TRI in the neonatal development of CD4⁺Foxp3⁺ thymocytes

We first sought to determine whether ablation of TGF- signaling influences the development of natural CD4⁺Foxp3⁺ T_reg cells in the thymus. For this, we bred mice with a loxP-flanked gene encoding TRI (Tgfbr1) with mice expressing Cre under control of the proximal Lck promoter^{39, 40}, thus creating mice with T cell lineage–specific deletion of TRI (called 'Tgfbr1^f/fLck-Cre⁺ mice' here; Supplementary Fig. 1a,b online). These mice developed lethal inflammation (Supplementary Fig. 1c–e) resembling that of mice lacking TGF-1 systemically or lacking TRII in T cells^{23, 37, 38}. We examined CD4⁺CD25⁺Foxp3⁺ thymocytes 3–5 d after birth because that is a period critical for the initial generation of natural T_reg cells in wild-type mice^{41, 42}. In addition, observations made during this time period are less likely to be complicated by secondary effects arising from systemic inflammation. We noted no significant differences in total thymocyte numbers or the distribution of CD4 and CD8 expression on total thymocytes for Tgfbr1^f/fLck-Cre⁺ and age-matched wild-type or Tgfbr1^f/+Lck-Cre⁺ control mice (Fig. 1a and data not shown). There were similar percentages of CD4⁺Foxp3⁺ thymocytes in Tgfbr1^+/+Lck-Cre⁺ and Tgfbr1^f/+Lck-Cre⁺ mice (data not shown). However, compared with control mice, Tgfbr1^f/fLck-Cre⁺ mice had substantially fewer and a smaller proportion of Foxp3⁺ thymocytes during this time period (Fig. 1a,b). At day 3, Tgfbr1^f/fLck-Cre⁺ mice lacked detectable Foxp3⁺ T cells in the CD4⁺CD8⁺ double-positive and CD4⁺CD8^- single-positive thymocyte populations, whereas Tgfbr1^f/+Lck-Cre⁺ control mice had 0.03–0.05% and 0.6–0.9% Foxp3⁺ cells in the double-positive and CD4⁺CD8^- single-positive thymocyte populations, respectively (Fig. 1a,b and data not shown). Even at days 4 and 5, when Foxp3⁺ thymocyte numbers increased to 'adult' numbers in wild-type mice⁴², the number of CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice remained at background amounts (Fig. 1a,b). The deficiency in CD4⁺Foxp3⁺ thymocytes was not due to a generalized defect in CD4⁺ T cell development, as the numbers of CD4⁺CD8^- single-positive and CD4⁺CD25⁺Foxp3^- thymocytes were similar in Tgfbr1^f/fLck-Cre⁺ and control mice (Fig. 1a and data not shown). The lower number of CD4⁺CD25⁺Foxp3⁺ cells in the Tgfbr1^f/fLck-Cre⁺ thymus persisted until about 1 week and then increased from the onset of the inflammatory syndrome (Supplementary Fig. 2a online and data not shown). These data indicate that ablation of TGF- signaling in thymocytes blocks the development of natural Foxp3⁺ T_reg cells in the thymus.

Figure 1: Deletion of TRI blocks CD4⁺CD25⁺Foxp3⁺ thymocyte development.

(a) Flow cytometry of whole thymocytes (top row) and gated CD4⁺CD8^- thymocytes (bottom row). Numbers in quadrants indicate (top row) percent CD4⁺CD8^- cells (top left) or CD4^- CD8⁺ cells (bottom right) or (bottom row) percent Foxp3⁺CD25^- cells (top left), Foxp3⁺CD25⁺ cells (top right) or Foxp3^- CD25⁺ cells (bottom right). Each plot is of one mouse representative of four (day 3, Tgfbr1^f/fLck-Cre⁺), seven (day 3, Tgfbr1^f/+Lck-Cre⁺), ten (days 4–5, Tgfbr1^f/fLck-Cre⁺) or fourteen (days 4–5, Tgfbr1^f/+Lck-Cre⁺). (b) Percent Foxp3⁺CD4⁺ thymocytes (mean s.d.) in the mice in a. ^*, P = 0.000005; ^**, P = 0.00000015. Data represent two to three independent experiments.

Full size image (58 KB)

Next we used single or mixed bone marrow transplantation to further confirm the function of TGF- signaling in the thymic development of natural CD4⁺CD25⁺Foxp3⁺ T_reg cells. We first injected bone marrow cells isolated from Tgfbr1^f/fLck-Cre⁺ or control littermates into sublethally irradiated recombination-activating gene 1–deficient (Rag1^–/–) recipient mice. We isolated and analyzed thymocytes from the chimeras at 5 weeks after transplantation. Thymi reconstituted with control bone marrow contained a significantly higher proportion of CD25⁺Foxp3⁺ cells than did thymi transplanted with Tgfbr1^f/fLck-Cre⁺ bone marrow (Fig. 2a,b). As the total number of thymocytes in the control and Tgfbr1^f/fLck-Cre⁺ chimeras was similar, the absolute number of CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ recipients was also lower (data not shown). We consistently obtained similar results with sublethally irradiated Rag1^–/– mice reconstituted with a mixture of C57BL/6 (CD45.1⁺) and Tgfbr1^f/fLck-Cre⁺ (CD45.2⁺) bone marrow or a mixture of C57BL/6 (CD45.1⁺) and Tgfbr1^f/+Lck-Cre⁺ (CD45.2⁺) bone marrow. We analyzed CD4⁺CD25⁺Foxp3⁺ 3–4 weeks after transplantation (the time when T cells normally begin to appear in the blood). Notably, Tgfbr1^f/fLck-Cre⁺ bone marrow contributed less than did coinjected C57BL/6 bone marrow to the CD4⁺CD25⁺Foxp3⁺ thymocyte pool, whereas coinjected Tgfbr1^f/+Lck-Cre⁺ and C57BL/6 showed no significant difference in their contribution to the CD4⁺CD25⁺Foxp3⁺ thymocyte population (Fig. 2c,d). These observations collectively demonstrate a previously unrecognized specific function for TGF- signaling in the induction of the differentiation of Foxp3⁺ T_reg cells in the thymus.

