Bim/Bcl-2 balance is critical for maintaining naive and memory T cell homeostasis

Because loss of a single Bim allele prevents the lethality and kidney disease of Bcl-2 deficiency (26), we reasoned that Bim^+/−Bcl-2^−/− mice could be used to determine whether Bcl-2 is required for T cell survival in vivo when Bim expression is reduced by 50%. Further, study of Bim^−/−Bcl-2^−/− mice could determine the specific role of Bcl-2 in antagonizing Bim. As previously reported for Bcl-2^−/−Bim^+/+ mice (23–25), T cell development in 28-d-old Bim^+/−Bcl-2^−/− mice was not grossly disturbed compared with Bim^+/+Bcl-2^+/+ controls. Bim^+/−Bcl-2^−/− mice had increased percentages and total numbers of CD4 single-positive (SP) thymocytes, but no substantial difference in the percentages or numbers of CD8 SP thymocytes (Fig. 1, A–D). The slight increase in CD4 SP thymocytes was also observed in Bim^+/−Bcl-2^+/− (not depicted) and Bim^−/−Bcl-2^−/− mice (Fig. 1, A and C). There were no significant differences in percentages or numbers of double-negative (CD4⁻CD8⁻) or double-positive (CD4⁺CD8⁺) thymocytes in Bim^+/−Bcl-2^+/− and Bim^−/−Bcl-2^−/− mice (unpublished data). Numbers and percentages of LN naive CD4⁺ and CD8⁺ T cells were, again, fairly normal in 28-d-old Bim^+/−Bcl-2^−/− mice, with only a slight decrease in numbers of naive CD8⁺ T cells compared with Bim^+/+Bcl-2^+/+ or double-deficient mice (Fig. 1, E and F). Similar results were obtained for spleen (Fig. 1, G and H). Bim^+/−Bcl-2^+/− mice had normal numbers of naive CD4⁺ and CD8⁺ T cells in the LN and spleen (unpublished data). Thus, initial T cell development and seeding of secondary lymphoid organs does not require Bcl-2, at least under conditions of Bim heterozygosity.

At 3 mo, the percentages of CD4 and CD8 SP thymocytes were similar for Bim^+/−Bcl-2^−/−, Bim^+/+Bcl-2^+/+, and Bim^+/−Bcl-2^+/− (not depicted) mice (Fig. 1, A–D), except for a less than twofold decrease in CD4 and CD8 SP thymocyte numbers in Bim^+/−Bcl-2^−/− mice (Fig. 1, A–D). In contrast, Bim^−/−Bcl-2^−/− mice had significantly increased percentages and numbers of CD4 and CD8 SP thymocytes (Fig. 1, A–D). In the periphery, Bim^+/−Bcl-2^−/− mice had significant decreases in the numbers of naive CD4⁺ (∼2–3-fold decrease) and naive CD8⁺ (∼10-fold decrease) T cells compared with wt or Bim^+/−Bcl-2^+/− mice (Fig. 1, E and F), whereas Bim^−/−Bcl-2^−/− mice had near normal numbers of peripheral T cell numbers in LN (Fig. 1, E and F) and spleen (Fig. 1, G and H). Again, Bim^+/−Bcl-2^+/− mice had normal numbers of naive CD4⁺ and CD8⁺ T cells in the LNs and spleen (unpublished data). The decreased numbers of mature naive T cells in the periphery of Bim^+/−Bcl-2^−/− mice in the face of their near normal numbers of SP thymocytes suggest that Bcl-2 promotes the survival of naive T cells. Further, the increase in SP thymocytes and the near normal number of peripheral T cells in Bim^−/−Bcl-2^−/− mice suggests that a single Bim allele controls T cell numbers in the periphery by promoting apoptosis and/or decreasing thymic production.

We used in vitro and in vivo approaches to further evaluate Bim and Bcl-2 effects on T cell homeostasis. We cultured naive peripheral T cells in vitro in the absence of growth and/or survival factors and measured cell survival over time. Both CD4⁺ and CD8⁺ T cells from Bim^+/−Bcl-2^−/− mice died very quickly ex vivo compared with either Bim^+/+ Bcl-2^+/+ or double-heterozygous controls (Fig. 2, A and B). Further, loss of the remaining Bim allele in double-deficient mice resulted in survival slightly greater than that observed with Bim^+/+Bcl-2^+/+ mice (Fig. 2, A and B). Because activated T cells die much more readily than naive T cells ex vivo (27), the increased death of the Bim^+/−Bcl-2^−/− T cells could have been caused by the presence of an increased proportion of activated T cells. However, although there were slight increases in activated/memory phenotype T cells in Bim^+/−Bcl-2^−/− mice (8.18 ± 1.23% of CD4⁺ T cells were CD44^hi in Bim^+/+Bcl-2^+/+ vs. 16.4 ± 2.61% in Bim^+/−Bcl-2^−/− mice), this increase could not account for the death of most Bim^+/−Bcl-2^−/− T cells in culture. Nonetheless, we directly examined the death of naive phenotype (CD44^lo) CD4⁺ and CD8⁺ T cells from Bim^+/−Bcl-2^−/− mice in culture and found that it was significantly increased compared with naive T cells from either Bim^+/+Bcl-2^+/+ or Bim^+/−Bcl-2^+/− mice (Fig. 2, C and D). Thus, naive T cells from Bim^+/−Bcl-2^−/− mice die at an accelerated rate in vitro during withdrawal from survival factors that are normally present in vivo, and this rate decreases to wt levels with loss of the remaining Bim allele.

