Main

Immune tolerance towards self antigens is dependent on the ability of the immune system to discriminate between self and non-self2,3. This occurs mainly through clonal deletion in the thymus of self-reactive T lymphocytes at an early stage of development. However, the immune system is also exposed to extrathymic self antigens and to repetitive stimulation by non-pathogenic antigens through inhalation and ingestion of foreign substances. To avoid chronic cell activation and inflammation, the immune system must develop unresponsiveness to such stimuli. Several mechanisms of peripheral tolerance have been proposed, including T-cell anergy4,5, T-cell deletion2, and active immune suppression1. Understanding these mechanisms of tolerance induction is clinically relevant for the treatment of autoimmune diseases and in transplantation, where the graft must ultimately be recognized as self.

The induction of anergy5 and cell deletion by apoptosis2 have been documented both in vitro and in vivo, but the analysis of active immune suppression mediated by T cells has been hampered by the inability to generate and clone these cells in vitro . The importance of cytokines in the development of specialized Th cells is now clear. Th1 cells are induced by activation in the presence of IL-12, whereas IL-4 drives the differentiation of Th2 cells6. Here we show that repetitive stimulation of CD4+T cells in the presence of IL-10 induces the differentiation of a unique subset of T cells with immunoregulatory properties.

Ovalbumin (OVA)-specific naive CD4+T cells obtained from the αβ T-cell antigen receptor (TCR) DO11-10 transgenic mice7 repeatedly stimulated with splenic antigen-presenting cells (APCs) and OVA peptide in the presence of IL-10, or IL-4 and IL-10, displayed a cytokine profile distinct from that of the classical Th0, Th1 or Th2 phenotype8. These T cells produce high levels of IL-10 and IL-5 and low levels of IL-2 and IL-4 (Fig. 1). Another characteristic of these T-cell populations is that they proliferated poorly in response to antigenic stimulation (Fig. 2a). In contrast, CD4+T cells isolated from OVA-specific TCR transgenic mice repeatedly stimulated with OVA peptide in the presence of IL-4 alone displayed the typical Th2-type profile, secreting IL-4, IL-5 and IL-10 (Fig. 1)9.

Figure 1: Generation of murine Tr1 cells in the presence of IL-10.
figure 1

Naive (MEL14 bright) CD4+T cells from BALB/c mice transgenic for the DO11.10 αβ TCR7 were cultured with OVA peptide (0.6 µM) and irradiated splenic APCs in the presence or absence of: IL-4 (200 U ml−1), IL-10 (100 U ml−1), or both. The stimulation, under the same conditions, was repeated weekly for three consecutive weeks9. Cells were collected, washed and restimulated with immobilized anti-CD3 (10 µg ml−1) and anti-CD28 mAb (10 µg ml−1) for 4 h at 37 °C. Brefeldin A (10 µg ml−1) was added for the last 2 h of culture29. The cells were fixed and stained for detection of intracellular cytokines using FITC or PE-conjugated monoclonal antibodies specific for IFN-γ, IL-2, IL-4, IL-5 and IL-10, or an isotype-matched control mAb (isotype CT), as indicated. Cells were also stained with the anti-TCR clonotype-specific mAb KJ1-26 (TCR) to show that all cells were positive for the transgene after 3 weeks of culture. One of three experiments is shown.

