The Apoptotic Protease-Activating Factor 1-Mediated Pathway of Apoptosis Is Dispensable for Negative Selection of Thymocytes1

Negative selection is a process to delete potentially autoreactive clones in developing thymocytes. Programmed cell death or apoptosis is thought to play an important role in this selection process. In this study, we investigated the role of apoptotic protease-activating factor 1 (Apaf1), a mammalian homologue of CED-4, in programmed cell death during the negative selection in thymus. There was no developmental abnormality in thymocytes from newborn Apaf1−/− mice in terms of CD4 and CD8 expression pattern and thymocyte number. Clonal deletion by endogenous male H-Y Ag of Apaf1-deficient thymocytes with transgenic expression of H-Y Ag-specific TCRs (H-Y Tg/Apaf1−/− thymocytes) was normally observed in lethally irradiated wild-type mice reconstituted with fetal liver-derived hemopoietic stem cells. Clonal deletion induced in vitro by a bacterial superantigen was also normal in fetal thymic organ culture. Thus, Apaf1-mediated pathway of apoptosis is dispensable for the negative selection of thymocytes. However, H-Y Tg/Apaf1−/− thymocytes showed partial resistance to H-Y peptide-induced deletion in vitro as compared with H-Y Tg/Apaf1+/− thymocytes, implicating the Apaf1-mediated apoptotic pathway in the negative selection in a certain situation. In addition, the peptide-induced deletion was still observed in H-Y Tg/Apaf1−/− thymocytes in the presence of a broad spectrum caspase inhibitor, z-VAD-fmk, suggesting the presence of caspase-independent cell death pathway playing roles during the negative selection. We assume that mechanisms for the negative selection are composed of several cell death pathways to avoid failure of elimination of autoreactive clones.

A poptosis or programmed cell death (PCD) 4 is essential for the normal development of the body and the precise regulation of homeostasis in multicellular organisms (1). Apoptosis is also critical for the development and homeostasis of T cells (2,3). In the thymus, CD4CD8 double-positive (DP) thymocytes bearing TCRs that fail to recognize the self MHC molecules die rapidly through a process termed death by neglect. In contrast, recognition of self MHC structure with bound peptide can trigger either functional differentiation (positive selection) or apoptosis (negative selection) of DP cells. If positively selected, immature DP thymocytes develop into mature single-positive (SP) T cells expressing either CD4 or CD8. DP thymocytes bearing TCRs that strongly react with relatively abundant thymic self Ags undergo negative selection, i.e., the clonal deletion of potentially autoreactive T cells. Thus, the process of negative selection results in the PCD of over 97% of developing thymocytes (4,5). The molecular mechanisms of apoptosis involved in these thymic selection processes, however, remain unclear. In the periphery, selfreactive T cells as well as Ag-stimulated mature T cells are deleted by a mechanism of activation-induced cell death, which is mainly mediated by Fas-mediated apoptosis (6). Apoptosis in these situations prevents autoimmune disease and inappropriate accumulation of activated lymphocytes.
A recent advance has shown that mitochondria play essential roles in apoptosis (7)(8)(9)(10). While mitochondria produce metabolic energy in the form of ATP, they contain and release proteins that are involved in the apoptotic cascade, such as cytochrome c (cyto c) and some of caspases (11). Cyto c, an essential component of the respiratory chain of the mitochondria, is released in response to various apoptotic stimuli (12,13) and binds the apoptotic proteaseactivating factor 1 (Apaf1), a mammalian homologue of CED-4, leading to the formation of apoptosome, which then proteolytically activates caspase 9. The activated caspase 9 cleaves the downstream caspases, including caspases 3, 6, and 7, to execute apoptotic cell death by digesting essential cellular proteins (14,15). Thus, deficiency of one of the essential components of the mitochondrial apoptotic pathways renders the cells remarkably resistant to apoptotic stimulation, as shown in gene-disrupted mice (16 -21). Apaf1-deficient (Apaf1 Ϫ/Ϫ ) mice die perinatally, and those embryos have defects in PCD in various tissues whose development is regulated by PCD, including removal of the interdigital webs, formation of the palate, control of the number of neurons, and development of the lens and retina (19). Apaf1 Ϫ/Ϫ embryonic stem cells and primary embryonic fibroblasts showed remarkable resistance to various apoptotic stimuli. Thymocytes deficient for Apaf1 are likewise resistant to various apoptotic stimuli, including dexamethasone, staurosporine, etoposide, and ␥-irradiation. However, it has yet to be elucidated whether the key molecule Apaf1 plays a role in apoptosis during thymic development.