Figure 2: Cell-autonomous requirement for TRI in the development of CD4⁺CD25⁺Foxp3⁺ thymocytes.

Flow cytometry of thymocytes isolated from sublethally irradiated Rag1^–/– mice 3–5 weeks after transfer () of bone marrow from Tgfbr1^f/fLck-Cre⁺ (CD45.2⁺) mice or Tgfbr1^f/+Lck-Cre⁺ (CD45.2⁺) mice injected alone (a,b) or mixed with C57BL/6 (CD45.1⁺) bone marrow at a ratio of 1:2 (c,d). (a) Gated CD4⁺CD8^- thymocytes of one mouse representative of three (recipients of Tgfbr1^f/+Lck-Cre⁺ bone marrow) or two (recipients of Tgfbr1^f/fLck-Cre⁺ bone marrow). (b) Percent CD4⁺Foxp3⁺ thymocytes (mean s.d.) in the mice in a. ^*, P = 0.03. (c) Gated CD4⁺CD8^- thymocytes of one mouse representative of five (recipients of C57BL/6 plus Tgfbr1^f/+Lck-Cre⁺ bone marrow) or four (recipients of C57BL/6 plus Tgfbr1^f/fLck-Cre⁺ bone marrow). (d) Percent CD4⁺Foxp3⁺ thymocytes (mean s.d.) in the mice in c. ^*, P = 0.026 (one-tail); ^**, P = 0.002. This experiment was repeated twice with similar results.

Full size image (53 KB)

TRI in CD4⁺ thymocyte activation and proliferation

Despite the profound deficiency of CD4⁺CD25⁺Foxp3⁺ thymocytes in 3- to 5-day-old Tgfbr1^f/fLck-Cre⁺ mice, after 1 week of age, Tgfbr1^f/fLck-Cre⁺ mice showed population expansion of CD4⁺CD25⁺Foxp3⁺ thymocytes (Supplementary Fig. 2a,b), whereas the frequency of CD4⁺Foxp3⁺ cells remained extremely low in the periphery (Supplementary Fig. 2b,d). By 3–4 weeks of age, when most of the Tgfbr1^f/fLck-Cre⁺ mice developed systemic inflammatory syndrome, the proportion of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice was even higher than that in control mice (Fig. 3a,b and Supplementary Fig. 2). CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice had higher expression of the T_reg cell markers CD25, GITR and intracellular CTLA-4 (Fig. 3c,d). CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice also showed upregulation of CD69 and CD45RB and downregulation of CD62L (Fig. 3d). Thus, in a setting of immune activation, CD4⁺Foxp3⁺ thymocytes had an activated phenotype in the absence of TGF- signaling. Notably, however, CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice expressed this set of activation markers even before the inflammatory syndrome appeared (data not shown). Foxp3 protein and transcript expression per cell was also slightly higher in CD4⁺Foxp3⁺CD25⁺ thymocytes from Tgfbr1^f/fLck-Cre⁺ mice than in those from control mice, but these increases failed to reach statistical significance (data not shown). Thus, these data indicate that ablation of TGF- signaling results in activation of CD4⁺CD25⁺Foxp3⁺ thymocytes.

Figure 3: Recovery and activated phenotype of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice.

(a) Expression of CD25 and Foxp3 by gated CD4⁺CD8^- thymocytes from Tgfbr1^f/fLck-Cre⁺ mice (n = 11) and control mice (n = 13) at the age of 3–4 weeks. (b) Percent CD4⁺Foxp3⁺ thymocytes (mean s.d.) in the CD4⁺ population of the mice in a. ^*, P = 0.0011. 'Tgfbr1^+/+ or Tgfbr1^f/+Lck-Cre⁺' includes Tgfbr1^+/+Lck-Cre⁺ and Tgfbr1^f/+Lck-Cre⁺ control mice. Data are representative of three experiments. (c) CD25 expression by CD4⁺Foxp3⁺ thymocytes from 2- to 4-week-old Tgfbr1^f/fLck-Cre⁺ and control mice, presented as mean fluorescence intensity (MFI) s.d. of five mice (Tgfbr1^f/fLck-Cre⁺) or seven mice (control). ^*, P = 0.012. This experiment was repeated four times with similar results. (d) Expression of activation markers on gated CD4⁺Foxp3⁺ thymocytes. Data are from one representative mouse from each group of at least three mice; this experiment was repeated more than three times.

