Fas Ligand Induces Cell-autonomous NF-κB Activation and Interleukin-8 Production by a Mechanism Distinct from That of Tumor Necrosis Factor-α*

Fas ligand (FasL) has been well characterized as a death factor. However, recent studies revealed that FasL possesses inflammatory activity. Here we found that FasL induces production of the inflammatory chemokine IL-8 without inducing apoptosis in HEK293 cells. Reporter gene assays involving wild-type and mutated IL-8 promoters and NF-κB- and AP-1 reporter constructs indicated that an FasL-induced NF-κB and AP-1 activity are required for maximal promoter activity. FasL induced NF-κB activation with slower kinetics than did TNF-α, yet this response was cell autonomous and not mediated by secondary paracrine factors. The death domain of Fas, FADD, and caspase-8 were required for NF-κB activation by FasL. A dominant-negative mutant of IKKγ inhibited the FasL-induced NF-κB activation. However, TRADD and RIP, which are essential for the TNF-α-induced NF-κB activation, were not involved in the FasL-induced NF-κB activation. Moreover, CLARP/FLIP inhibited the FasL- but not the TNF-α-induced NF-κB activation. These results show that FasL induces NF-κB activation and IL-8 production by a novel mechanism, distinct from that of TNF-α. In addition, we found that mouse FADD had a dominant-negative effect on the FasL-induced NF-κB activation in HEK293 cells, which may indicate a species difference between human and mouse in the FasL-induced NF-κB activation.

The tumor necrosis factor (TNF) 1 receptor family is a still growing group of cytokine receptors that regulate cell prolifer-ation, differentiation, and death. A subset of this family called death receptors possesses a characteristic cytoplasmic region named death domain (DD) (1). Activation of these receptors induces recruitment of the death-inducing signaling complex, which consists of adaptor molecules (such as FADD and/or TRADD) and upstream caspases (such as caspase-8 and -10); this complex in turn initiates the activation cascade of caspases, which eventually results in apoptotic cell death. Like other members of the TNF receptor family, all the death receptors (TNF receptor type I (TNFR1), Fas, DR3, DR4, DR5, and DR6) have been reported to induce the activation of NF-B (2)(3)(4)(5)(6)(7)(8), one of the most important transcription factors for the activation and regulation of the immune system. However, how these receptors activate NF-B is still obscure.
Fas (Apo1/CD95), a prototype of the death receptors, directly recruits FADD and strongly induces apoptosis in a variety of cell types upon its ligation by FasL (1). The Fas-FasL system plays pivotal roles in various aspects of immune regulation and function, such as self-tolerance, and cell-mediated cytotoxicity (9). It has been proposed that FasL is expressed in "immune privileged" organs (such as the eye and testis) and protects them from destructive inflammation by counterattacking inflammatory cells (10,11). In contrast, ectopic expression of FasL by genetic engineering induces inflammation accompanied by massive neutrophil infiltration in animals (12)(13)(14)(15). Furthermore, FasL seems to play detrimental roles in various inflammatory diseases such as hepatitis, graft-versus-host diseases, and pulmonary fibrosis (16 -18).
To clarify how FasL induces inflammation, we have been investigating the molecular mechanism of FasL-induced inflammation. We previously reported that FasL simultaneously induces apoptosis and the conversion of inactive pro-IL-1␤ into its active form; both of these processes are mediated by caspases (15). This active IL-1␤ plays an important role in the FasL-induced inflammation. On the other hand, several reports have shown that some normal and transformed cell lines produce IL-8, a chemokine for neutrophils, upon Fas ligation by an anti-Fas monoclonal antibody (mAb) or FasL (19 -26). Activation of NF-B is correlated with the FasL-induced IL-8 production; however, there is no direct evidence that NF-B activation is essential for this response. Furthermore, it has not been investigated whether FasL induces NF-B activation directly or indirectly through the induction of another cytokine, such as IL-1.