As an independent test of the roles of Bim and Bcl-2 in the maintenance of peripheral T cells, we used a synthetic Bcl-2 antagonist, which binds to Bcl-2, Bcl-xL, and Bcl-w, but not to Mcl-1 or A1 (28). Culture of wt T cells from naive mice with ABT-737 led to substantial induction of cell death (Fig. 2 E). Further, ABT-737 required Bim for its ability to kill T cells (Fig. 2 E). Additionally, administration of ABT-737 in vivo led to a significant loss of CD4⁺ and CD8⁺ T cells (and, to a lesser extent, B cells) compared with vehicle-treated mice (Fig. 2 F). Because naive T cells normally express little Bcl-xL (29) and do not express Bcl-w (unpublished data), the effects of ABT-737 in vitro and in vivo are likely caused by its ability to bind to Bcl-2. Further, because ABT-737 requires Bim for its ability to kill T cells, these data suggest that Bcl-2 is critical to antagonize Bim to promote naive T cell survival, even in T cells that have developed in mice with normal amounts of Bim and Bcl-2.

We next tested whether cytokines that increase Bcl-2 expression and promote naive T cell survival in vitro and in vivo (9, 12, 30, 31) require Bcl-2 to promote survival. Groups of resting, untreated Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^+/−, and Bim^+/−Bcl-2^−/− mice were killed, and their LN T cells were cultured with various doses of cytokines (IL-7, -4, -6, or -15) that can prevent naive T cell death in vitro (9, 12, 30–32). Both CD4⁺ and CD8⁺ T cells from unmanipulated Bim^+/− Bcl-2^−/− mice were significantly less responsive to rescue by IL-4 and -7 than those from either Bim^+/+Bcl-2^+/+ or Bim^+/− Bcl-2^+/− mice (Fig. 3, A and C). On the other hand, Bcl-2 was not required for IL-6–driven survival of CD4⁺ T cells or for most CD8⁺ T cells (Fig. 3 B). As expected, IL-15 was unable to rescue CD4⁺ T cells from death, but largely required Bcl-2 to prevent the death of CD8⁺ T cells (Fig. 3 D). The effect of Bcl-2 loss was consistently greater for CD8⁺ than CD4⁺ T cells, which is consistent with the greater loss of CD8⁺ over CD4⁺ T cells in vivo observed in Bim^+/−Bcl-2^−/− mice (Fig. 3). Impaired cytokine-driven rescue was likely not caused by lack of cytokine receptor expression on the surface of the T cells in Bim^+/−Bcl-2^−/− mice because these cells express normal levels of CD127 (IL-7Rα) and CD132 (cytokine receptor common γ chain [γ_c]; Fig. 3, E and F). Further, levels of CD122 (IL-2Rβ), CD124 (IL-4Rα), and IL-15Rα were not decreased on Bim^+/−Bcl-2^−/− T cells (unpublished data). Collectively, these data show that Bcl-2 is required downstream of several γ_c cytokines to optimally promote T cell survival.

The restoration of naive T cell numbers in Bim^−/−Bcl-2^−/− mice (compared with Bim^+/−Bcl-2^−/− mice) could be caused by the absence of the proapoptotic effects of Bim on thymocytes, causing increased T cell production, or on mature peripheral naive T cells, causing increased survival. We performed two independent sets of experiments to differentiate between these possibilities. First, we injected groups of Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, or Bcl-2^−/−Bim^−/− mice with neutralizing anti–IL-7 mAb to limit in vivo availability of IL-7 and determined peripheral T cell numbers 1 wk later, relative to mice treated with an isotype-matched control antibody. The efficacy of IL-7 neutralization in vivo for each mouse was confirmed by demonstrating the loss of immature (IgM⁺B220^dull) B cells in the BM (unpublished data) (33). After IL-7 neutralization, Bim^+/+Bcl-2^+/+ mice had 65–70% of their peripheral naive CD4⁺ and CD8⁺ T cells remaining, whereas only 45–50% were left in Bim^+/−Bcl-2^−/− mice (Fig. 4, A and B). In contrast, there was essentially no decline in the numbers of naive CD4⁺ or CD8⁺ T cells in Bcl-2^−/−Bim^−/− mice.

