Apoptosis and the homeostatic control of immune responses

Introduction

During infections, antigen specific cells can increase to become a huge proportion of all the lymphocytes in the infected animals (> than 50% of all CD8 T cells can be specific for a single virus for example [1]). However, the antigen activated T and B cells die in large numbers as immune responses wane. This death allows the host to maintain its numbers of lymphocytes at reasonable levels in spite of the fact that the animal probably confronts hundreds of different infectious organisms in its lifetime.

Activated T cell death is readily observed in vivo and can only be subverted, to some extent, by life prolonging reagents such as adjuvants. In vitro, experiments have attempted to mimic the death of activated T cells through repetitive TCR stimulation. It is our view that these experiments have to be interpreted with caution, however. For example, early in vitro experiments suggested that Fas/Fas ligand interaction was a driving force behind activated T cell death [2–4]. However, this has turned out not to be the case in vivo, where other mechanisms may supercede the death receptors [5, 6]. Nevertheless, Fas clearly plays a role in control of lymphoproliferation and autoimmune disease. This may be due to non-lymphocyte effects of Fas signaling [7] since recent experiments suggest that Fas and other survival proteins may affect autoimmunity in part via the otherwise quite short lived dendritic cells (DCs). These and other issues are discussed in detail below.

Table 1 is a summary of the expression of mRNA for genes that might be involved in the death of lymphocytes and dendritic cells. Remarkably, the relative expression of these genes is quite similar among the T and B cells at various stages of maturation and activation. As expected, levels of Bcl-2 are decreased with activation [5], while Bim levels were increased slightly even though we have not observed increases in Bim protein levels after in vivo activation [5]. Strikingly, the two most highly induced genes following activation were A1 and survivin (Table 1). As the majority of the activated cells from which the RNA was taken are destined to die, it is unclear why high levels of these anti-apoptotic genes fail to rescue these cells. Perhaps these high levels were important for the survival of proliferating T cells, rather than survival of “post-peak” cells as has been suggested by recent studies [8, 9]. Of all the genes, only the death receptors, Fas and the TNFRs appear to be markedly differently expressed in T and B cells. Dendritic cell gene expression, by contrast, is quite different from that of lymphocytes, with greatly increased expression of the Bcl-2-related pro-apoptotic messangers, such as Puma, NOXA and BNIP3.

Table 1. Expression of Bcl-2 related genes in lymphocytes and dendritic cells.

C57BL/6 mice were either uninjected or injected with Staphylococcal enterotoxin B (SEB) for 48 hours prior to sacrifice. Single cell suspensions from pooled lymph nodes and spleens were stained with antibodies against TCRβ, CD4, CD8, IA^b, CD69, and CD25 for na�ve cells (sorted for TCR+CD4+ or TCR+CD8+ and IA^b-CD69−CD25−) and Vb8.x CD4, CD8, IA^b for SEB activated cells (sorted for Vβ8+CD4+ or Vβ8+CD8+ and IA^b-). RNA was extracted and sent to the Gene Microarray core at Cincinnati Children’s Hospital and run on Affymetrix mouse genome 430.2 arrays. Analysis was performed using Genespring software after normalization. Data shown reflect averages of multiple probe sets (whose values were above 50) or single probe sets depending upon the gene.

