Transmembrane TNF-α promotes activation-induced cell death by forward and reverse signaling

Secretory tumor necrosis factor-alpha (sTNF-α) is known to mediate activation- induced cell death (AICD). However, the role of tmTNF-α in AICD is still obscure. Here, we demonstrated that tmTNF-α expression significantly increased accompanied with enhanced apoptosis during AICD in Jurkat and primary human T cells. Knockdown or enhancement of tmTNF-α expression in activated T cells suppressed or promoted AICD, respectively. Treatment of activated T cells with exogenous tmTNF-α significantly augmented AICD, indicating that tmTNF-α as an effector molecule mediates AICD. As tmTNF-α can function as a receptor, an anti-TNF-α polyclonal antibody was used to trigger reverse signaling of tmTNF-α. This antibody treatment upregulated the expression of Fas ligand, TNF-related apoptosis-inducing ligand and tmTNF-α to amplify AICD, and promoted activated T cells expressing death receptor 4, TNF receptor (TNFR) 1 and TNFR2 to enhance their sensitivity to AICD. Knockdown of TNFR1 or TNFR2 expression totally blocked tmTNF-α reverse signaling increased sensitivity to sTNF-α- or tmTNF-α-mediated AICD, respectively. Our results indicate that tmTNF-α functions as a death ligand in mediation of AICD and as a receptor in sensitization of activated T cells to AICD. Targeting tmTNF-α in activated T cells may be helpful in facilitating AICD for treatment of autoimmune diseases.


INTRODUCTION
Activation-induced cell death (AICD) refers to activated T cells expressing death receptors and ligands that triggers caspase cascade, inducing suicide and fratricide by apoptosis upon persistent or repeated stimulation of their T cell receptor (TCR). AICD plays a critical role in maintenance of peripheral immune tolerance and protection against autoimmune diseases through deletion of overactivated or autoreactive T cells in the periphery [1]. AICD is induced in activated T lymphocytes via the interaction of death receptors and their cognate ligands, including Fas (CD95)/FasL, tumor necrosis factor-alpha (TNF-α)/TNF Receptor 1 (TNFR1), and TNF-related apoptosis-inducing ligand (TRAIL)/Death receptor 4 or 5 (DR4/DR5). Activation of these death receptors results in recruitment of TRADD, or/and FADD, and caspase 8, forming death-inducing signaling complexes and leading to apoptosis of activated T lymphocytes [2][3][4][5].
Although Fas/FasL-mediated apoptosis is a major contributor to ACID, TNF-α is also one of effector molecules in AICD. It has been reported that sTNF-α can mediate mature T-cell receptor-induced apoptosis, inducing the death of most CD8 + T cells, whereas FasL mediates the death of most CD4 + T cells [4]. In addition, sTNF-α has also been shown to participate in apoptosis of CD4 + T cell [9].
tmTNF-α acts not only as a ligand, transmitting 'forward signaling' to target cell via TNFR, but also as a receptor, delivering 'reverse signaling' to tmTNF-α expressing cell via itself [10][11][12]. It has been reported that infliximab, an monoclonal antibody binding to both forms of TNF-α, induced lamina propia and peripheral lymphocytes apoptosis [13][14][15], suggesting that tmTNF-α may be involved in AICD via reverse signaling. However, etanercept, a soluble TNFR2-Fc fusion protein, failed to induce apoptosis of activated lymphocytes [14]. The function of tmTNF-α in AICD still remains to be elucidated.
Our previous study demonstrated that tmTNF-α increases NK-mediated cytotoxicity through upregulation of multiple cytotoxic effector molecules including perforin, granzyme B, FasL and both forms of TNF-α via reverse signaling [11]. We assumed that tmTNF-α may also play a role in AICD through its forward and reverse signaling. To test this hypothesis, we investigated the changes of endogenous tmTNF-α in AICD, and the impact of down-or up-regulation of tmTNF-α expression on AICD. We used a polyclonal antibody to trigger reverse signaling of tmTNF-α in T cells and observed its effect on AICD. We found that tmTNF-α mediated AICD via forward signaling and sensitized activated T cells to apoptosis via reverse signaling.

