The pro-apoptotic protein Bim is a convergence point for cAMP/protein kinase A- and glucocorticoid-promoted apoptosis of lymphoid cells

The mechanisms by which cAMP mediates apoptosis are not well understood. In the current studies, we used wild-type (WT) S49 T-lymphoma cells and the kin(-) variant (which lacks protein kinase A (PKA)) to examine cAMP/PKA-mediated apoptosis. The cAMP analog, 8-CPT-cAMP, increased phosphorylation of the cAMP response element-binding protein (CREB), activated caspase-3, and induced apoptosis in WT but not in kin(-) S49 cells. Using an array of 96 apoptosis-related genes, we found that treatment of WT cells with 8-CPT-cAMP for 24 h induced expression of mRNA for the pro-apoptotic gene, Bim. Real-time PCR analysis indicated that 8-CPT-cAMP increased Bim RNA in WT cells in <2 h and maintained this increase for >24 h. Bim protein expression increased in WT but not kin(-) cells treated with 8-CPT-cAMP or with the beta-adrenergic receptor agonist isoproterenol. Both apoptosis and Bim expression were reversible with removal of 8-CPT-cAMP after <6 h. The glucocorticoid dexamethasone also promoted apoptosis and Bim expression in S49 cells. In contrast, both UV light and anti-mouse Fas monoclonal antibody promoted apoptosis in S49 cells but did not induce Bim expression. 8-CPT-cAMP also induced Bim expression and enhanced dexamethasone-promoted apoptosis in human T-cell leukemia CEM-C7-14 (glucocorticoid-sensitive) and CEM-C1-15 (glucocorticoid-resistant) cells; increased Bim expression in 8-CPT-cAMP-treated CEM-C1-15 cells correlated with conversion of the cells from resistance to sensitivity to glucocorticoid-promoted apoptosis. Induction of Bim appears to be a key event in cAMP-promoted apoptosis in both murine and human T-cell lymphoma and leukemia cells and thus appears to be a convergence point for the killing of such cells by glucocorticoids and agents that elevate cAMP.     

Identification and characterization of the interactive proteins with cytotoxic T-lymphocyte antigen-2α

Cytotoxic T-lymphocyte antigen-2α (CTLA-2α) is a potent inhibitor of cathepsin L-like cysteine proteases. Recombinant CTLA-2α is known to be a potent, competitive inhibitor of cathepsin L-like cysteine proteases. In this study, cathepsin L, cathepsin C, and tubulointerstitial nephritis antigen-related protein 1 (TINAGL1) were identified as novel interactive proteins of CTLA-2α by the yeast two-hybrid screening system. The direct interactions and co-localization of these proteins with CTLA-2α were confirmed using co-immunoprecipitation and immunofluorescence staining, respectively. The disulfide-bonded CTLA-2α/cathepsin L complex was isolated from mouse tissue. CTLA-2α was found to be specific and consistently expressed on the maternal side of the mouse placenta. Double immunofluorescence analysis showed that CTLA-2α was co-localized with cathepsin L, cathepsin C, and TINAGL1 in placenta. A simple cell-based fluorescence assay revealed that CTLA-2α exhibited inhibitory activity toward cathepsin C in live cells, which indicated that CTLA-2α is a novel endogenous inhibitor of cathepsin C.     

Proteomic and Metabolic Analyses of S49 Lymphoma Cells Reveal Novel Regulation of Mitochondria by cAMP and Protein Kinase A