Full size image
Figure 2: Proliferative responses of Tr1 cells.
figure 2

a, Naive (MEL14 bright) CD14+T cells from BALB/c mice transgenic for the DO11.10 αβ TCR7 were differentiated (Diff) in vitro as described in Fig. 1 and restimulated with OVA peptide (0.6 µM) and irradiated splenic APCs for 3 days. One OVA-specific Th0 clone (A-7) and two OVA-specific Tr1 type clones (A-10-9) and A-10-11) were stimulated using the same conditions. b, The human Th1 (JDV 305) and Tr1 (JDV24) alloantigen-specific CD4+T-cell clones or the non-alloantigen-specific Tr1ell clones (JDV15 and JDV 308) were stimulated with either allogeneic irradiated monocytes for 5 days or with immobilized anti-CD3 (10 µg ml−1) and anti-CD28 mAb (1 µg ml−1) for 3 days. The proliferative responses were measured by 3[H]-TdR incorporation during the last 12 h of culture. Cultures of human or mouse T cells were carried out in the presence of an isotype-matched control mAb (GL113, rat IgG1 at 20 µg ml−1+ mouse IgG1 from Caltag (Burlingam, CA) at 20 µg ml−1; black bars); a neutralizing anti-IL-10 mAb (12G8 at 5 µg ml−1for human T cells or JES5-2A5 at 10 µg ml−1for mouse T cells; white hatched bars); a neutralizing anti-TGF-β (from Genzyme (Cambridge, MA) at 10 µg ml−1; black hatched bars); a combination of the two antibodies (white bars); or a neutralizing anti-IL-2 mAb (BG-5, 10 µg ml−1for human cells; grey bars). A representative Th1 (JDV 305) and a representative Tr1(JDV24) are shown for antigen-specific response. One of four experiments is shown.

Full size image

Analysis of T-cell clones isolated from the mouse CD4+T-cell populations that were repeatedly stimulated with OVA peptide and splenic APCs in the presence of IL-10 showed that half of the CD4+T-cell clones obtained displayed this cytokine profile. They produced high levels of IL-10 and undetectable levels of IL-2 and IL-4, whereas the levels of production of interferon (IFN)-γ and transforming growth factor (TGF)-β were comparable with those of Th0 and Th1 clones, respectively (Table 1). These T-cell clones also proliferated poorly in response to antigen-specific stimulation (Fig. 2a), which may explain the previous failure to isolate these cells. Only Th1, Th2 and Th0 clones were isolated from T-cell populations cultured in the absence of IL-10 (Table 1). Thus chronic activation of mouse CD4+T cells in the presence of IL-10 results in the generation of a T-cell subset with a unique cytokine profile and low proliferative capacity. These cells are designated Tr1 cells on the basis of the their function.

Table 1 Cytokine production by murine OVA-specific CD4+T-cell clones
Full size table

Tr1 cells can also be generated from human peripheral blood. Both alloantigen-specific CD4+T-cell clones (JDV24) and non- alloantigen-specific T-cell clones (JDV15, JDV308 and JDV94) displaying the Tr1 cell cytokine profile were derived from CD4+T-cell populations that had initially been stimulated with allogeneic monocytes in the presence of IL-10 (Table 2). In contrast, only Th1-type clones were obtained when the CD4+T cells were stimulated by allogeneic monocytes in the absence of IL-10. The most striking characteristic of the human Tr1 cell clones was their unusually high level of IL-10 production (Table 2). They also produced very low levels of IL-2, and failed to secrete detectable levels of IL-4, whereas their levels of IL-5, IFN-γ and TGF-β production were similar to those of human Th0 clones. Taken together, these results demonstrate that antigen-specific Tr1 cells can be generated in vitro from naive populations of both human and mouse CD4+T cells.

Table 2 Cytokine production by human Tr1 cells
Full size table

Human Tr1 clones stimulated with either immobilized anti-CD3 and anti-CD28 monoclonal antibodies or allogeneic monocytes (for JDV24) showed low proliferative responses that were sustained by secretion of IL-2, as the addition of an anti-IL-2 monoclonal antibody completely blocked the growth of the Tr1 cells (Fig. 2b). Kinetic studies showed that IL-10 is produced rapidly after activation of the human Tr1 clones. IL-10 is detectable in the supernatants of Tr1 clones 8 h after activation (data not shown), which indicates that IL-10 production by Tr1 clones generally occurs before or at the same time as IL-2 production. In contrast, production of IL-10 by Th0, Th1 and Th2 clones established from the same donor in the same experiment was not detectable until 20 h after activation, which is compatible with previous findings10.