Elevation of tmTNF-α expression in activated T cells in AICD
To test whether there is an association of tmTNF-α with AICD, a human leukemia T cell line, Jurkat cell, was stimulated with PHA to induce AICD [16][17][18]. We found that PHA stimulation induced apoptosis of Jurkat cells in a dose-dependent manner. The apoptosis rate of Jurkat cells was elevated gradually with increasing doses of PHA, and reached 63% and 83% in a dose of 5 μg/ml or 20 μg/ml, respectively ( Figure 1A). We chose 5 μg/ml of PHA to observe the kinetics of tmTNF-α expression and AICD. As shown in Figure 1B-1D, apoptosis rate increased gradually with the time after PHA stimulation, which was accompanied with raised tmTNF-α expression, indicating an association of tmTNF-α with AICD. As tmTNF-α expression and apoptosis rate reached nearly a plateau 24 h after stimulation with PHA, we chose 5μg/ml of PHA to activate Jurkat cells for 24 h to induce AICD in the following experiments.
In addition, the same phenomenon was observed in another AICD model in that primary human T cells isolated from PBMC were preactivated with PHA (5μg/ ml), then re-stimulated with αCD3 mAb to induce AICD [19,20]. The apoptosis of T cells was significantly increased (over 60%) in AICD upon αCD3 restimulation detected by both Annexin V/PI ( Figure 1E) and Hochest/PI double staining ( Figure 1F). tmTNF-α expression was also markedly enhanced in primary T cells after restimulation ( Figure 1G). These results strongly suggest that tmTNF-α expressed in activated T cells may be involved in AICD.

tmTNF-α is involved in AICD
In order to verify the involvement of tmTNF-α in AICD, we used two methods: knockdown of TNF-α expression by TNF-α AS or increase of tmTNF-α expression by silence of TACE expression to suppress the cleavage of tmTNF-α into sTNF-α. As expected, transfection of TNF-α AS inhibited tmTNF-α expression in activated Jurkat cells (Figure 2A and 2D), and the apoptosis rate was notably decreased consequently ( Figure 2A). In contrast, transfection of TACE AS, which suppressed TACE mRNA transcription and protein expression ( Figure 2B), markedly increased tmTNF-α expression ( Figure 2C and 2D). Consequently, the apoptosis rate was significantly enhanced (p<0.01), compared with PHA alone treatment ( Figure 2C). The similar results were observed in primary T cells treated with either TNF AS or a TACE inhibitor TAPI-1 ( Figure 2E and 2F). Of note, PHA alone did not affect TACE mRNA transcription, but significantly increased TACE expression on the cell surface ( Figure 2B). These data suggested that tmTNF-α expressed in activated T cells may be an effector molecule to induce AICD.
To test the possibility of tmTNF-α-mediated AICD, we observed whether exogenous tmTNF-α could induce apoptosis of activated T cells. tmTNF-α highly expressing (45.8%) Jurkat cells activated by PHA for 24 h were fixed with 1% paraformaldehyde and used as tmTNF-α-bearing effector cells. To exclude the effect of endogenous tmTNF-α in activated T cells, we used Jurkat cells activated by PHA for 3 h as target cells, because this time point was early stage of AICD at which the apoptosis rate was lower (20.8%) and the activated T cells expressed very low level (5.21%) of tmTNF-α. We co-cultured effector cells and target cells at a ratio of 10:1 for 48 h followed by the Annexin V/PI detection and found that tmTNF-α overexpressing Jurkat cells induced significant apoptosis of 3 h-PHA activated T cells (p<0.001). Neutralizing tmTNF-α in fixed effector cells by a specific antibody evidently blocked tmTNF-α-mediated apoptosis in 3 h-PHA activated T cells ( Figure 2G). For primary T cells, tmTNF-α overexpressing T cells restimulated with αCD3 for 24 h were fixed with 1% paraformaldehyde and cocultured for 48 h with PHA-preactivated T cells. In line with the results in Jurkat cells, tmTNF-α did induce apoptosis in preactivated T cells ( Figure 2H). Our results indicate that tmTNF-α-mediated apoptosis contributes to AICD via forward signaling, as tmTNF-α functioned as a ligand in this case.