However, in vivo IL-7 neutralization can retard thymocyte development in addition to its effects on peripheral T cell survival (33, 34). Thus, IL-7–driven decreases in pre– T cell numbers could have contributed to the observed loss of peripheral T cells. In a second experiment, we studied the role of Bcl-2 in naive T cell survival in vivo without varying T lymphopoiesis by performing adult thymectomy to block new T cell production. Groups of Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, and Bim^−/−Bcl-2^−/− mice were either thymectomized or sham thymectomized. 4 wk later, their peripheral naive T cells were analyzed by flow cytometry. When compared with their respective sham thymectomy controls, both Bim^+/+Bcl-2^+/+ and Bim^+/−/Bcl-2^−/− mice had ∼50% of their naive CD4⁺ T cells remaining, whereas ∼75% were left in Bim^−/−Bcl-2^−/− mice (Fig. 4 C). These data suggest that molecules other than Bcl-2 may antagonize Bim in CD4⁺ T cells. Conversely, for CD8⁺ T cells, Bim^+/+Bcl-2^+/+ mice had ∼50% of their naive CD8⁺ T cells remaining, slightly more (∼65%) naive CD8⁺ T cells remained in Bim^−/− Bcl-2^−/− mice, whereas only ∼25% were left in Bim^+/− Bcl-2^−/− mice (Fig. 4 D). These data suggest that, in vivo, a major function of Bcl-2 within naive CD8⁺ T cells is to antagonize Bim and that naive CD8⁺ T cells in Bim^+/−Bcl-2^−/− mice are largely maintained by thymic output rather than peripheral survival.

Although the high levels of Bcl-2 on memory CD4⁺ and CD8⁺ T cells have been suggested to contribute to their survival in vivo (9–12, 35), the contribution of Bcl-2 to memory T cell survival in vivo has never been experimentally demonstrated. Consequently, the slightly increased numbers of memory phenotype T cells in the absence of exogenous antigenic stimulation in older Bim^+/−Bcl-2^−/− mice suggested that perhaps memory T cells do not require Bcl-2 for survival. To test this, we inoculated groups of Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, and Bim^−/−Bcl-2^−/− mice with LCMV and used MHC class I and II tetramers to enumerate endogenously responding T cells 141 d later. The total numbers of D^bgp33⁺ memory CD8⁺ T cells were similar for Bim^+/+ Bcl-2^+/+, Bim^+/−Bcl-2^−/−, and Bim^−/−Bcl-2^−/− mice 141 d after inoculation (Fig. 5 A). The number of D^bnp396⁺ CD8⁺ T cells was decreased by approximately twofold in Bim^+/−Bcl-2^−/− mice and increased by approximately twofold in Bim^−/−Bcl-2^−/− mice relative to Bim^+/+Bcl-2^+/+ mice (Fig. 5 A). Despite the similar normal numbers of tetramer⁺ T cells for all three mouse lines, the percentage of CD8⁺ T cells that were tetramer⁺ was significantly increased in Bim^+/−Bcl-2^−/− mice compared with control Bim^+/+Bcl-2^+/+ mice for both epitopes (Fig. 5 B). Numbers and percentages of D^bnp396-specific T cells were increased in Bim^−/−Bcl-2^−/− mice relative to control Bim^+/+Bcl-2^+/+ mice (Fig. 5, A and B), likely reflecting the role of Bim in limiting the numbers of effector T cells during the response (36, 37). Similarly, Bim^+/−Bcl-2^−/− mice had a two- to threefold decrease in total numbers of memory CD4⁺ IA^bgp61-specific T cells compared with Bim^+/+Bcl-2^+/+ mice, whereas Bim^−/−Bcl-2^−/− mice exhibited an intermediate phenotype (Fig. 5 A, inset). Thus, compared with their substantial reductions in naive T cells, Bim^+/−Bcl-2^−/− mice maintain relatively normal numbers of LCMV-specific memory T cells.

It was possible that, in Bim^+/−Bcl-2^−/− mice, an exaggerated expansion of T cells could have contributed to their increased numbers of memory T cells. To address this, we quantified LCMV-SP T cell responses at several times after LCMV infection in both Bim^+/−Bcl-2^−/− and wt mice. For CD8⁺ T cells, numbers of LCMV-SP T cells at the peak of the response was actually significantly decreased in Bim^+/−Bcl-2^−/− mice compared with Bim^+/+Bcl-2^+/+ mice (Fig. S1). Interestingly, the overall contraction of the response was decreased in Bim^+/−Bcl-2^−/− mice compared with Bim^+/+ Bcl-2^+/+ mice, which was reflected by an increased percentage of tetramer-staining T cells amongst the CD8⁺ population in Bim^+/−Bcl-2^−/− mice (Fig. S1). For IA^bgp61-SP CD4⁺ T cells, expansion and contraction of the response was minimally affected in Bim^+/−Bcl-2^−/− mice (Fig. S1). Thus, the similar numbers of memory T cells in Bim^+/−Bcl-2^−/− mice compared with Bim^+/+Bcl-2^+/+ mice was not caused by an exaggerated expansion in the former. Instead, the expansion of the response was actually less in Bim^+/−Bcl-2^−/− mice.

