Regulatory CD4+CD25+ T cells (Treg) are mandatory for maintaining immunologic self-tolerance. We demonstrate that the cell-cell contact–mediated suppression of conventional CD4+ T cells by human CD25+ Treg cells is fixation resistant, independent from membrane-bound TGF-β but requires activation and protein synthesis of CD25+ Treg cells. Coactivation of CD25+ Treg cells with Treg cell–depleted CD4+ T cells results in anergized CD4+ T cells that in turn inhibit the activation of conventional, freshly isolated CD4+ T helper (Th) cells. This infectious suppressive activity, transferred from CD25+ Treg cells via cell contact, is cell contact–independent and partially mediated by soluble transforming growth factor (TGF)-β. The induction of suppressive properties in conventional CD4+ Th cells represents a mechanism underlying the phenomenon of infectious tolerance. This explains previously published conflicting data on the role of TGF-β in CD25+ Treg cell–induced immunosuppression.
The process of TCR generation, that is based on random rearrangements, as well as the promiscuity of the resulting receptor expressed by mature T cells requires central and peripheral mechanisms of tolerance induction (1). CD25+ T regulatory (Treg) cells play a central role in the maintenance of peripheral tolerance as the depletion of these cells in a murine adoptive transfer model leads to various autoimmune diseases (2, 3). These CD4+ T cells express CD25 but do not or only marginally proliferate after polyclonal or antigen-specific activation. However, although they seem to be immunologically inert, activated CD25+ Treg cells strongly suppress the proliferation of coactivated conventional CD4+CD25− T cells in vitro (4, 5). We and others characterized a human equivalent of murine CD25+ Treg cells that has comparable properties and can be isolated from human peripheral blood (6–8). The suppressive capacity of these human CD25+ Treg cells depends on direct cell–cell contact but the mechanism of suppression is largely unknown. Several studies implied that signaling through CTLA-4 might be responsible for the inhibitory potency of CD25+ Treg cells, whereas others could not confirm this finding (2). Studies using a murine IBD model suggest that TGF-β and IL-10 are potent mediators of suppression (9, 10). However, neither anti–IL-10 nor anti–TGF-β antibodies could abrogate the suppressive capacity of human CD25+ Treg cells in vitro (6). On the other hand, it was shown that blocking anti–TGF-β antibodies abrogated suppressor activity in vivo (9, 11, 12). Hence, there are a lot of conflicting and contradictory findings regarding the suppressive mechanisms of CD25+ Treg cells. Herein we demonstrate that coculture of human CD25+ Treg cells with CD25−CD4+ T cells results in the development of an additional CD4+ T suppressor cell population. These induced CD4+ Treg cells emerge from the CD25−CD4+ T cell population and suppress the proliferation of freshly isolated conventional CD4+ T cells. This process is partially mediated by soluble TGF-β.
Materials And Methods
X-VIVO-15 supplemented with 1% autologous plasma was used for culture of dendritic cells (DCs) and without plasma for culture of T cells (BioWhittaker).
All cytokines used in this study were recombinant human proteins. Final concentrations: 800 U/ml GM-CSF (Leukomax), IL-4, 1,000 U/ml IL-6 (Strathmann Biotech GmbH), IL-1β, 10 ng/ml TNF-α (Strathmann), and PGE2 (Minprostin; Pharmacia-Upjohn) 1 μg/ml. The DCs were generated from buffy coats of healthy volunteers as described previously (13). For preactivation of T cells: IL-2 (Proleukin; Chiron) 10 U/ml.
These mAbs were used for the staining of MACS®-sorted T cells: FITC- or PE-conjugated CD3, CD4, CD25, CTLA-4, anti–HLA-A,B,C, and FITC- and PE-conjugated mouse IgG (Beckman Coulter and Immunotech). These mAbs were used for the staining of membrane-bound TGF-β: LAP (biotinylated mAb, R&D Systems, used according to the manufacturer's instructions) detected by streptavidin-Cy5 (Dianova). The anti–TGF-β (R&D Systems, used according to the manufacturer's instructions) and anti–IL-10 mAb (JES-19F1.1.1; American Type Culture Collection (ATCC), blocking capacity tested in proliferation assays using IL-10 receptor transfected Baf3 cells) were used for blocking experiments (14). For sorting of preactivated HLA-A2–positive conventional CD4+ T cells, purified anti–HLA-A2–positive-specific mAb (BB7.2; ATCC) and anti–mouse IgG microbeads (Miltenyi Biotec) were used. Labeled T cells were analyzed by flow cytometry (FACScalibur™ and CELLQuest™; Becton Dickinson). Necrosis versus apoptosis were determined by propidium iodide and annexin-V staining according to the manufacturer's instructions (BD PharMingen).