sTNF-α participates in AICD
As mentioned above, downregulation of tmTNF-α expression decreased apoptosis rate, while upregulation of tmTNF-α expression increased apoptosis rate. However, Primary human T cells were preactivated by PHA (5μg/ml) for 3 days and cultured in the presence of IL-2 (50 U/ ml) for 7 days. The preactivated T cells were restimulated for 24 h with plate-coated αCD3 (OKT3, 10μg/ml). The apoptosis was measured by Annexin V/PI (E, left) and its quantitative analysis was shown in the histogram (E, right). Hochest/PI double staining in situ for apoptosis (F). The tmTNF-α expression was determined by flow cytometry (G, left) and its quantitative analysis was shown in the histogram (G, right). All the quantitative data represent means ± S.D. of at least three independent experiments. *p<0.05, ** p<0.01, ***p<0.001. in these two cases, sTNF-α in the culture supernatant declined significantly, compared with PHA alone treatment either in Jurkat cells or in primary T cells ( Figure 3A and 3B). To investigate whether sTNF-α participate in AICD, commercial sTNF-α (50U/ml) was added and cultured with Jurkat cells for 24 h. As shown in Figure 3C, sTNF-α exerted cytotoxicity not only to unstimulated Jurkat cells but also to PHA-activated T cells, compared with control or PHA alone treatment (p<0.001). However, sTNF-α treatment could only boost AICD in primary T cells . (E, F) PHA-preactivated primary T cells were transfected with 10 μM of TNF-α AS or NS for 48 h, and then treated with αCD3 (10 μg/ml) for 24 h. A TACE inhibitor TAPI-1 (10 μM) was added simultaneously with αCD3 treatment, and vehicle DMSO served as a control. Apoptosis was evaluated by Annexin V/PI and tmTNF-α expression was detected by flow cytometry. (G, H) tmTNF-α overexpressing 24 h-PHA activated and fixed Jurkat cells were cocultured with 3 h-PHA activated Jurkat cells at an effector/target ratio of 10:1 for 48 h (G). For primary T cells, tmTNF-α overexpressing, αCD3-restimulated and fixed T cells were co-cultured with PHApreactivated T cells (preactivated T) at an effector/target ratio of 10:1 for 48h (H). For neutralization of tmTNF-α, effector cells were treated with TNF-α pAb for 30 min and then washed prior to the addition to the target cells. Apoptosis was determined by the Annexin V/PI. All the quantitative data represent means ± S.D. of at least three independent experiments. *p<0.05, **p<0.01, ***p<0.001. www.impactjournals.com/oncotarget reactivated with αCD3 but not in PHA-preactivated T cells ( Figure 3D), suggesting that sTNF-α participates in AICD.

tmTNF-α-mediated reverse signaling enhances AICD
In addition to mediation of apoptosis as a ligand in AICD, tmTNF-α may also function as a receptor in promotion of AICD. To test this hypothesis, anti-TNF-α polyclonal antibody (TNF-α pAb) and soluble TNFR1 (sTNFR1) were used to trigger reverse signaling of tmTNF-α, respectively. Indeed, TNF-α pAb augmented apoptosis in PHA activated Jurkat cells (p<0.01), compared with PHA alone treatment ( Figure 4A). Neither TNF-α pAb alone could induce apoptosis in Jurkat cells, nor normal rabbit serum had influence on AICD, indicating that TNF-α pAb could activate the reverse signaling of tmTNF-α to amplify AICD, rather than directly induce apoptosis. This enhancement effect was further confirmed in primary T cell AICD model, showing an increased apoptosis in αCD3 re-stimulated T cells induced by TNF-α pAb ( Figure 4B).
To confirm that tmTNF-α enhances AICD via reverse signaling, we used another activator sTNFR1 to initiate the reverse signaling of tmTNF-α in different concentrations. Similarly, treatment with sTNFR1 led to an obvious increase in apoptosis of PHA activated Jurkat cells in a dose-dependent manner (p<0.001) ( Figure  4C). These data suggest that tmTNF-α-mediated reverse signaling may increase sensitivity of activated T cells to AICD.

