A major role for Bim in regulatory T cell homeostasis

Abstract

We have previously shown that regulatory T cells (T_reg) accumulate dramatically in aged animals and negatively impact the ability to control persistent infection. However, the mechanism(s) underlying the age-dependent accrual of T_reg remain unclear. Here, we show that T_reg accumulation with age is progressive and likely not the result of increased thymic output, increased peripheral proliferation, nor from enhanced peripheral conversion. Instead, we found that T_reg from aged mice are more resistant to apoptosis than T_reg from young mice. Although T_reg from aged mice had increased expression of functional IL-7Rα, we found that IL-7R-signaling was not required for maintenance of T_reg in vivo. Notably, aged T_reg exhibit decreased expression of the pro-apoptotic molecule Bim compared to T_reg from young mice. Further, in the absence of Bim, T_reg accumulate rapidly, accounting for more than 25% of the CD4⁺ T cell compartment by 6 months of age. In addition, accumulation of T_reg in Bim-deficient mice occurred after the cells left the transitional recent thymic emigrant compartment. Mechanistically, we show that IL-2 drives preferential proliferation and accumulation of Bim^lo T_reg. Combined, our data suggest that chronic stimulation by IL-2 leads to preferential expansion of T_reg having low expression of Bim, which favors their survival and accumulation in aged hosts.

Introduction

Considerable changes in the function of T cells, in particular decreased responsiveness of CD4⁺ and CD8+ T cells, have been reported in aged hosts (mice and humans). T cells from aged mice and humans exhibit serious defects in TCR-mediated activation, including changes in the cascades of serine/threonine and tyrosine phosphorylation and trouble forming functional immune synapses with antigen-presenting cells (APC) (1–4). Furthermore, many functional defects have been reported in vivo, as evidenced by the decreased ability of aged hosts to control many infections (rev. in (5)). Although a number of the defects described in senescent CD4⁺ and CD8⁺ T cells appear to be intrinsic (rev. in (6)), recent lines of evidence have suggested that the changed balance between effector and regulatory T cells may play a major role in decreased T cell responses in aged hosts.

Regulatory T cells (T_reg), a subset of CD4⁺ T cells, are critical for maintaining self-tolerance (7). T_reg dampen effector responses against persistent infections (8) and tumors (9). T_reg function by decreasing the level of activation, proliferation and cytokine production of effector T cells in mice and humans, and also by controlling the stimulatory functions of dendritic cells (10) (11). T_reg are characterized by their expression of the transcriptional factor FoxP3 (Forkhead box P3), which is critical for their differentiation, function and maintenance (12).

We and other investigators have reported that the frequency of FoxP3⁺CD4⁺ T cells significantly increases in multiple lymphoid tissues from aged mice (13–15), and in the blood of or tissues of healthy elderly humans (15–17). Also, we showed that FoxP3⁺ cells from aged mice have a greater in vitro suppressive activity on a per cell basis than their young counterparts (15). Further, T_reg play a critical role in the increased disease severity and reactivation of chronic Leishmania major infection in old mice (15). Notably, depletion of T_reg in aged mice leads to enhanced effector T cell responses and increased control of L. major, suggesting that, if relieved from T_reg suppression, effector T cells remain functional in vivo (15). Similarly, another group has also reported that T_reg accumulation in aged mice inhibits protective anti-tumor responses (13). Thus, increased T_reg frequency in aged animals likely contributes to age related immunosuppression. However, the mechanisms underlying their accumulation are not well understood. Here we explored the mechanism(s) contributing to altered T_reg homeostasis during aging.

Materials and Methods

Mice

C57BL/6 mice or FoxP3-IRES-GFP knock-in Balb/c were purchased from either Taconic Farms, Jackson Laboratories, or were purchased from the National Institutes of Aging Colony. Bim-deficient (Bim KO) mice were a generous gift from Drs. P. Bouillet and A. Strasser and were backcrossed for 19–20 generations to C57BL/6 (Walter and Eliza Hall Institute, Melbourne, Australia). Rag2p-GFP mice (18) were originally a gift from Dr. Michel Nussenzweig (Rockefeller University, New York) and were backcrossed 10 or more generations to C57BL/6 and then mated to Bim KO mice. Mice were housed under specific pathogen free conditions. All animal protocols were reviewed and approved by our Institutional Animal Care and Use Committee.

Flow Cytometry

Single cell suspensions were generated from thymi, lymph nodes, or spleens and 10⁶ cells/tube were stained with antibodies against CD4, CD44, CD127 (BD Biosciences) and intracellularly for FoxP3 (eBiosciences), Bcl-2 (generated in house), Mcl-1 (Rockland Biochemicals) and Bim (Cell Signaling Technologies). For detection of Mcl-1 and Bim secondary anti-rabbit antibodies (Jackson Immunoresearch) were used. Staining for FoxP3 was performed according to the manufacturer’s instructions. Data were acquired on a FacsCalibur or an LSRII Flow Cytometer and analyzed using CellQuestPro or FacsDIVA software (BD Biosciences).