To investigate the role of Apaf1 in the negative selection of thymocytes, in this study, we examined the negative selection of Apaf1 Ϫ/Ϫ thymocytes in lethally irradiated wild-type mice reconstituted with fetal liver-derived hemopoietic stem cells (fetal livertransferred chimeric mice) that bear transgenic (Tg) TCRs specific for male H-Y Ag (22). We also examined the negative selection induced by bacterial superantigen in fetal thymic organ culture (FTOC) (23). Our results showed that clonal deletion in these systems was normally executed in Apaf1-deficient thymocytes, demonstrating that Apaf1-dependent apoptotic pathway is dispensable for PCD during the negative selection process. However, we also showed that Apaf1-deficient thymocytes are more resistant to the peptide-induced cell death in vitro, implicating Apaf1-mediated apoptotic pathway in the negative selection of thymocytes. In addition, we demonstrate that Apaf1-independent caspase activation and cell death that were not inhibited by a broad spectrum caspase inhibitor, z-VAD-fmk, occurred during the peptide-induced cell death in vitro. Taken together, these data indicate that the cell death mechanisms of negative selection are composed of several pathways, which presumably play synergistic and mutually compensatory roles.

Mice
The mice bearing Tg TCR specific for male H-Y Ag peptide on H-2D b (H-Y Tg mice) were provided from the Amgen Institute (Toronto, Ontario, Canada) and maintained on C57BL/6 background. Mice positive for the Tg were typed by clonotypic TCR expression using the specific mAb (T3.70; specific for ␣-chain of H-Y TCR) (24). Apaf1 Ϫ/Ϫ mice were generated as described previously (19), and were backcrossed into the C57BL/6 background more than six times before crossing with H-Y Tg mice. H-Y Tg/ Apaf1 ϩ/Ϫ mice were generated by crossing H-Y Tg mice with Apaf1 ϩ/Ϫ mice; male H-Y Tg mice were also used for crossing to obtain mice with the Y chromosome derived from C57BL/6 (but not from 129) background. Mice were confirmed for H-2 b/b phenotype. C57BL/6 mice were purchased from Japan SLC (Shizuoka, Japan). These mice were maintained in a specific pathogen-free condition.

Genotyping
Genotyping of Apaf1 ϩ/ϩ , Apaf1 ϩ/Ϫ , and Apaf1 Ϫ/Ϫ mice or fetuses was performed using PCR analysis of tail DNA. Two PCR primer sets were used for genotyping. One primer set for detecting the wild-type allele is 5Ј-CCA TCC CTG GTC CTC TGT AAG-3Ј and 5Ј-AAC ACG GAG GCG GTC TTT-3Ј. The other primer set for detecting the mutant type allele is 5Ј-GGG CCA GCT CAT TCC TC-3Ј and 5Ј-CAC TCT ATG GTC CAG GCT ATC-3Ј.
Generation of lethally irradiated wild-type mice reconstituted with fetal liver-derived hemopoietic stem cells H-Y Tg/Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ fetuses at embryonic day (E) 14.5 were obtained from H-Y Tg/Apaf1 ϩ/Ϫ mice intercrosses. The sex of fetuses was determined individually under a microscope, and liver cell suspensions were prepared from each fetus. Eight-week-old male or female C57BL/6 mice were irradiated (900 rad), and approximately 2-5 ϫ 10 6 of fetal liver cells from each fetus were transferred i.v. to each irradiated, sex-matched mouse. Thymocytes isolated from these fetal liver-transferred chimeric mice were used for analyses at least 6 wk after transfer. A schematic protocol is shown in Fig. 4A.
