxp3+ Treg.

Mice with a T cell-specific deletion of the TGF-βRII have been generated and develop a severe, lethal autoimmune syndrome [12]. The pathology in these mice was mediated by CD4+ and CD8+ T cells that produce high amounts of IFN-γ and Fas ligand. These cells produced pathology even in the presence of wild-type Treg cells. It remains possible that the inflammatory environment in these mice inactivates Treg, but in mixed bone marrow chimeras, wild-type Treg controlled wild-type T cells, but not T cells that lacked the TGF-βRII. It appears that the disease in the TGF-βRII −/− mice is not secondary to a failure of Treg-produced TGF-β to control effector cells, but to a cell-intrinsic mechanism of TGF-β-mediated control of T cell reactivity.

TGF-β and IL-2 are Essential for Induction of Foxp3+ Treg In Vitro

Stimulation of naive T cells in the presence of TGF-β results in the induction of Foxp3 expression (Fig. 2) and T regulatory activity in vitro and in vivo [13]. IL-2 plays a nonredundant role in TGF-β-induced Foxp3 expression, as other common γ-chain-utilizing cytokines were unable to induce Foxp3 in T cells from IL-2 −/− mice [14]. Foxp3 expression was stable in vitro and in vivo in the absence of IL-2. TGF-β-induced Treg (iTreg) resemble thymic-derived Treg, as they are anergic and suppressive and produce very low levels of effector cytokines. Some studies have suggested that iTreg lack regulatory function, even though they express Foxp3 and they rapidly lose Foxp3 expression after transfer in vivo. We tested whether iTreg would also be effective in rescuing Scurfy mice [15]. We found that polyclonal iTreg generated from peripheral CD4+ Foxp3 T cells could completely suppress all the pathologic manifestations of the severe autoimmune disease that develops in scurfy mice in both lymphoid sites and in tissues. Thus, the T cell receptor (TCR) repertoire of the iTreg generated from normal adult mice appears to be sufficient to prevent autoimmune disease in the scurfy mouse. The iTreg maintained high levels of Foxp3 expression in the inflammatory environment, whereas the same cells, when transferred to normal mice, lost expression of Foxp3, suggesting that certain factors, such as proinflammatory cytokines in the sick scurfy mouse, might act as survival or expansion signals for the iTreg. When naive T cells specific for the gastric parietal cell autoantigen, H+/K+ adenosine triphosphatase, were converted to iTreg, they maintained Foxp3 expression for long periods of time in vivo and inhibited the development of AIG. The inhibitory effect of the antigen-specific iTreg in this model was directed against autoantigen presenting dendritic cells (DC), as DC exposed to iTreg in vivo were reduced in their ability to present antigen to an autoantigen-specific T cell line [16].
Fig. 2

TGF-β is a potent differentiation factor for the induction of Foxp3+ Treg. CD4+Foxp3 T cells from a TCR transgenic mouse on a RAG −/− background were stimulated with plate-bound anti-CD3 and IL-2 in the absence (left panel) or in the presence of TGF-β. Foxp3 expression was detected by intracellular staining after 96 h of culture

TGF-β Production by T Cells is Critical for Maintenance of Peripheral Tolerance

The Cre-loxP system has been used to specifically inactivate TGF-β in peripheral CD4+ and CD8+ T cells [17]. These mice were healthy until 4 months of age when they developed wasting disease and diarrhea with severe, lethal IBD developing at 6 months of age. T cell-produced TGF-β was dispensable for the maintenance of peripheral Treg cells, but was essential for Treg-mediated suppression of IBD. TGF-β −/− Treg cells exhibited normal suppressive activity in vitro. Treg cell maintenance is likely to depend on other cellular sources of TGF-β. TGF-β −/− Treg cells were defective in inhibiting colitis induced by transfer of naive T cells. The differences between these results and some of the results seen with Treg from mice with a global deficiency of TGF-β can best be explained by the different gut flora present in the different institutions. Mice that completely lack Treg cells such as the scurfy mouse have a much more severe phenotype than mice with a selective deficiency of TGF-β in T cells. Thus, secretion of TGF-β is only one mechanism by which Treg can mediate their suppressive effects. One other important observation in this report was that T cell produced TGF-β was essential for the differentiation of adjuvant-induced Th17 cells and the development of experimental allergic encephalomyelitis.

An autoimmune phenotype very similar to that seen in mice with a selective deficiency of TGF-β production in T cells was observed in mice with a T cell conditional deletion of furin, the proprotein convertase, that is required for processing of TGF-β and the generation of active TGF-β [18]. Furin −/− Treg cells were markedly impaired in their ability to prevent gut pathology and weight loss induced by transfer of naive CD4+ wild-type T cells to RAG −/− mice. Thus, furin plays a nonredundant role in controlling the levels of bioavailable active TGF-β.