In this study, we have found that FasL induces cell-autonomous NF-B activation and IL-8 production without inducing detectable apoptosis in HEK293 cells. Subsequent experiments using this cell line revealed that FasL induces NF-B activation by a mechanism distinct from that of TNF-␣.
Reporter Assays-HEK293 or 293T cells were transfected with one of the firefly luciferase reporter plasmids described above and pRL-TK using the TransIT-LT1 reagent (TAKARA, Otsu, Japan). Stable transfectants of HEK293 cells harboring the NF-B-Luc construct and a hygromycin B resistance gene was also generated using the TransIT-LT1 reagent according to the manufacturer's protocol. Jurkat and Jurkat-derived cell lines were transfected with pNF-B-Luc and pRL-TK using the LipofectAMINE PLUS reagent (Invitrogen). In some experiments, cells were cotransfected with one of the tester plasmids described above and/or an empty vector as a control. The total amount of transfected DNA per culture was kept constant within an experiment. Cells were harvested 24 h after the transfection, and then assayed for luciferase activity using the Dual-Luciferase Reporter Assay System (Promega) in a LB9501 Lumat luminometer (Berthold, Bad Wildbad, Germany). Firefly luciferase activity was normalized to the Renilla luciferase activity. In some experiments, HEK293 cells were transfected with pNF-B-GFP and the internal control plasmid pCMV-DsRed-Express (Clontech) using the LipofectAMINE PLUS reagent. Twenty-four hours after the transfection, cells were subjected to flow cytometry analyses using a FACSCalibur® (BD Biosciences) equipped with a 488-nm argon laser. The DsRed-positive cells were gated and their mean fluorescent intensity of GFP and DsRed was calculated using CELLQuest software. Relative NF-B activity was assessed as follows; mean fluorescent intensity of GFP/mean fluorescent intensity of DsRed ϫ 100.

RESULTS
FasL Induces Delayed IL-8 Expression at the mRNA Level-We tested various human cell lines for their potential to produce IL-8 in response to FasL stimulation, and found that HEK293 cells exhibit this response without showing any evidence of apoptosis (Fig. 1, A and B). In addition, U937 and HeLa cells produced IL-8 upon FasL stimulation only when apoptosis was inhibited by a pan-caspase inhibitor z-VAD-fmk (data not shown). A neutralizing mAb against FasL inhibited the FasL-induced but not the TNF-␣-induced IL-8 production (Fig. 1A), confirming that FasL was responsible for this response. The artificial soluble mFasL (27) used in these experiments has strong cytotoxic activity, resembling the membrane-bound form rather than the soluble form of FasL. To investigated which form of FasL is responsible for its IL-8inducing activity, we transfected HEK293 cells with expression vectors carrying a FasL cDNA producing both membranebound and soluble forms (FasLDC), the membrane-bound form only (FasLDC2), or a soluble form only (FasLS) (31) (Fig. 1C, left panel). As expected, the cytotoxicity of FasL was detected in culture supernatant of the cells expressing FasLDC and FasLS but not FasLDC2, which is not released into the culture medium (Fig. 1C, right panel). Transfection of FasLDC and FasLDC2 but not FasLS induced IL-8 production, indicating that the membrane-bound form mediates the IL-8-inducing activity. TNF-␣ induced significant IL-8 production within 1.5 h, which lasted more than 12 h; in contrast, significant IL-8 production was detected starting 6 h after the FasL addition (Fig. 1D). Consistent with this, strong IL-8 mRNA expression was detected as early as 1.5 h after TNF-␣ stimulation, whereas weaker IL-8 mRNA expression was detected beginning 3 h after the FasL addition by RT-PCR analyses (Fig. 1E). These results indicate that FasL, compared with TNF-␣, induces IL-8 expression more slowly at the mRNA level.