Early endogenous IL-10 production accounts in part for the low proliferative capacity of the human and mouse Tr1 clones in response to TCR/CD3 stimulation (Fig. 2), as proliferation of the Tr1 clones was augmented by the addition of a neutralizing anti-IL-10 monoclonal antibody, which is consistent with previous observations indicating that IL-10 prevents or inhibits T-cell proliferation11. In contrast, anti-IL-10 monoclonal antibody had no effect on the proliferative responses of control Th1, Th2 or Th0 clones (Fig. 2). Proliferation of both human and mouse Tr1 clones was also partly augmented by a neutralizing anti-TGF-β monoclonal antibody (Fig. 2), whereas the proliferation of the control Th clones was unaffected (Fig. 2). The combination of anti-IL-10 and anti-TGF-β antibodies had additive effects and almost completely restored the proliferative responses of Tr1 clones (Fig. 2).

The observation that Tr1 cells secrete high levels of the immunosuppressive cytokine IL-10, and low levels of the T-cell growth-promoting cytokines IL-2 and IL-4, suggested that antigen-specific activation of Tr1 cells may result in inhibition of antigen-specific proliferation of other T cells. Indeed, co-culture experiments using a transwell system confirmed this notion for both human and mouse T cells. Resting human CD4+T cells were stimulated with irradiated allogeneic monocytes in the bottom compartment, whereas alloantigen-specific Tr1 clone JDV24 or non-antigen-specific Tr1 clone JDV308 were stimulated with the same allogeneic monocytes in the top compartment. Alloantigen-specific stimulation of the Tr1 clone JDV24 strongly reduced the alloantigen-specific proliferative responses of autologous human CD4+T cells, whereas the non-alloantigen-specific Tr1 clone JDV308 was ineffective (Fig. 3a). A slight increase in CD4+T-cell proliferation was noticed following co-culture with the alloantigen-specific T-cell clone JDV305, which belongs to the Th1 subset (Fig. 3a).

Figure 3: Tr1 cells inhibit the antigen-specific proliferative responses of naive CD4+T cells.
figure 3

a, Purified human CD4+T cells from donor JDV were stimulated by allogeneic purified irradiated monocytes for 5 days in the bottom compartment of a transwell system. The resting T cells were co-cultured with various autologous T-cell clones contained in the upper transwell compartment; the Tr1 clone JDV 24 and the Th1 clone JDV 305 are alloantigen-specific, whereas the Tr1 clone JDV 308 is non-alloantigen-specific. After 5 days the basket was removed and the proliferative response of the CD4+T cells was analysed by 3[H]-TdR incorporation during the last 12 h of culture. b, Naive (MEL14 bright) CD4+T cells from BALB/c mice transgenic for the DO11.10 αβ TCR were stimulated in the bottom compartment of a transwell system with OVA peptide (0.6 µM) and irradiated APCs for 3 days. These naive T cells were co-cultured with the indicated populations contained in the upper transwell basket: OVA-specific CD4+T-cell populations differentiated (diff) in vitro by three consecutive restimulations in the presence of IL-4 (IL-4 Diff), IL-10 (IL-10 Diff) or IL-4 and IL-10 (IL4/10 Diff), respectively, or OVA-specific CD4+T-cell clones A-10-9, A-10-11 (Tr1 type clones) or A-7 (Th0 type clone). After 3 days the basket was removed and the proliferative response of the naive CD4+T cells was analysed by 3[H]-TdR incorporation during the last 12 h of the culture. Cultures of human and murine cells were carried out in the presence of an isotype-matched control (GL113, rat IgG1 at 20 µg ml−1+ mouse IgG1 from Caltag Laboratories (Burlingame, CA) at 20 µg ml−1; black bars); a blocking anti-IL-10 mAb (12G8 at 5 µg ml−1for human cells or JES5-2A5 at 10 µg ml−1for mouse cells; white hatched bars); a blocking anti-TGF-β (from Genzyme (Cambridge, MA) at 10 µg ml−1; black hatched bars); or a combination of the two antibodies (white bars). Results represent one of three experiments performed.