Isolation and Stimulation of T Cell Populations.
Conventional CD4+ Th cells and CD4+CD25+ regulatory T cells were isolated from buffy coats of healthy volunteers as described previously (6). Briefly, CD4-MACS®-Multisort-Beads (Miltenyi Biotec) were used for isolation of CD4+ T cells. After detaching, cells were washed and CD4+CD25+ T cells were positively selected according to the instructions of the manufacturer using CD25 microbeads. For some experiments, CD4+CD25+ T cells and conventional CD4+ T cells were preactivated with 0.5 μg/ml anti-CD3 mAb at 37°C for 20 h in the presence of 10 U/ml IL-2. Aliquots of the cultures were used for proliferation assays performed in X-VIVO-15 and in the presence of different numbers of allogeneic DCs or anti-CD3/CD28 stimulation in 96-well plates. T cell proliferation was measured after 3–4 d of incubation and an additional 16-h pulse with 3[H]Tdr (37 kBq/well) using a liquid scintillation counter.
Isolation of Precultured CD4+ T Cells.
106 freshly isolated HLA-A2-negative CD4+ T cells and 106 HLA-A2-positive CD25+ Treg from a different donor were coactivated with anti-CD3 (1 μg/ml) and anti-CD28 (2 μg/ml) mAb. After 6 d of coculture the CD25+ Treg cells were stained with a HLA-A2-specific mAb and depleted from the anergized CD4+ T cells using anti–mouse IgG microbeads (purity >95%). The anergized CD4+ T cells were restimulated with anti-CD3/CD28 mAb as described previously. 24–72 h after restimulation aliquots were used for detection of TGF-β by ELISA (DRG-Instruments, detection limit of biologically active TGF-β: 4.7 pg/ml).
Transwell experiments were done in 24-well plates as described previously (14). Briefly, 106 CD4+ conventional T cells or CD4+CD25+ T cells were stimulated with anti-CD3 (1 μg/ml) and anti-CD28 (2 μg/ml) in 1.5 ml. Additionally, 106 CD4+CD25+ T cells or precultured CD4+ T cells (Thsup) were either directly added to cultures of activated conventional CD4+ T cells or were placed in transwell chambers (Millicell, 0.4 μm; Millipore) in the same well. After 3 d of culture, activated T cells (200 μl per well) were transferred to 96-well plates in triplicates. Proliferation was measured after an additional 16-h pulse with 3[H]Tdr using a liquid scintillation counter.
Fixation of Freshly Isolated and Preactivated T Cells.
For some experiments, conventional CD4+ T helper cells and CD25+ Treg were fixed with 1% paraformaldehyde for 10 min in PBS, either directly after isolation or 20 h after preactivation with 0.5 μg/ml anti-CD3 mAb and 10 U/ml IL-2. Additionally, CD25+ Treg were preactivated in the presence of 10 μg/ml cycloheximide or 1 μg/ml monensin. These concentrations of cycloheximide and monensin showed no toxic effects but blocked the upregulation of CTLA-4 on CD25+ Treg and CD25 on conventional CD4+ T cells during preactivation (unpublished data). Fixed cells were washed intensively with RPMI 1640 plus 10% FCS and used for coculture with freshly isolated CD4+ T cells in proliferation assays.
Results And Discussion
CD25− conventional CD4+ T cells and CD25+ Treg cells were isolated from buffy coats of healthy volunteers as described previously (6). Freshly isolated, CD25+ Treg cells showed a fourfold increased expression of membrane-bound TGF-β as compared with conventional CD4+ Th cells (Fig. 1 A). However, polyclonal activation using anti-CD3 in combination with anti-CD28 antibodies resulted in an upregulated surface expression of TGF-β on conventional CD4+ T cells, whereas TGF-β on CD25+ Treg cells was downregulated. Both populations, CD25+ Treg cells and conventional CD4+ Th cells, either resting or activated, showed no significant production of biologically active soluble TGF-β (see Fig. 4 C).