NF-B and AP-1 Binding Elements Are Required for the IL-8 Promoter Activation by FasL and NF-B Activation Is Required for the FasL-induced IL-8 Production-
The IL-8 promoter region contains three important cis-acting elements for IL-8 gene transcription, namely NF-B, AP-1, and NF-IL-6 binding sites (29), and IL-8 transcription required either the combination of NF-B and NF-IL-6 or that of NF-B and AP-1 sites, depending on the type of cells or stimulation (37)(38)(39). The contribution of these individual elements to the FasL-induced IL-8 promoter activity was examined using a luciferase reporter gene construct controlled under the minimal essential promoter region of the IL-8 gene (Ϫ133 to ϩ44 bp) containing all three elements (40) and a series of constructs with mutation in each of the elements ( Fig. 2A). FasL induced strong luciferase activity in HEK293 cells transiently transfected with the construct containing the wild-type IL-8 promoter. The mutation in the NF-B-binding site completely abolished the FasL-induced luciferase activity. The mutation in the AP-1 binding site inhibited it strongly but not completely. In contrast, the mutation in the NF-IL-6 binding site had no effect on it. These results strongly suggest that the NF-B site is essential, and the AP-1 site is partially required for the FasL-induced IL-8 promoter activation. Consistent with these results, FasL induced the expression of a luciferase gene controlled under the NF-B or AP-1 elements in HEK293 cells (Fig. 2, B and C). In agreement with the results shown in Fig. 1, addition of an anti-FasL mAb completely abolished the FasL-induced and NF-B-driven expression of luciferase (Fig. 2B), and FasL induced slow NF-B activation compared with TNF-␣ (data not shown). Furthermore, proteasome inhibitor lactacystin and expression of proteasome-resistant IB␣ mutant (IB␣ super-repressor) inhibited both NF-B activation and IL-8 production induced by FasL or TNF-␣ (Fig. 2, D-G and data not shown). Considering the transfection efficiency in these experimental conditions (about 70%), the inhibition of IL-8 production by IB␣ superrepressor is almost completely. These results indicate that NF-B activation is essential for the FasL-induced IL-8 production. that there was no stable secondary mediator in the supernatant. TNF-␣-treated supernatant induced NF-B activity because of the carryover of TNF-␣, serving as a positive control. However, these results do not rule out the possibility that the secondary mediator was unstable or a membrane-bound. To exclude this possibility, we mixed two HEK293 cell cultures that had been separately transfected with mFas cDNA and either the NF-B reporter or IL-8 promoter reporter construct, and then cultured them with or without Jo2, the agonistic mAb specific for mFas (Fig. 3B). Jo2 did not induce luciferase activity when the reporter constructs were transfected alone, because Jo2 cannot activate the endogenous hFas of HEK293 cells (lane 2). However, the cotransfection of mFas and reporter constructs induced luciferase activity (lanes 5 and 7) and Jo2 enhanced it (lanes 6 and 8). In contrast, significant luciferase activity was not detected when mFas cDNA and a reporter construct were separately transfected, even after Jo2 stimulation (lanes 9 and 10). Consistent with reporter assays shown in Fig. 2B, FasL as well as TNF-␣ induced NF-B-specific DNA binding activity in the nuclear extract of HEK293 cells, as revealed by the electrophoretic mobility shift assay (Fig. 3C). Importantly, pretreatment with actinomycin D or cycloheximide had no effect on NF-B DNA binding activity induced either FasL or TNF-␣. Therefore, neither transcriptional nor translational events are required for FasL-and TNF-␣-induced NF-B activation. These results indicate that no secondary mediator is essential for the FasL-induced NF-B activation. Consistent with a previous report (25), addition of anti-p65 and -p50 antibodies resulted in the supershift of the nuclear NF-B induced by either FasL or TNF-␣ (data not shown), suggesting that FasL and TNF-␣ induce a similar NF-B complex.