	Naïve CD4+ T cells	Activated CD4+ T cells	Memory Phenotype CD4+ T cells	Naïve CD8+ T cells	Activated CD8+ T cells	Memory Phenotype CD8+ T cells	(Follicular) B cells	Germinal Center B cells	Memory B cells	Plasma Cells	Bone marrow derived dendritic cells	Spleen CD4− CD8− dendritic cells
Bcl-2	+++++	++++	++	+++++	+++	+++++	++	++	+++	−	++	+
Bcl-xL	++	++	+++	+	++	++++	+	+	+	+	+++	+
Mcl-1	+++++	+++++	+++++	+++++	+++++	+++++	+++++	+++++	+++++	+++++	+++++	+++
A1	+++	+++++	+++++	+++	+++++	+++++	+++	++++	+++	+	+++++	+++++
Bim	+++	++++	++	++	+++	+++	++	++	+	++	+++++	++
Bad	+	++	+	+	++	+	−	−	−	−	−	+
PUMA	+	+	nd	+	+	++++	−	−	−	−	++	nd
NOXA	+	+	nd	++	+++	++++	−	−	−	−	++++	nd
Hrk	−	−	−	−	+	−	−	−	−	−	+	−
Bid	+	+	−	−	−	++	+	++	++	++	+++	+
Boo/DIVA	−	−	−	−	−	+	−	−	−	−	−	−
Bik	−	−	nd	−	−	+	−	−	−	−	−	−
NIX	+++++	+++++	nd	+++++	+++++	+++++	++++	++	+++	++	+++++	+
BNIP3	++	++	++	+	+	+++	++	+	+	−	+++++	−
Bax	+++	+++	++++	+++	+++	+++++	+	+	+	+	+++++	+++
Bak	+	++	++++	+	++	++++	−	−	−	−	++++	+
Bok	−	−	++	−	−	+	−	−	−	−	++	+
FAS	+++	+++	++	+++	+++	++	−	++	+	−	+++	−
Tnfrsf1a	+++	++	+++	++	++	++	−	−	−	+	+++++	++
Tnfrsf1b	+++	++++	+++++	+++	+++	++	(+)	−	+	−	+++++	++

Cells were purified and analyzed on Affymetrix chips. All cells were from mice. Germinal center B cells, Memory B cells and plasma cells were from the spleens of mice immunized with NP-CGG 11-14 days, 10–12 weeks and 7 days previously respectively. Data on T cells are unpublished from our laboratories or from Swanson et al. Immunity 17:605, 2002. Data on B cells are unpublished from our laboratories or from Weissman, I. depositied in the Geo data base, 1/30/06. Data on bone marrow derived dendritic cells are from D.Stetson depositied in the Geo data base 3/1/05. Data on spleen CD4− CD8− dendritic cells are from Chaussabel, deposited in the Geo Data base 6/13/03. To help readers, values are expressed as: −, 0–50; + 51–200; ++, 201–500; +++, 501–1000; ++++, 1001–2000; +++++, >2000.

Life and death of antigen presenting dendritic cells

Dendritic cells, regardless of their subtype, have relatively short half-lives [10, 11]. Nevertheless these are long enough under most circumstances to achieve successful antigen presentation to T cells. The normal life expectancy of DCs is thought to be controlled by balances between Bcl-2-related proteins and not via the action of death receptor proteins such as Fas [11–13]. Even so, DCs that express the baculovirus caspase inhibitor, p35 (which primarily blocks death receptors) are resistant to death in vivo and increase in numbers, albeit only after one year of age [14]. Mice containing such DCs have increased numbers of activated T cells and anti-nuclear-antigen antibody production [14]. In contrast, mice whose DCs overexpress a hBcl-2 transgene or lack Bim have increased steady-state levels of DCs in relatively young mice, which can elicit increased T cell responses and autoantibody production [11, 15]. Thus, both death receptors and Bcl-2 family members contribute to the demise of DCs following their activation, suggesting that DCs with extended life times can enhance T and B cell responses.

It is not obvious why longer life expectancies for DCs would promote autoimmunity. The DCs are always present, after all, and are always presenting self peptides, so why should their life spans affect the outcome? One obvious possibility is that activation of DCs or their interactions with activated T cells significantly shortens their lifespan, which in turn effectively removes T cell substrates. Indeed, it has long been known that survival of DCs is curtailed by cytotoxic T cells [16], which probably act both via perforin [17] and Fas [14, 18]. This killing of DCs may also explain the profound immunosuppression observed after certain viral infections such as measles [19]. Conversely, perhaps Fas signaling has an additional effect on DCs, as indicated by induction of chemokine production by DCs stimulated with anti-Fas [20]. If this is so, autoimmunity induced by the DCs in these experiments may be caused by increased DC activity induced by the very signals which are normally dampened by their ability to also drive DC death. Overall, these experiments suggest that the death of DCs (especially activated DCs) may be an important contributor to the resolution of immune responses and to the development of autoimmunity. Certainly this is a subject which we expect to see more of in the next few years.