We also examined the total numbers of all memory T cells in the 141-d-old LCMV-infected mice to determine whether memory T cells accumulated at the expense of naive cells in Bim^+/−Bcl-2^−/− mice. Total numbers of CD44^hi memory CD8⁺ T cells were reduced by less than twofold in both Bim^+/−Bcl-2^−/− and Bim^−/−Bcl-2^−/− mice (Fig. 5 C). However, as assessed on a percentage basis, memory T cells had nearly completely overtaken the spleen, as >80% of the CD8⁺ T cells expressed high levels of CD44 (Fig. 5 D). Numbers of memory CD4⁺ T cells were decreased by approximately two- to threefold in Bim^+/−Bcl-2^−/− mice compared with Bim^+/+Bcl-2^+/+ mice and were intermediate in Bim^−/−Bcl-2^−/− mice (Fig. 5 C). Percentages of activated/memory CD4⁺ T cells were also increased in Bim^+/−Bcl-2^−/− mice compared with Bim^+/+Bcl-2^+/+ mice, although not as dramatically as CD8⁺ T cells (Fig. 5 D). Collectively, these results suggest that Bcl-2 is required less for the generation and/or maintenance of T cell memory than for the maintanance of naive T cells and is not absolutely required to combat Bim in memory T cells.

It was also possible that increased viral persistence might contribute to memory T cell accumulation in Bim^+/−Bcl-2^−/− mice. Thus, we assessed viral titers in the livers of mice at multiple times after inoculation. We were unable to detect virus in any mouse at any time by plaque assay (unpublished data). Further, because LCMV persistence is associated with decreased functionality of T cells (38, 39), we assessed the ability of splenic T cells from Bim^+/−Bcl-2^−/− mice to produce IFN-γ after brief stimulation in vitro. Both CD4⁺ and CD8⁺ LCMV-specific T cells from Bim^+/−Bcl-2^−/− mice were fully capable of producing IFN-γ and IL-2 after restimulation in vitro (unpublished data), suggesting that memory T cells in Bim^+/−Bcl-2^−/− mice are functional. Thus, it is unlikely that viral persistence contributes to accumulation of memory T cells in Bim^+/−Bcl-2^−/− mice.

We next examined whether memory T cell accumulation in Bim^+/−Bcl-2^−/− mice was caused by increased survival or proliferation. To accomplish this, we evaluated the ability of IL-7 and -15 to promote endogenous memory T cell survival in vitro. LN cells from uninfected Bim^+/+Bcl-2^+/+ and Bim^+/−Bcl-2^−/− mice were cultured with varying concentrations of IL-7 or -15, stained with anti-CD44 mAb to track endogenous activated/memory T cells and with Annexin V to assess apoptosis, and analyzed by flow cytometry. Although CD44^hi T cells from Bim^+/+Bcl-2^+/+ mice were efficiently rescued from death by either IL-7 or -15, survival of CD44^hi T cells from Bim^+/−Bcl-2^−/− mice was barely affected by either cytokine (Fig. 6, A and B). These data suggest that, similar to their naive counterparts, endogenous memory T cells require Bcl-2 for cytokine-driven survival in vitro.

We next examined proliferative renewal of memory CD8⁺ T cells in Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, and Bim^−/− Bcl-2^−/− mice, using in vivo BrdU labeling. LCMV-infected mice from the experiment in Fig. 5 were injected daily with 1 mg of BrdU during the 14 d before sacrifice. Mice were killed 141 d after LCMV inoculation, and the percentage of LCMV-specific T cells that were BrdU⁺ was assessed by intracellular immunofluorescence staining and flow cytometry. 13% of the CD62L^hi D^bgp33⁺ and 17% of the CD62L^hi D^bnp396⁺ T cells in Bim^+/+Bcl-2^+/+ mice were BrdU⁺, whereas 26% of the CD62L^hi D^bgp33⁺ and 35% of the CD62L^hi D^bnp396⁺ T cells were BrdU⁺ in Bim^+/−Bcl-2^−/− mice (Fig. 6, C and D). Loss of the remaining Bim allele restored levels of proliferative renewal to normal in both the D^bgp33⁺ and D^bnp396⁺ memory T cell populations (Fig. 6, C and D). Thus, enhanced proliferation of LCMV-specific memory T cells in Bim^+/−Bcl-2^−/− mice also contributed to their maintenance at near normal levels.

The lymphopenic environment in Bim^+/−Bcl-2^−/− mice could have contributed to the enhanced proliferative renewal and subsequent accumulation of memory T cells in these mice. We addressed this issue in three independent experiments. First, we attempted to generate mixed BM chimeras in which lineage-depleted BM cells from either Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, or Bim^−/−Bcl-2^−/− mice were mixed 50:50 with lineage-depleted BM from Ly5.1 congenic mice and transplanted into Ly5.1 congenic mice. This would allow us to track cells that had originated from the mutant mice based on their expression of Ly5.2. Although BM from Bim^+/+Bcl-2^+/+mice was able to establish chimerism in recipients, BM from Bim^+/−Bcl-2^−/− or Bim^−/−Bcl-2^−/− mice showed only a small degree of chimerism at 5 and 8 wk in recipient mice. (Fig. S2). Thus, under competitive conditions, Bcl-2 is required for lymphoid reconstitution, which unfortunately precluded study of memory T cell homeostasis.