Induction of NF-B Activity and IL-8 Promoter Activity by FasL Is a Cell-autonomous Response and Neither Transcriptional nor Translational Events
The DD of Fas and FADD Play a Critical Role for the FasLinduced NF-B Activation-To clarify which cytoplasmic region of the Fas receptor was responsible for the FasL-induced NF-B activation, we expressed a set of mutants of hFas in HEK293 cells. Comparable expression of the wild-type and mutant Fas on the cell surface of transfectants was confirmed by FACS analysis (see Supplemental Fig. S1). Consistent with previous reports, overexpression of hFas activated NF-B (Fig.  4A). Deletion of the C-terminal 15 amino acids from hFas up-regulates its ability to induce apoptosis (32). However, this deletion did not affect its capacity to induce NF-B activation. On the other hand, the further deletion of Fas up to a part of the DD (FD7 and FD2) or the lpr cg -type point mutation (Val 238 to Asn) in the DD (FP1), which abolishes its apoptosis-inducing capacity (32), also abrogated its ability to activate NF-B. These results indicate that the C-terminal 15 amino acids of hFas are dispensable, but the DD of Fas is indispensable for its ability to activate NF-B.
We next addressed the role of FADD, which is essential to recruit caspase-8 and to induce apoptosis upon Fas ligation (28,41), in the FasL-induced NF-B activation. As shown in Fig.   FIG. 2

. Activation of NF-B and AP-1 activity by FasL is required for the maximum IL-8 promoter activation, and the NF-B activation is essential for the FasL-induced IL-8 production.
A, HEK293 cells were transiently transfected with 50 ng of pRL-TK and 100 ng of a reporter plasmid carrying a luciferase gene driven by the wild-type IL-8 promoter (Ϫ133-luc) or this promoter mutated at the NF-B-, AP-1-, or NF-IL-6-binding site. Seventeen hours after the transfection, FasL (500 units/ml) was or was not added to the culture medium, and the cells were cultured for 7 more hours. 4B, overexpression of a dominant-negative mutant of FADD (FADD-DD) efficiently blocked the FasL-but not TNF-␣-induced NF-B activation. Because overexpression of FADD-DD might interfere with Fas to recruit signaling molecules other than FADD, we sought to suppress FADD expression specifically using siRNA. The effect of siRNA to endogenous FADD was validated by semi-quantitative RT-PCR (Fig. 4C). FADDtargeting siRNA but not a control siRNA with the reverse sequence of the FADD-targeting siRNA reduced the expression of FADD gene. In addition, the specificity of our FADD-targeting siRNA was confirmed by its ability to diminish the expression of transiently transfected FADD but not the control p84-DD construct (Fig. 4D). This siRNA inhibited NF-B activation induced by FasL but not that induced TNF-␣ (Fig. 4E). The control siRNA had no effect on these responses. These results show that FADD is essential for the FasL-but not TNF-␣-induced NF-B activation.
Consistent with previous reports (42,43), transient transfection of hFADD potently activated NF-B in HEK293 cells (Fig.  4F, left panel, lane 2). FADD consists of a death effector domain (DED), a DD, and a C-terminal domain (CTD). It has been suggested that the CTD of FADD is involved in cell proliferation (44 -46). However, a deletion mutant of hFADD lacking the CTD (hFADD⌬C) is equivalent to its wild type in NF-Bactivating potential (Fig. 4F, left panel, lane 3), indicating that the CTD of hFADD is not required for it to activate NF-B.
It was previously reported that Fas ligation induces apoptosis but not NF-B activation in L929 mouse fibroblast cell line (47). We also failed to observe the FasL-induced NF-B activation using several mouse cell lines including RAW264.7 (macrophage cell line), BaF/3 (pre-B cell line), and MEF (mouse em-bryonic fibroblast cell line). Because mFas induced NF-B activation in HEK293 cells, we tested the potential of mFADD to activate NF-B. Interestingly, the NF-B-activating potential of mFADD was much lower than that of hFADD in HEK293 cells (Fig. 4F, left panel, lanes 2 and 4), although mFADD induced apoptosis as potently as hFADD in Jurkat cells (data not shown). Similarly, mFADD failed to induce NF-B activation in MEF, although hFADD induced this response weakly but significantly in the same cell line. Thus, it is possible that the inability of mFADD to induce NF-B activation explains why FasL does not induce strong NF-B activation in mouse cell lines. Remarkably, mFADD inhibited FasLbut not TNF-␣-induced NF-B activation (Fig. 4F, middle and  right panels, lane 4). We then generated chimeric molecules between hFADD and mFADD to clarify the region of hFADD responsible for NF-B activation. The m/hFADD1 (consisting of the DED of mouse origin and the DD plus CTD of human origin) induced NF-B activation as potently as did hFADD. In contrast, the m/hFADD2 (consisting of the DED plus N-terminal half of DD from mFADD and the C-terminal half of DD plus CTD from hFADD) induced only minimal NF-B activation and had a dominant-negative effect on the FasL-induced NF-B activation, similar to mFADD. These results suggested that the N-terminal half of hFADD-DD is critical for the hFADD-mediated NF-B activation.