Life and death of activated B cells

Research over many years has suggested that activated B cells die via Fas (reviewed in [21]). However, the death of some types of activated B cells is affected by the balance of pro and anti-apoptotic members of the Bcl-2 family, for example, Bcl-2 itself (of course!), Bax and Bak and Bim [22–27]. BAFF is the principle survival factor for activated B cells and helps to break the unresponsiveness of autoimmune B cells [25, 26]. In this regard, the discussion of the effects of Bcl-2 related proteins on the activities of DCs is quite interesting since DCs are themselves are a potential source of BAFF (reviewed in [28]). Thus, breakage of B cell tolerance in autoimmunity could be due to effects on DCs which, in turn via BAFF, may be manifested on B cells.

The BH3-only molecule, Bim, is critical for apoptotic contraction of T cell responses

The pro-apoptotic BH3-only molecule, Bim, plays a critical role in culling activated effector T cells [5, 6, 29–31]. A major factor controlling Bim activation resides not with Bim but with its major antagonist, Bcl-2. At the peak of response, levels of Bcl-2 are significantly decreased within antigen-specific T cells [5, 32, 33]. In unstimulated T cells, Bim is complexed to Bcl-2 [34], the downregulation of Bcl-2 that occurs during T cell activation may result in increased amounts of free Bim. Bim that is unopposed by Bcl-2 could then either directly or indirectly lead to the activation of Bax and/or Bak. Indeed, Bax/Bak conformational changes are not observed in activated Bim^−/− T cells [31]. Other factors also may contribute to Bim activation as T cells lacking an adjuvant-induced survival factor, Bcl-3 [35], accumulate cytosolic Bim which may then drive cell death [36]. However, additional processes such as release from microtubules, variation in expression of isoforms or phosophorylation may also affect Bim activity [37, 38]. Further work is required to understand how these mechanisms are integrated following T cell activation in vivo.

Other BH-3 proteins substitute poorly for Bim

If T cells express multiple BH3-only molecules (see Table 1), why does Bim play such a dominant role in T cell death and what is the function of other BH-3-only molecules in T cells? It is possible that other BH3-only molecules may act as “messengers” of apoptotic signals, then transmitted by Bim. Alternatively, as suggested by others, different BH3-only molecules may act as sensors of divergent pro-apoptotic stimuli [39]. For example, Bim appears to protect cells from “growth factor withdrawal” induced death [40], while Puma and Noxa are direct transcriptional targets of the tumor suppressor p53 and are critical for DNA-damage-induced apoptosis [41]. Puma is particularly intriguing because its overall structure, with respect to its arrangement of BH and transmembrane domains, is similar to that of Bim. However, only minor effects, if any, of Puma on activated T cell death have been observed, even in activated T cells that lack Bim [42]. Absence of Noxa likewise has no effect on activated T cell death (in spite of the fact that it is well expressed in these cells, Table 1) indicating that termination of T cell responses is not driven by a DNA-damage response [36]. Overall, Bim appears to be the dominant and largely non-redundant BH3-only molecule critical for apoptotic contraction of T cell responses in vivo.

How do some T cells survive and become memory cells?

