Structure/Function analysis of p55 tumor necrosis factor receptor and fas-associated death domain. Effect on necrosis in L929sA cells.

Tumor necrosis factor (TNF) induces a typical apoptotic cell death program in various cell lines by interacting with the p55 tumor necrosis factor receptor (TNF-R55). In contrast, triggering of the fibrosarcoma cell line L929sA gives rise to characteristic cellular changes resulting in necrosis. The intracellular domain of TNF-R55 can be subdivided into two parts: a membrane-proximal domain (amino acids 202-325) and a C-terminal death domain (DD) (amino acids 326-413), which has been shown to be necessary and sufficient for apoptosis. Structure/function analysis of TNF-R55-mediated necrosis in L929sA cells demonstrated that initiation of necrotic cell death, as defined by swelling of the cells, rapid membrane permeabilization, absence of nuclear condensation, absence of DNA hypoploidy, and generation of mitochondrial reactive oxygen intermediates, is also confined to the DD. The striking synergistic effect of the caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp(OMe)-fluoromethylketone on TNF-induced necrosis was also observed with receptors solely containing the DD. TNF-R55-mediated necrosis is not affected by the dominant negative deletion mutant of the Fas-associated death domain (FADD-(80-205)) that lacks the N-terminal death effector domain. Moreover, overexpression of FADD-(80-205) in L929sA is cytotoxic and insensitive to CrmA, while the cytotoxicity due to overexpression of the deletion mutant FADD-(1-111) lacking the DD is prevented by CrmA. These results demonstrate that the death domain of FADD can elicit an active necrotic cell death pathway.

Tumor necrosis factor (TNF) 1 can induce cell death by necro-sis or apoptosis, depending on the cell line (1)(2)(3)(4) and/or the intracellular ATP concentration (5). Apoptosis is morphologically characterized by membrane blebbing, shrinking of the cell and its organelles, and internucleosomal degradation of DNA (6). Finally, the cell disintegrates and apoptotic bodies are cleared by phagocytosis, in most cases without any detrimental effects on the surrounding tissue (7,8). In contrast, cell death by necrosis is often accompanied by inflammation due to massive release of the cytoplasmic cell content. Necrosis is characterized by swelling of the cell and its organelles and an immediate loss of the plasma membrane integrity (1).
A key step in the pathway to apoptosis is activation of procaspases. Activation of these cysteinyl aspartate-specific proteases is initiated by formation of a death-inducing signaling complex (DISC) after oligomerization of the p55 TNF receptor (TNF-R55) or the Fas receptor by the respective ligands (9,10). Both death receptors contain a homologous C-terminal cytoplasmic death domain (DD) involved in apoptosis (11,12). After binding of TNF to TNF-R55, the clustered DDs recruit the TNF-R55-associated DD-containing protein TRADD (13)(14)(15). TRADD in turn recruits Fas-associated DD protein (FADD) by DD-DD interaction (13). In contrast, the DD of Fas does not require the TRADD adaptor protein but serves as a direct docking surface for FADD (16 -18). Besides its C-terminal DD, FADD contains a death effector domain (DED) implicated in the recruitment of procaspase-8 (19,20). Oligomerization of procaspase-8 leads to proximity-induced autocatalytic activation followed by direct or indirect downstream activation of executionary caspases, which cleave substrates involved in apoptotic morphology (21,22).
The initial intracellular molecular events responsible for necrosis are less well understood. Leist and co-workers (5) proposed a model in which low cellular ATP concentrations give rise to a necrotic cell death process, whereas in the presence of high ATP concentrations the apoptotic, caspase-dependent pathway becomes apparent. On the other hand, it was shown that mitochondria are crucial in the necrotic process (reviewed in Ref. 4). Depletion of mitochondria protects L929sA cells from necrotic cell death (23), and TNF induces the production of mitochondrial reactive oxygen intermediates (ROI) (24). This oxidative phosphorylation-linked ROI production is required for TNF-induced necrotic cell death, since addition of butylated hydroxyanisole (BHA), an oxygen radical scavenger and inhibitor of oxidative phosphorylation (25), blocks TNF cytotoxicity (24). Recently, we demonstrated that inhibition of caspase ac-* This work was supported in part by the Interuniversitaire Attractiepolen. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  tumor necrosis factor receptor; TRADD, tumor necrosis factor receptorassociated death domain; zVAD-fmk, benzyloxycarbonyl-Val-Ala-Asp-(OMe)-fluoromethylketone; FACS, fluorescence-activated cell sorter. tivity by the caspase inhibitors zVAD-fmk or CrmA resulted in enhanced ROI formation and considerably increased the sensitivity to TNF-induced necrosis (26), suggesting a modulator role for caspases in the oxidative metabolism.