Caspase-8 Is Essential for the FasL-induced NF-B Activation-A role for caspase-8 in NF-B activation has been suggested mainly because transient overexpression of caspase-8 induces NF-B activation (42,43). However, whether caspase-8 is essential for FasL-induced NF-B activation or not has not been directly investigated. We approached this question by reducing caspase-8 expression in HEK293 cells using siRNA. As shown in Fig. 5A, the siRNA-targeting caspase-8 effectively suppressed endogenous caspase-8 expression in HEK293 cells. This siRNA inhibited NF-B activation induced by FasL but not by TNF-␣ (Fig. 5B). Consistently, Z-VAD-fmk and Z-IETDfmk (pancaspase and caspase-8 inhibitor, respectively) diminished the FasL-induced NF-B activation, whereas Z-DEVDfmk, Z-YVAD-fmk, and Z-AAD-fmk (inhibitors for caspase-3, caspase-1, and granzyme B, respectively) showed no effect (Fig.  5C). None of these inhibitors showed a significant effect on the TNF-␣-induced NF-B activation (data not shown). Importantly, the increase of Z-VAD-fmk up to 50 M did not cause further inhibition (Fig. 5D). This is a striking contrast to that the proteasome inhibitor lactacystin completely abolished the FasL-induced NF-B activation (Fig. 5C).
To further examine the role of caspase-8 in the FasL-induced NF-B activation, we used the caspase-8-deficient Jurkat cell line JB-6. Consistent with previous report (28), JB-6 cells were completely resistant to FasL-induced apoptosis, while wildtype Jurkat cells were highly sensitive to it (data not shown). When Jurkat cells were treated with FasL in the presence of pan-caspase inhibitor Z-VAD-fmk, apoptosis was blocked and NF-B activity was observed (Fig. 5E). Although TNF-␣-in-duced NF-B activation was detected in both Jurkat and JB-6 cells, the FasL-induced NF-B activation was completely abolished in JB-6 cells in the presence or absence of Z-VAD-fmk ( Fig. 5E and data not shown). Transient as well as stable transfection of an expression vector encoding wild-type caspase-8 restored the FasL-induced NF-B activation in JB-6 cells, demonstrating that the absence of caspase-8 was the basis for this defect in JB-6 cells. Transient expression of a catalytically inactive mutant of caspase-8 (active site C377S) also restored the FasL-induced NF-B activation in JB-6 cells in the presence of Z-VAD-fmk (Fig. 5E). Taken together, these results clearly indicate that caspase-8 plays an important role in the FasL-induced NF-B activation, and suggest that there are both catalytic activity-dependent and -independent pathways for this response.