Since Bim is critical for activated T cell death, the cells that survive apoptosis to become memory cells must somehow avoid Bim-driven death. One possibility is that the phenomenon is stochastic, and depends on random over-expression of death-avoiding proteins in the activated cells. Alternatively, cells that survive may be those which receive or inherit different signals or proteins after TCR engagement. Such an idea is supported by the increased conversion to memory of daughter cells which are distal rather than proximal to the contact point between T cells and antigen presenting cells [43]. As distal daughter cells express more IL-7Rα, they may be more able to receive the life giving signals (such as Bcl-2 induction) of IL-7 [43–45]. Thus, IL-7Rα^hi effector T cells may compete for limiting amounts of IL-7, and may be selected to survive and become memory T cells. These data suggest that early differentiation decisions result in the formation of “memory precursors” and a pool of terminally differentiated effectors. Given this scenario, it would be expected that interference with contraction of the response at the level of apoptosis would not increase memory T cell numbers, assuming that apoptosis is downstream of differentiation decisions. But, data from studies using Bim^−/− mice argue against this possibility because effector T cells from these mice die poorly and give rise to significantly increased numbers of memory T cells [29]. These data suggest that effector T cells are somewhat flexible in their ability to form memory cells.

Further, other pieces of data suggest that contraction of the response is not solely dictated by limiting amounts of IL-7. First, significant numbers of viral-specific IL-7Rα^lo T cells survive contraction of the response [29]. Second, following peptide immunization, substantial numbers of antigen-specific IL-7Rα^hi T cells die [46]. Moreover, during chronic infections, IL-7Rα^lo T cells persist and these cells can re-express IL-7Rα once the virus is cleared [47]. Finally, neutralization of IL-7 (or Bcl-2) during viral infection does not exacerbate contraction of the response, suggesting that IL-7 is not required for T cells to survive contraction of the response [48]. It is possible that IL-15 plays a redundant role with IL-7 because IL-15-deficient mice have a slightly exacerbated contraction of CD8+ T cell responses [49, 50]. Thus, while decreased IL-7Rα expression may mark some effector T cells destined to die via Bim, expression of IL-7Rα is not the only determinant of death decisions within the effector population.

Role of Bcl-2 family members in controlling long-term memory T cell survival

Following the contraction phase of the T cell response, in general, a long-lived memory population persists. Memory CD8+ T cells are maintained after the contraction phase by the pro-survival and proliferative actions of IL-7 and IL-15, respectively, and do not appear to require interactions with class I MHC [50–52]. Memory CD4+ T cells are largely maintained by IL-7 [53] and possibly by interactions with class II MHC, although this is controversial [54, 55]. Nonetheless, because memory T cells divide regularly, but generally do not accumulate in number, apoptosis is critical to maintain memory T cell homeostasis.

Memory T cells express higher levels of Bcl-2 compared to naïve T cells [32] (Table 1). This may help memory T cells combat Bim. In addition, TNF receptor-associated factor-1 (TRAF-1) signals may suppress Bim expression to allow memory cell survival [56]. Also, in human memory T cells, lowered levels of Bim, perhaps because of Fox03a expression, may allow better survival of central versus effector memory cells [57]. Consistent with these observations, viral-specific, effector memory T cells accumulate more in Bim^−/− mice compared to controls (D.H. personal observations). However, the overall numbers of memory T cells in Bim^−/− mice slowly wane over time and approach levels observed in wild type mice [29], suggesting that factors other than Bim act to keep memory T cells in check. These may involve other BH3-only proteins since, unlike the situation in Bim^−/− mice, endogenous memory T cells accumulate rapidly and spontaneously in Bax/Bak double-deficient animals [30].

Discussion

Bcl-2 family members play a critical role in the regulation of immune system homeostasis. These molecules control the survival of both innate and adaptive immune cells. Interestingly, the effects of Bcl-2 family members and death receptors on innate immune cells may affect the magnitude of T cell responses, through how this happens is not clear. On the other hand, the ability of Bim to limit effector T cell survival substantially limits the numbers of cells available for the memory pool. Thus, novel therapies aimed at targeting the Bcl-2 family which are being developed as cancer therapeutics may also prove useful in manipulating and regulating immune homeostasis.