In this paper we demonstrate that the DD of TNF-R55 as such is sufficient for mediating necrotic signaling pathways of TNF. We show that the membrane-proximal region, upstream of the DD, is required neither for necrosis nor for formation of mitochondrial ROI. Furthermore, the strong sensitization of TNF-induced necrosis in the presence of caspase inhibitors is also confined to the DD. In contrast to apoptotic systems, overexpression of FADD-(80 -205) lacking the DED is cytotoxic for L929sA cells in a CrmA-insensitive way, while overexpression of a FADD-(1-111) mutant containing the DED is cytotoxic in a CrmA-inhibitory way. This indicates that the death domain of FADD might be implicated in TNF-R55-mediated necrosis.
Cytokines, Antibodies, and Reagents-Recombinant murine TNF was produced in our laboratory and was purified to at least 99% homogeneity. The specific activity amounted to 2.2 ϫ 10 8 IU/mg, as determined in a standardized cytotoxicity assay on L929sA cells. htr1 and htr9 are agonistic mouse monoclonal antibodies directed against the extracellular domain of the p55 human tumor necrosis factor receptor (hTNF-R55) and was generously provided by Dr. M. Brockhaus (Hoffmann-La Roche, Basel, Switzerland) (27). The agonistic anti-human Fas antibody, clone 2R2, was purchased from Cell Diagnostica (Mü nster, Germany). BHA (Sigma) was dissolved in ethanol and used at 100 M. zVAD-fmk (Enzyme Systems Products, Dublin, CA) was dissolved in ethanol and used at 25 M. Propidium iodide (PI; Sigma) was prepared in phosphate-buffered saline (3 mM) and was used at 30 M. Dihydrorhodamine 123 (DHR123; Molecular Probes, Eugene, OR) was dissolved at 5 mM in dimethyl sulfoxide and was used at 1 M.
Plasmids-Constitutive expression of hTNF-R55 and various mutants thereof were obtained by cloning the cDNA after the early SV40 promoter in pSV25S as described previously (29). pSV2neo, containing the neomycin resistance gene, was used as a selection vector. Mutants were generated by standard cloning procedures and subsequently verified by sequence analysis. For transient transfection assays, receptor variants were cloned into pCDM8 (Invitrogen, Carlsbad, CA). The expression vectors for CrmA (cDNA was a gift from D. Pickup, Durham, NC) and human Fas (the cDNA was a gift from S. Nagata, Department of Genetics, Osaka University Medical School, Suita, Japan) have been described previously (30). Mouse RIP and procaspase-8 (31) were cloned via reverse transcription-polymerase chain reaction and introduced into the mammalian expression vector pCDNA1 and pCDNA3 (Invitrogen, Carlsbad, CA), respectively. The human FADD and TRADD genes were also picked up by reverse transcription-polymerase chain reaction and cloned into pCDNA1. FADD-(80 -205), encoding a DED-deficient FADD molecule and reported as a dominant negative molecule in many apoptotic assays (32), FADD-(1-111), containing the entire DED but lacking the DD, were made by standard cloning procedures. All sequences were verified by sequence analysis.
FACS Analysis of Receptor Expression-Cells were cultured in uncoated 24-well suspension plates. At day 1, cells were seeded at 5 ϫ 10 5 /well and incubated at 37°C in a humidified air incubator. 1 ϫ 10 6 cells were incubated on ice for 1 h with 200 l of primary anti-hTNF-R55 antibody solution (htr9 at 1 ng/l). Fluorescein isothiocyanateconjugated goat anti-mouse Ig (Harlan Sera-Lab, Crawley Down, UK) was used as secondary antibody. Fluorescein isothiocyanate fluorescence intensity (measured at 525 nm) was analyzed on a FACScalibur flow fluorocytometer (Becton Dickinson, Sunnyvale, CA), equipped with a 488 nm argon ion laser.