Full size image (64 KB)

We then investigated whether TRI ablation influenced the survival of CD4⁺CD25⁺Foxp3⁺ thymocytes. CD4⁺CD25⁺ and CD4⁺CD25^- thymocytes freshly isolated from Tgfbr1^f/fLck-Cre⁺ mice had a slightly higher proportion of cells positive for 7-amino-actinomycin D (7-AAD⁺) than did those from control mice, but this difference failed to reach statistical significance (Supplementary Fig. 3a online). Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ thymocytes had lower expression of the antiapoptotic protein Bcl-2 than did their control counterparts (Supplementary Fig. 3b). However, we also noted this lower Bcl-2 expression in CD4⁺CD25^- and CD8⁺CD4^- thymocytes from Tgfbr1^f/fLck-Cre⁺ mice. We labeled CD4⁺ thymocytes from control or Tgfbr1^f/fLck-Cre⁺ mice with the cytosolic dye CFSE and adoptively transferred them into lethally irradiated C57BL/6 mice. At 3 d after transfer, Foxp3⁺ and Foxp3^- splenocytes and lymph node cells derived from Tgfbr1^f/fLck-Cre⁺ mice had undergone more cell divisions than had their control counterparts (Supplementary Fig. 3c and data not shown). Consistent with the findings obtained with freshly isolated Tgfbr1^f/fLck-Cre⁺ thymocytes, CD4⁺ splenocytes derived from Tgfbr1^f/fLck-Cre⁺ mice had a higher proportion of 7-AAD⁺ T cells (Supplementary Fig. 3d) and lower Bcl-2 expression (Supplementary Fig. 3e) than did their control counterparts. Thus, in the absence of TGF- signaling, CD4⁺ thymocytes undergo slightly more apoptosis but show enhanced activation and proliferation.

Defective immunosuppression by TRI-deficient T_reg cells

We next examined the ability of CD4⁺CD25⁺Foxp3⁺ T_reg cells from Tgfbr1^f/fLck-Cre⁺ or control mice to inhibit the proliferation of CD4⁺CD25^- T cells in an established in vitro coculture assay⁴³. CD4⁺CD25⁺Foxp3⁺ T_reg cells from Tgfbr1^f/fLck-Cre⁺ thymus and spleen showed a defect in suppressing TCR-driven CD4⁺CD25^- T cell proliferation, as determined by either CFSE dilution (Supplementary Fig. 4 online) or ³H incorporation (data not shown). This defect was not attributable to the amount of Foxp3 expression per cell, because compared with their control counterparts, Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺Foxp3⁺ thymocytes and T cells had similar or higher expression of Foxp3 (data not shown).

Because of the technical limitations of T_reg cell isolation, we were unable to exclude the possibility that a few CD4⁺CD25⁺Foxp3^- activated T cells contaminated the CD4⁺CD25⁺ cell preparation; such contaminants might have been responsible for the slightly larger amount of IL-2 noted in the Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^hi population (Supplementary Fig. 5 online). However, both Tgfbr1^f/fLck-Cre⁺ and control CD4⁺CD25^+hi populations contained similar proportions of Foxp3⁺ T cells (85–90%) and were anergic to TCR stimulation in vitro in the absence of exogenous IL-2 (data not shown). Thus, a minor contamination of CD4⁺CD25⁺Foxp3^- activated and/or effector cells could not entirely account for the defective suppressive ability of Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ cells. Although Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ populations contained a slightly higher proportion of 7-AAD⁺ cells, analysis of coculture systems showed similar frequencies of 7-AAD⁺ cells in Tgfbr1^f/fLck-Cre⁺ and control CD4⁺CD25⁺ populations after 24 or 72 h of culture (data not shown). Thus, CD4⁺CD25⁺ cell death was unlikely to have been the chief cause of defective suppressive activity. These findings are consistent with a published report showing that TGF- signaling in CD4⁺Foxp3⁺ T_reg cells may also be required for their immunosuppressive activity³⁵, although the detailed mechanisms remain to be elucidated.

Proliferation of natural T_reg cells and IL-2 responsiveness

Next we set out to understand how CD4⁺CD25⁺Foxp3⁺ cells were replenished in Tgfbr1^f/fLck-Cre⁺ mice during later and inflammatory stages. As CD4⁺Foxp3⁺ cells in the Tgfbr1^f/fLck-Cre⁺ thymus had an activated phenotype, we first reasoned that these cells might have a higher rate of population expansion. We pulse-labeled Tgfbr1^f/fLck-Cre⁺ or control mice with the thymidine analog BrdU for 16 h and noted that Tgfbr1^f/fLck-Cre⁺ mice had two- to threefold more BrdU⁺ CD4⁺Foxp3⁺ thymocytes than did control mice, whereas CD4⁺Foxp3^- Tgfbr1^f/fLck-Cre⁺ thymocyte populations had similar or even slightly lower proportions of BrdU⁺ cells than did their control counterparts (Fig. 4a,b).

Figure 4: CD4⁺CD25⁺Foxp3⁺ thymocytes show more cycling and proliferation in response to IL-2 in the absence of TRI.

(a,b) Flow cytometry of thymocytes isolated from 2- to 3-week-old mice injected with BrdU for 16 h. (a) Gated CD4⁺CD8^- thymocytes in one representative mouse of each group. Numbers in quadrants indicate percent cells in each. (b) Percent BrdU⁺ cells (mean s.d.) among the CD4⁺Foxp3⁺ or CD4⁺Foxp3^- thymocyte populations (n = 3 mice per group). ^*, P = 0.0011; ^**, P = 0.0022; NS, not significant. This experiment was repeated twice with similar results. (c) Real-time PCR of IL-2 mRNA in CD4⁺CD25^- thymocytes, relative to that in freshly isolated control Tgfbr1^+/+Lck-Cre⁺ CD4⁺CD25^- thymocytes, set as 1. Data are the mean s.d. of pooled cells from two mice in each group, and this experiment was repeated two to three times at each time point with similar results. (d) [³H] incorporation by CD4⁺CD25⁺ thymocytes from 2- to 3-week-old mice, stimulated for 3 d with medium (Med), anti-CD3 (-CD3) or IL-2 in the presence of splenic APCs, presented as the mean s.d. of triplicate samples. This experiment was repeated twice with similar results. (e) Flow cytometry of CD4⁺CD25⁺ thymocytes labeled with CFSE and cultured with various stimuli (above plots). Data show gated CD4⁺Foxp3⁺ thymocytes; quadrants were set according to the CFSE fluorescence of live Foxp3⁺ thymocytes cultured with medium alone. Numbers in quadrants indicate percent undivided (CFSE^hi) cells in the designated gate. Data represent pooled cells from two to three mice in each group; this experiment was repeated three times with similar results.