In vitro conversion

Spleen and pLN were recovered from FoxP3-IRES-GFP knock-in Balb/c mice. Pan-T Cell Isolation kit was used to purify T cells (Miltenyi Biotec). To purify naïve and memory CD4⁺ T cells, T cells were sorted on their expression of CD4, CD62L and CD44 with a FACSAria (BD Biosciences). 5 × 10⁵ cells/well were stimulated with immobilized anti-CD3e (eBioscience, 2.5 ug) and soluble anti-CD28 (eBioscience, 2 ug), in presence of 100 U/ml IL-2 (NIH) and 5 ng TGF-β1 (PeproTech Inc., Rocky Hill, NJ). As controls, cells were stimulated without TGF-β1 or cultured in IL-2-containing medium. GFP expression was analyzed by flow cytometry in gated CD4⁺ cells after 5 days.

BrdU administration

C57BL/6 and Bim^−/− mice were injected i.p. daily with BrdU (0.6 mg/mouse) for the indicated periods of time. Mice not injected and served as negative controls for BrdU staining. Spleen cells were stained for CD4, BrdU (BD Biosciences), FoxP3 and CD44 expression, and analyzed by flow cytometry.

In vivo cytokine administration and blockade

Human IL-7 was purchased from R&D systems and anti-IL-7 (M25 hybridoma) was grown as ascites in Balb/c mice and purified as described previously (19, 20). Immune complexes (IC) were generated by mixing IL-7 with anti-IL-7 at a 1:2 molar ratio for 5 minutes at room temperature. IC were then diluted in PBS and 200ul (the equivalent of 2.5ug IL-7 and 12.5ug anti-IL-7 per mouse) was injected intraperitoneally every other day for one week. Anti-mouse IL-7Ra antibody (A7R34) or control antibody was purchased from BioXcell (Dartmouth, New Hampshire) and 3mg was injected i.p. into mice every other day for one week. Mouse IL-2 was obtained from the Biological Resource Branch of the Research Reagents Program of the National Cancer Institute. IL-2 was resuspended in PBS and 3×10⁴ U was injected i.p. twice daily for various periods of time. IL-2 “pulse-chase” experiments were performed by injecting groups of C57BL/6 or BimKO mice with IL-2 for 3 days and then were injected with BrdU 3 times over the last 36 hours of IL-2 treatment. On days 1 and 7 after the last IL-2/BrdU injection, mice were sacrificed and spleen cells stained with antibodies against CD4, FoxP3 and intracellularly for BrdU and data acquired by flow cytometry. The percent FoxP3+ BrdU+ cells remaining was calculated as follows: % FoxP3+ cells that were BrdU+ on day 7 / % FoxP3+ cells that were BrdU+ on day 1.

Results

T_reg frequency and survival is increased in aged mice

For these studies, young animals were between 1.5–3 months, middle-aged animals were between 9–15 months, and old animals were more than 18 months of age. As we previously showed (15), the frequency and total numbers of T_reg are increased in spleens and mesenteric lymph nodes in aged mice (Fig. 1A, Table S1). T_reg accrual could be due to increased thymic output, increased peripheral conversion, increased proliferation, and/or altered survival. Although thymic output is directly proportional to thymic size as mice age (21), it remained possible that thymic output of T_reg could be selectively increased with age. However, percentages of FoxP3⁺ CD4⁺ single positive (SP) thymic T cells were not significantly different in old and young mice (p = 0.35, Mann-Whitney test, Fig. 1B). As expected, the absolute number of aged CD4⁺FoxP3⁺ SP thymocytes was reduced to approximately one third compared to young animals, due to decreased thymic cellularity in aged mice (Fig 1B). These data suggest that altered T_reg thymic output is not responsible for increased peripheral T_reg in aged mice.

Figure 1. T_reg accumulate in aged mice and exhibit increased survival in vitro.