Cytotoxicity Assay-Cells were seeded on day 1 at 2 ϫ 10 4 /well in 96-well plates. The next day, cells were treated with inhibitors, cyto-kines, and/or antibodies, as mentioned. Generally, cells were pretreated for 2 h with inhibitor, followed by 18 -24-h treatment with TNF or htr1 agonistic antibody. Next, medium was removed by flicking the microtiter plate to discard detached dead cells. Crystal violet staining on the remaining adherent cells was used to monitor the extent of cell viability. The percentage of cell survival was calculated as follows: (A 595 treated cells Ϫ A 595 blank well)/(A 595 untreated cells Ϫ A 595 blank well) ϫ 100. In case of combined addition of TNF or htr1 with zVADfmk, the percentage of cell survival was compared with the condition with caspase inhibitor alone.
Measurement of ROI Production and Cell Death by FACS-To obtain L929sA cells in suspension, cells were cultured in bacterial-grade Petri dishes or uncoated 24-well plates. At day 1, cells were seeded at 5 ϫ 10 5 /ml and incubated overnight at 37°C in a humidified air incubator. DHR123 was added together with TNF, and samples were taken at different time points. Simultaneously, PI fluorescence (excitation at 488 nm and detection at 610 nm) was measured to exclude interference by dead cells. Rhodamine 123 fluorescence, as a result from DHR123 oxidation, was excited at 488 nm and was detected at 525 nm on 3000 PI-negative cells. Changes in rhodamine 123 fluorescence are shown by subtracting the basal mean fluorescence of untreated cells from the fluorescence of treated cells at a given time point (24).
Analysis of Cell Death by Transient Transfection-L929sA, 24T2.5, or HeLa H21 cells were seeded 24 h before transfection at 40,000 cells/24-well plate. Transient transfection was done using the Lipo-fectAMINE PLUS transfection system (Life Technologies, Inc.). The cytotoxicity in transient transfection assays with TNF-R55 and FADD constructs was evaluated by cotransfecting the pUT651 reporter gene construct, containing the ␤-galactosidase gene fused to a nuclear localization signal under control of the cytomegalovirus promoter (Cayla, Toulouse, France). The total amount of DNA per 24-well plate was 400 ng. Immediately after removal of the transfection mix, cells were either left untreated or treated with 100 ng/ml htr1 for 24 h, after which cells were lysed to measure ␤-galactosidase activity by chemiluminescence using the Galacto Light kit according to manufacturer's instructions

RESULTS
The DD of hTNF-R55 Is Required for Induction of Necrosis-L929sA cells were stably transfected with cDNAs encoding different hTNF-R55 variants (Fig. 1A) and a pSV2neo selection plasmid. After selection, several clones were screened for plasma membrane expression of hTNF-R55 and the different mutants. FACS analysis revealed constitutive cell membrane expression of full-length hTNF-R55, of the deletion mutants hR55⌬203-304, hR55⌬327-426, and hR55⌬243-383, as well as of the hR55-L351A mutant mimicking the DD-inactivating lpr cg mutation originally found in Fas (Fig. 1, A and B).
Specific triggering of the membrane-associated hTNF-R55 mutants was achieved by treatment of the cells with the agonistic antibody htr1. The L929sA transfectants expressing those two receptor variants containing an active DD, viz. hTNF-R55 and hR55⌬203-304, displayed htr1-dependent death ( Fig. 2A). In contrast, oligomerization of hR55⌬327-426, hR55-L351A, or hR55⌬243-383 with htr1 revealed the inability of TNF-R55 variants lacking the death domain to trigger cell death ( Fig. 2A). Next, we excluded the possibility of having selected for TNF-resistant L929sA cell clones (27). Therefore, we treated these murine cells with human TNF, which interacts both with human and murine TNF-R55, and with agonistic anti-murine TNF-R55 polyclonal antibodies. We found that triggering of endogenous TNF-R55 was still cytotoxic (data not shown).