Full size image (48 KB)

Because almost all Tgfbr1^f/fLck-Cre⁺ CD4⁺Foxp3⁺ thymocytes had high expression of CD25 (Fig. 3) and IL-2 is crucial for the maintenance and expansion of natural Foxp3⁺ T_reg cell populations^{20, 21, 22, 28}, we hypothesized that the greater proliferation of Tgfbr1^f/fLck-Cre⁺ CD4⁺Foxp3⁺ thymocytes was attributable to an overproduction of and 'supersensitive' response to IL-2 in the thymus. Indeed, thymocytes freshly isolated from Tgfbr1^f/fLck-Cre⁺ mice 2–3 weeks of age had five- to sixfold more IL-2 mRNA than did control thymocytes (data not shown). Specifically, without exogenous treatment or stimuli, Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^- thymocytes had 50-fold and 100- to 200-fold more IL-2 mRNA than did control CD4⁺CD25^- thymocytes for mice 1–2 weeks and 3–4 weeks of age, respectively (Fig. 4c). Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^lo but not CD4⁺CD25^hi thymocytes also produced large amounts of IL-2 (Supplementary Fig. 5). Analysis of other subsets of thymocytes showed that CD8⁺ cells from Tgfbr1^f/fLck-Cre⁺ mice also expressed slightly more IL-2 than did wild type control cells (Supplementary Fig. 5). Neither CD4⁺CD8⁺ nor CD4^- CD8^- thymocytes from Tgfbr1^f/fLck-Cre⁺ mice had higher IL-2 expression. Thymic CD11c⁺ dendritic cells from Tgfbr1^f/fLck-Cre⁺ and control mice produced similar amounts of IL-2 (Supplementary Fig. 5). Peripheral CD4⁺ T cells from Tgfbr1^f/fLck-Cre⁺ mice also produced large quantities of IL-2 (data not shown).

To determine whether Tgfbr1^f/fLck-Cre⁺ CD4⁺Foxp3⁺ thymocytes were hypersensitive to IL-2 stimulation, we cultured CD4⁺CD25⁺Foxp3⁺ thymocytes with IL-2 in the presence or absence of TCR stimulation. Notably, in the absence of exogenous TCR stimulation, IL-2 alone caused vigorous proliferation of Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ thymocytes but minimal proliferation of control CD4⁺CD25⁺ thymocytes (Fig. 4d,e). The addition of exogenous TGF-1 suppressed the IL-2-mediated growth of control but not Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ thymocytes (Fig. 4e). Stimulation with both IL-2 and CD3-specific antibodies resulted in similar proliferation of control and Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺ thymocytes (Fig. 4e). Consistent with their 'supersensitive' response to IL-2 stimulation, Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25⁺Foxp3⁺ thymocytes showed greater phosphorylation of STAT5, a critical mediator of IL-2 signal transduction (Supplementary Fig. 6 online). These data further strengthen the idea that IL-2 is involved in the expansion of Tgfbr1^f/fLck-Cre⁺ thymocyte populations.

Because activated T cells may recirculate back to the thymus⁴⁴ and Tgfbr1^f/fLck-Cre⁺ mice showed T cell activation and a systemic inflammatory syndrome, the late replenishment of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice might have been the result of recirculation of peripheral CD4⁺CD25⁺Foxp3⁺ cells back to the thymus. To examine that possibility, we first adoptively transferred 10 10⁶ CFSE-labeled peripheral CD4⁺ T cells isolated from 3- to 4-week-old Tgfbr1^f/fLck-Cre⁺ mice into Tgfbr1^f/fLck-Cre⁺ mice of a similar age or into lethally irradiated syngeneic C57BL/6 mice; the latter recipients served as a positive control for T cell recirculation to the thymus⁴⁴. Consistent with published observations⁴⁴, at 3–4 d after transfer, peripheral CD4⁺ T cells recirculated back to the thymus, especially in the lethally irradiated C57BL/6 recipients (Supplementary Fig. 7a online). However, CFSE⁺ CD4⁺ T cells accounted for almost 40% of total CD4⁺ cells in the spleen but less than 1% of total CD4⁺ thymocytes in the irradiated C57BL/6 recipients (Supplementary Fig. 7a,b). We noted similar trends for Tgfbr1^f/fLck-Cre⁺ recipients, but the percentages of donor T cells present in all organs were much lower (Supplementary Fig. 7a,b). Thus, very few donor Tgfbr1^f/fLck-Cre⁺ CD4⁺ T cells recirculated to the thymus in the inflammatory environment of Tgfbr1^f/fLck-Cre⁺ recipients. In addition, Foxp3⁺ cells constituted only a small fraction of the few CD4⁺ T cells that did recirculate to the thymus. These data suggested that most CD4⁺Foxp3⁺ thymocytes in 3- to 4-week-old Tgfbr1^f/fLck-Cre⁺ mice were generated and proliferated locally. To strengthen that conclusion, we examined thymic recirculation 12 d after adoptive transfer of 9 10⁶ CD45.2⁺ peripheral CD4⁺ T cells from Tgfbr1^f/fLck-Cre⁺ mice into sublethally irradiated C57BL/6 (CD45.1⁺) mice. Whereas almost half of the splenic CD4⁺ T cells were donor derived, only 2–4% of the Foxp3⁺ thymocytes were derived from the injected Tgfbr1^f/fLck-Cre⁺ T cells (data not shown). We also used a CD4-specific antibody to selectively deplete 5- to 10-day-old Tgfbr1^f/fLck-Cre⁺ mice of peripheral CD4⁺ T cells (over 80–90% depletion for at least 2–3 weeks) but spare CD4⁺ thymocytes. Two weeks later, the proportion of CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice treated with antibody to CD4 (anti-CD4) was not significantly lower than that in untreated Tgfbr1^f/fLck-Cre⁺ mice (data not shown).