Fetal thymic organ culture FTOC were performed as described previously (23). Briefly, the thymic lobes were obtained from Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ fetuses at E14.5, and cultured on polycarbonate filters (pore size, 4.5 m; Millipore, Bedford, MA) floating on complete RPMI 1640 medium supplemented with 10% FCS in a humid atmosphere with 5% CO 2 . The lobes were cultured for 5 days, and followed by further cultivation in the presence or absence of 1 g/ml staphylococcal enterotoxin B (SEB; Sigma, St. Louis, MO) for 2 days. For harvesting, lobes were ground between frosted glass slides in PBS, washed, and used for flow cytometric analysis.

T cell proliferation assay
Nylon wool-nonadherent lymph node cells containing 3 ϫ 10 4 T3.70 ϩ CD8 ϩ cells from male or female H-Y Tg mice or fetal liver-transferred chimeric mice were cultured in the presence of 5 ϫ 10 5 of irradiated (3000 rad) spleen cells from male or female C57BL/6 mice as a stimulator. The cultures were pulsed with 1 Ci/well [ 3 H]thymidine on day 4, followed by harvesting 20 h later.

Measurement of cyto c release
Thymocytes from wild-type mice with or without peptide stimulation were homogenized in ice-cold preparation buffer (10 mM Tris-HCl, pH 7.5, and 0.3 M sucrose with protease inhibitors) and supernatants collected after centrifugation at 10,000 ϫ g for 60 min. The amounts of cyto c in the supernatants were measured with cyto c ELISA assay kit (MBL), according to the manufacturer's direction.

Antigenic peptide-induced dissipation of ⌬⌿m and cyto c release
We next investigated whether stimulation of thymocytes with physiological antigenic peptide leads to mitochondrial alterations, the upstream events of the Apaf1-mediated apoptotic pathway. To do so, we took advantage of the mice bearing Tg TCR specific for male H-Y Ag peptide. The H-Y TCRs recognize a male-specific H-Y Ag in the context of H-2D b class I molecule (22). Thymocytes from female H-Y Tg mice were stimulated with the peptide at 0 -10 M for 8 h, and ⌬⌿m was measured. The Tg thymocytes showed a peptide dose-dependent dissipation of the ⌬⌿m, as shown in Fig. 2A. However, the degree of dissipation was unexpectedly small when approximately 70% of the thymocytes were H-Y TCR ϩ (T3.70 ϩ ) and CD8 ϩ (data not shown). We also examined the release of cyto c, the trigger of the initiation of the Apaf1-mediated apoptotic pathway in response to the peptide stimulation. The peptide stimulation induced the release of cyto c in a dose-dependent manner (Fig. 2B). Thus, it appears that thymocyte stimulation with relevant Ag induces mitochondrial alterations.

SEB-induced deletion of Apaf1-deficient thymocytes
Since mitochondrial alteration occurred in response to the peptide stimulation, an impairment of negative selection in the absence of Apaf1, an immediate downstream molecule of the mitochondrial damages, was examined. To investigate whether Apaf1-mediated apoptotic pathway is involved in PCD during the negative selection, we first examined the clonal elimination of thymocytes from Apaf1 Ϫ/Ϫ fetuses induced by addition of a bacterial superantigen, SEB, into FTOC, which has been used as a model of negative selection based on apoptotic clonal deletion (23). As shown in Fig.  3, SEB-reactive V␤8 ϩ cells in CD4 ϩ CD8 Ϫ population were eliminated similarly in both Apaf1 ϩ/Ϫ and Apaf1 Ϫ/Ϫ thymocytes by addition of SEB (11-14.9% to 4.3-7.2% in Apaf1 ϩ/Ϫ and 10.8 -13.8% to 4.4 -6.6% in Apaf1 Ϫ/Ϫ in three independent experiments). Percentage of SEB nonreactive V␤6 ϩ cells was not affected. Thus, Apaf1 deficiency did not affect the superantigeninduced elimination of thymocytes in this system of negative selection.