TNF-R55 DD and FADD Signaling to Necrosis
Microscopic evaluation of treated cells revealed that both hTNF-R55 and hR55⌬203-304 induced characteristic necrotic swelling of the cell, resulting in loss of membrane integrity and finally cell lysis (Fig. 2B). Staining with PI further demonstrated the absence of nuclear condensation (Fig. 2B). Triggering of hR55⌬327-426, hR55⌬243-383, or hR55-L351A did not result in cell death or in any morphological changes (data not shown). Hence, the DD of TNF-R55 is required and sufficient for TNF-R55-mediated necrosis.
zVAD-fmk Increases DD-mediated Necrosis-Recently, we demonstrated that overexpression of CrmA, which is a specific inhibitor of caspase-1 and caspase-8 (33), strongly increased TNF-induced necrosis in L929sA cells, instead of blocking it. A similar observation was made when cells were pretreated with zVAD-fmk (26). To elucidate the mechanism of zVAD-fmkinduced synergy, the different TNF-R55 mutants were triggered in the presence of this caspase inhibitor. As shown in Fig.  3A, htr1-induced necrosis by clustering hTNF-R55 was 100-fold sensitized in the presence of zVAD-fmk. In contrast, the necrotic inactive mutants hR55-L351A, hR55⌬327-426, or hR55⌬243-383 remained insensitive to htr1 treatment, even in the presence of zVAD-fmk (Fig. 3, B, D, and E). However, these clones retained the ability to respond to a combined treatment of human TNF and zVAD-fmk, indicating that the endogenous zVAD-fmk-sensitive pathway(s) were still intact in these cells (data not shown). Necrotic cell death induced by hR55⌬203-304, on the other hand, was enhanced as strongly by zVAD-fmk as the full-length receptor (Fig. 3C). This demonstrates that the synergistic action of caspase inhibitors to necrotic cell death occurs independently of the membrane-proximal region of hTNF-R55.
Induction of DD Necrosis Is Accompanied by ROI Production-The production of mitochondrial ROI by TNF has been shown to be crucial for necrotic cell death of L929sA cells (24). Nevertheless, it is still not clear which signaling pathways originating from TNF-R55 are involved in the formation of ROI. When hTNF-R55 was triggered by htr1, the generation of ROI could be detected by the accumulation of oxidized DHR123 in PI-negative cells (Fig. 4A). Simultaneously, necrotic cell death was monitored by the uptake of PI (Fig. 4B). After 3 h of incubation, about 50% of the cells were dead, whereas the remaining plasma membrane-intact cells produced twice as much ROI. Aggregation of hR55⌬203-304 resulted in a delayed cell death, as reported previously (34). However, the extent of ROI production in PI-negative cells at 50% cell death was exactly the same as with full-length receptor. Treatment of cells expressing hR55⌬327-426 or hR55-L351A did not result in the production of ROI (data not shown). Thus the DD alone is sufficient to generate a full oxidative response.
To examine whether an increase in ROI is required for ne- TNF-R55 DD and FADD Signaling to Necrosis crotic cell death, cells were incubated in the presence of the hydrophobic radical scavenger and inhibitor of oxidative phosphorylation BHA (25). Fig. 5 shows that BHA strongly delayed the formation of PI-positive cells, both in cells expressing hTNF-R55 and hR55⌬203-304. Furthermore, BHA abrogated almost completely the synergistic effect of zVAD-fmk, confirming the involvement of mitochondrial ROI production in zVADfmk-synergized necrotic cell death (26).