Finally, we injected an antibody specific for CD49d (integrin ₄)⁴⁵ into Tgfbr1^f/fLck-Cre⁺ mice. Antibody treatment efficiently blocked the recirculation of donor peripheral Tgfbr1^f/fLck-Cre⁺ CD4⁺ T cells to the thymus but not the spleen in irradiated C57BL/6 recipients (Supplementary Fig. 8a,b online) but did not significantly affect the proportion of CD4⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice (Supplementary Fig. 8c,d). These data collectively suggest that the replenishment of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice is mainly attributable to IL-2-mediated accelerated local proliferation in the absence of TGF- signaling.

Deletion of IL-2 abrogates CD4⁺CD25⁺Foxp3⁺ replenishment

The aforementioned results indicated that TGF- signaling was critical for the development of natural CD4⁺CD25⁺Foxp3⁺ thymocytes and that IL-2 was the main driving force behind the accelerated population expansion of the few CD4⁺Foxp3⁺CD25⁺ thymocytes that nevertheless appeared in the absence of TGF- signaling. If that conclusion were correct, then Tgfbr1^f/fLck-Cre⁺ mice lacking IL-2 should have had persistently and profoundly lower proportions of CD4⁺CD25⁺Foxp3⁺ thymocytes. To test that hypothesis, we generated Tgfbr1^f/+Lck-Cre⁺ mice on an Il2^–/– background. The resulting Tgfbr1^f/fLck-Cre⁺Il2^–/– mice seemed healthy at birth and were phenotypically indistinguishable from wild-type Tgfbr1^+/+Lck-Cre⁺Il2^+/+, Tgfbr1^+/+Lck-Cre⁺Il2^–/– or Tgfbr1^f/fLck-Cre⁺Il2^+/+ mice in terms of total thymic cellularity and CD4-versus-CD8 thymocyte profile (Fig. 5a). Consistent with published reports^{20, 21}, deletion of IL-2 alone resulted in only slightly fewer CD4⁺Foxp3⁺ thymocytes, although CD25 expression was downregulated (Fig. 5). As expected, at the age of 3–4 weeks, CD4⁺Foxp3⁺ thymocytes from Tgfbr1^f/fLck-Cre⁺Il2^+/+ mice had higher expression of CD25 than did cells from control mice (Fig. 5). Notably, Tgfbr1^f/fLck-Cre⁺Il2^–/– mice, even at the age of 3–4 weeks, completely lacked CD4⁺CD25⁺Foxp3⁺ thymocytes and had only a small number of CD4⁺CD25^- Foxp3⁺ thymocytes (Fig. 5), similar to the baseline number of thymic CD4⁺Foxp3⁺ T cells in 3- to 5-day-old Tgfbr1^f/fLck-Cre⁺Il2^+/+ mice (Fig. 1). We detected no Foxp3⁺ T cells in the CD4⁺CD8⁺ population of Tgfbr1^f/fLck-Cre⁺Il2^–/– mice (data not shown). As the total number of thymocytes was similar in Tgfbr1^f/fLck-Cre⁺Il2^–/– and control mice, the absolute number of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice was also negligible (data not shown).

Figure 5: Deletion of IL-2 results in a complete lack of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice.

(a) Flow cytometry of thymocytes isolated from 3- to 4-week-old mice: top, whole thymocytes; bottom, gated CD4⁺CD8^- thymocytes. Data are from one mouse representative of eight (Tgfbr1^+/+Lck-Cre⁺Il2^+/+), four (Tgfbr1^f/fLck-Cre⁺Il2^+/+), five (Tgfbr1^+/+Lck-Cre⁺Il2^–/–) or four (Tgfbr1^f/fLck-Cre⁺Il2^–/–). Numbers in quadrants indicate percent cells in the designated gate. (b) Percent CD4⁺Foxp3⁺ thymocytes (mean and s.d.) in the mice in a. ^*, P = 0.00002; ^**P = 0.0000001; ^***P = 0.0003. Data are representative of three independent experiments.

Full size image (64 KB)

To exclude the possibility that the absence of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice was due to a defect in Foxp3 induction in the absence of IL-2, we examined CD4⁺Foxp3⁺ thymocytes in Il2^–/– mice at neonatal days 4–5, a critical period for natural T_reg cell development. In contrast to the profound deficiency in Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice (Fig. 1a,b), Il2^–/– mice had slightly but not significantly fewer Foxp3⁺ thymocytes (data not shown). These data further support the idea that TGF- signaling, not IL-2 signaling, is the main factor driving the induction of Foxp3 expression in thymic Foxp3^- precursors.