Cyclic AMP (cAMP), acting via protein kinase A (PKA), regulates many cellular responses, but the role of mitochondria in such responses is poorly understood. To define such roles, we used quantitative proteomic analysis of mitochondria-enriched fractions and performed functional and morphologic studies of wild-type (WT) and kin⁻ (PKA-null) murine S49 lymphoma cells. Basally, 75 proteins significantly differed in abundance between WT and kin⁻ S49 cells. WT, but not kin⁻, S49 cells incubated with the cAMP analog 8-(4-chlorophenylthio)adenosine cAMP (CPT-cAMP) for 16 h have (a) increased expression of mitochondria-related genes and proteins, including ones in pathways of branched-chain amino acid and fatty acid metabolism and (b) increased maximal capacity of respiration on branched-chain keto acids and fatty acids. CPT-cAMP also regulates the cellular rate of ATP-utilization, as the rates of both ATP-linked respiration and proton efflux are decreased in WT but not kin⁻ cells. CPT-cAMP protected WT S49 cells from glucose or glutamine deprivation, In contrast, CPT-cAMP did not protect kin⁻ cells or WT cells treated with the PKA inhibitor H89 from glutamine deprivation. Under basal conditions, the mitochondrial structure of WT and kin⁻ S49 cells is similar. Treatment with CPT-cAMP produced apoptotic changes (i.e. decreased mitochondrial density and size and loss of cristae) in WT, but not kin⁻ cells. Together, these findings show that cAMP acts via PKA to regulate multiple aspects of mitochondrial function and structure. Mitochondrial perturbation thus likely contributes to cAMP/PKA-mediated cellular responses.     

Activation of the CAMP signaling pathway increases apoptosis in human B-precursor cells and is associated with downregulation of Mcl-1 expression

During B- and T-cell ontogeny, extensive apoptosis occurs at distinct stages of development. Agents that increase intracellular levels of cAMP induce apoptosis in thymocytes and mature B cells, prompting us to investigate the role of cAMP signaling in human CD10⁺ B-precursor cells. We show for the first time that forskolin (which increases intracellular levels of cAMP) increases apoptosis in the CD10⁺ cells in a dose-dependent manner (19%–94% with 0–1,000 μM forskolin after 48 hours incubation, IC₅₀ = 150 μM). High levels of apoptosis were also obtained by exposing the cells to the cAMP analogue 8-chlorophenylthio-cAMP (8-CPT-cAMP). Specific involvement of cAMP-dependent protein kinase (PKA) was demonstrated by the ability of a cAMP antagonist, Rp-isomer of 8-bromo-adenosine- 3′, 5′- monophosphorothioate (Rp-8-Br-cAMPS), to reverse the apoptosis increasing effect of the complementary cAMP agonist, Sp-8-Br-cAMPS. Furthermore, we investigated the expression of Bcl-2 family proteins. We found that treatment of the cells with forskolin or 8-CPT-cAMP for 48 hours resulted in a fourfold decline in the expression of Mcl-1 (n = 6, P = 0.002) compared to control cells. The expression of Bcl-2, Bcl-x_l , or Bax was largely unaffected. Mature peripheral blood B cells showed a smaller increase in the percentage of apoptotic cells in response to 8-CPT-cAMP (1.3-fold, n = 6, P = 0.045) compared to B-precursor cells, and a smaller decrease in Mcl-1 levels (1.5-fold, n = 4, P = 0.014). Taken together, these findings show that cAMP is important in the regulation of apoptosis in B-progenitor and mature B cells and suggest that cAMP-increased apoptosis could be mediated, at least in part, by a decrease in Mcl-1 levels. J. Cell. Physiol. 180:71–80, 1999. © 1999 Wiley-Liss, Inc.     

Bcl-2 protects lymphoma cells from apoptosis but not growth arrest promoted by cAMP and dexamethasone

Glucocorticoids orincreases in cellular cAMP promote apoptosis in many celltypes, including murine S49 cells. We examined the impact of Bcl-2, anantiapoptotic protein, on S49 cell growth and death promoted by theglucocorticoid dexamethasone or agents that increase cAMP:isoproterenol (a -adrenergic agonist) + 3-isobutyl-1-methylxanthine (a phosphodiesterase inhibitor) andforskolin (diterpene). These agents promoted apoptosis (i.e.,increased expression of annexin V) of wild-type (WT) S49 cells, butBcl-2-overexpressing S49 cells were protected from this response. Bcl-2overexpression did not protect cells from G₁ growth arrestbut did allow cells to grow longer in culture and protected cells fromculture-dependent necrosis. Commitment to and reversal fromapoptosis vs. G₁ growth arrest byisoproterenol + 3-isobutyl-1-methylxanthine showed different kinetics. Although both processes required several hours to develop, removal of agonists readily reversed growth arrest, but notapoptosis. Thus commitment to apoptosis is lessreversible than G₁ growth arrest. The findings alsoindicate that glucocorticoid- and cAMP-mediated G₁ growtharrest is unaffected by Bcl-2 overexpression, even though increasedBcl-2 allows these lymphoma cells to resist necrosis and apoptosis.