tmTNF-α-mediated reverse signaling promotes AICD through upregulation of FasL/Fas and TRAIL/DR4
Death receptors and its cognate ligands such as FasL/Fas and TRAIL/DR4/5 are involved in AICD [21]. We hypothesized that the reverse signaling of tmTNF-α may enable to upregulate the expression of FasL/Fas, stimulated with PHA-P (5μg/ml) for 24 h. For primary T cells, PHA-preactivated T cells transfected with TNF-α AS for 48 h were restimulated with αCD3 (10 μg/ml) for 24 h; or PHA-preactivated T cells simultaneously treated with TAPI-1 (10μM) and αCD3 (10 μg/ml) for 24 h. Concentration of sTNF-α in supernatants was detected by ELISA. (C, D) Recombinant human sTNF-α (50U/ml) was incubated with PHA-P activated Jurkat cells or αCD3-restimulated primary T cells for 24 h and the apoptosis was detected by Annexin V/PI. Nonstimulated Jurkat T cells or preactivated primary T cells served as a control. All the quantitative data represent means ± S.D. of at least three independent experiments. ** p<0.01, *** p<0.001. www.impactjournals.com/oncotarget TRAIL/DR4/5 to promote AICD. To test this idea, the expression of FasL/Fas and TRAIL/DR4/5 was detected in both AICD models by flow cytometry. As expected, treatment with TNF-α pAb to trigger tmTNF-α-mediated reverse signaling significantly upregulated expression of FasL and Fas in PHA activated Jurkat cells ( Figure 5A). But in primary T cell AICD model, FasL was evidently increased by TNF-α pAb treatment and Fas expression remained unchanged at very high levels, similar to that in PHA-preactivated T cells and αCD3 retactivated T cells ( Figure 5B). These results indicate that the reverse signaling of tmTNF-α promotes the expression of FasL and/or Fas that are a pair of major contributors to AICD.
In addition, TNF-α pAb could greatly enhance expression of TRAIL and DR4, but not DR5, in PHA activated Jurkat cells ( Figure 5C). However, in primary T cell AICD model, the expression of DR4, but not TRAIL and DR5, was markedly upregulated by TNF-α pAb treatment, although TRAIL was slightly increased without statistical significance ( Figure 5D).
To further explore whether tmTNF-α-mediated reverse signaling promotes AICD via upregulation of these death ligands, we used siRNA to silence of FasL and TRAIL expression, respectively. As expected, knockdown of FasL and TRAIL expression ( Figure 5E) not only inhibited PHAinduced AICD, but also totally abolished the enhanced effect of TNF-α pAb on the apoptosis ( Figure 5F). The data suggest that reverse signaling of tmTNF-α increases the sensitivity of activated T cells to AICD through upregulating the expression of FasL/Fas and TRAIL/DR4.

tmTNF-α-mediated reverse signaling enhances the sensitivity of activated T cells to TNF-αinduced AICD by upregulating TNFR1 and TNFR2
AICD can be mediated by TNFR1 and TNFR2 [4,22]. It is possible that the reverse signaling of tmTNF-α may increase tmTNF-α expression by a positive feedback to amplify AICD and upregulate the expression of TNFR1 and TNFR2 to enhance the sensitivity of activated T cells to TNF-α-mediated AICD. To test this hypothesis, we determined the expression of tmTNF-α and two types of TNFR in both AICD models after treatment with TNF-α pAb. As expected, tmTNF-α expression was dramatically increased in PHA-activated Jurkat cells and in αCD3reactivated primary T cells by treatment with TNF-α pAb ( Figure 6A and 6B). Interestingly, the expression of both TNFR1 and TNFR2 increased obviously by TNF-α pAb treatment in both AICD models ( Figure 6C and 6D).  To observe whether upregulation of TNFR expression by tmTNF-α-mediated reverse signaling could increase the sensitivity of activated T cells to both forms of TNF-α-induced apoptosis, we compared TNF-αmediated cytotoxicity to 3 h-PHA-activated Jurkat cells in the absence and presence of TNF-α pAb. The fixed TNF-α stably transfected NIH3T3 cells in that FasL is not expressed [23]were used as tmTNF-α-bearing effector cells. Indeed, either sTNF-α-or tmTNF-α-mediated apoptosis was significantly enhanced even in 4 hours to 3 h-PHA-activated Jurkat cells with activation of reverse signaling of tmTNF-α, in which two type of TNFR were upregulated ( Figure 6G), compared with the cytotoxicity of both forms of TNF-α to those T cells without activation of tmTNF-α reverse signaling ( Figure 6E and 6F). In contrast, knockdown of TNFR1 expression ( Figure 6G) totally blocked the increased sensitivity induced by tmTNF-α reverse signaling to sTNF-α-, but not to tmTNFα-mediated apoptosis ( Figure 6H), while knockdown of TNFR2 expression ( Figure 6G) entirely abolished the increased sensitivity to tmTNF-α-, but not to sTNF-αinduced apoptosis ( Figure 6I). The data strongly indicate that tmTNF-α-mediated reverse signaling increases the sensitivity of activated T cells to sTNF-α-induced apoptosis via upregulation of TNFR1 expression and to tmTNF-α-induced apoptosis via upregulation of TNFR2 expression.