Normal negative selection of Apaf1 Ϫ/Ϫ thymocytes bearing H-Y TCRs in fetal liver-transferred chimeric mice
We then examined the negative selection by endogenous self Ag in Apaf1 Ϫ/Ϫ thymocytes. To do so, we generated and analyzed C57BL/6 radiation chimeras reconstituted with fetal liver-derived hemopoietic stem cells from H-Y Tg mice with Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ genotype ( Fig. 4A; H-Y Tg/Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice, respectively). In male mice, the FIGURE 1. Normal thymocyte development in Apaf1 Ϫ/Ϫ mice. Thymocytes from 10-day-old Apaf1 ϩ/ϩ , Apaf1 ϩ/Ϫ , and Apaf1 Ϫ/Ϫ mice were stained for CD4 and CD8 and analyzed by flow cytometry. Genotypes and percentages of thymocyte subpopulation are indicated. Flow cytometric analysis of thymocytes from these fetal liver-transferred chimeric mice showed that negative selection of Tg-positive T3.70 ϩ CD8 ϩ thymocytes in male mice was complete in the Apaf1 Ϫ/Ϫ background (Fig. 4B, top). Positive selection in female mice was likewise unchanged in thymocytes from H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice (Fig. 4B, bottom). These results suggest that PCD pathway for the elimination of self-reactive clones during thymic negative selection does not require Apaf1.

Lack of self-reactive T cell population in the periphery of H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice
In male H-Y Tg mice, there are many T3.70 ϩ CD8 ϩ T cells in the periphery. We also found T3.70 ϩ CD8 ϩ T cells in the periphery of male H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice. Although these T3.70 ϩ CD8 ϩ T cells in male H-Y Tg mice reportedly develop extrathymically and are unresponsive to male H-Y Ag (24,27,28), it was of importance to examine whether selfreactive T cells were actually eliminated from periphery of the H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice. To address this question, we examined anti-male Ag response of T3.70 ϩ CD8 ϩ lymph node T cells from the fetal liver-transferred chimeric mice and compared it with those from male or female H-Y Tg mice. In H-Y Tg mice, peripheral T3.70 ϩ CD8 ϩ T cells from female mice showed strong proliferative response to male spleen cells, while those from male mice did not respond to male cells, confirming that T3.70 ϩ CD8 ϩ cells with reactivity to self male Ag are eliminated in thymus by negative selection (Fig. 5). Similarly, peripheral T cells from female H-Y Tg/Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice showed strong proliferative response to male Ag. However, no proliferative response to male Ag was observed in cells from either Apaf1 ϩ/Ϫ or Apaf1 Ϫ/Ϫ male chimeric mice. These results showed that self-reactive T cells are virtually absent in the periphery of male H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice, indicating complete negative selection in the thymus of these mice.

Apaf1-dependent and Apaf1-independent cell death of thymocyte induced by antigenic peptides
Normal negative selection of Apaf1-deficient thymocytes in SEBinduced and H-Y Ag-induced negative selection system was unexpected, when involvement of caspase 3 activation in the TCRinduced negative selection both in vitro and in vivo (29,30) and the mitochondrial alterations by antigenic stimulation (Fig. 2) are taken into consideration. Therefore, to examine whether or not the susceptibility of Apaf1 Ϫ/Ϫ thymocytes to TCR stimulation nonetheless differs from that of Apaf1 ϩ/Ϫ thymocytes, we performed  Deletion of V␤8 ϩ Apaf1 Ϫ/Ϫ thymocytes by SEB treatment. E14.5 fetal thymic lobes from Apaf1 ϩ/Ϫ and Apaf1 Ϫ/Ϫ mice were cultured for 5 days without SEB, and further cultured for 2 days in the presence or absence of SEB (1 g/ml). Cell suspensions from these thymic lobes were stained for V␤8 or V␤6, CD4, and CD8, and the expression of V␤8 or V␤6 in CD4 SP cell population was analyzed by flow cytometry. Percentages of positive cells are indicated in the histograms. Representative data from three independent experiments are shown.
antigenic peptide-induced deletion assay in vitro using thymocytes from female H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeras, as a surrogate for negative selection in vivo. Addition of H-Y peptide, which is specifically recognized by H-Y TCRs, to the thymocyte culture induces the deletion of peptide-specific T3.70 ϩ CD8 ϩ cells. As shown in Fig. 6, even in Apaf1 Ϫ/Ϫ thymocytes, cell viability of T3.70 ϩ CD8 ϩ cells was lost by peptide stimulation in a dose-dependent manner, and substantial cells were proven dead at 24 h after stimulation with 10 M peptide. However, Apaf1 Ϫ/Ϫ T3.70 ϩ CD8 ϩ thymocytes were more resistant to peptide stimulation than Apaf1 ϩ/Ϫ T3.70 ϩ CD8 ϩ thymocytes at 10 M of the peptide. These results indicate the possibility that Apaf1-mediated apoptotic pathway may contribute at least partially to PCD pathway of negative selection in a certain condition.