To analyze the potential inhibitory effect of membrane-bound TGF-β on the surface of freshly isolated human CD25+ Treg cells, as postulated for murine CD25+ Treg cells (15), we stimulated freshly isolated CD4+ Th cells in coculture with human CD25+ Treg cells and in the presence or absence of blocking antibodies against TGF-β. As shown in Fig. 1 B, the presence of anti–TGF-β antibodies could not reverse the cell contact–dependent inhibitory effect of the CD25+ Treg population on the proliferation of CD4+ Th cells, neither upon allogeneic nor polyclonal stimulation of Th cells. Thus, these data indicate that membrane-bound TGF-β is not responsible for the suppressive effects of freshly isolated human CD25+ Treg cells. In addition, the fact that highly proliferative CD4+ Th cells express the same amounts of TGF-β as highly suppressive CD25+ Treg cells (Fig. 1 A) also strongly argues against the assumption that membrane-bound TGF-β is responsible for the suppressive capacity of human CD25+ Treg cells.
To analyze the suppressive mechanism in more detail, we stimulated CD4+ T cells in the presence of fixed CD25+ Treg cells. In coculture experiments, freshly isolated human CD25+ Treg cells inhibited the proliferation and cytokine production of coactivated conventional CD4+ T cells in a dose-dependent manner. If CD25+ Treg cells were fixed directly after isolation, no suppressive activity could be detected (Fig. 2 A). However, if CD25+ Treg cells were preactivated overnight with anti-CD3 antibodies (0.5 μg/ml) and 10 U/ml IL-2 before fixation, they showed a comparable suppressive capacity for CD4+ T cells as unfixed CD25+ Treg cells. In contrast, Treg-depleted and activated conventional CD4+ T helper cells did not exert any suppressive activity, although such cells express comparable amounts of TGF-β (Fig. 1 A). These data suggest that the inhibitory function of human CD25+ Treg cells is activation dependent. Additional experiments revealed that the induction of suppressor activity requires protein synthesis as it can be inhibited by the presence of cycloheximide or monensin (Fig. 2 B). However, once activated, the suppressive activity of human CD25+ Treg cells is fixation-resistant (Fig. 2 A). These data also strongly corroborate our findings that the inhibitory function of human CD25+ Treg cells is independent of soluble mediators (6). Furthermore, the activation of suppressor function of human CD25+ Treg is also independent of costimulation, since the addition of mature DCs or soluble anti-CD28 antibodies during overnight preactivation with anti-CD3 antibodies did not alter/enhance the functional activities of human CD25+ Treg cells (unpublished data).
We and others could not detect any influence of TGF-β on the cell contact–mediated inhibitory function of CD25+ Treg cells in vitro (2, 6), whereas several groups demonstrated that neutralizing antibodies against TGF-β abrogated the suppressive function of murine CD25+ Treg cells in vivo (9, 15). One possible explanation for this discrepancy could be that CD25+ Treg cells exert an additional suppressive mechanism that is presumably also involved in the phenomena of bystander suppression or infectious tolerance (10, 16) but is not directly mediated by the CD25+ Treg cells themselves. Consequently, we analyzed the functional properties of the anergized conventional CD4+ Th cells after coculture with human CD25+ Treg cells. To distinguish both T cell populations, HLA-mismatched CD25+ Treg cells were used and depleted of the anergized CD4+ T cells after coculture with the aid of HLA-specific mAbs as illustrated in Fig. 3 A. Coculture of HLA-mismatched conventional CD4+ T cells served as a control. As shown in Fig. 3 B, the purified CD4+ T cells, anergized by CD25+ Treg cells, revealed a strong suppressive activity for conventional CD4+ Th cells that was comparable to the inhibitory capacity of CD25+ Treg cells. Thus, human CD25+ Treg cells, in a cell contact–dependent fashion, suppress the proliferation of conventional CD4+ T cells and simultaneously induce suppressive activity in these CD4+ Th cells (Fig. 3 B). In contrast, conventional CD4+ T cells preactivated in the presence of HLA-mismatched conventional CD4+ T cells showed no suppressive activity for freshly isolated Th cells (Fig. 3 B, lowest bar).