Second, we avoided complications of lymphopenia by transferring T cells from the various groups of mice into Bim^+/+Bcl-2^+/+ Ly5.1 congenic recipients. As expected, significantly fewer naive CD4⁺ and CD8⁺ T cells remained in recipients of Bim^+/−Bcl-2^−/− cells compared with recipients of either Bim^+/+Bcl-2^+/+ or Bim^−/−Bcl-2^−/− T cells (Fig. 6, E and F). Interestingly, similar to naive T cells, significantly fewer memory CD4⁺ and CD8⁺ T cells remained in recipients of Bim^+/−Bcl-2^−/− cells compared with recipients of Bim^+/+ Bcl-2^+/+ T cells (Fig. 6, E and F). Loss of the remaining allele of Bim increased the recovery of memory CD4⁺, but not CD8⁺ T cells (Fig. 6, E and F).

Third, to avoid complications of memory T cells arising in the various genetic backgrounds and potential minor antigenic differences in the KO mice that could influence the adoptive transfer experiments, we compared the in vivo effects of ABT-737 on naive versus memory T cells. Purified T cells from C57BL/6 mice were transferred into recipient mice, which were treated with either vehicle or with ABT-737 for 10 d, and then killed. The numbers of donor naive (CD44^lo) and memory (CD44^hi) T cells were quantified. Similar to our previous data, ABT-737 caused a substantial reduction in naive CD4⁺ and CD8⁺ T cells (Fig. 6, G and H). ABT-737 also decreased the numbers of transferred memory CD4⁺ and CD8⁺ T cells, but to a lesser extent than what we observed for naive T cells (Fig. 6, G and H). Collectively, these data suggest that naive T cells and, to a somewhat lesser extent, memory T cells depend on Bcl-2 to combat Bim to maintain survival.

Our data demonstrate that Bcl-2 is required to maintain naive T cell survival in vivo and does this largely by inhibiting the proapoptotic effects of Bim. Although Bcl-2 is important for survival of T cell precursors (40), our studies suggest that the defects in naive T cell survival in Bcl-2–deficient mice are largely intrinsic to mature peripheral T cells and reflect unopposed Bim activity. First, naive peripheral T cells from Bim^+/−Bcl-2^−/− mice die readily in vitro, and this death is prevented by loss of the remaining Bim allele. Second, Bim^+/− Bcl-2^−/− mice have nearly normal numbers of thymocytes, and thymectomy results in greater loss of naive CD8⁺ T cells in Bim^+/−Bcl-2^−/− mice than in Bim^+/+Bcl-2^+/+ mice. Both observations are more compatible with a defect in peripheral survival than a defect in thymic output. On the contrary, these data suggest that, in Bim^+/−Bcl-2^−/− mice, CD8⁺ T cells are maintained by continual thymic output rather than peripheral survival. A specific role for Bcl-2 in suppressing Bim-induced apoptosis is supported by previous evidence that genetic loss of other proapoptotic Bcl-2 family members, such as Bax or Bik, does not prevent the lethality or defects in peripheral T cell homeostasis observed in Bcl-2^−/− mice (41, 42). For example, the absence of Bax can mostly correct the loss of thymocytes in Bcl-2^−/− mice and IL-7R^−/− mice, but does not restore peripheral T cell numbers (43). Together with our observations, these data suggest that peripheral naive T cell survival requires suppression of Bim function either directly by IL-7 or by IL-7–driven induction of Bcl-2. This observation for peripheral naive T cells is consistent with previous observations that Bim drives the death of activated T cells (36) and that Bim and Bcl-2 are bound to one another in resting T cells (44). Bim that is not complexed to Bcl-2 would be free to promote apoptosis by activating Bax and/or Bak.

Although antagonism between Bim and Bcl-2 is crucial for the regulation of T cell homeostasis, other Bcl-2 family members also play roles. The following four observations indicate the importance of additional proapoptotic family members: a) T cell homeostasis is more severely disturbed in Bax/Bak double-deficient mice than in Bim^−/− mice; b) although naive Bim^−/−Bcl-2^−/− T cells die slower than Bim^+/− Bcl-2^−/− T cells in vitro, they still die quite readily; c) total loss of Bim (in Bcl-2^−/− mice) does not completely restore numbers of naive or memory T cells to normal; and d) Bim^−/− Bcl-2^−/− mice have significantly increased numbers of SP thymocytes, but normal or slightly decreased numbers of peripheral T cells, suggesting that other proapoptotic Bcl-2 family members can limit the size of the naive T cell compartment. Our preliminary data indicate that Bad and Noxa mRNA levels appear to be increased in naive Bim^−/−Bcl-2^−/− T cells. It is possible that, in the absence of Bcl-2, either Bad or Noxa can limit peripheral naive T cell survival.