DISCUSSION
In the present study, we found that tmTNF-α expression in activated T cells was upregulated, accompanied with enhanced apoptosis. This intramembrane molecule functioned not only as a ligand in mediation of AICD in T cells, similar to its soluble form, but also as a receptor on activated T cells in promoting them to express FasL/ Fas, TRAIL/DR4 and tmTNF-α/TNFR to increase their sensitivity to AICD (Figure 7).
We found originally that tmTNF-α expression is significantly upregulated in activated T cells, which is associated with increased apoptosis either in PHA-induced AICD in Jurkat cells or in αCD3 reactivation-induced AICD in primary human T cells. Although TNF-α and TNFR deficient mice show normal AICD [24], FasL/Fas and other effector molecules may compensate in this case. In addition, lpr/lpr or gld/gld mice with spontaneous mutations in the fas gene or the Figure 7: tmTNF-α promotes AICD via forward and reverse signaling. tmTNF-α expression is upregulated in activated T cells during AICD. tmTNF-α as a death ligand mediates apoptosis of activated T cells (1). Meanwhile, tmTNF-α as a receptor increases the sensitivity of activated T cells to sTNF-α-or tmTNF-α-induced apoptosis through upregulating TNFR1 or TNFR2, respectively (2). It can also augment the expression of FasL/Fas and TRAIL/DR4 to increase the sensitivity of activated T cells to AICD via reverse signaling (3). fas ligand gene cannot abolish AICD [4,25,26], suggesting that other death-inducing molecules including TNF-α are involved in AICD. Our results showed that changing expression levels of tmTNF-α in activated T cells affected AICD consequently, as increasing tmTNF-α expression by TACE AS to suppress the cleavage of this transmembrane molecule led to an enhancement of AICD. On the contrary, inhibiting tmTNF-α expression by TNF-α AS resulted in a decrease of AICD. This phenomenon suggests that tmTNF-α may function as a death ligand in AICD. This is supported by our evidence that tmTNF-α overexpressing, fixed activated T cells could obviously increase apoptosis in both 3 h-PHA activated Jurkat T cells and PHA-preactivated primary T cells and this tmTNF-α-mediated AICD could be blocked by neutralization of tmTNF-α on fixed activated T cells with a specific antibody. As tmTNF-α-mediated apoptosis in activated T cells was prohibited by siTNFR2, but not by siTNFR1, it is likely that tmTNF-α overexpressed on activated T cells, as a death ligand, mediates AICD via TNFR2.
Of note, TACE is a key enzyme in regulation of tmTNF-α processing. We found that PHA increased TACE expression on the cell surface of activated T cells, but did not affect TACE transcription. It has been reported that PHA stimulation leads to the phosphorylation of ERK [27] that mediates phosphorylation of Thr735 in TACE, inducing translocation of TACE to the cell surface [28,29]. Therefore, PHA-induced increase of TACE expression is due to promotion of translocation of TACE to the cell surface rather than TACE production.
As activated T cells not only express tmTNF-α, but also express TNFR1 and TNFR2 whose expression is strictly regulated, it is possible that tmTNF-α binds TNFR transmitting dual signaling to TNFR-carrying neighboring activated T cells and to tmTNF-α-bearing activated T cells simultaneously. Our results demonstrated that activation of reverse signaling of tmTNF-α by TNF-α pAb or sTNFR1 significantly increased AICD in both activated Jurkat cells and primary T cells. This is consistent with a report showing that infliximab, a monoclonal antibody against TNF-α, increases apoptosis of activated T lymphocytes in the gut mucosa of patients with Crohn's disease [13,14], although it is possible that infliximab induced apoptosis of activated T cells by antibody-dependent cell-mediated cytotoxicity or complement-dependent cytotoxicity in vivo, because our results showed an overexpression of tmTNF-α in activated T cells ( Figure 1C and 1G).