Caspase-like Apoptosis Regulatory Protein (CLARP) Inhibits NF-B Activation by FasL-CLARP/FLIP is a cellular caspase-8 analog that inhibits the death receptor-mediated apoptosis (48). CLARP is expressed mainly in two alternatively spliced forms; the long form (CLARP-L) consists of two DEDs and a caspase-like domain, and the short form (CLARP-S) has the DEDs only. Both forms of CLARP interact with FADD and caspase-8 and inhibit FasL-induced apoptosis. Consistent with previous reports (42,43), transfection of a large amount of CLARP-L expression plasmid induced NF-B activation in HEK293 cells, while CLARP-S expression plasmid (up to 300 ng per 1 ϫ 10 5 cells) induced little NF-B activation (data not shown). In contrast, we found that expression of a small amount of CLARP-L or -S inhibited FasL-but not TNF-␣induced NF-B activation (Fig. 6A). CLARP-S exhibited stronger inhibitory effect than CLARP-L. An anti-apoptotic Bcl-2 family member, Bcl-XL had no effect on both FasL-and TNF-␣-induced NF-B activation. In addition, both CLARP-L and -S inhibited the FasL-but not TNF-␣-induced NF-B activation in Jurkat cells (Fig. 6B). These results suggest that CLARP is an inhibitor rather than a mediator for the FasL-induced NF-B activation, although it may induce NF-B activation under other circumstances.
Effect of Dominant-negative Mutants of TRADD, RIP, and IKK␥ on the FasL-induced NF-B Activation-TRADD, RIP, and IKK␥ are involved in the TNFR1-mediated NF-B activation. As shown in Fig. 7A, a dominant-negative mutant of IKK␥ inhibited both FasL-and TNF-␣-induced NF-B activation, whereas a dominant negative form of TRADD inhibited the TNF-␣-but not FasL-induced NF-B activation. The DD of p84 protein had no such inhibitory effect on NF-B activation by FasL or TNF-␣. Although RIP has been suggested to be involved in the FasL-induced NF-B activation (49), a dominant negative mutant of RIP (RIP-DD) that inhibited the TNF-␣induced NF-B activation had little effect on the FasL-induced NF-B activation (Fig. 7B). Taken together, our results indicate that FasL and TNF-␣ induce NF-B activation through distinctive signal transduction pathways. DISCUSSION It has recently been demonstrated that a wide variety of cell types produce IL-8 in response to Fas ligation, and activation of NF-B has been correlated with the FasL-induced IL-8 production (19 -26). However, there has been no direct evidence that NF-B activation is essential for this response, or that the FasL-induced NF-B activation is a cell-autonomous. We demonstrated here that the NF-B binding site is essential for the FasL-induced reporter gene activity controlled under the IL-8 promoter (Fig. 2A). In addition, the IB␣ super-repressor and a proteasome inhibitor lactacystin that suppresses NF-B activation inhibited FasL-induced IL-8 production (Fig. 2, D-G). These results indicate that NF-B activation is indispensable for the FasL-induced IL-8 production. Although the kinetics of IL-8 production by FasL was considerably slower than those of the IL-8 production induced by TNF-␣ (Fig. 1, D and E), experiments involving transfer of culture supernatant or mixed culture of cells transfected separately with the mFas or reporter gene clearly indicated for the first time that NF-B activation, and hence IL-8 production by FasL stimulation is cell autonomous and is not mediated by a secondary factor (Fig. 3, A and B).
The signal transduction pathway from a death receptor leading to NF-B activation has been best studied for TNFR1 (50). The most widely accepted pathway involves TRADD, RIP, and TRAF2. TRADD directly interacts with the DD of TNFR1 and provides a platform to recruit RIP and TRAF2. These molecules coordinately activate the IKK complex, which in turn induces the phosphorylation and proteasome-dependent degradation of IB␣. On the other hand, how FasL induces NF-B activation is still obscure, although several reports have described the molecular mechanisms of NF-B activation by ligation or overexpression of Fas (2,3,25,42,51). Judging from previous reports, Fas seemed to utilize similar machinery to TNFR1 to activate NF-B. However, our results described here clearly indicated that a different set of signaling molecules are used for the FasL-induced NF-B activation from those used for the TNF-␣-induced one. Notably, FADD and caspase-8, important molecules for the FasL-induce apoptosis, also mediated NF-B activation induced by FasL but not by TNF-␣. Therefore, the branch point of the Fas signal transduction pathways leading to apoptosis and NF-B activation must be downstream of caspase-8. This is remarkably similar to the Drosophila IMD pathway, in which DREDD, the Drosophila homolog of caspase-8, plays an important role in both apoptosis and NF-B activation (52), and clearly different from the TNFR1 system, in which the branch point is TRADD; i.e. the TRADD-FADD interaction leads to apoptosis, while the TRADD-RIP interaction causes NF-B activation. On the other hand, TRADD and RIP were not essential for the FasL-induced NF-B activation (Fig. 7). It has been reported that a dominant-negative form of RIP inhibits NF-B activation induced by overexpression of FADD or caspase-8 (43). Because both FADD and caspase-8 possibly have intracellular functions that are separate from the death receptor-signaling pathways, based on the phenotype of knock-out mice of these molecules (53)(54)(55)(56), the overexpression of these molecules may not always reflect cooperative function with death receptors. In fact, Harper et al. (57) recently reported that Fas did not bind TRADD and RIP. Nonetheless, a dominant-negative mutant of IKK␥ inhibited both the FasLand TNF-␣-induced NF-B activation, suggesting that common downstream molecules are used in these responses. Currently, our efforts are aimed at clarifying what molecules connect caspase-8 to IKK␥.