(A) The percentages of FoxP3⁺CD4⁺ T cells were determined in the spleens and peripheral LNs of 7 young (1.5-month old) and 6 old (22-month old) C57Bl/6 mice. Lines represent means. Statistical differences were evaluated using unpaired t-tests. (B) Thymocytes from 7 young (1.5-month old) and 6 old (22-month old) mice were analyzed by flow cytometry. Cellularity was assessed for each thymus. The percentage of FoxP3⁺ cells was analyzed in CD4⁺ SP thymocytes. Lines represent medians. Statistical differences were evaluated using Mann-Whitney tests. (C). Naïve (CD62L⁺CD44⁻GFP⁻CD4⁺), effector memory (EM, CD62L⁻CD44⁺GFP⁻CD4⁺) and central memory (CM, CD62L⁺CD44⁺GFP⁻CD4⁺) non-T_reg were sorted from a pool of spleen and pLN from young (2-month old, N=3, full line) and middle-aged (12-month old, N=4, dotted line) FoxP3-IRES-GFP knock-in Balb/c mice. 5×10⁵ naïve, EM or CM cells were stimulated with immobilized anti-CD3 Ab and soluble anti-CD28, in presence of IL-2 and TGF-β1. GFP expression was analyzed by flow cytometry after 5 days of culture in all cell populations. (D) Young (1.5-month old) and old (18-month old) mice were injected with BrdU every day for 3 days. The percentage of BrdU⁺ cells was analyzed in gated splenic T_reg (CD4⁺FoxP3+). Values shown are mean % (± SE) of BrdU⁺ cells in each group. (E) LN cells were harvested from 12 young (1.5-month old) and 11 old (20-month old) mice and cultured in S-MEM media containing 10% FBS, in the absence of exogenous cytokines. After 24 hrs in culture, cells were stained for CD4, CD25 and Propidium Iodide (PI). The percentage of dead cells (PI⁺) was analyzed in gated non-T_reg (CD4⁺CD25⁻) and T_reg (CD4⁺CD25⁺). Values shown are mean % (± SE) of PI⁺ cells in each group. p values represent the difference between old and young mice (unpaired t-tests).

Alternatively, enhanced conversion of non-T_reg in the periphery could explain increase T_reg frequency in aged mice. TCR-mediated stimulation in presence of TGF-β1 is efficient at inducing T_reg conversion from naïve non-T_reg in young animals (22). We therefore assessed conversion of naïve non-T_reg from middle-aged mice. In vitro conversion was actually less efficient in T cells from middle-aged compared to young mice (21.8% versus 36.7% in middle-aged versus young mice, respectively, Fig. 1C). Such conversion required TGF- β because less than 0.5% GFP⁺ (FoxP3⁺) cells were detected when non-T_reg were cultured without TGF- β1 (data not shown). It was possible that FoxP3+ T cells in aged mice were converted from memory T cells; however no accrual of FoxP3⁺ cells was found after 5 days of in vitro culture of memory (both effector and central) T cells from middle-aged mice (Fig. 1C) as found previously in young mice (23, 24). Combined, these data show that although in vitro conversion of non-T_reg could occur in aged mice, it was less efficient than in young mice.

The increased frequency and overall numbers of T_reg in aged hosts could also be due to an increased rate of peripheral proliferation. There are few data available on T_reg in vivo turnover rates in non-lymphopenic animals, particularly in aging. Therefore, to assess T_reg proliferation in vivo, we injected mice with bromodeoxy-uridine (BrdU) for 3 days and measured the frequency of T_reg that had incorporated BrdU. The percentage of T_reg that had incorporated BrdU⁺ was similar in young and old mice (Fig. 1D, p<0.45). Thus, these data show that enhanced T_reg proliferation in aged mice is unlikely to significantly contribute to their accumulation in aging.

It is also possible that increased T_reg frequency in old mice is due to enhanced survival of T_reg compared to non-T_reg. To test this, T_reg survival was determined ex vivo, by quantification of the percentage of dead T_reg after 24 hrs in culture. For activated and resting T cells, we found that this assay correlates with in vivo cell survival (19). T_reg from old animals died significantly less over the course of this assay than did T_reg from young mice (Fig. 1E, p = 0.008). In contrast, non-T_reg from old animals died significantly more than non-T_reg from young mice (Fig. 1E, p = 0.001). Importantly, the ratio of dying T_reg/non-T_reg, was significantly decreased in old mice (mean ratio of 1-fold versus 2-fold in old versus young mice, p < 10⁻⁵). We note that in old mice, most of the non-T_reg are likely endogenous memory T cells while in young mice most of the non-T_reg are naïve T cells. Thus, increased survival of T_reg in aged mice, combined with the decreased survival of non-T_reg, likely contributes to the increase in T_reg frequency in aged mice.