To elucidate the mechanism of this conveyed suppressive activity, we analyzed the properties of the anergized CD4+ T cells (subsequently termed Thsup) in comparison to the original CD25+ Treg population in transwell stimulations. The semipermeable transwell membrane prevents direct cell–cell contact between the responsive Th cell population and the suppressor T cells. As shown recently, the physical separation by the transwell membrane abrogated the cell contact–dependent suppression of CD4+ Th cells by CD25+ Treg cells (6). In contrast, the conveyed suppressor activity of Thsup cells could not be abrogated by the semipermeable membrane (Fig. 4 A). However, fixation of such Thsup cells completely abrogated their suppressor function while it could not inhibit the suppressive properties of activated CD25+ Treg cells (Fig. 4 B). These data strongly support our primary finding that the induced secondary suppressor activity of Thsup cells is cell contact independent. In addition, these results were further confirmed by the fact, that the suppressive activity of Thsup cells could be partially abrogated upon the addition of neutralizing anti–TGF-β antibodies (Fig. 4 A), whereas neutralizing anti–IL-10 antibodies showed no effect (data not shown). These data strongly suggest that the suppressive activity of Thsup cells is mediated partially by biologically active TGF-β. Therefore, the production of TGF-β was assessed by using an ELISA that specifically detects biologically active material. As shown in Fig. 4 C, Thsup cells produce biologically active TGF-β immediately after polyclonal stimulation. In contrast, neither conventional CD4+ Th cells nor CD25+ Treg cells, alone or in coculture, produced detectable amounts of TGF-β. This finding again strongly supports our thesis that resident human CD25+ Treg cells convey a suppressor function to conventional CD4+ Th cells that is completely cell contact independent and mediated mainly by TGF-β.
In summary, our findings help to explain several conflicting and contradictory results published previously. In accordance with published data on the properties of resident murine and human CD25+ Treg in vitro (2, 6, 8), we show that the primary functional activity of CD25+ Treg is strictly activation- and cell contact–dependent and independent of soluble mediators. Moreover, our data clearly indicate that classical human CD25+ Treg cells can, in a cell contact–dependent manner, direct the differentiation of conventional CD4+ Th cells toward an additional population of regulatory T cells (Thsup). Both regulatory T cell populations can inhibit Th cell proliferation through distinct suppressor mechanisms (cell contact–dependent versus soluble mediators). Thus, these results imply that human CD25+ Treg cells can amplify their suppressive capacity through the recruitment of an additional population of suppressor T cells that exert a distinct but complementary suppressor mechanism. Data from murine in vivo studies suggest that the suppressive properties of CD25+ Treg cells are mediated via soluble mediators such as TGF-β (2). In contrast, in vitro studies clearly revealed that murine and human CD25+ Treg cells inhibit T cell proliferation via cell contact and independent from soluble factors (6, 8). These apparently discordant results could be reconciled by the assumption that in vitro a local cell contact–dependent mechanism is sufficient to inhibit the proliferation of the responding T cell population. However, in vivo, e.g., in the murine model of IBD (9), additional soluble suppressive mediators such as TGF-β, derived from anergized CD4+ Th cells, are important for the systemic control of autoreactive T effector cells (Fig. 5). Finally, the regulation of immune responses that includes bystander suppression and infectious tolerance (10, 16) obviously requires a transfer of tolerizing potencies from resident or induced CD25+ Treg cells to another T cell population with a different antigen specificity. Hence, we would like to speculate that the spreading of suppression from human CD25+ Treg cells to conventional CD4+ Th cells is one of the fundamental mechanisms that is involved in the induction or maintenance of peripheral tolerance.
The authors are grateful to Dr. E. Rüde and Dr. K. Mahnke for critical reading of this manuscript and helpful discussions. We also thank L. Paragnik, A. Kandemir, S. Fondel, and P. Hoelter for expert technical assistance.
This work was supported by the Deutsche Forschungsgemeinschaft, grant A6SFB548 (to E. Schmitt) and grant A3SFB548 (to A.H. Enk).