of whole cell cultures in U-bottomed wells of a 96-well plate are shown. 2

Supplementary online material Supplementary figure legends Supplementary Figure 1 Exposure to T reg cells causes loss of T resp cells in co-cultures. T resp cells were stimulated with CD3+CD28 alone or in co-cultures with T con or T reg cells for 3 d. Photomicrographs of whole cell cultures in U-bottomed wells of a 96-well plate are shown. 2 independent experiments showed similar results. Supplementary Figure 2 T reg cells do not affect early activation of T resp cells. (a) CD25 expression on CFSE-labeled T resp cells 24 h after CD3+CD28 stimulation with T con (blue) or T reg (red) cells as indicated (gated on CFSE + T resp cells). Grey line shows the unstimulated control cells. (b) Intracellular IFN-γ expression in T resp cells, 24 h after stimulation in culture with T con or T reg cells as indicated (gated on live CFSE + T resp cells). The numbers reflect the percentages of cells within gates. 3 independent experiments showed similar results. Supplementary Figure 3 T reg cells induce apoptosis independently of proliferation. CFSElabeled T resp cells were left unstimulated or were stimulated with CD3+ CD28 in co-cultures for 3 d alone or in the presence of pre-activated T reg cells in the absence or presence of cell cycle inhibitors aphidicolin (Aphid.) or nocadazole (Noca.). Proliferation and death was analyzed using CFSE dilution and propidium iodide (PI) staining. Numbers indicate percentages of live cells within the gates. Data represent 2 independent experiments. 1

Supplementary Figure 4 T reg cell-induced apoptosis is independent of cytolytic mechanisms but is abrogated by IL-7. (a) Wild-type T resp cells were stimulated in co-cultures with T con or T reg cells from wild-type, gld or perforin-deficient (Prf1 / ) mice. TNFR-Fc or an inhibitor of granzyme B (GzmB) was added at the beginning of cultures where indicated. T resp cell apoptosis is normalized to the death of T resp cells in T con cell co-cultures under each condition. Individual graphs depict independent experiments. Data represent 2 independent experiments. (b) T resp cells were stimulated in co-cultures with indicated ratios of T reg cells for 3 d. T resp cell apoptosis is normalized to the death of T resp cells in T con cell co-cultures under each condition. 5 independent experiments showed similar results. Supplementary Figure 5 T reg cells express IL-4R-α and take up IL-4 in vitro. (a) IL-4R-α expression on T con or T reg cells stimulated for 72 h with CD3+CD28 was assessed by flow cytometry. Numbers indicate the fraction of events within the gates illustrated. 2 independent experiments showed similar results. (b) Freshly isolated T reg or T con cells were stimulated as above for 4 d in the presence or absence of human IL-4-Fc fusion protein and the percentages of cells containing intracellular human IL-4-Fc was assessed by flow cytometry. Data represent 2 independent experiments. Supplementary Figure 6 T reg cells suppress late cytokine production. CFSE-labeled T resp cells were stimulated in co-cultures with T con or T reg cells for 3 or 4 d as indicated. After restimulation with PMA + ionomycin for 5 h, cytokine expression was assessed by intracellular staining and flow cytometry. Plots are gated on CFSE + cells. Data represent 3 independent experiments. 2

Supplementary Figure 7 T reg cell-mediated suppression of late cytokine production is partially reversed by IL-2. CFSE-labeled T resp cells were stimulated as indicated in co-cultures for 1 or 3 d in the presence or absence of exogenous IL-2. Percentages of CFSE + T resp cells producing cytokines in CD3+CD28 (left panel) or CD3+APC (right panel) co-cultures were analyzed by intracellular staining and flow cytometry following re-stimulation with PMA + ionomycin for 5 h. 2 independent experiments showed similar results. Supplementary Figure 8 T reg cells ameliorate IBD and induce apoptosis of colitogenic CD4 + CD45RB hi cells. (a) Thy1.2 + C.B-17 scid mice received 4 1 5 fresh Thy1.1 + CD25 CD45RB hi CD4 + congenic cells (n =1). One week later, mice received PBS (CD4 + CD45RB h +PBS) or 1 1 6 fresh T reg cells (CD4 + CD45RB hi +T reg ). Five control mice received PBS in both the injections (PBS+PBS). H&E staining of colon sections with original magnification 5 (left panel) or 2(right panel) are shown (b) Photographs of two different colon samples, prepared 8 wk after initial transfer from mice in each group. Data represent 2 experiments. (c) Apoptosis of infiltrating cells in colons of mice in (a,b). TUNEL-Digoxigenin (red) histological staining was performed in colonic sections 6 d after T reg cell transfer (Original magnification 4). Data in all the panels represent at least 2 independent experiments showing similar results. Supplementary Figure 9 T reg cells home to MLN and kill effector cells in colon without suppressing their early proliferation in MLN (a) Thy1.2 + C.B-17 scid mice received 4-5 1 5 fresh Thy1.1 + CD25 CD45RB hi CD4 + congenic cells in the first injection. (n =6). Six days later, they received a second transfer of pre-activated CD4 + cells (pt con ) or 1 1 6 pre-activated T reg 3