To further confirm that observation, we created bone marrow chimeras by injecting bone marrow cells from Tgfbr1^f/fLck-Cre⁺Il2^–/– or control mice into sublethally irradiated Rag1^–/– mice. Consistent with the data obtained with unmanipulated Tgfbr1^f/fLck-Cre⁺Il2^–/– mice, the thymi of the mice reconstituted with the Tgfbr1^f/fLck-Cre⁺Il2^–/– bone marrow showed a complete deficiency in CD4⁺CD25⁺Foxp3⁺ cells at 5 weeks after transfer (Fig. 6). In contrast, mice reconstituted with control bone marrow had substantial numbers of CD4⁺Foxp3⁺ thymocytes (Fig. 6). These data collectively show that TGF- signaling is key in the induction of CD4⁺CD25⁺Foxp3⁺ thymocyte populations, whereas IL-2 is crucial for their expansion.

Figure 6: Requirement for both TGF- signaling and IL-2 in CD4⁺CD25⁺Foxp3⁺ thymocyte development, demonstrated by bone marrow chimeras.

(a) Flow cytometry of thymocytes isolated from sublethally irradiated Rag1^–/– mice 5 weeks after transplantation of bone marrow from Tgfbr1^f/fLck-Cre⁺Il2^–/– or Tgfbr1^+/+Lck-Cre⁺Il2^+/+ mice: top, total thymocytes; bottom, gated CD4⁺CD8^- thymocytes. Numbers in quadrants indicate percent cells in the designated gate. Data are from one mouse representative of three (Tgfbr1^+/+Lck-Cre⁺Il2^+/+) or four (Tgfbr1^f/fLck-Cre⁺Il2^–/–). (b) Percent CD4⁺Foxp3⁺ thymocytes (mean s.d.) in the mice in a. Data are representative of two experiments.

Full size image (48 KB)

Deletion of TRI and IL-2 results in complete T_reg cell deficiency

We then examined peripheral T_reg cells in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice. Notably, like Tgfbr1^f/fLck-Cre⁺Il2^+/+ mice, conditional TRII-knockout mice³⁷, and TGF--knockout mice⁴⁶, Tgfbr1^f/fLck-Cre⁺Il2^–/– mice had many more CD4⁺ T cells than did control Tgfbr1^+/+Lck-Cre⁺Il2^+/+ mice, which resulted in an increase in the CD4⁺/CD8⁺ ratio (Fig. 7a and data not shown). However, Tgfbr1^f/fLck-Cre⁺Il2^–/– mice showed a complete lack of CD4⁺CD25⁺Foxp3⁺ T_reg cells in the spleen (Fig. 7a,b), lymph nodes and nonlymphoid tissues such as lungs and liver (data not shown). Tgfbr1^+/+Lck-Cre⁺Il2^–/– mice had fewer but detectable CD4⁺Foxp3⁺ T_reg cells than did Tgfbr1^+/+Lck-Cre⁺Il2^+/+ mice in the periphery (Fig. 7a,b). Peripheral CD4⁺Foxp3^- and CD8⁺ T cells in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice had an activated CD44^pos–hiCD45RB^pos–hiCD69⁺CD62L^neg–lo phenotype (Fig. 7c and data not shown). The mean fluorescence intensity of CD44 and CD45RB in peripheral Tgfbr1^f/fLck-Cre⁺Il2^–/– T cells was higher than that in the wild-type control T cells (data not shown). Despite the complete lack of CD4⁺Foxp3⁺ T cells and the activation of the remaining CD4⁺Foxp3^- cells, the vital organs of Tgfbr1^f/fLck-Cre⁺Il2^–/– mice had an extent of inflammatory cell infiltration similar to that of Tgfbr1^f/fLck-Cre⁺Il2^+/+ and Tgfbr1^+/+Lck-Cre⁺Il2^–/– mice (data not shown).

Figure 7: Peripheral CD4⁺CD25⁺Foxp3⁺ T cells are absent from and CD4⁺Foxp3^- T cells have an activated phenotype in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice.

(a) Flow cytometry of whole splenocytes (top) and gated CD4⁺ T cells (bottom). Numbers in quadrants indicate percent cells in the designated gates. Data are from one mouse representative of eight (Tgfbr1^+/+Lck-Cre⁺Il2^+/+), five (Tgfbr1^+/+Lck-Cre⁺Il2^–/–) or four (Tgfbr1^f/fLck-Cre⁺Il2^–/–). (b) Percent CD4⁺Foxp3⁺ T cells (mean s.d.) among CD4⁺ cells (top) and percent CD4⁺ T cells among whole splenocytes (bottom), of the cells in a. ^*, P = 0.005; ^**, P = 0.0000014; ^***, P = 0.002. Data (a,b) are representative of three experiments. (c) Expression of activation markers on splenocytes, assessed as gated CD4⁺ T cells. Numbers in quadrants indicate percent cells in designated gates. This experiment was repeated twice with similar results.

Full size image (49 KB)

We then used the bone marrow chimeras to verify the cell-autonomous nature of the peripheral T_reg cell deficiency in Tgfbr1^f/fLck-Cre⁺Il2^–/– mice. Rag1^–/– mice reconstituted with bone marrow from Tgfbr1^f/fLck-Cre⁺Il2^–/– mice had no detectable CD4⁺Foxp3⁺ T cells in the spleen, whereas the spleens of mice reconstituted with control bone marrow had abundant CD4⁺Foxp3⁺ T_reg cells (Fig. 8). These data collectively demonstrate that TGF- signaling in combination with IL-2 is essential for the development of natural T_reg cells and for maintenance of their homeostasis in the periphery.