Although the wild-type and active center mutant (C377S) of caspase-8 have comparable capacity to induce NF-B activation (42,43), we reproducibly observed about 50% inhibition of the FasL-but not TNF-␣-induced NF-B activity following the addition of 10 M pan-caspase or caspase-8-specific inhibitor, but not those inhibitors for caspase-3, or caspase-1 (Fig. 5C). Increasing Z-VAD-fmk to 50 M did not cause further inhibition (Fig. 5D). Therefore, it is likely that there are both caspase activity-dependent and -independent mechanisms for the FasL-induced NF-B activation in HEK293 cells. The expression of a catalytically inactive mutant of caspase-8 in JB-6 cells restored the FasL-induced NF-B activation (Fig. 5E), indicating that a caspase activity-independent pathway exists in Jurkat cells. In addition, the FasL-induced NF-B activation was completely abolished in JB-6 cells, indicating that caspase-8 protein is also required for the caspase activity-independent pathway. Because we observed the FasL-induced NF-B activation in Jurkat and Jurkat-derived cell lines only in the presence of Z-VAD-fmk, it is unclear whether the caspase activitydependent pathway exists in Jurkat cells.
Interestingly, mFADD had a dominant-negative effect on FasL-induced NF-B activation in HEK293 cells. Although hFADD has a very similar amino acid sequence to mFADD except for the very end of its C terminus, experiments using their chimeric mutants suggested that the N-terminal half of DD of hFADD is important for it to activate NF-B (Fig. 4F). Thus, mFADD may be an effective tool to investigate the signal transduction pathway for the FasL-induced NF-B activation in human cells.
T cells from FADD-but not Fas-or FasL-deficient mice show impaired proliferation (54), suggesting that FADD plays an essential role in a growth signal transduction downstream of receptors other than Fas. Recently, it was reported that Ser194 in the CTD of hFADD is phosphorylated by a 70-kDa cell cycle-regulated kinase (44), and that overexpression of C-FADD (the N-terminal-truncated form of hFADD, amino acids 80 -208) but not its S194A mutant, inhibits cell growth (45). Furthermore, it was suggested that phosphorylation of Ser 191 in mFADD, which corresponds to Ser 194 in hFADD is involved in the regulation of T cell proliferation (46). However, we found that the CTD of hFADD, containing this phosphorylation site, is not required for its ability to activate NF-B (Fig.  4F). Taken together, these results not only indicate that the CTD of hFADD is not essential for FADD's ability to induce NF-B activation, but also suggest that NF-B activation is not sufficient or required for the FADD-dependent proliferation.
In conclusion, our results clearly demonstrated that FasL can induce an inflammatory and angiogenic mediator, IL-8, by activating NF-B in a cell-autonomous manner through a signal transduction pathway that is distinct from that of TNF-␣. Further characterization of this pathway will help us to understand and, hopefully, to control the FasL-induced inflammation.