Full size image

Similar data were obtained with mouse Tr1 clones (Fig. 3b). The proliferation of naive CD4+T cells in response to OVA peptide and splenic APCs was dramatically reduced following co-culture in the transwell system with OVA peptide-activated Tr1 clones or with activated Tr1ell populations generated in the presence of IL-10. In contrast, OVA-specific Th0 clones or Th2-cell populations generated in the presence of IL-4 had no suppressive effects, but rather enhanced OVA-induced proliferation of naive CD4+T cells.

Addition of neutralizing anti-IL-10 and anti-TGF-β antibodies augmented the proliferative responses of naive human and mouse CD4+T cells in the presence of Tr1 cells (Fig. 3). These data demonstrate that IL-10 induces the in vitro differentiation of a new regulatory CD4+T-cell subset, which suppresses antigen-specific T-cell responses in both mice and man. These suppressive activities are predominantly mediated by IL-10 and TGF-β. Moreover, similar to the data previously reported for T cells stimulated in the presence of exogenous IL-10 (ref. 12), naive T cells stimulated in the presence of Tr1 clones fail to proliferate and to secrete IL-2, IL-4, IL-5 and IFN-γ after re-stimulation (not shown).

We sought to establish whether these Tr1 clones were functional in vivo and could regulate a pathogenic T-cell response. IL-10 is important for the maintenance of T-cell tolerance in the gut, as IL-10-deficient mice develop inflammatory bowel disease, which is thought to be mediated by normal enteric antigens13. In addition, administration of IL-10 to young IL-10-deficient mice prevents inflammatory bowel disease14. IL-10 also significantly inhibits the development of inflammatory bowel disease (IBD) induced by the transfer of CD4+CD45RBhiD4+T cells into SCID mice15. Here we show, in this latter model of IBD induction, that transfer in SCID mice of as few as 2 × 105 OVA-specific Tr1 cells (A-10-9 (Fig. 4) or A-10-11 (data not shown)) prevents IBD induced by pathogenic CD4+CD45RBhisplenic T cells. Not surprisingly, the Tr1 cells are effective only upon stimulation in vivo by feeding the mice with OVA (Fig. 4). No difference in the pathogenicity of the CD4+CD45RBhisplenic T cells or in the protective effects of the CD4+CD45RBloT cells16 was observed after feeding the mice with OVA in the absence of Tr1 cells (Fig. 4). These results demonstrate that Tr1 clones can prevent a T-cell-mediated disease in vivo . Furthermore, they indicate that the Tr1 cells must be activated in vivo to be effective. Finally, because the function of Tr1 cells is mediated by soluble factors, the present findings suggest that they can suppress active immune responses to unknown antigens in the microenvironment by an antigen-driven bystander suppression mechanism.

Figure 4: Co-transfer of Tr1 clones prevents colitis.
figure 4

A, Groups of 5 scid mice were reconstituted with 4 × 105 CD4+CD45RBhisplenic T cells from 6-week-old BALB/c mice either alone (white squares) or in the presence of either 2 × 105 CD4+CD45RBloT cells (black triangles) or 2 × 105 OVA-specific Tr1 clones (A-10-9) (white diamonds for animal not fed with ovalbumin, and black diamonds for animals fed with ovalbumin). Half of the groups were fed with ovalbumin diluted in the drinking water (100 ng ml−1). Feeding with ovalbumin did not change the pathogenic effect of CD4+CD45RBhiT cells or the protective effect of CD4+CD45RBloT cells, and only animals fed with ovalbumin are shown for these two groups. Data represent mean plus s.e.m. of 5 animals per group. Results are representative of two experiments. B, Severe colitis in mice restored with CD4+CD45RBhiT cells (a) and with CD4+CD45RBhiT cells and a Tr1 clone (A-10-9) but not fed with ovalbumin (b). Protection from colitis in mice reconstituted with either CD4+CD45RBhiT cells plus CD4+CD45RBloT cells (c) or in micereconstituted with CD4+CD45RBhiT cells and a Tr1 clone and fed with ovalbumin (d).