FADD-(80 -205) Induces CrmA-insensitive Cell Death in L929sA Cells-The TNF-R55 adapter molecules TRADD and FADD have been shown to be required for TNF-R55 induced apoptosis (13). Also RIP is recruited in the TNF-R55 complex and its overexpression induces apoptotic cell death (35). We examined the influence of FADD-(80 -205), a dominant negative mutant for TNF-R55-induced apoptosis (32), on TNF-R55, TRADD, RIP, and FADD cytotoxicity in the necrotically dying L929sA cells. Surprisingly, in L929sA cells transient expres-sion of FADD-(80 -205) alone resulted already in massive cell death (Fig. 6A). As a control, overexpression of FADD-(80 -205) in 24T2.5 (Figs. 6B and Fig. 7B) or HeLa H21 cells (Fig. 8C) prevented TNF-R55-, TRADD-, and RIP-induced apoptosis. Overexpression of hTNF-R55 or hR55⌬203-304 induced already substantial cell death in both L929sA and 24T2.5 cells (Fig. 6, A and B). Addition of htr-1 agonistic antibody further enhanced cell death in this transfection system. Receptors lacking an active death domain (hR55-L351A and hR55⌬327-426) were incapable of activating any cell death program in both cell lines. Thus, the transient transfection cytotoxic assays with TNF-R55 mutants reflect the data obtained in stable transfected cell lines (Fig. 2A). The strong cytotoxic effect of FADD-(80 -205), that lacks the DED and is not able to recruit procaspase-8, prompted us to distinguish whether FADD is at the bifurcation of necrotic or apoptotic cell death in L929sA cells. Therefore, we tested whether cell death by transient overexpression of human TNF-R55 mutants, TRADD, FADD, and RIP was affected by cotransfection with the caspase-8 inhibitor CrmA. Clearly, overexpression of CrmA is not able to block TNF-R55-, hR55-link-326 -426-, TRADD-, and RIP-induced cell death in L929sA cells (Fig. 8A). Furthermore, also FADDinduced cell death is insensitive to CrmA-mediated inactivation of caspases. This indicates that FADD-induced cell death occurs despite the presence of CrmA. To elaborate further on the ability of FADD to induce cell death in the presence of a caspase-8 inhibitor, we tested the influence of CrmA overexpression on the cytotoxicity by FADD mutants that either lacked the DED domain or the DD domain, FADD-(80 -205) and FADD-(1-111), respectively. As shown in Fig. 8B, FADDand FADD-(80 -205)-induced cell death is not affected by the presence of CrmA. In contrast, cell death induced by FADD-(1-111) in the same cells is blocked by the presence of CrmA, indicating a role for caspase-8. As a control, the same constructs were transfected in HeLa H21 cells (Fig. 8C). In these cells both FADD-and FADD-(1-111)-induced cell death is counteracted by cotransfecting CrmA, whereas FADD-(80 -205) has no killing capacity at all, as expected (Fig. 8C).

DISCUSSION
Both hTNF-R55 and Fas mediate apoptosis via their socalled DD motif, which allows aggregation with other DDcontaining proteins (36). The important role of the TNF-R55 DD in apoptotic cell death has been demonstrated in various cell lines. In contrast, the specific receptor domains involved in TNF-R55-induced necrosis remain to be characterized. Therefore, we performed a structure/function analysis of hTNF-R55 in respect with cell killing in the fibrosarcoma cell line L929sA, which dies necrotically upon exposure to TNF (1,26,37).
We observed that TNF-R55 molecules lacking an active DD were incapable of inducing cellular necrosis. The typical necrotic morphology seen in hR55⌬203-304-mediated cell death demonstrates that the DD as such is sufficient to generate necrosis. Besides the ability to induce characteristic morphological features of necrosis, stimulation of hTNF-R55 or hR55⌬203-304 showed a typical pattern of diploidy and tetraploidy, without indication of internucleosomal degradation of DNA ( Fig. 2B; data not shown). Similarly, in the transient transfection assays, the receptor hR55-link-326 -426, lacking the FAN binding site (38), also induced necrosis in these cells. Hence, no secondary signal, generated by the membrane-proximal region of TNF-R55, is required for necrotic cell death in L929sA cells. Fas, which only contains a short membraneproximal region (12), normally mediates apoptosis. However, in the presence of caspase inhibitors (30) or in the absence of procaspase-8 (39), Fas-induced apoptosis is switched to necrosis. This suggests that in the absence of caspase activation a No membrane blebbing, characteristic for apoptosis, is observed. Right panels, PI was added to the cells and nuclear morphology of the dying cells was analyzed. PIpositive, necrotic cells revealed no nuclear condensation and no marginal chromatin localization, a characteristic of apoptotically dying cells (37). The swollen cell (open arrow) in the picture of the hTNF-R55 transfectant just started to die as PI uptake was hardly visible, confirming that cell swelling precedes loss of cellular membrane integrity.