Figure 8: Requirement for both TGF- signaling and IL-2 in the development of peripheral CD4⁺CD25⁺Foxp3⁺ T cells, as shown by bone marrow chimeras.

(a) Flow cytometry of splenocytes from Rag1^–/– chimeric mice that received Tgfbr1^f/fLck-Cre⁺Il2^–/– or Tgfbr1^+/+Lck-Cre⁺Il2^+/+ bone marrow 5 weeks earlier: top, whole splenocytes; bottom, gated CD4⁺ T cells. Data are from one mouse representative of three (Tgfbr1^+/+Lck-Cre⁺Il2^+/+) or four (Tgfbr1^f/fLck-Cre⁺Il2^–/–). (b) Percent CD4⁺Foxp3⁺ T cells (mean s.d.) among CD4⁺ cells (top) and percent CD4⁺ T cells among whole splenocytes (bottom), of the cells in a. Data are representative of two experiments.

Full size image (52 KB)

Deletion of TRI blocks the conversion of T_reg cells

We next studied molecular mechanisms that might account for the defect in the generation of CD4⁺CD25⁺Foxp3⁺ thymocytes in Tgfbr1^f/fLck-Cre⁺ mice. We focused on the phosphorylation and nuclear translocation of the signal transducers Smad2 and Smad3, a critical step in TGF- signal transduction⁴⁷ in T cells. Phosphorylated Smad2 and Smad3 translocate into the nucleus to mediate TGF- signaling, whereas nonphosphorylated Smad2 and Smad3 are retained in the cytoplasm. We cultured CD4⁺CD25^- thymocytes from Tgfbr1^f/fLck-Cre⁺ and control mice for 2 h with anti-CD3 in the presence and absence of TGF-. Without TGF-, neither Tgfbr1^f/fLck-Cre⁺ nor control thymocytes showed detectable Smad2 phosphorylation, although all populations expressed nonphosphorylated Smad2 and Smad3 (Supplementary Fig. 9a online). The addition of TGF- upregulated Smad2 phosphorylation in control but not Tgfbr1^f/fLck-Cre⁺ thymocytes. To visualize the intracellular location of Smad2 and Smad3, we stained thymocytes with the DNA-intercalating dye DAPI and antibody to Smad2 and Smad3 and analyzed the localization of Smad2 and Smad3 by confocal microscopy. TGF- treatment led to accumulation of Smad2 and Smad3 in the nuclei of control but not Tgfbr1^f/fLck-Cre⁺ thymocytes (Supplementary Fig. 9b). Thus, the abrogation of TGF- signaling in Tgfbr1^f/fLck-Cre⁺ thymocyte may be attributed at least in part to impaired phosphorylation and nuclear localization of Smad2 and Smad3.

We then assessed whether Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^- single-positive thymocytes, in the absence of phosphorylation and nuclear translocation of Smad2 and Smad3, could differentiate into Foxp3⁺ T_reg cells in vitro and in vivo. Stimulation with anti-CD3 in the presence of TGF- induced the conversion of control but not Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^- thymocytes into Foxp3⁺ T_reg cells in vitro^{28, 31} (Supplementary Fig. 9c). To assess conversion in vivo, we adoptively transferred Tgfbr1^f/fLck-Cre⁺ or control CD4⁺CD25^- thymocytes into Rag1^–/– mice and analyzed them 8 weeks later. Control but not Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^- thymocytes produced substantial Foxp3⁺ T_reg cell populations in the spleens and lymph nodes of recipient mice (Supplementary Fig. 9d). The defect was specific for Foxp3 induction, as the same Tgfbr1^f/fLck-Cre⁺ thymocytes produced similar or even greater numbers of interferon--positive T cells in vitro and after injection into Rag1^–/– mice (data not shown). Thus, the complete block of phosphorylation and nuclear translocation of Smad2 and Smad3 in Tgfbr1^f/fLck-Cre⁺ CD4⁺CD25^- Foxp3^- thymocytes may account for their impaired conversion into Foxp3⁺ T_reg cells and the deficit of T_reg cells in Tgfbr1^f/fLck-Cre⁺ mice. Given all of the data we have presented here, we propose a model in which TGF- signaling in combination with IL-2 is essential for the development of natural T_reg cells (Supplementary Fig. 10 online).

Mice.

The generation of Tgfbr1^f/f mice has been described⁴⁰. Lck-Cre–transgenic mice were from the Jackson Laboratory. Genotypes were determined by PCR as reported⁴⁰. Littermates with wild-type alleles (Tgfbr1^+/+Lck-Cre⁺) or one loxP-flanked allele (Tgfbr1^f/+Lck-Cre⁺) were used as control mice. Tgfbr1^f/fLck-Cre⁺ mice on the 'pure' C57BL/6 background were obtained by backcrossing for more than six generations. Il2^–/– mice and C57BL/6 CD45.1⁺ (ptprc^a) mice were from the Jackson Laboratory. All animal studies were done according to National Institutes of Health guidelines for the use and care of live animals and were approved by the Institutional Animal Care and Use Committee of National Institute of Dental and Craniofacial Research.

Antibodies and reagents.