cells (pt reg ). Left, MLN were isolated 1 d after the second transfer and analyzed by flow cytometry (gating on donor Thy1.1 + cells) ( **, P <.1). Right, colons were snap frozen and fixed and stained with anti-thy1.1 followed by TUNEL staining. Cryosections were analyzed by confocal microscopy. The frequency of apoptotic Thy1.1 + cells are shown in both analyses. ( ***, P <.7). The data are representative of 3 independent experiments with similar results (b) Thy1.2 + C.B-17 scid mice received 4 1 5 fresh Thy1.1 + CD25 CD45RB hi CD4 + congenic cells in the first injection (n=1). One week later, they received a second injection of PBS or 1 1 6 fresh T reg cells (CD4 + CD45RB hi +T reg ) together with BrdU. Mice received another injection of BrdU 24 h later. One day later, colons and MLNs were isolated and anti-brdu staining was performed. Data represent 2 independent experiments. (c) CD45.1 + Rag1 / mice were reconstituted with naïve CD45.2 + congenic cells from WT or Bim-deficient mice (n= 4-5 per group). Six days later, mice received PBS or 1 1 6 fresh CD45.1 + T reg cells. Two control mice received PBS in both injections (PBS+PBS). MLNs were isolated 6 days after T reg cell transfer, and single cell suspensions were fixed and stained for Foxp3. The numbers within gates show the percentage of CD45.2 Foxp3 + T reg cells before (left panel) transfer into mice. In (right panel) the percentage of CD45.2 Foxp3 + in MLN of the recipient mice are shown. Data in all the panels represent 3 independent experiments showing similar results. Discussion Model of apoptotic suppression of CD4 + T cells by Tregs. On priming CD4 + T cells produce IL- 2 and /or other growth factors and expand responding to these cytokines. If Tregs are present during this early phase of IL-2 production, CD4 + T cells are deprived of IL-2 leading to CD4 + T cell death. However, cells that are not in direct/close proximity to Tregs, may remain in the system and continue to expand. Because of the low levels of IL-2 in their mileu, their cytokine 4

receptor expression and p-akt levels are low and their effector function and cytokine production may be impaired. In the absence of Bim, because the effector cell progenitors (cytokine producers) are not killed (thus not reduced in numbers), the overall cytokine levels in their immediate environment will be high enough to expand the cells and impart appropriate effector functions despite cytokine consumption by Tregs. Whether the cytokine deprivation apoptosis is caused solely by consumption of cytokines in vivo remains to be investigated. It may be also caused by the inability of the cells to produce cytokines at later phases of the response. Even though we believe that suppression of cytokine production may have been caused by the disruption of cytokine positive feedback loop, the effects of other suppressive cytokines such as TGF-β in vivo should also be considered. TGF-β does not impact cytokine deprivation in vitro. Cultured T cells may be more sensitive to cytokine consumption than in vivo T cells and thus may be killed by cytokine deprivation even before the suppressive effects of TGF-β becomes apparent. Therefore, the effect of TGF-β on Bim dependent death and suppression of late cytokine production on in vivo remains to be investigated. We believe that TGF-β could be an important component, because, it is evident that TGF-β can cause cell death in Bim dependent fashion 1, 2. Our model also proposes another interesting possibility that death of many effector cells by Treg mediated cytokine consumption may cause the contracting population that remains to be more susceptible to the effects of other suppressive molecules such as CTLA-4, IL-1 and TGF-β (dependent or independent of Tregs). Because we observed that Tregs may have a minor delayed effect on Bim deficient T cells much later than on WT cells (later than 6 weeks in IBD model), Tregs may have additional effects independent of direct cytokine deprivation effects. Note that after 6 weeks, Bim deficient recipients tend to 5