Full size image

Transfer of CD4+T-cell clones isolated from the gut of SJL mice after oral tolerance induction with myelic basic protein (MBP)17 or from the pancreas of NOD mice18 has been shown to suppress experimental autoimmune myelitis and diabetes, respectively. These inhibitory activities were relieved by anti-TGF-β, suggesting that these clones mediated their suppression through TGF-β17,18. Although these T-cell clones appear to have similar functions to Tr1 cells, their cytokine production profiles are different. The T-cell clones isolated from SJL and NOD mice produce low levels of IL-10 and high levels of biologically active TGF-β, whereas the T-cell clones isolated from MBP transgenic mice fed with MBP (called Th3 cells) exclusively produced active TGF-β and no IL-10, IL-4, IL-2 or TNF-α19. In contrast, Tr1 cells consistently produce low levels of active TGF-β and high levels of IL-10. Another difference is that both cytokines contribute to the suppressive activity of Tr1 cells. Taken together, these data suggest that Th3 and Tr1 cells are distinct types of CD4+regulatory T cells.

The isolation of Tr1 cells from the peripheral blood of SCID patients, in which high levels of IL-10 in vivo are associated with tolerance after allogeneic stem-cell transplantation10, and from anergic T cells stimulated by major histocompatability complex (MHC) class II-positive melanoma cells20, supports the hypothesis that Tr1 cells are generated in vivo by chronic exposure of naive T cells to alloantigens in the presence of IL-10. Furthermore, based on the in vitro immunosuppressive activities of Tr1 cells, it is possible that these Tr1 cells specific for the host antigens have similar properties in vivo, and therefore may be responsible for transplantation tolerance in man. The observation that IL-10 production is associated with tolerance in murine models of organ and stem-cell transplantation is consistent with this notion21,22.

However, administration of IL-10 failed to induce T-cell tolerance in other experimental models of transplantation23,24,25,26 and autoimmune disease, such as experimental autoimmune encephalomyelitis27 or autoimmune haemolytic anaemia28, which could be explained by our observations suggesting that the potential induction of tolerance through differentiation of Tr1 cells is a lengthy process in vivo, requiring chronic antigen-specific stimulation in the presence of IL-10. Thus systemic administration of IL-10 after the onset of the disease may not impede the effector phase of the immune response, and may therefore not result in tolerance.

Our findings indicate that stimulation of CD4+T cells in the presence of IL-10 generates a subset of CD4+T cells that inhibits antigen-specific immune responses through the secretion of IL-10 and TGF-β. The cytokine profile appears to be stable, as Tr1 clones cultured for more than 12 months produced the same cytokine profile after activation. All regulatory cells described previously have been isolated after in vivo priming with antigens. The ability to generate Tr1 cells after exposure to antigen in vitro may be an advantage for further functional analysis and potential clinical applications, particularly because we have demonstrated their regulatory capacity in vivo . Moreover, because Tr1 cells suppress immune responses directed against other antigens in the microenvironment, they could potentially be used to regulate T-cell-mediated disease, even when the pathogenic antigen is unknown.

Methods

Cloning of Tr1 cells. Human Tr1 cells were generated from peripheral blood CD4+T cells stimulated with allogeneic monocytes in the presence of exogenous IL-10. After 10 days, the CD4+T cells were cloned by flow cytometry. Cells were labelled with an anti-CD4-FITC monoclonal antibody (Becton Dickinson, Mountain View, CA) and the viable cells were sorted at one cell per well in wells pre-coated with anti-CD3 monoclonal antibody (SPV-T3 (ref. 12) 100 µg ml−1in 100 µl Tris, 0.1 M, pH 9.4). After sorting, a mixture of irradiated feeder cells (JY, 105 per ml and PBMC, 106per ml) and 10 U ml−1of rIL-2 in 100 µl were added. Clones were expanded with IL-2 (10 U ml−1) and by repetitive stimulation with high doses of crosslinked anti-CD3 monoclonal antibody (100 µg ml−1). The cloning of the mouse Tr1 cells was performed by limiting dilution at 0.3 cells per well in wells coated with 50 µg ml−1of anti-CD3 monoclonal antibody (2C11, PharMingen, San Diego) in PBS and containing irradiated total splenocytes (107per ml) and 20 U ml−1IL-2. Clones were expanded by culture with IL-2 (20 U ml−1) and IL-4 (20 U ml−1), OVA peptide (2 µM) and irradiated splenic APCs (107per ml).