TNF-R55 DD and FADD Signaling to Necrosis
hidden necrotic pathway becomes apparent. Thus, both the DD of hTNF-R55 and Fas seem to initiate necrosis in a caspaseindependent way (30). Leist and co-workers (5) identified the cellular ATP concentration as a crucial parameter in the decision between apoptosis and necrosis. In human T cells depleted of ATP, default apoptotic triggers such as staurosporine or Fas, switched from apoptosis to necrosis, indicating that the energy homeostatic condition of the cell determines the kind of cell death process activated (5). Whether the concentration of ATP is at the decision point between TNF-induced necrosis and Fas-induced apoptosis in the same cellular context of L929sA cells is not clear yet. In the L929sA system the absence of caspases clearly facilitates necrosis (26,30). If ATP concentration would be implicated in the decision between apoptotic (high ATP) and necrotic cell death (low ATP), one would predict that a cell death signal in the absence of caspases would favor somehow ATP depletion. Otherwise, it is also possible that depending on the trigger or the cell line used, different mechanisms, such as reactive oxygen generation and/or ATP con-  TNF-R55 DD and FADD Signaling to Necrosis centration, might initiate or promote the necrotic process.
Apoptosis by TNF-R55 is the result of ligand-induced formation of a death-inducing signaling complex leading to procaspase-8 activation (19). In this receptosome complex, TRADD is recruited by the oligomerized DDs of TNF-R55 (13-15). Next, overexpression studies showed that TRADD serves as a docking molecule for FADD. Dominant negative FADD, FADD-(80 -205), prevents TNF-induced procaspase-8 activation (20). We were unable to demonstrate any caspase activation in L929sA cells after TNF stimulation (30). Nevertheless, inhibition of caspase activity by CrmA or zVAD-fmk strongly enhanced the TNF-induced necrotic process (30), which suggests that TNF might activate nondetectable levels of caspase activity. To examine whether procaspase-8 recruitment is implicated in necrotic cell death, we evaluated the effect of the dominant negative mutant of FADD-(80 -205) (13,16,17) in L929sA cells. Unexpectedly, transient overexpression of FADD-(80 -205) was already highly cytotoxic in L929sA cells. Moreover, in several attempts we were not able to generate stable transfectants of L929sA cells overexpressing FADD-(80 -205) due to strong counter selection. 2 Cytotoxicity enhancing effects of the dominant negative mutant of FADD has also been reported for TNF-induced necrosis in NIH3T3 cells in the presence of caspase inhibitors or protein synthesis inhibitors (40). However, in this particular system, FADD-(80 -205) on itself was not cytotoxic, indicating that L929sA cells might be more prone to necrotic cell death. It was also found that absence of caspase-8 favors FADD-induced necrosis in Jurkat cells (33). Recently, it was also reported that FADD-(80 -205) or a caspase-8-specific inhibitor sensitizes TNF-induced cell death in NIH3T3 cells (41). A similar mechanism might occur during TNF-induced necrosis in L929sA cells. Inefficient recruitment of FADD and/or procaspase-8 in the TNF-R55 DISC would result in low levels of caspase-8 facilitating necrotic signaling. In contrast, efficient recruitment of procaspase-8 in the Fas DISC complex in the same cells allows apoptotic signaling (30). The molecular mechanism for this inefficient recruitment and/or activation of caspases by TNF-R55 in L929sA cells is not clear, but it is not due to absence of endogenous TRADD. 3 The strong synergism of CrmA or zVAD-fmk on TNF-induced necrosis (26), the observation that absence of caspases favors necrotic cell death (30,39,41) and that FADD dominant negative mutants facilitate TNF-mediated cell death (40,41), suggest that caspases might be implicated in anti-necrotic mechanisms (4,26,30). In this paper we demonstrate that the dichotomy between necrotic and apoptotic cell death might be situated at FADD, viz.