We then investigated the state of caspase 3 activation in Apaf1 Ϫ/Ϫ thymocytes and its relationship with cell viability during negative selection in our in vitro stimulation system. After induction of negative selection by peptide stimulation, we evaluated for dead cells by PI staining and cells positive for active caspase 3 by PhiPhiLuxG2D2 in CD8 ϩ T3.70 ϩ thymocytes. PhiPhiLuxG2D2 is a cell-permeable fluorogenic caspase substrate with specificity for caspase 3 (and, possibly, related caspases), and detects caspase activation in apoptotic cells without fixation (26). As shown in Fig.   7A, positive signal for caspase activation was detected in most of Apaf1 ϩ/Ϫ CD8 ϩ T3.70 ϩ thymocytes after 10 M peptide stimulation, and most of them underwent cell death. In Apaf1 Ϫ/Ϫ CD8 ϩ T3.70 ϩ thymocytes, cells positive for caspase activation signal remarkably decreased compared with Apaf1 ϩ/Ϫ CD8 ϩ T3.70 ϩ thymocytes. Concomitant with this decreased caspase activity, the percentage of dead cells in Apaf1 Ϫ/Ϫ CD8 ϩ T3.70 ϩ thymocytes also decreased. Thus, it is demonstrated that TCR ligation induces caspase activation in an Apaf1-dependent way, further substantiating the possible involvement of Apaf1 in negative selection. However, substantial caspase activation was still observed in Apaf1-deficient cells after peptide stimulation, and cell death was also induced, indicating caspase activation via Apaf1-independent apoptotic pathway also took part in the negative selection. Western blot analysis of caspase activation revealed caspase 3 activation in peptide-stimulated wild-type and Apaf1-deficient thymocytes (Fig.  7B). There was no detectable level of caspase 7 and 6 activation both in wild-type and Apaf1-deficient thymocytes in response to the peptide stimulation, indicating caspase 3 was the dominant caspase activated by the stimulation.
Additionally, to examine the requirement of caspase activity in the negative selection, we also took advantage of a broad spectrum caspase inhibitor, z-VAD-fmk, in this assay. Notably, even in the presence of z-VAD-fmk, although the dead cells remarkably decreased compared with the culture without the inhibitor, peptide stimulation apparently induced cell death of Apaf1 Ϫ/Ϫ thymocytes when caspase 3 activation was completely inhibited. Increasing dose of z-VAD-fmk up to 200 M did not prevent the cell death (data not shown), indicating that cell death observed in the presence of z-VAD-fmk is caspase independent. However, with relatively weak inhibitory activities of z-VAD-fmk toward other caspases than caspase 3 along with its short t 1/2 , it is not formally excluded that the cell death in the presence of z-VAD-fmk is still caspase dependent. In any case, these data, taken together, indicate that the cell death during the negative selection is caused dominantly by the Apaf1-independent pathway and partially or supplementarily by Apaf1-mediated pathway in a certain situation; the former pathway may consist of caspase-dependent (z-VAD-inhibitable) one and caspase-independent (z-VAD-uninhibitable) one.