Our data showing the IL-7 dependence of some Bim^+/− Bcl-2^−/−, but not Bim^−/−Bcl-2^−/−, naive T cells suggests that IL-7 may have direct or indirect effects on combating Bim. One potential candidate for an indirect effect may be via IL-7's effects on Mcl-1, which also appears to be a critical mediator of naive T cell homeostasis (45). Deletion of Mcl-1 results in substantial loss of both thymocytes and peripheral T cells. Physical interactions between Mcl-1 and Bak, analogous to those between Bcl-2 and Bim, prevent Bak homodimerization, mitochondrial pore formation, and spontaneous induction of apoptosis (18) that may bypass Bim. Alternatively, in the absence of Bcl-2, Mcl-1 may interfere with Bim directly. Although Bcl-xL, which is up-regulated in activated T cells (46–49), may be another potential candidate molecule, recent work has shown that Bcl-xL is not normally required for memory T cell survival (50). Further, previous studies have shown that enforced expression of a human Bcl-2 or Bcl-xL transgene were not sufficient to enhance survival of memory T cells after LCMV infection (51, 52). However, we have observed that, in the hBcl-2 transgenic mice used by Razvi et al. (51), the levels of endogenous Bcl-2 are significantly decreased (unpublished data). Thus, potential effects of transgenes on endogenous molecules could have complicated the interpretation of these data.

In contrast to its crucial role in maintaining the survival of peripheral naive T cells, Bcl-2 did not appear to be strictly required for the generation or maintenance of memory T cells in Bim^+/−Bcl-2^−/− mice. However, three key pieces of data argue for a prosurvival role for Bcl-2 in memory T cells under normal conditions. First, IL-7 and -15 were unable to rescue the in vitro apoptosis of Bim^+/−Bcl-2^−/− memory T cells. Second, adoptive transfer studies showed that, similar to naive T cells, memory T cells from Bim^+/−Bcl-2^−/− mice were quickly lost when forced to compete with wt T cells in a lymphoreplete environment. Third, Bcl-2 inhibition studies with ABT-737 in wt T cells show that although memory T cells are more resistant to the effects of ABT-737, they are still decreased by the drug. The apparent discrepancy between the in vivo accumulation of memory T cells in Bim^+/− Bcl-2^−/− mice and their rapid loss in normal recipient mice is resolved by the increased proliferation of the LCMV-SP memory T cells in lymphopenic Bim^+/−Bcl-2^−/− mice. In addition, the increased proliferation of LCMV-SP T cells likely contributed to the lessening of the contraction of the T cell response in Bim^+/−Bcl-2^−/− mice. Thus, although memory T cells appear to be slightly less affected by the loss of Bcl-2 than naive T cells, under normal conditions, Bcl-2 is critical to maintain both memory and naive T cell homeostasis. In summary, we have investigated for the first time the involvement of Bcl-2 in regulating survival of naive and memory T cell populations and have shown that Bcl-2 promotes T cell survival by antagonizing Bim.

Bim^−/− mice were a gift from P. Bouillet and A. Strasser (Walter and Eliza Hall Institute, Melbourne, Australia) and have been backcrossed to C57BL/6 mice for at least 14 generations. Breeding pairs of Bcl2^Tm1sjk mice were obtained from The Jackson Laboratory and were mated to Bim^−/− mice to generate Bim^+/−Bcl-2^+/−, Bim^+/−Bcl-2^−/−, and Bim^−/−Bcl-2^−/− mice. Mice were genotyped by PCR of tail DNA (36). B6.SJL-Ptprc^a/BoAiTac (Ly5.1 congenic) mice used as recipients in adoptive transfer experiments were purchased from Taconic Farms or were bred in-house. All animal protocols were reviewed and approved by our Institutional Animal Care and Use Committee.

M25 antibody was purified from ascitis fluid harvested from nude mice that had been injected with M25 hybridoma cells. Antibody was purified from ascitic fluid by precipitation with ammonium sulfate, followed by DEAE-cellulose column chromatography. Fractions containing mouse IgG2b were pooled, dialyzed against 0.15 M NaCl, and concentrated by pressure filtration. Mice were injected intraperitoneally every other day with 3 mg of M25 antibody or with MPC-11 isotype control.

6–9-wk-old mice were anesthetized with 200 mg/Kg avertin and given 0.03 mg/kg atropine i.p. to reduce bradycardia. Under anesthesia, mice were immobilized on a surgical tray ventral side up. Using sterile technique, a small incision was made through the skin of the neck extending down to the middle of the ribcage with a sterile scalpel. The submandibular glands were pushed aside to expose the thin layer of tissue, which was cut to expose the thoracic cavity. A second midline incision was made to the level of the second rib. One side of the ribcage was lifted with a sterile forceps, and a suction bell constructed from a glass Pasteur pipette was inserted into the thoracic cavity. In sham-thymectomized mice, the suction bell was then removed from the thoracic cavity and the opening was sutured with surgical staples. In thymectomized mice, suction was applied to draw one lobe of thymus into the bell; this was repeated for the other thymic lobe. After removal of both thymic lobes, the skin was sutured with sterile surgical staples. Mice were then placed under a heat lamp and monitored until they awoke from anesthesia.