Expression of Bim strikingly decreases in T_reg with aging

The pro-apoptotic molecule, Bim, is critical for survival and homeostasis of conventional non-T_reg (19, 25, 26). As age-related changes in Bim expression in T_reg could contribute to their increased frequency, we quantified Bim expression in T_reg in aged and young mice, using a Bim-specific antibody (Fig. 2A). Strikingly, Bim expression was decreased by ~2-fold in old T_reg compared to young T_reg (Fig. 2B, C), whereas its expression was only slightly decreased in old non-T_reg (Fig 2B, C). For non-T_reg cells, we focused on CD4+ CD44^hi FoxP3− cells as this population is probably the most comparable to T_reg. This process is gradual, as Bim expression was already decreased in FoxP3⁺ CD4⁺ T cells from middle-aged mice (Fig. 2D). We also found that levels of the 2 critical anti-apoptotic molecules in T cells, Bcl-2 and Mcl-1 (19, 27), were both decreased ~2-fold in T_reg from old mice, suggesting that neither Bcl-2 nor Mcl-1 contributes to increased survival of aged T_reg (Table 1).

Figure 2. Bim expression is decreased in T_reg from old and middle-age mice.

(A) Spleen cells from either BimKO or C57BL/6 (WT) mice were stained for CD4, FoxP3 and intracellularly for Bim, and analyzed by flow cytometry. Histogram shows the fluorescence signal for Bim in BimKO (light line) and WT (dark line) mice. (B, C) Spleen cells from 12 young (1.5-month old) and 11 old (20-month old) mice were stained for CD4, FoxP3 and Bim, and analyzed by flow cytometry. Bim MFI was analyzed in gated T_reg (CD4⁺FoxP3⁺) versus gated non-T_reg (CD4⁺FoxP3⁻CD44^hi). Representative Bim staining in (A) T_reg and (B) non-T_reg in young (thin lines) and old (thick lines) mice. (C) Mean (± SE) Bim MFI in T_reg and non-T_reg. p values represent differences in Bim expression in young versus old mice, in each cell subset (unpaired t-tests). (D) Spleen cells from 3 young (6–8 wk old) and 3 middle-aged (10-month old) mice were stained for CD4, FoxP3, and Bim. Bim expression in gated T_reg (CD4⁺FoxP3⁺) is shown in representative animals. The p value represents the difference between young and middle-aged mice (unpaired t-test). Results are representative of two independent experiments.±

Table 1.

Levels of Bcl-2 and Mcl-1 decrease in T_regs from old mice

	Bcl-2 a	Mcl-1 a
Young (n= 6)	118.9 ± 2.0	388.7 ± 18.3
Old (n=5)	57.3 ± 4.8	168.4 ± 8.0
t-test	p < 10−6	p < 10−5

Spleen cells from young and old mice were stained for CD4, FoxP3, CD25 and Bcl-2, or CD4, FoxP3, CD25 and Mcl-1. Bcl-2 or Mcl-1 MFI were determined in gated CD4⁺FoxP3⁺ cells.

^aresults are expressed as mean MFI ± SE

Decreased expression of Bim contributes to T_reg accumulation in aged mice

If Bim plays a central role in age-related T_reg accumulation, the absence of Bim should lead to a more rapid accumulation of T_reg. We thus quantified T_reg frequency and total numbers in wild-type (WT) and Bim KO mice at either 2 or 6 months of age. T_reg frequency was strikingly increased in 6-month old Bim KO animals (Fig. 3A), compared to age-matched WT mice. While the total numbers of T_reg were increased in Bim KO animals at 2 months of age, they accumulated even further by 6 months of age (Fig. 3B). In contrast, the numbers of non-T_reg did not increase in Bim KO mice between 2 and 6 months of age (Fig 3C).

Figure 3. Increased frequency of FoxP3⁺ T cells in middle-aged Bim KO mice.

(A, B, C) Groups of 2 month and 6 month old C57BL/6 (N=9 total; white symbols) and Bim KO mice (N==11 total; black symbols), were injected i.p. with BrdU for 6 days. Results show the mean (±SE) (A) frequency and total numbers of (B) CD4⁺ FoxP3⁺ vs (C) CD4+FoxP3⁻ T cells in the spleen. p-values represent the differences between WT and Bim KO mice (unpaired t-test). (D,E) Groups of Rag-2p-GFP transgenic mice (N=3) and Bim KO-Rag-2-GFP (N=3) mice were bled every 2 weeks. Results show the mean (±SE) percent of CD4⁺ FoxP3⁺ cells that were GFP+ from Rag-2GFP or Bim KO-Rag-2-GFP in the peripheral blood. (E) Mice were sacrificed at 17 weeks of age. Results show the percent (±SE) of GFP+ versus GFP− CD4+ LN T cells that were FoxP3⁺. (F) Results show the percent (±SE) of FoxP3⁺ cells that are BrdU⁺ in the spleens of Bim KO versus C57BL/6 mice after in vivo treatment with BrdU.