increase their weight in Fig. 8e. However, Tregs deficient in CTLA-4, IL-1 and TGF-β are shown to suppress effector cells in IBD model 3. Therefore, it would be interesting to test if presence of Foxp3 + cells in vivo induces the expression of the suppressive molecules by other cell types (possibly due to cytokine deprivation effects), which in turn may be involved in the indirect suppressive effects of Tregs. Thus, it is plausible that Tregs do not have to express TGF-β but the effector cells have to be responsive to the suppressive cytokine. This possibilty is also consistent with the finding by Von Andrian s group, that showed that effector CD8 cells lacking TGF-β receptor expression are resistant to the effects of Tregs 4. We believe that our results regarding IL- 2 transcription differ from previous analyses because the latter did not separate Tresps from Tregs prior to IL-2 mrna analyses, (which would dilute the total Tresp RNA with Treg RNA which would be devoid of IL-2 RNA) and no analyses of IL-2 gene locus was carried out 5. Thus, our model supports the idea that Tregs do not impact initial priming of CD4 cells but decrease maintenance/survival phase of primed cells, which may further influence other effector responses in several ways. Our apoptosis model of Tregs solves several discrepancies in Treg literature regarding various mechanisms of suppression and opens a whole new avenue for further investigations. A central tenet of immunobiology is that cells, especially lymphocytes, are the elemental units that are increased or decreased in number, to control the magnitude of specific immune responses 6. We now show that Tregs exert their suppressive effect in good measure through this same fundamental approach. References 1. Ramjaun, A. R., Tomlinson, S., Eddaoudi, A. & Downward, J. Upregulation of two BH3- only proteins, Bmf and Bim, during TGFbeta-induced apoptosis. Oncogene (26). 6

2. Wildey, G. M., Patil, S. & Howe, P. H. Smad3 potentiates transforming growth factor beta (TGFbeta )-induced apoptosis and expression of the BH3-only protein Bim in WEHI 231 B lymphocytes. J Biol Chem 278, 1869-77 (23). 3. Kullberg, M. C. et al. TGF-beta1 production by CD4+ CD25+ regulatory T cells is not essential for suppression of intestinal inflammation. Eur J Immunol 35, 2886-95 (25). 4. Mempel, T. R. et al. Regulatory T cells reversibly suppress cytotoxic T cell function independent of effector differentiation. Immunity 25, 129-41 (26). 5. Thornton, A. M. & Shevach, E. M. CD4+CD25+ immunoregulatory T cells suppress polyclonal T cell activation in vitro by inhibiting interleukin 2 production. J Exp Med 188, 287-96 (1998). 6. M. Burnet, Cellular Immunology, Cambridge University Press, Cambridge, 1969. Supplementary methods Cell transfers in vivo. Thy1.2 + C.B-17 scid mice received 4 1 5 fresh Thy1.1 + CD25 CD45RB high CD4 + congenic cells in the first intra peritoneal (i.p.) injection. After 6 or 7 days, they received an i.p. injection of PBS or.8-1 1 6 freshly isolated Thy1.2 + T reg cells. Control mice received PBS in both the injections (PBS+PBS). C57BL/6 Rag1 / or Rag2 / mice (CD45.1 + or CD45.2 + ) were used as recipients when wild-type or Bim-deficient (C57BL/6) donor cells were transferred. After 6 or 7 days, they received an i.p. injection of PBS or 1 1 6 freshly isolated CD45.1+ Treg cells. On day 6-1 after Treg cells were transferred, Annexin-V and PI staining was performed to measure apoptosis. For 5CC7 TCR transgenic experiments, WT-B1A mice received 6 1 5 fresh CFSE-labeled CD45.1+ CD4 + congenic cells from 5CC7 TCR tg mice. They also received PBS or.8 1 6 fresh Treg cells from donor 5CC7 TCR tg mice followed by 1 µg of MCC peptide 24 h later. Unimmunized control mice received PBS in both the injections. On day 4 after immunization we analyzed the death of the donor CD45.1+ cells. To measure the proliferation of transferred cells,.8 mg-1 mg of BrdU in PBS (1mg/ml) was injected during the injection with Treg cells or PBS. A day later, mice received another injection of BrdU (1mg). On day 2 after PBS or Treg cell injection, we measured the BrdU uptake in cells isolated from the MLN and colon using APC-BrdU detection kit as per manufacturer s instructions. 7