Discussion
Mitochondria-dependent apoptotic pathway, i.e., the apoptotic pathway via mitochondria3cyto c3 Apaf13caspase 93caspase 3, has been demonstrated in many reports to be critical for development of the body and maintenance of homeostasis of various tissues (16 -21). Since PCD plays a critical role in thymocyte selection, we examined the role of Apaf1, a central element in the mitochondria-dependent apoptotic pathway, in the development of thymocytes. Although Apaf1 Ϫ/Ϫ thymocytes show resistance to a FIGURE 5. Proliferative response of peripheral T cells from fetal liver-transferred chimeric mice. Nylon wool-nonadherent lymph node cells (3 ϫ 10 4 /well) from male or female H-Y Tg mice, and H-Y Tg/Apaf1 ϩ/Ϫ or H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimeric mice were cultured for 4 days with irradiated spleen cells from female (open columns) or male (closed columns) C57BL/6 mice (5 ϫ 10 5 /well). Cell proliferation was measured by counting the incorporation of [ 3 H]thymidine for the final 20 h. Data are expressed as the means of triplicates Ϯ SD. Representative data from three independent experiments are shown. wide range of apoptotic stimuli (19), our results clearly showed that Apaf1-mediated apoptotic pathway is not essential for development of thymocytes (Fig. 1). Similar observations have been reported for the thymocytes of caspase 3-or caspase 9-deficient mice (16 -18). Thus, mitochondria-mediated apoptosis appears to be unnecessary for development of thymus. In this study, we first examined the involvement of the Apaf1-dependent apoptotic pathway in thymic negative selection using Apaf1 Ϫ/Ϫ thymocytes in H-Y Tg and SEB-induced deletion system in FTOC, and demonstrated that this pathway is not essential for the PCD during this selection process. However, although our data clearly showed that Apaf1 is unnecessary for the negative selection, it is not excluded that another Apaf1-like molecule(s), which has yet to be identified, is possibly at play in PCD during the negative selection.
Bcl-2 and Bcl-x L are members of the Bcl-2 family with antiapoptotic function. It has been shown that these molecules exert their antiapoptotic role by inhibiting mitochondrial membrane disruption, thereby working upstream of Apaf1 (31,32). Thus, when overexpressed, these molecules render the resistance of cells to apoptosis induced by mitochondria-damaging stimuli. Actually, thymocytes of Bcl-2 or Bcl-x L Tg mice show resistance to deletion induced in vivo by anti-CD3 Ab (33,34). Nevertheless, it has been reported that the negative selection of thymocytes, induced by endogenous superantigen or by H-Y Ag in H-Y Tg model, is normal in these mice (33)(34)(35). However, in contrast to these reports, Strasser (36) also reported that Tg expression of bcl-2 Tg mice diminished self Ag-reactive H-Y Tg T cells. Thus, although controversial, antiapoptotic members of Bcl-2 family may play roles during the negative selection of the thymocytes.
In this study, we demonstrated that thymocyte stimulation with relevant antigenic peptide caused mitochondrial alterations, such as dissipation of ⌬⌿m and cyto c release, although not at a striking degree. We, however, also showed Apaf1 is not necessary for execution of the negative selection. In addition, Kuida et al. (16) demonstrated normal susceptibility to TCR stimulation-induced apoptosis of caspase 3-deficient thymocytes, while Hakem et al. (18) showed normal caspase 3 activation in caspase 9-deficient thymocytes. These results suggest that all the apoptotic events commencing with mitochondrial damage do not play roles during the physiological negative selection process, although partial involvement of the pathway is not excluded. In this context, another view of the function of antiapoptotic Bcl-2 family members should be of note. Strasser et al. (37) proposed a model in which antiapoptotic proteins (such as Bcl-2) keep adaptor proteins (Apaf1 or Apaf1-related molecule(s)) from activating caspases, as observed in the complex formation of CED-4 and CED-9 in Caenorhabditis elegans (38). In this model, adaptor proteins may exert their proapoptotic effect when freed from antiapoptotic proteins, and the effect may be independent from mitochondrial alterations.
Ligation of the death receptors, or receptors for TNF family members, such as Fas, TNFR1, etc., has been shown also to induce apoptosis of thymocytes (6). Binding of ligands to these death receptors on thymocytes has been shown to trigger the activation of caspase 8 through the adopter molecule Fas-associated death domain (Mort1) (6). The activated caspase 8 directly cleaves and activates caspase 3, thus inducing apoptosis that is not mediated by mitochondria (39). Indeed, Apaf1 Ϫ/Ϫ thymocytes are still sensitive to Fas-induced apoptosis (19). As such, death receptor-mediated apoptotic pathway is a candidate for the PCD during negative selection besides the mitochondrial pathway. However, it has been demonstrated that thymic clonal deletion is apparently normal in mice lacking the functional Fas and Fas ligand system (such as lpr/lpr, gld/gld, or Fas-null mice) (40) and mice Tg for dominantnegative Fas-associated death domain (41,42). In addition, Smith et al. (43) demonstrated that inhibition of caspase 8 activity by Tg expression of CrmA did not impair the deletion of self-reactive T lymphocytes. These lines of evidence suggest that death receptormediated apoptosis pathway is also dispensable for this process, as is the mitochondria-mediated apoptosis.