Defects in negative selection in Bim KO mice (28) could contribute to increased T_reg production. To test whether defects in negative selection affected T_reg production, we crossed Bim KO mice to Rag2p-GFP transgenic mice and tracked thymic output of T_reg in Bim KO and WT mice. In Rag2p-GFP transgenic mice, GFP expression is an excellent marker of recent thymic emigrants (RTE) (29). No significant difference was found in the frequency of circulating CD4⁺FoxP3⁺ T cells that were recent thymic emigrants (GFP⁺) between WT-Rag2p-GFP and Bim KO-Rag2p-GFP mice (Fig. 3D). Further, in the LN of Bim KO-Rag2p-GFP mice, the increase in FoxP3⁺ cells was confined to the GFP⁻ cells (Fig. 3E), suggesting that T_reg start accumulating in Bim KO mice in the periphery, after they left the transitional RTE compartment.

As the decreased proportion of GFP⁺ FoxP3⁺ cells in Bim KO-Rag2p-GFP mice could be due to their enhanced proliferation, and thus GFP dilution, we monitored in vivo incorporation of BrdU in T_reg. However, T_reg from Bim KO mice proliferate significantly less than their WT counterparts (Fig. 3F). Together, these data show that Bim controls normal T_reg homeostasis by mechanisms that do not involve increased thymic output, nor peripheral proliferation, but instead likely involve survival.

T_reg exhibit dysregulated expression of CD127 with aging

IL-7 is critical for survival of peripheral CD4+ T cells (30). However, few studies have examined CD127 expression on, and the role of IL-7 in, maintaining T_reg in aged mice. In young animals, some FoxP3⁺ T cells were CD127^hi (Fig. 4A), similar to previous reports (31, 32), but, the frequency of FoxP3⁺ T cells expressing CD127 increased by more than 2-fold in aged mice (Fig. 4A). Increased CD127 expression was found in peripheral LN, spleen and mesenteric LN. Most of the FoxP3⁺CD127⁺ T cells in aged mice also expressed CD25 (82.3 ± 3.9% in the peripheral LN, 51.9 ± 6.8% in the spleen, 69.9 ± 4.4% in the mesenteric LN).

Figure 4. CD127 expression is increased and functional on T_reg from aged mice but is not required for their survival.

Spleen cells from 7 young (1.5-month old) and 6 old (22-month old) mice were stained with antibodies against CD4, FoxP3, CD127, and CD25 and analyzed by flow cytometry. (A) Mean % (± SE) of CD127⁺ cells in gated T_reg (CD4⁺FoxP3⁺) are shown. p-value denotes significant difference between young and old mice (student’s t-tests). (B, C) Old (18-month old, n=5) mice were treated with IL-7 IC every other day for one week. Results show the mean (± SE) absolute numbers of CD4⁺FoxP3⁻ and CD4⁺FoxP3⁺ cells. Spleen cells were harvested and stained with antibodies against CD4, FoxP3 and Bim and analyzed by flow cytometry. (C) Results show the mean fluorescence intensity of the Bim stain in CD4+ FoxP3-vs CD4+FoxP3+ cells from PBS (open bars) vs IL-7-treated (filled bars) mice. p-values represent the difference between untreated and IL-7 IC-treated mice (student’s t-test). Results are representative of 2 independent experiments. (D, E) Groups of 2 month (N=4) or 12 month old (N=4) C57BL/6 mice were injected with either anti-IL7R (black bars) or control antibody (ctl Ab; white bars). Results show the mean (±SE) total numbers of splenic (D) CD4⁺FoxP3⁺ and (E) CD4⁺FoxP3⁻. p-values represent the difference between control and anti-IL-7R-treated mice (student’s t-test).

We next determined whether the CD127 expressed by old T_reg was functional by treating 18-month old mice with IL-7 immune complexes (IC), a method to deliver IL-7 in vivo (19). As expected, the absolute number of splenic CD4⁺ T cells was significantly increased in IL-7 IC-treated old mice (Fig. 4B). Importantly, IL-7 IC treatment also significantly increased T_reg absolute number (Fig. 4B), although it did not alter their frequency (25.1 ± 1.7% and 27.1 ± 1.7% of CD4⁺ T cells expressed FoxP3 in untreated and treated animals, respectively; p = 0.23, unpaired t-test). Similarly, IL-7 significantly improved in vitro T_reg survival after 24 hrs (25.6 ± 6.9 % increased survival compared to untreated cultures, p = 0.003). However, in vivo IL-7 administration did not affect expression of Bim in either T_reg or non-T_reg (Fig. 4C).

