Preferential control of induced regulatory T cell homeostasis via a Bim/Bcl-2 axis

Apoptosis has an essential role in controlling T cell homeostasis, especially during the contraction phase of an immune response. However, its contribution to the balance between effector

and regulatory populations remains unclear. We found that Rag1−/− hosts repopulated with Bim−/− conventional CD4+ T cells (Tconv) resulted in a larger induced regulatory T cell (iTreg)

population than mice given wild-type (WT) Tconv. This appears to be due to an increased survival advantage of iTregs compared with activated Tconv in the absence of Bim. Downregulation of

Bcl-2 expression and upregulation of Bim expression were more dramatic in WT iTregs than activated Tconv in the absence of IL-2 in vitro. The iTregs generated following Tconv reconstitution

of Rag1−/− hosts exhibited lower Bcl-2 expression and higher Bim/Bcl-2 ratio than Tconv, which indicates that iTregs were in an apoptosis-prone state in vivo. A significant proportion of the

peripheral iTreg pool exhibits low Bcl-2 expression indicating increased sensitivity to apoptosis, which may be a general characteristic of certain Treg subpopulations. In summary, our data

suggest that iTregs and Tconv differ in their sensitivity to apoptotic stimuli due to their altered ratio of Bim/Bcl-2 expression. Modulating the apoptosis pathway may provide novel

therapeutic approaches to alter the balance between effector T cells and Tregs.

Regulatory T cells (Tregs) are an important subpopulation of CD4+ T cells that have a critical role in maintaining immune homeostasis and tolerance. Tregs can be divided into natural Tregs

(nTreg), that develop in the thymus, and induced Tregs (iTreg), that develop from conventional CD4+ T cells (Tconv) in the periphery during an immune response. iTregs can be induced in

response to either increased expression of self-antigen or as a consequence of infection. Therefore, their homeostasis must be tightly controlled. Insufficient differentiation of iTregs may

result in immune pathogenesis, whereas prolonged survival of these cells may compromise the establishment of immune memory. Although significant insight has been obtained concerning Treg

development and differentiation,1 little is known about the molecular mechanism(s) controlling nTreg and iTreg survival.2, 3, 4, 5

The BH3-only family member Bim (encoded by Bcl2l11) is a pro-apoptotic molecule that controls the coordinated contraction of the immune T cell response.6, 7, 8, 9 It directly interacts with

and deactivates the anti-apoptotic regulators Bcl-2 and Bcl-xl. Thus, the fine balance between Bim and Bcl-2/Bcl-xl expression levels represents an important checkpoint for cell survival.10,

11 Despite enhanced peripheral T cell expansion, Bim−/− mice do not develop aggressive T cell-mediated autoimmune disease.12 This may be due in part to an increased Treg population in

moderately aged Bim−/− animals compared with their wild-type (WT) littermates.13, 14 Interestingly, Bim expression naturally declines in peripheral WT CD4+ T cells during aging,15 which is

co-incident with increasing numbers of Tregs. This raises the possibility that Bim has an important role in Treg homeostasis.

As iTregs and nTregs cannot be distinguished in Bim−/− animals, the impact that Bim depletion has on either Treg cell subtype remains unclear. It appears that depletion of Bim may have a

limited impact on nTreg development, as the emigration rate of nTregs from the thymus is similar in WT and Bim−/− mice.13 The observation that Bim−/− nTregs are more proliferative is,

however, contradictory to the observation that Bim−/− T cells are hypo-responsive upon TCR activation.16 In this study, we ask if the balance between Bim and Bcl-2 acts as a critical

survival parameter that controls iTreg homeostasis.

To directly investigate the influence of Bim depletion on iTreg homeostasis in vivo, we induced inflammatory bowel disease (IBD) in mice by adoptively transferring CD4+CD25−CD45RBhi Tconv

into Rag1−/− hosts. In this model, transferred Tconv repopulate the host lymphoid tissue, leading to intestinal inflammation in a certain percentage of Tconv recipients. As only Tconv are

transferred (Supplementary Figure S1a), essentially all Foxp3+ T cells recovered from the recipients are de novo-generated iTregs. Although Bim is important for T-cell apoptosis, we only

observed a slight difference between Bim−/− and WT Tconv recipients in both percentage and absolute number of CD4+ T cells recovered from the spleens and mesenteric lymph nodes (MLN)

(Figures 1a and b). Surprisingly, there was a significant increase in the percentage (∼2–3-fold) and number (∼3–4-fold) of Foxp3+CD4+ iTregs recovered from Bim−/− Tconv recipients compared

with WT Tconv recipients in the spleen, MLN and a combined population of intraepithelial lymphocytes and lamina propria lymphocytes (IEL/LPL) isolated from the intestines (Figures 1c and d).

This increase of iTregs did not appear to be due to differing levels of Foxp3 protein expression, as iTregs recovered from both Bim−/− and WT Tconv recipients exhibited comparable levels of

Foxp3 expression (spleen=1022/958 (WT/Bim−/− MFI); MLN=1649/1443; intestine=1272/1258; Figure 1e). Nor was the difference due to an increased ability of Bim−/− Tconv to expand, compared

with WT Tconv following transfer into a lymphopenic host, as the percentage of BrdU uptake by CD4+Foxp3− cells was similar in the spleen and surprisingly higher in WT cells in the MLN

(Figure 1f and Supplementary Figure S1b). This difference, however, was minimal in terms of the number of CD4+Foxp3−BrdU+ cells (Supplementary Figure S1c). Interestingly, there was an

increased percentage and number of Bim−/− CD4+Foxp3+BrdU+ cells in the spleen (Figure 1f and Supplementary Figure S1c). This may partially explain the larger fold difference between the

number of Bim−/− iTregs and WT iTregs in the spleen (∼3-fold difference) compared with MLN (∼2-fold difference).

Enhanced in vivo iTreg induction and decreased incidence of IBD in the absence of Bim. Tconv were sorted from the spleen and lymph nodes of C57BL/6 (WT) or Bim−/− mice and injected i.v. into

the tail veins of Rag1−/− recipients (0.4 × 106 cells per mouse). Four weeks post-transfer, spleen, MLN and IEL/LPL (intestine) were analyzed by flow cytometry. Percentage and number of

CD4+ (a and b) and CD4+Foxp3+ (c and d) cells. Data are the mean±S.E. of two to three individual experiments with five to seven mice per group (a−d), and statistical analysis performed using

standard unpaired t-test (*P