Caspases are critical mediators and effectors of apoptosis (44,45). It has been shown that caspase 3 is activated in the thymocytes during apoptosis induced in vitro by dexamethasone, anti-CD3 FIGURE 7. Apaf1-dependent and Apaf1-independent caspase activation and caspase-independent cell death in thymocyte induced by specific antigenic peptide. A, Thymocytes (5 ϫ 10 5 /well) from female H-Y Tg/ Apaf1 ϩ/Ϫ or H-Y Tg/Apaf1 Ϫ/Ϫ fetal liver-transferred chimera were cultured with 10 M H-Y peptide for 24 h in the presence or absence of 100 M z-VAD-fmk. T3.70 ϩ CD8 ϩ cells were analyzed for dead cells using PI staining (upper, closed histograms) and caspase 3 activity using PhiPhiLuxG2D2 staining (lower, open histograms). Cells incubated at 4°C were used as a negative control. Values in the histograms indicate the percentages of cells positive for PI or active caspase 3. Representative data from three independent experiments with similar results are shown. B, Thymocytes from Apaf1 wild-type or Apaf1-deficient fetal liver-transferred chimeric mice were stimulated with the antigenic peptide at 10 M or staurosporine (STS) at 0.5 M for 24 h. Cell lysates were electrophoresed, and caspases 3, 6, and 7 were visualized with respective Abs. Chevrons and arrowheads show full-length and activated fragments of the caspases, respectively. mAb, or specific antigenic peptide, and that inhibition of this enzymatic activity by addition of caspase inhibitor z-VAD-fmk prevents cell death (29,30). Actually, Izquierdo et al. (46) showed that negative selection in vivo of thymocytes triggered by two exogenous Ags, SEB and an antigenic peptide in the F5 TCR Tg model, was specifically inhibited in mice Tg for baculovirus p35 protein that is a broad-range caspase inhibitor. However, the contribution of caspases in cell death during negative selection is controversial. Contrary to Izquierdo's report, a recent report by Doerfler et al. (47) showed that caspase inhibition by p35 Tg did not block negative selection induced by antigenic peptide in vitro in OT1 Tg model, and by endogenous Ag in H-Y Tg model. This discrepancy may arise from the strength and/or duration of Ag stimulation, since Izquierdo et al. also reported that the expression of p35 was not able to inhibit thymocyte deletion induced by high Ag concentrations or by chronic Ag treatment, and negative selection by endogenous superantigens. Thus, in the physiological environment in which there is sustained stimulation by endogenous Ags, caspase-dependent pathway is activated to eliminate possibly autoreactive T cells, and the blockage of this pathway may be compensated for by other mechanisms. In agreement with this assumption, we showed that Apaf1-deficient thymocytes bearing H-Y Ag-specific TCRs were completely deleted in vivo in the male chimeric mice (Figs. 4B and 5) while showing more resistance to Ag stimulation-induced cell death in vitro at the highest concentration of the peptide (Figs. 6 and 7A). The physiological relevance of the concentration of antigenic peptide used is unclear, however. In any case, the Apaf1-independent cell death pathway per se is sufficient to complete the negative selection. Actually, a stress-activated protein kinase pathway involving mitogen-activated protein kinase kinase 63p38 activation, known to induce apoptosis in response to various stress, is reportedly sufficient for providing negative selection signal (48). Thus, this pathway is a strong candidate for this caspase-independent pathway of cell death in negative selection.
In summary, our results demonstrated that Apaf1-dependent apoptotic pathway is not essential to PCD during negative selection. However, this apoptotic pathway may contribute at least partially to the negative selection. Beside this pathway, Apaf1-independent pathways dominantly contribute to cell death in the negative selection. Therefore, we suggest that the process of negative selection is composed of several death pathways and these pathways collaboratively work for the completion of negative selection in thymocytes, presumably compensating each other. Involvement of multipathways in the negative selection is reasonable to avoid autoimmune diseases as a result of a failure in negative selection, because a defect in one pathway can be compensated for by other pathways.