We next determined if IL-7 was required for T_reg survival in middle-aged animals by preventing IL-7 signaling in vivo. Blockade of IL-7R signaling in vivo did not decrease the numbers of T_reg in young or old mice (Fig. 4D). As expected, IL-7R blockade significantly decreased the numbers of non-T_reg in young mice (Fig. 4E) and substantially decreased immature B cells in the bone marrow (Fig. S1). Interestingly, IL-7R blockade did not decrease non-T_reg in middle-aged mice (Fig. 4E). Together, these data suggest that in both young and middle-aged mice short-term survival of T_reg does not strictly require IL-7.

Homeostatic and IL-2-driven proliferation promotes accrual of T_reg cells that have decreased expression of Bim

IL-2 is a major survival/proliferative factor for T_reg (33) and has been reported to control Bim expression (34). Therefore, we first examined whether levels of Bim would be altered in cells undergoing normal homeostatic proliferation. Interestingly, levels of Bim were significantly decreased in FoxP3+BrdU+ compared to FoxP3+BrdU− cells, whilst in FoxP3− cells, Bim levels were not different between BrdU+ and BrdU− populations (Figure 5A). We next determined the effect of exogenous IL-2 on T_reg homeostasis and Bim expression. IL-2 drove a progressive 7-fold increase in CD4⁺FoxP3⁺ T cells (Fig. 5B), compared to a 3-fold increase in CD4⁺FoxP3⁻ T cells (Fig. 5C). Notably, within the T_reg population, IL-2 preferentially increased the numbers of Bim^lo T_reg (Fig. 5D). The effect of IL-2 on Bim was clearly seen only after repeated IL-2 injection (Fig. 5D). As cell proliferation in response to IL-2 may have contributed to the accumulation of Bim^lo T_reg, we assessed their in vivo proliferation by BrdU labeling along with intracellular staining for Bim. We found that IL-2 drove substantial BrdU incorporation in FoxP3⁺ compared to FoxP3⁻ cells (Fig. 5E). In addition, similar to our previous data under basal homeostatic conditions, most proliferating T_reg expressed low levels of Bim (Fig. 5F).

Figure 5. Proliferation and accrual of Bim^lo T_reg.

(A). Young BL/6 mice were injected with BrdU for 3 days, sacrificed and spleen cells analyzed for Bim expression and BrdU incoporation by intracellular flow cytometric staining. Results show the mean fluorescence intensity of the Bim signal in either CD4+FoxP3+ or CD4+FoxP3− cells that were either BrdU− (open bars) or BrdU+ (closed bars) (± SE) (B–D) Young BL/6 mice (N = 3/group) were injected i.p., with either 3×10⁴ U hIL-2 or PBS twice daily for 4 days. Results show the mean (±SE) absolute number of (B) CD4⁺ FoxP3⁺ T cells; (C) CD4⁺FoxP3⁻ T cells; or (D) CD4⁺FoxP3⁺ T cells that are either Bim^lo or Bim^hi. (E, F) Groups of 8 week old C57BL/6 mice were injected twice daily with either PBS or IL-2 for 3 days and on the day before sacrifice were injected twice with 0.6mg BrdU. (E) Results show percent (±SE) of FoxP3+ versus FoxP3− cells that are BrdU+ (F). Results show the percent (±SE) of BrdU⁻ or BrdU⁺ cells that were Bim^lo in either FoxP3⁻ (open bars) versus FoxP3⁺ (closed bars) T cells in IL-2 treated mice. (G). Groups of BL/6 and BimKO mice (n=4 mice/group/timepoint) were injected with IL-2 twice daily for three days and with BrdU the last 36 hours of IL-2 treatment. Groups of BL/6 and BimKO mice were sacrificed either one or seven days after the last IL-2/BrdU injection. Results show the percentage of FoxP3+BrdU+ cells (± SE) remaining on day 7 relative to the amount observed on day 1 after cessation of IL-2/BrdU treatment.

We next determined whether deficiency in Bim promoted T_reg survival following brief treatment with IL-2, by performing an IL-2/BrdU “pulse-chase” experiment. We treated groups of C57BL/6 and BimKO mice with IL-2 and BrdU and then sacrificed groups of mice one or 7 days after the last IL-2/BrdU treatment. One day after cessation of IL-2/BrdU, there was no significant difference in the frequency of T_reg that were BrdU+ between BL/6 and BimKO mice (42.8 +/− 2.8% versus 39.4 +/− 1.9%, respectively; p<0.365). However, one week later, significantly more BrdU+ T_reg remained in BimKO compared to C57BL/6 mice (Figure 5G). Together, our data suggest that the progressive accrual of T_reg in aged animals is due to increased IL-2-driven proliferation, combined with enhanced survival of Bim^lo T_reg.