Phycoerythrin-conjugated anti-CD45RB (16A), anti-CD44 (IM7), anti-CD69 (H1.2F3), and anti-CD62L (MEL-14), purified anti-CD3 (145-2C11; no azide and low endotoxin), anti-CD28, and anti-FcRII/III (to CD32 and CD16; 2.4G2), fluorescein isothiocyanate–, phycoerythrin- or allophycocyanin-conjugated anti-CD25 (7D4 and PC61), and peridinin chlorophyll protein– or fluorescein isothiocyanate–conjugated anti-CD4 (RMA4-5) and anti-CD8 (53-6.7) were from BD Pharmingen. Anti-CD49d (PS/2) was from Southern Biotechnology Associates. Recombinant human TGF-1 was from R&D Systems. A phycoerythrin- or fluorescein isothiocyanate–conjugated anti-mouse/rat Foxp3 Staining Set (FJK-16s, eBioscience) was used for intracellular Foxp3 staining according to the manufacturer's recommendations. DNeasy tissue kits and RNeasy mini kits were from Qiagen. Staining with 7-AAD was done as described⁴⁶. For Bcl-2 staining, T cells were stained first with the appropriate antibodies to surface markers, followed by staining for intracellular Bcl-2 with the phycoerythrin-conjugated Bcl-2 Antibody Reagent Set (BD Parmingen).

BrdU labeling.

Mice were injected intraperitoneally with a single dose of 50 mg BrdU (5-bromodeoxyuridine; Becton Dickinson) per kg body weight; 12–16 h later, cells incorporating BrdU were detected by intracellular staining. Cells were stained for surface markers and were then fixed for 30 min with Cytofix solution (BD Pharmingen). Fixed cells were incubated with DNase I and were stained with allophycocyanin-conjugated anti-BrdU in Cytoperm solution (BD Pharmingen) according to the manufacturer's manual.

Cell culture and T cell proliferation assay.

Purified CD4⁺CD25⁺ thymocytes (5 10⁴ cells) from Tgfbr1^f/fLck-Cre⁺ or control mice were cultured for 3 d at 37 °C in 5% CO₂ with IL-2 (5 ng/ml) and wild-type splenic APCs (2 10⁵ cells) in the presence or absence of anti-CD3 (0.5 g/ml) or TGF-1 (2 ng/ml). For standard ³H-incorporation assays, cultured cells were labeled with 1 Ci [³H]thymidine per well for the final 16 h and then collected, then radioactivity was measured with a liquid scintillation counter. In some experiments, T cells were prelabeled with a solution of 5 M CFSE (carboxyfluorescein diacetate succinimidyl diester; Invitrogen), followed by culture for 3 d with various stimuli. CFSE dilution was analyzed on a FACSCalibur. In vitro coculture suppression assays were done as reported before^{27, 43}. In some cultures, CD45.1⁺ C57BL/6 CD4⁺CD25^- T cells were used as responder T cells.

Mixed bone marrow chimeras.

Bone marrow was isolated from 2- to 3-week-old CD45.2⁺Tgfbr1^f/fLck-Cre⁺ or Tgfbr1^f/+Lck-Cre⁺ mice or 5- to 6-week-old CD45.1⁺ C57BL/6 wild-type mice. Bone marrow was then depleted of T cells with anti-CD4 and anti-CD8 magnetic beads with a large depletion column (Miltenyi). Donor bone marrow from CD45.2⁺Tgfbr1^f/fLck-Cre⁺ mice or Tgfbr1^f/+Lck-Cre⁺ mice (1 10⁶ cells) was mixed with bone marrow from CD45.1⁺ C57BL/6 mice (2 10⁶ cells) and was injected intravenously into 6- to 7-week-old sublethally irradiated (450 rads) Rag1^–/– C57BL/6 mice. In some experiments, sublethally irradiated Rag1^–/– C57BL/6 mice were injected with 3 10⁶ T cell–depleted bone marrow from donor mice of a single genotype. At 3 or 5 weeks after reconstitution, thymi, spleens and mesenteric lymph nodes were collected and analyzed by flow cytometry.

Foxp3⁺ T_reg cell–conversion experiments.

Freshly isolated CD4⁺CD8^- CD25^- thymocytes were cultured for 3–5 d with plate-bound anti-CD3 and soluble anti-CD28 in the presence or absence of TGF-1 as described^{27, 31}. T cells were then collected and stained for intracellular Foxp3 and surface markers. For adoptive transfer experiments, CD4⁺CD8^- CD25^- thymocytes were injected into Rag1^–/– C57BL/6 mice. Then, 8 weeks later, spleens and mesenteric lymph nodes were collected from the recipients and stained for CD4, CD25 and Foxp3. Aliquots of cells were stained with anti-CD4 and anti-interferon-. Cells were analyzed on a FACSCalibur (BD).

Additional Methods.

Information on genotyping, cell purification and flow cytometry, real-time PCR and immunoblot analysis and immunofluorescence is available in the Supplementary Methods online.

Statistical analysis.

Student's t-tests (two-tailed) were used to analyze the significance of data comparison, except where otherwise indicated.

Accession codes.

UCSD-Nature Signaling Gateway (http://www.signaling-gateway.org): A002750, A002271, A002273 and A002272.

Note: Supplementary information is available on the Nature Immunology website.

Author contributions

Y.L. designed and did experiments, analyzed data and contributed to the writing of the manuscript; P.Z. designed and did experiments; S.P. and J.L. did experiments; A.B.K. provided critical materials and helped analyze data; and W.C. initiated and directed the research, designed experiments, analyzed data and wrote the manuscript.