We proved for the first time that one of the molecular mechanisms underlying the enhancement of AICD by tmTNF-α-mediated reverse signaling is up-regulating the expression of death ligands FasL, TRAIL and tmTNF-α itself. Nuclear factor kappa B (NF-κB) is one of transcription factors to induce FasL expression [30]. Our previous study demonstrated that NF-κB pathway can be activated by tmTNF-α-mediated reverse signaling [12]. It is not surprising that expression of TNF-α and FasL, as target genes of NF-κB, can be upregulated by treatment with TNF-α pAb. Our results showed that knockdown of expression of FasL or TRAIL blocked the promotion effect of tmTNF-α via reverse signaling on AICD, indicating an amplification effect of tmTNF-α-mediated reverse signaling on AICD.
The second mechanism of reverse signaling of tmTNF-α is up-regulating DR4, both types of TNFR or Fas to increase the sensitivity of activated T cells to AICD. Interestingly, initiation of tmTNF-α reverse signaling significantly increased sTNF-α-and tmTNF-α-mediated apoptosis in activated T cells, however, knockdown of TNFR1 or TNFR2 expression by siRNA abolished the promotion effect of tmTNF-α reverse signaling on sTNF-α-or tmTNF-α-mediated apoptosis, respectively. This indicates that tmTNF-α reverse signaling increases the sensitivity of activated T cells to ACID through upregulating TNFR1 expression to enhance sTNF-αinduced apoptosis and promoting TNFR2 expression to augment tmTNF-α-induced apoptosis. These results point out not only a closely cooperative action between both forms of TNF-α but also a synergistic effect of tmTNFα-mediated forward signaling and reverse signaling on AICD. Notably, in this experiment target cells were Jurkat cells activated by PHA and TNF-α pAb together for 3 h. Although 3 h-PHA alone activated T cells that expressed tmTNF-α at low level, the reverse signaling of tmTNF-α could be activated by TNF-α pAb during T cell activation. That might promote PHA-induced tmTNF-α expression in T cells, which in turn enhanced the reverse signaling of tmTNF-α to display its effect on AICD.
In summary, our results demonstrate biological activities of tmTNF-α in AICD. tmTNF-α mediates AICD as a ligand via forward signaling and increases the sensitivity of activated T cells to AICD as a receptor via reverse signaling. On the contrary, tmTNF-α reverse signaling also provides a co-stimulatory signal for T cell activation [31,32]. This transmembrane molecule is similar to other cytokine IL-2 that plays a pivotal role not only in activation of T cells during the early expansion phase after an antigen challenge, but also in controlling AICD through its upregulation of FasL and downregulation of c-FLIP, an anti-apoptotic molecule, during the contraction phase by restimulation [30,33]. Although the mechanism by which the effect of tmTNF-α can be switched from costimulation in apoptosis-resistant T cells to induction of AICD in apoptosis-sensitive T cells is unclear, targeting tmTNF-α may be beneficial in treatment of autoimmune diseases by promoting AICD of autoreactive lymphocytes.