Electron microscopy and qpcr. In co-cultures of CD45.1 + T resp and CD45.2 + T con or T reg cells, after 2 d of CD3 and CD28 stimulation T resp were labeled with APC-conjugated anti-cd45.1 and anti-apc microbeads to differentiate them from other cells in the co-culture. Cells were fixed and pelleted before electron microscopy analyses. For qpcr analyses of IL-2 mrna, CD45.1 + CD4 + CD25 cells were stimulated in co-cultures as above and microbead-labeled CD45.1 + cells were separated at different time points using Auto MACS sorter. RNA was prepared using Qiagen RNeasy Kit. IL-2 c-dna was quantified by q-pcr normalized to β-actin c-dna according to the manufacturer s instructions (Primer sequences: IL-2-fwd -5 -TTGATGGACCTACAGGAGCT-3, IL-2-rev -5 -AAATCCAGAACATGCCGCAG-3, β-actin-fwd- 5 - ACAGCTTCTTTGCAGCTCCT-3, β-actin-rev- AGTCCTTCTGACCCATTCC-3 ). Histology, intracellular staining, p-akt1 assay and immunoblot. For immunocytochemical H&E and TUNEL staining, tissues were washed with PBS, fixed with 1% formalin overnight and suspended in 7% ethanol to prevent over fixation. Paraffin sectioning and staining were performed by Histoserv Inc, MD. For immunofluorescent staining of tissue sections, they were snap frozen after three washes in PBS. Cryosectioning and staining of frozen sections were performed by Histoserv, Inc, MD. For single-cell staining, cells were cultured as above, washed in PBS, fixed with 4% paraformaldehyde, permeabilized with.5% Triton-X-1, incubated with respective antibodies and analyzed by flow cytometry or cytospun on slides and mounted for confocal micrscopy. CFSE labeling of responders distinguished them in co-cultures. Live cell gates based on forward and side scatter were used during flow cytometric analyses for intracellular staining. CytoFix/Cytoperm kit (BD BioSciences) was used for fixation and permeabilization for intracellular IL-2 staining. Before fixation, co-cultures were re- 8

stimulated with phorbol-myristate acetate (PMA) (1 ng/ml) and ionomycin (5 ng/ml) for 4-6 h, with Brefeldin-A (1µg/ml) added in last 2-2.5 hours. p-akt was measured in lysates using p-akt Pathscan ELISA (Cell Signaling Technologies) as per the manufacturer s instructions. Lysates were prepared from CFSE-labeled responders that were separated from co-cultures using fluorescence activated cell sorting (BD Biosciences) 2 d after CD3 and CD28 stimulation. For measuring p-bad, T resp cells were separated from 3 day co-cultures and lysates were prepared. Optical density readings are represented as arbitrary units. 9

Supplementary figures Fig.S1 Tresp Tresp+Treg Tresp+Tcon a b Count No stain Fig.S2 Tcon Treg 11 12 IFN-γ CD25 Fig.S3 CD3+CD28 Unstim. _ + Aphid. +Noca. PI _Treg 33 68 4 42 + Treg 31 21 6 CFSE 9

Fig.S4 a b Apoptosis(%) WT T reg WT T reg + TNFR-Fc Prf1 -/- T reg gld T reg 7 7 5 3 1 5 3 1 + T reg + T reg + Gz B inhibitor Apoptosis(%) 6 4 2 Apoptosis(%) 6 4 2-2 _ IL-7 +IL7 :1 1:4 1:3 1:2 1:1 T reg : T resp Fig.S5 a Count T con T reg 96 94 IL-4R-α b intra IL-4 Fc + (%) 9 6 3 _ IL-4-Fc T con T reg Fig.S6 T con T reg T con T reg T con T reg 7 57 38 11 44 9 CD3+CD28 Count 48 16 14 7 9 6 CD3+APC IL-2 IFN-γ IL-4

Fig.S7 Tcon Treg IL- 4+(%) IFN- γ+(%) IL-2+(%) CD3+CD28 CD3+APC 9 9 6 6 3 3 5 5 25 25 45 45 3 3 15 15 Time(d) 1 3 IL-2 - + - + 1 - + 3 - + Fig.S8 a b c PBS+PBS PBS+PBS 1 CD45RBhi+PBS CD45RBhi+PBS 2 3 CD45RBhi +Treg CD45RBhi +Treg 1) PBS+PBS 2) CD45RBhi+PBS 3) CD45RBhi+Treg

Fig.S9 a b +PBS +T reg Apoptosis(%) 6 4 2 ** pt con pt reg Apoptosis(%) 1 5 *** Thy 1.1 85 85 54 79 BrdU MLN Colon c WT CD45RB hi +PBS WT CD45RB hi +T reg Unstain control Fresh sorted T reg 1.65 6.41 Foxp3 2.2 87 Foxp3 BIM-KO CD45RB h +PBS BIM-KO CD45RB hi +T reg CD45.2 2.81 8.6 CD45.2