Discussion

Age-related immunosuppression is likely driven by several factors. While it is clear that cell-intrinsic mechanisms contribute, less is known about the contribution of T_reg to age-related immunosuppression. We previously showed that T_reg accumulate dramatically with age and that their depletion led to significantly increased effector T cell responses against Leishmania major (15). These data strongly suggest that, under the conditions of this in vivo model, T_reg contribute to age-related immunosuppression. In addition, other investigators have shown that T_reg were functionally suppressive in aged hosts (13, 17, 35). Collectively, these data incriminate the accrual of T_reg as one of the mechanisms involved in the immune suppression associated with aging.

However, the mechanisms underlying T_reg accumulation in aging are not well understood. Recent studies of T_reg biology have identified several critical steps. It is now well established that T_reg can emerge from two different pathways, either the generation of CD4⁺FoxP3⁺ cells directly in the thymus, or the peripheral induction of FoxP3 in naïve T cells that did not express FoxP3 (rev. in (36)). T_reg generated through these two pathways express similar phenotype, although thymic-derived T_reg appear to more specifically express the transcription factor Helios (37). Thus, T_reg accrual in aging could be due to changes in one or several of these pathways, i.e. increased thymic output, increased peripheral conversion, and/or altered homeostasis, such as a changed capacity to proliferate or a decreased susceptibility to death. In depth investigation of these pathways was thus the primary objective of our present studies.

Our data show that altered T_reg thymic output is not responsible for the increased frequency of peripheral T_reg in aged mice. Indeed, we did not find a selectively increased thymic output of T_reg with age, as percentages of FoxP3⁺ CD4⁺ SP thymic T cells were similar in old and young mice, a result in agreement with a previous study (14). Our data also do not support a major role for increased peripheral conversion in aged mice. This is in contrast to another study, which showed that both T_reg and non-T_reg from elderly CMV seropositive patients were enriched for Vβ2-bearing T cells. Although the authors of this study suggested that this commonality of T cells with similar TCRs resulted from conversion of conventional T cells into T_reg, direct evidence of peripheral conversion was not shown (35). An alternative explanation is that, within the thymic derived T_reg compartment, there exists Vβ2-bearing T_reg with specificity to CMV and these cells expanded during chronic CMV infection. Recent data have suggested that T_reg have a fairly broad TCR repertoire (38). To our knowledge, our study is the first one to examine the ability of naïve versus memory T cells from aged mice to undergo conversion. Our finding that naïve non-T_reg from old mice exhibited a lesser capacity to start expressing FoxP3, while old memory T cells remained resistant to conversion, do not support increased conversion as a major mechanism underlying age-associated T_reg accrual. Further, Thornton et al have shown that the frequency of FoxP3⁺ T cells expressing Helios (a marker of thymically derived T_reg) do not change with age, arguing against peripheral conversion being a major mechanism of T_reg accrual (37).

Importantly, our study shows that aged T_reg exhibit better ex vivo survival and strikingly decreased expression of Bim. We found that Bim levels in T_reg are progressively and normally decreased with age in wildtype mice and that T_reg accumulate significantly faster in Bim KO mice after they have left the RTE compartment, despite decreased in vivo proliferation. During our studies, it was shown that Bim also contributes to the longer lifespan of aged naïve CD4⁺ T cells (39). Our data are consistent with these data, although they also suggest that the effects of Bim are more profound on T_reg than on conventional non-T_reg. Our data also broaden and expand on a previous study showing increased T_reg frequency in Bim KO mice, although the age of the mice was not indicated in that study (40). Neither Bcl-2 nor Mcl-1 appeared to contribute to increased survival of aged T_reg, as their levels were also significantly decreased in old T_reg. Decreased Bcl-2 levels had already been described in T_reg (35); however, levels of Bim were not assessed in that study. The significance of Bcl-2 levels needs to be interpreted in concordance with Bim levels, as Bcl-2 functions as the major Bim antagonist in T cells (19). We acknowledge that molecules other than Bim may contribute to Treg homeostasis, however, our data strongly indicate that decreased expression of Bim contributes significantly to Treg accrual with age.

Bim is normally counteracted by IL-7 signaling in non-T_reg (26). The frequency of FoxP3⁺ T cells expressing CD127 increased by more than 2-fold in aged mice, and this increased CD127 expression was found in peripheral LN, spleen and mesenteric LN. As it had been shown that IL-7 signaling allows for survival of peripheral T_reg in IL-2Rβ KO mice (41), we determined whether increased CD127 expression was involved in enhanced survival of old T_reg. Although CD127 was clearly functional in old T_reg, because administration of IL-7 IC led to increased T_reg numbers in old mice; in vivo neutralization of IL-7 did not alter T_reg frequency. Furthermore, IL-7 signaling did not affect Bim expression in T_reg (Fig. 4C). Thus, despite T_reg increased expression of CD127, IL-7 is not required for their survival in old animals. However, these data do not necessarily exclude the possibility that IL-7 plays a redundant role with IL-2 in T_reg survival in aged mice.