LN and spleen cells from 2-mo-old mice were purified by negative selection using a Pan T cell isolation kit and an autoMACS cell separator (Miltenyi Biotech). Purity of T cells using this technique was >95%. Approximately 2 × 10⁶ purified T cells were labeled in vitro with 5-(and-6)-carboxyfluorescein diacetate and 0.5 μM CFSE and transferred into Ly5.1 recipient mice via i.v. injection. On days 1 and 10 after transfer, groups of recipient mice were killed and LNs (axillary, brachial, inguinal, and mesenteric) and spleens were removed and stained with antibodies against CD4 or CD8, Ly5.1, and CD44. Data were acquired on a FACSCalibur (BD Biosciences) flow cytometer and analyzed using CellQuest software (BD Biosciences). For analysis, cells were gated on Ly5.1⁺CFSE⁺ (0–2 divisions) before analysis of CD44 expression.

BM was harvested from the femurs of Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, Bim^−/−Bcl-2^−/−, and B6.SJL-Ptprc^a/BoAiTac mice. Lineage-negative BM cells were purified using a lineage cell depletion kit and an AutoMacs Separator (Miltenyi Biotech). Lineage-depleted BM cells from either Bim^+/+Bcl-2^+/+, Bim^+/−Bcl-2^−/−, or Bim^−/−Bcl-2^−/− mice were mixed in a 50:50 ratio with lineage-depleted BM cells from B6.SJL-Ptprc^a/BoAiTac mice and 10⁶ cells were injected i.v. into lethally irradiated (2 × 500 rads) recipient B6.SJL-Ptprc^a/BoAiTac mice. At 5 and 8 wk after transfer, mice were bled and chimerism was tested by staining the cells with antibodies against CD4, CD8, Ly5.2, and Ly5.1.

The Armstrong-3 strain of LCMV, a gift from R. Ahmed (Emory University, Atlanta, GA), was grown in BHK-21 cells; the number of plaque-forming units (pfu) was assayed on Vero cells as previously described (53). Mice were injected i.p. with 0.25 ml of 2 × 10⁵ pfu LCMV diluted in balanced salt solution.

Class II MHC tetrameric staining reagents were created as previously described (54, 55). In brief, recombinant baculoviral plasmids encoding I-A^b plus genetically linked peptide were transfected into SF-9 insect cells to produce baculoviral stocks. Baculoviral SF-9 supernatant was harvested and used to infect Hi-5 insect cells (Invitrogen). After infection, Hi-5 cells produced and secreted assembled monomeric I-A^b peptide complexes. Hi-5 cell supernatants were harvested and monomers were purified by affinity chromatography using anti–I-A^b antibody (M5-114) and eluted with a high pH buffer. After purification of monomeric MHC–peptide complexes, the complexes were buffer-exchanged into biotinylation buffer and biotinylated with the recombinant enzyme BirA (Avidity, Inc.). After biotinylation, unbound biotin was removed and buffer-exchanged into PBS using Centricon concentrators. Biotinylated monomers were coupled to streptavidin-phycoerythrin and subsequently subjected to size exclusion fast protein liquid chromatography on Superdex 200 (GE Healthcare) to separate large multimeric complexes from excess free monomeric complexes.

MHC I tetramers were provided by A. Zajac (University of Alabama, Birmingham, AL). The methodology for preparation of MHC class I tetramers was modified from the protocol described by Altman and coworkers (56). Recombinant MHC class I molecules fused to a BirA substrate peptide, and recombinant β₂-microglobulin were produced in Eschericia coli BL21 (DE3) cells as previously described (57). MHC–peptide complexes were generated by folding class I heavy chains in vitro with β₂-microglobulin and defined LCMV-derived peptide epitopes and enzymatically biotinylated with recombinant BirA enzyme. Biotinylated MHC monomers were purified over a Sepharose–G75 column (GE Healthcare), and fractions containing biotinylated class I–β₂-microglobulin–peptide complexes were pooled and exchanged into 20 mM Tris, pH 8.0, using centrifugal filtration devices (Millipore). The complexes were further purified by ion exchange chromatography using a QM–Sepharose column (GE Healthcare). Appropriate fractions were concentrated and dialyzed against PBS containing 2 mM EDTA, 1 μg/ml leupeptin, 1 μM pepstatin, and 200 μM PMSF (Sigma-Aldrich). The complexes were tetramerized by the step-wise addition of allophycocyanin-conjugated streptavidin (Invitrogen).