Cell lines and purification of primary human T cells and their stimulation
Jurkat cells, a human acute T lymphocyte leukemia cell line and TNF-α stably transfected NIH3T3 cells [12] were cultured at 37°C in 5% CO 2 in RPMI-1640 medium supplemented with 10% heat-inactivated pyrogen-free FCS, 1.0 mM sodium pyruvate, 2.0 mM L-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin. For AICD induction, Jurkat cells were incubated with PHA (5 μg/ ml) for 24 h.
Peripheral blood mononuclear cells (PBMC) from healthy volunteers were isolated by Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden) density centrifugation. Non-adherent cells were collected by adherence of PBMC at 37°C for 1 h. T cells were isolated through nylon columns (Corning, NY, USA), the purity of the CD3 + T cells was more than 90% as indicated by flow cytometry. For preactivation, resting T cells were cultured with 5 μg/ml PHA for 3 days. After washing, T cells were cultured for additional 7 days in the presence of 50 U/ ml IL-2. The preactivated T cells were restimulated with plate-coated anti-CD3 antibody OKT3 10 μg/ml for 24 h to induce AICD.

sTNFR 1 expression and purification
Human cDNA coding for the extracellular region of sTNFR1 (275-772) was constructed and cloned into the pET-28a (+) vector at the Bam HI and Hind III cloning site [11]. sTNFR 1 was expressed in Escherichia coli upon stimulation with 1 mM IPTG, and purified using a Ni2 + -NTA resin. The purity was 95%. Endotoxin was removed with a Detoxi-Gel endotoxinremoving column according to the manufacturer's instructions. Residual endotoxin concentration was <0.2 U/mg.

Flow cytometry
Cells were collected after stimulation and washed by pre-cold PBS for 3 times. The PE, APC or FITC-conjugated antibodies or unconjugated primary antibodies were then added and incubated at 4°C for 1 h. The incubation with primary antibodies was followed by staining at 4°C for 45 min with FITC-conjugated secondary antibody. The expression of tmTNF-α, Fas, FasL, TRAIL, DR4, DR5, TNFR1 and TNFR2 was analyzed on a FACS Calibur 440E flow cytometer (Becton Dickinson, San Jose, CA, USA).

Apoptosis detection
The apoptosis was evaluated by an Annexin V-FITC Apoptosis Detection Kit (BD biosciences), according to the manufacturer's instructions. Briefly, cells after stimulation were collected, washed twice with precold PBS and resuspended in 100 μl binding buffer. 5 μl of Annexin V-FITC and 10 μl of PI (50 μg/ml) were added into the suspension. Cells were then stained for 15 min at room temperature (RT) in the dark. Apoptosis was analyzed by flow cytometry. Apoptosis (%) = percentage of Annexin V positive cells + percentage of both Annexin V and PI positive cells.
For Hoechst 33258/PI double staining assay, primary human T cells after activation or reactivation were stained for 7 min at 37°C with Hoechst 33342 (5 μg/ml), then followed by PI (1 μg/ml) for 7 min at RT. Then the stained cells were observed under a fluorescence microscope (Nikon DXM1200 fluorescence microscope, Japan).

ELISA for sTNF-α
The concentration of sTNF-α in supernatants was determined by a Human TNF-α ELISA kit, according to the manufacturer's instructions. Briefly, the supernatant was collected after activation of T cells. A human monoclonal antibody specific to TNF-α was used to coat ELISA plates. After incubation with samples and the standard of TNF-α at RT for 2 h, abiotin-conjugated monoclonal anti-human TNF-α antibody was added and cultured for 1 h at RT, followed by the incubation with streptavidin-HRP for 30 min after washing. The color was developed for 15 min by addition of TMB substrate solution and the absorbance was measured at 450 nm on a microplate reader (Tecan, Groedig, Austria).

Western blot
Total protein was extracted by lysis of cells in pre-cold buffer A (10 mM HEPES, pH 7.8, 10 mM KCl, 0.1 mM EDTA, 1mM DTT) and a protease inhibitor cocktail (Sigma-Aldrich, St. Lous, MO, USA) on ice for 20 min. After centrifugation at 12,000 x g for 20 min at 4˚C, the total protein was collected. 50 μg of protein was electrophoresed on a SDS-polyacrylamide gels and transferred to PVDF membranes (Millipore, Billerica, MA, USA) using a semi-dry transfer system (BioRad Laboratories, Hercules, CA, USA). The membranes were blocked for 2 h at RT with 5% non-fat dry milk in PBS containing 0.05% Tween-20 and then probed overnight at 4°C with primary antibodies including anti-TNF-α and anti-β-actin, followed by horseradish peroxidaseconjugated anti-rabbit IgG secondary antibody at RT for 1 h. The bands were visualized using SuperSignal West Pico Chemiluminescence Substrate (Thermo, Waltham, MA, USA).

Statistical analysis
One-or two-way analysis of variance (ANOVA) was used for statistical analysis. Data are represented as mean ± S.D. p < 0.05 is considered to be statistically significant.