We show a novel intersection between IL-2 and Bim leading to the age-related persistence of T_reg in vivo. It is possible that IL-2 drives T_reg proliferation on one hand and regulates expression of Bim mRNA or Bim protein half-life in T_reg on the other (42). Regulation of Bim expression is complex and is controlled at both the transcriptional and post-transcriptional levels. Post-transcriptionally, Bim protein levels can be controlled by the microRNA 17–92 cluster in B cells (43) and by phosphorylation and turnover downstream of ERK signaling (44). At the transcriptional level, Foxo family members have been implicated as potential regulators of Bim expression (34, 45). Foxo activity is controlled by AKT-dependent phosphorylation, which sequesters Foxo molecules in the cytoplasm, preventing them from promoting Bim transcription (34, 45). However, additional, AKT-independent, mechanisms can modify Foxo regulation of transcriptional targets (46). Recent work has shown that IL-2-driven, AKT-activation in T_reg cells is impaired due to their high expression of PTEN (phosphatase and tensin homologue deleted on chromosome 10) (47), arguing that classical Foxo regulation of Bim in response to IL-2 in T_reg may involve additional pathways.

Alternatively, IL-2 may not directly control Bim expression, but instead could drive the selective accumulation of Bim^lo cells because these cells survive better than their Bim^hi counterparts. Three key pieces of data support this model. First, we did not observe changes in Bim expression after short-term treatment with IL-2; instead, Bim levels were decreased only after IL-2 has started driving T_reg expansion. If IL-2 were directly controlling Bim expression, we would have expected to observe changes in Bim prior to T_reg expansion. Second, under both homeostatic and IL-2-driven proliferation, the majority of the proliferating (BrdU+) cells were Bim^lo. Third, after IL-2 withdrawal in vivo, Bim-deficient cells survive better. Combined, these data are supportive of a model in which IL-2 promotes homeostatic proliferation of Bim^lo T_reg and these cells have a survival advantage over Bim^hi T_reg which leads to their progressive accumulation with age. We acknowledge that we cannot definitively determine if a small preexisting Bim^lo cell population has a proliferative advantage as well as a survival advantage as we would need to sort cells based on their expression of Bim to perform this experiment. In the absence of Bim-reporter mice, this is technically infeasible.

It is also possible that T_reg in old mice may not require constant exposure to IL-2 to maintain their Bim^lo status, and thus, survive. In fact our data is supportive of such a model as we showed that the lack of Bim increases survival of T_reg after IL-2 withdrawal in vivo. The source and availability of IL-2 in aged mice is unclear. Previous work has suggested that IL-2 levels decline with age, however, these data were based on T cell production of IL-2 after mitogenic stimulation in vitro (48). Whether or not in vivo IL-2 levels decline with age and whether T cells are a major source of such IL-2 is unclear. However, defining the active levels of IL-2 in vivo is complicated by the fact that bioactive IL-2 binds to heparan sulfate, a glycosaminoglycan found on cell surfaces and within extracellular matrices (49). Nonetheless, it is possible that IL-2 is critical to initiate the expansion of Bim^lo cells, but IL-2 is not critical for the maintenance of these cells thereafter. A potential scenario for such a model is that the accumulation of Bim^lo T_reg is due to genetic modifications that control Bim expression. For example, the Bim promoter has CpG-rich areas and Bim gene expression can be controlled by promoter methylation (50–52). Further, increased methylation of CpG-rich areas is a common finding in aged cells (53, 54). As such methylation is an inherited trait, methylation of the Bim promoter in T_reg would be inherited and selected for, as T_reg from Bim-deficient mice survive better than WT T_reg (40). Control of Bim expression may be indirect as well, transcription factors controlling basal Bim expression may themselves be regulated and therefore not be available to promote Bim expression in aged mice. Future work will uncover the mechanistic interplay between IL-2, T_reg accrual, and regulation of Bim expression in T_reg.

In summary, our data strongly suggest that age-related T_reg accumulation is not due to overproduction of T_reg late in life, nor increased peripheral conversion, but rather due to enhanced survival of T_reg. Together, these data have implications for the immunosuppression that occurs with aging. Limiting or reducing T_reg frequency in the elderly may enhance their ability to maintain immune competence. On the other hand, limiting T_reg in the elderly may enhance autoimmune responses. Nonetheless, with a better understanding of cytokine-mediated regulation of T_reg homeostasis, acute and/or chronic manipulation of T_reg numbers may become feasible when increased immune function is required to counteract infectious diseases or tumors in the elderly.

Non-standard abbreviations

FoxP3

Forkhead box P3

IC

immune complexes

T_reg

regulatory T cells