Spleens, LNs (axillary, inguinal, and brachial), and thymuses were harvested, and single-cell suspensions were counted with a Coulter Counter (model Z2; Beckman Coulter). 10⁶ cells were stained with fluorochrome-labeled antibodies (spleen and LNs [anti-CD4, -CD8, -CD44, and -CD62L]; thymus [anti-CD4, -CD8, -CD44, -CD25]), and data was acquired using a FACSCalibur flow cytometer. Data were analyzed with CellQuest software. To detect antigen-specific CD4⁺ T cells, 2 × 10⁶ LN or spleen cells/well was stained in triplicate with I-A^bgp61-80 tetrameric staining reagents for 2 h at 37°C. During the last 45 min of incubation, cells were stained with anti–CD4-allophycocyanin, anti-CD16/32 biotin, and anti–CD44-FITC antibodies. Cells were washed twice, stained with SA-Cychrome, washed twice again, and fixed with 2% paraformaldehyde. CD4⁺, CD16/32⁻ cells were gated and analyzed by dot plots that displayed CD44 versus tetramer to reveal small populations of LCMV-SP T cells. LCMV-SP. CD8⁺ T cells were detected by staining 2 × 10⁶ LN or spleen cells/well in triplicate with D^bgp33, D^bnp396, or K^bnp205 tetrameric staining reagents for 90 min at 4°C. During the last 45 min of incubation, cells were stained with antibodies directed against various surface markers (CD8, CD44, CD62L, and CD127 from BD Biosciences, eBioscience, or produced in-house), washed, and fixed with 2% paraformaldehyde. Gates were drawn around tetramer⁺ populations based on background staining of uninfected animals (generally <0.05% for class II tetramer staining and <0.1% for class I tetramer staining). A minimum of 5 × 10⁵ events were acquired on a FACSCalibur flow cytometer and analyzed using CellQuest software.

To assess naive T cell viability, spleen or LN cells were cultured in vitro in Kappler-Marrack complete tumor medium with or without the indicated concentrations of survival cytokines (IL-4, -6, -7, and -15, all purchased from R&D Systems) as previously described (31, 58). At the indicated times after culture, cells were stained with antibodies against TCR-β, CD4 or CD8, and propidium iodide (5 μg/ml), and cell viability was assessed by flow cytometry. To assess viability of memory versus naive T cells, cells were stained with antibodies against CD4, CD8, CD44, and AnnexinV-FITC (R&D Systems) and analyzed by flow cytometry.

Assessment of in vivo proliferation was performed using a BrdU Flow kit (BD Biosciences). In brief, mice were inoculated i.p. with 2 × 10⁵ pfu LCMV and injected i.p. with 1 mg of BrdU daily for 14 d, beginning 127 d after inoculation. Mice were killed 141 d after inoculation, and single-spleen cell suspensions were surface stained with antibodies against CD8 and CD62L and MHC tetramers, followed by intracellular staining with anti-BrdU antibody according to the instructions in the kit. Stained cells were analyzed with a FACSCalibur flow cytometer and CellQuest software.

Intracellular cytokine staining was performed as previously described (57). In brief, spleen cells from LCMV-infected or -uninfected mice were cultured at 37°C for 4–5 h with or without various LCMV peptides (10 μg/ml GP61-80; 1 μg/ml GP33-41, NP396-404) and 10 μg/ml Brefeldin A. After culture, cells were stained for cell surface markers (CD4, CD8, CD44, and antibodies from BD Biosciences) for 45 min at 4°C. Cells were washed, fixed, and permeabilized with 0.03% saponin and stained intracellularly with PE-labeled anti–IFN-γ or anti–IL-2 antibody (BD Biosciences). A minimum of 5 × 10⁵ events were acquired on a FACSCalibur flow cytometer and analyzed using CellQuest software.

ABT-737 was produced and provided by Abbott Laboratories, as previously described (28), and dissolved and diluted in 30% polyethylene glycol, 5% Tween-80, and 65% of a 5% dextrose in water solution. Mice were injected i.p. once a day with 75 mg/kg in a volume of 0.2 ml.

Statistical analyses were performed using a 2-sample Student's t test with Minitab for Windows Software (Release 14).

Fig. S1 shows the kinetics of the LCMV-specific CD4⁺ and CD8⁺ T cell responses in Bim^+/−Bcl-2^−/− compared with Bim^+/+Bcl-2^+/+ mice. Fig. S2 shows that Bcl-2 is critical to combat proapoptotic molecules other than Bim to allow BM reconstitution when done in a competitive manner.

The authors thank Drs. Thomas Mitchell and Claire Chougnet for critical reading of the manuscript. The authors thank Abbott Laboratories for the generous gift of ABT-737. The authors also thank Dr. Alan Zajac for the class I MHC tetrameric staining reagents.

This work was supported by Public Health Service grants AI057753 (to D. Hildeman), CA105152 (to H.L. Grimes.), and AI55848 (to F.D. Finkelman), and a Veteran's Association Merit award (to F.D. Finkelman).

The authors have no conflicting financial interests.