Foxp3+ Treg Confer Infectious Tolerance in a TGF-β-Dependent Manner

Infectious tolerance refers to a process where a tolerance-inducing state is transferred from one cell population to another [19]. The demonstration that TGF-β is a potent inducer of Foxp3 expression and functional Treg cells together with studies demonstrating that Treg production of TGF-β is required for protection against autoimmune disease under certain conditions raised the possibility that TGF-β produced by Treg might be capable of inducing Treg from naive precursors thus generating infectious tolerance [20]. Freshly explanted Treg did not express LAP on their cell surface, but after activation via their TCR, almost all Foxp3+ T cells became LAP positive. We could not identify free TGFβ on the surface of these cells. As activated Treg express LAP on their cell surface, we tested whether Treg could induce Foxp3 expression in Foxp3 T cells. Activated Foxp3+ T cells were cocultured with naive Foxp3 T cells in the presence of splenic DC, anti-CD3, and IL-2. Following this co-culture, 10–30% of the responder cells expressed Foxp3. Cell contact was required for induction of Foxp3 expression and induction of Foxp3 was inhibited by the addition of recombinant LAP (Fig. 3). Treg cells from TGF-β −/− mice or from the furin T cell conditional −/− mice were incapable of inducing Foxp3 expression in responder T cells, indicating that the Treg are the source of the TGF-β. Responder cells that have been induced to express Foxp3 were isolated from the cocultures and were shown to be anergic and suppressive in vitro. More importantly, the converted cells were capable of inhibiting the development of colitis when cotransferred with CD4+Foxp3 T cells to RAG −/− mice. Antigen-specific activated Foxp3+ T cells were also capable of inducing Foxp3 expression in naive antigen-specific T cells when transferred into normal mice followed by immunization with the cognate antigen. Thus, Treg-mediated infectious tolerance can occur in a physiological setting in vivo.
Fig. 3

Treg can induce Treg in a TGFβ-dependent manner. In a first culture, purified CD4+CD25 or CD4+CD25+ T cells from normal mice were activated for 72 h by stimulation with plate-bound anti-CD3 and IL-2. These cells were then washed and mixed with freshly explanted CFSE-labeled CD4+Foxp3 T cells and then stimulated with plate-bound anti-CD3 and IL-2 for 96 h. Foxp3 expression of the CFSE-labeled cells was then assayed by intracellular staining. In one set of cultures, the activated CD4+CD25+ T cells were separated from the CD4+Foxp3 T cells by a transwell, and in another culture (far right panel), recombinant LAP was added to the coculture of the activated CD4+CD25+ T cells and the CD4+ Foxp3 responders

Conclusions and Speculations

The relationship between TGF-β and Foxp3+ Treg development and function are complex. Although several earlier studies raised the possibility that TGF-β secretion contributed significantly to the in vitro suppressive capacity of Treg, more recent studies using Treg from mice with a global deficiency of TGF-β or from mice with a T cell-specific deletion of TGF-β have shown that TGF-β production by Treg plays only a minor, if any, role in the in vitro suppressive function of Treg. The role of TGF-β in protection of mice from IBD induced by transfer of naive T cells to RAG −/− recipients is clear. However, considerable controversy still remains as to the source of the TGF-β. Several studies have shown that Treg cells from mice with a global deficiency of TGF-β can protect from IBD, but do so in the TGF-β-dependent manner. The nature of the non-Treg TGF-β-producing cell type has not been defined. However, it is also clear that, under certain circumstances, Treg production of TGF-β may be required to protect against IBD. Differences in the flora in the animal colonies in which these studies have been performed may account for some of the differences in the experimental findings. Colonies with more “pathogenic” flora and a greater degree of inflammatory response may require that Treg in the intestinal tract themselves produce the TGF-β in addition to their use of other suppressor mechanisms.

It is also difficult to integrate the immunosuppressive function of TGF-β with its other major role as the major cytokine involved in the induction of Foxp3+ Treg in peripheral lymphoid tissues, at least in the mouse [21]. Our studies on the ability of Foxp3+ Treg to induce Foxp3+ Treg de novo from naive precursors also raise the possibility that one of the major functions of Treg-produced TGF-β is not related to its immunosuppressive capacity but to its capacity to induce the differentiation of Treg. One possibility is that Treg-mediated induction of Foxp3-expressing Treg is most prominent at sites of inflammation such as the gastrointestinal tract or at the site of allograft rejection [22]. Simultaneous stimulation of antigen-specific Treg and auto- or alloantigen-specific T cells might occur on the surface of the same DC in the form of a three-cell interaction (Fig. 4). As a single DC might be capable of presenting several organ- or graft-derived antigens, it is possible that the Treg-TGF-β pathway of Treg induction would result in expansion of the Treg repertoire.
Fig. 4

Proposed model for infectious tolerance mediated by activated Foxp3+ Treg cells. During the course of an inflammatory response at the site of an autoimmune disease or at the site of an allograft, Foxp3+ Treg become activated by recognition of their cognate peptide major histocompatibility complex (MHC) class II antigen complex presented by a DC and then express LAP-TGF-β on their cell surface. CD4+Foxp3 effector cells are also activated on the surface of the same DC by recognition of either the same or a different MHC class II peptide complex. During the course of this three-cell interaction, active TGF-β is generated and leads to conversion of the CD4+Foxp3 T cells to CD4+Foxp3+ Treg resulting in expansion of the number of Foxp3+ Treg and potentially in the broadening of the Foxp3+ Treg repertoire, if the two CD4+ T cells recognize different peptide MHC class II antigen complexes

Copyright information

© Springer Science + Business Media, LLC 2008