TNF-R55 DD and FADD Signaling to Necrosis
ing of an intact death domain would initiate cell death that is not prevented by CrmA (caspase-independent necrosis). This would argue that the decision between necrosis and apoptosis is taken in the receptosome complex. As FADD has been implicated in TNF-R55, Fas, TRAIL-R1, and TRAIL-R2 signaling (42,43), one can postulate from all these death domain receptors necrotic signaling could be initiated. However, this does not exclude that there are deviations between necrotic and apoptotic cell death at other levels in the cell death pathway. Conditions that favor necrotic cell death are inactivation of caspases (44) or low levels of ATP (5).
Oxidative phosphorylation and concomitant oxygen radical production are indispensable in TNF-induced necrosis (23,24). However, the link between receptor and production of radicals in the mitochondria remains unresolved. In this report, we demonstrate that TNF-R55 DD-induced signaling pathways are required and sufficient to generate mitochondrial radical production. Also the strong synergistic effect of zVAD-fmk is confined to DD-initiated signaling components and involves enhanced receptor induced production of ROI as the radical scavenger BHA counteracts the effect of zVAD-fmk. Furthermore, reports of Khwaja (40) and Lü schen (41) demonstrated increased ROI production and protection by BHA in their cell models. Lü schen and co-workers (41) question the causality of ROI production in the observed cell death because other radical scavengers such as BHT and N-acetylcysteine could not mimic the effect of BHA. This might reflect the mechanism of action of BHA, which besides its properties as a direct oxygen radical scavenger, also possess inhibitory activities at the level of oxidative phosphorylation (25). These combined features might explain the strong anti-necrotic properties of BHA. The observation that complex I inhibitors such as amytal and complex II inhibitors such as TTFA reduce TNF-induced cell death in L929sA cells (23) underline the important contribution of the oxidative phosphorylation in the necrotic process. It is also possible that the specific inhibitory action of BHA reflects that involved ROI are formed and act in a hydrophobic environment, viz. at mitochondrial membranes, where BHA can penetrate but not most other hydrophilic, reducing agents.
How does addition of zVAD-fmk sensitize DD-mediated ROI production and consequent necrosis? This property is also shared with CrmA, since CrmA-overexpressing cells exhibited a 1000-fold sensitization to TNF-induced necrosis (26,30). An obvious target for inhibition would be caspase-8 activation in the receptosome complex. Inhibition or very low levels of active caspase-8 might allow the formation of a more efficient necrotic DISC complex. In this respect it was shown that caspase-8 is able to proteolyze members of the receptosome complex such as RIP (45). Furthermore, inhibition of procaspase-8 recruitment by FLIP allows the conversion of a proapoptotic signal to Fasinduced proliferation in T cells (46). This demonstrates that modulation of levels of active caspase-8 might regulate the outcome of a ligand-induced signal transduction pathway.
However, our results do not exclude that zVAD-fmk and CrmA might also operate at other levels of the cell death pathway. An intriguing possibility is that zVAD-fmk-or CrmAinhibited proteases/caspases are implicated in a surveillance system for damaged mitochondria (47). If this removal system would be blocked, accumulation of damaged mitochondria might occur, which would further increase ROI production in an autoamplifying way (4). In support of this hypothesis is the observation that zVAD-fmk synergistically enhances TNF-induced ROI production (26) and that preincubation with zVADfmk or overexpression of CrmA results in higher levels of spontaneous ROI production (26). 2 Moreover, zVAD-fmk alone, in the absence of any ligand, is able to evoke some necrotic cell death in cells overexpressing Fas, hTNF-R55,or hTNF-R55⌬203-304 (data not shown). These results suggest that zVAD-fmk-or CrmA-inhibitable proteases/caspases might be implicated in the regulation of the basal oxidative metabolism.
Finally, we can conclude that TNF is able to activate directly a necrotic pathway initiated from the DD of TNF-R55. This necrotic pathway might include recruitment of FADD and is sensitized in the presence of zVAD-fmk, suggesting an antinecrotic role for caspase-8. FADD would be the point of bifurcation between apoptotic and necrotic signaling, since FADD-