Flow cytometry analysis of intracellular cytokines. Analysis of intracellular cytokines by flow cytometry was performed as described29. Cells (106per ml) were activated with immobilized anti-CD3 and anti-CD28 monoclonal antibodies (PharMingen, San Diego) for 4 h. Brefeldin A (Sigma, St Louis) was added at 10 µg ml−1, and 2 h later cells were collected, washed and fixed with formaldehyde (2%). For intracellular staining, cells were incubated with the following monoclonal antibodies (PharMingen, San Diego): anti-IL-4-FITC or PE (11B11, 5 µg ml−1), anti-IFN-γ-PE (XMG1.2, 5 µg ml−1), anti-IL-5-PE (TRFK5-2, 5 µg ml−1), anti-IL-10-FITC (JES-16E3, 5 µg ml−1), anti-IL-2-PE (JES6-5H4, 2.5 µg ml−1) or the anti-clonotype KJ1-26-FITC (5 µg ml−1). Samples were analysed on a FACScan (Becton Dickinson, Mountain View, CA).

Cytokine analysis. Human CD4+T-cell clones or populations at 1 × 106 per ml were stimulated by immobilized anti-CD3 (10 µg ml−1) and anti-CD28 monoclonal antibodies (1 µg ml−1) for 48 h. The production of IL-2, IL-4, IL-5, IL-10 and IFN-γ was measured by immunoenzymatic assays, as described12. The amounts of IL-2, IL-4, IL-10 and IFN-γ produced by murine CD4+T-cell clones was measured by ELISA in supernatants collected 48 h after culture of the T cells (106per ml) stimulated with OVA-peptide (1 µM) and irradiated total splenic APCs, as described9. For TGF-β measurements, cells were cultured for 72 h in Yssel's medium30 without serum and the amount of TGF-β was measured by ELISA (R and D System, Minneapolis, MN) after an acid activation, according to the manufacturer's instructions.

Proliferation assays. All proliferation assays were carried out in Yssel's medium30 supplemented with 10% FCS and 1% human serum for human cells or with 10% FCS for murine cells. Alloantigen-specific stimulation was performed by culturing human CD4+T-cell clones (106per ml) with purified irradiated (4,000 rad) monocytes (106per ml) in 200-µl round-bottomed 96-well plates (Linbro, ICN Biomedicals, Aurora, OH). T-cell proliferation was measured after 5 days incubation at 37 °C and 5% CO2 and a 12-h pulse with 3[H]-TdR. Activation of human CD4+T cells with immobilized anti-CD3 and anti-CD28 mAb was performed as described12. In brief, antibodies were diluted at the indicated concentration in 0.1 M Tris buffer, pH 9.4, and plates incubated for a week at 4 °C. After washing the plates, CD4+T cells were added at a concentration of 5 × 104 cells per well and proliferation was measured after 3 days by 3[H]-TdR uptake. Similarly, murine CD4+T-cell populations or T-cell clones were stimulated at a concentration of 5 × 104 cells per well with OVA peptide (0.5 µM) and irradiated (4,000 rad) total splenocytes (5 × 105 cells per well) for 3 days.

Inflammatory bowel disease. The induction of IBD was performed as described16. In brief, C.B-17 scid mice 8–12 weeks old. (Simonsen Laboratories, Gilroy, CA) were injected intraperitoneally with 100 µl of PBS containing 4 × 105 CD4+CD45RBhisorted splenic T cells with or without either 2 × 105 CD4+CD45RBlosplenic T cells or 2 × 105 Tr1 clones. Half of the mice were fed with OVA diluted in the drinking water at 100 ng ml−1. Weight gain or loss, as a percentage of original starting weight, was scored weekly for 9 weeks. For preparation of tissue for histopathologic analysis, approximately 10-mm segments of the distal large intestine were removed and fixed in a solution of 10% formalin. Fixed tissues were frozen and sections were prepared and stained with haematoxylin.