     

EBV infection renders B cells resistant to growth inhibition via adenylyl cyclase

Cyclic AMP (cAMP) is an important physiological growth inhibitor of lymphoid cells, and the cAMP/protein kinase A (PKA) pathway is disrupted in several immunological disorders and cancers. Epstein Barr virus (EBV) infection of B lymphocytes is responsible for the development of lymphoproliferative disease as well as certain B-lymphoid malignancies. Here we hypothesized that EBV infection might render B lymphocytes resistant to cAMP/PKA-mediated growth inhibition. To test this, we assessed the growth-inhibitory response of cAMP-elevating compounds such as forskolin and isoproterenol, as well as the PKA activator 8-CPT-cAMP in normal B lymphocytes, EBV-infected B cells and in the EBV-negative B lymphoid cell line Reh. We could demonstrate that EBV infection indeed abolished cAMP-mediated growth inhibition of B cells. The defect was pinpointed to defective adenylyl cyclase (AC) activation by forskolin and isoproterenol, resulting in reduced formation of cAMP and lack of PKA activation and CREB phosphorylation. In contrast, 8-CPT-cAMP which directly activates PKA was able to inhibit EBV-infected B cell growth. The physiological implications of these results were underlined by the observation that the ability of forskolin to inhibit camptothecin-induced apoptosis was abolished in EBV-infected B cells. We conclude that EBV infection of B cells abrogates the activation of AC and thereby cAMP formation, and that this dysfunction renders the cells resistant to growth inhibition via the cAMP/PKA pathway.     

Anti-proliferative effects of 8-chloro-cAMP and other cAMP analogs are unrelated to their effects on protein kinase A regulatory subunit expression

Conflicting reports have attributed 8-chloro-cAMP (Cl-cAMP)-mediated inhibition of tumor cell growth to either a toxic 8-chloro-adenosine (Cl-AdR) breakdown product or a Cl-cAMP-mediated decrease in ratio of Type I to Type II regulatory (R) subunits of protein kinase A (PKA). Using the MCF-7 human breast cancer and S49 mouse lymphoma cell lines as models, we show that the effects of Cl-cAMP and other cAMP analogs on growth and R subunit expression are unrelated. MCF-7 cell growth was insensitive to most analogs and inducers of cAMP, but was potently inhibited by Cl-cAMP acting through uptake and phosphorylation of its Cl-AdR breakdown product. Possible roles of adenosine receptors or P(2) purinoceptors in these Cl-cAMP-mediated growth effects were ruled out by studies with agonists and antagonists. Cholera toxin markedly decreased the ratio of Type I to Type II R subunits in MCF-7 cells without affecting growth, while growth inhibitory concentrations of Cl-cAMP or Cl-AdR had insignificant effects on this ratio. In S49 cells, where PKA activation is known to inhibit cell growth, PKA-deficient mutants retained sensitivity to both Cl-cAMP and the related 8-bromo-cAMP. Adenosine kinase (AK)-deficient S49 cells were inhibited only by higher concentrations of these 8-halogenated cAMP analogs. Of the commonly used cAMP analogs, only 8-(4-chlorophenylthio)-cAMP acted purely as a cyclic nucleotide-having no effect on PKA-deficient cells, but strongly inhibiting both wild-type and AK-deficient cells. Where growth inhibitory concentrations of most cAMP analogs reduced RI expression in the AK-deficient mutant, a functionally equivalent concentration of (N(6), O(2'))dibutyryl-cAMP maintained or increased this expression.     

Anesthetic Barbiturates Enhance Gsα-Dependent Cyclic AMP Production in S49 Mouse Lymphoma Cells

Abstract: Cyclic AMP (cAMP) regulates many important physiological processes. Barbiturates influence cAMP regulation, possibly through effects on G proteins. This study used intact S49 mouse lymphoma cells to characterize the role of G proteins in the effect of barbiturates on cAMP regulation. cAMP accumulation was determined in intact S49 WT (wild-type) and S49 cyc^? cells (the G_sα-deficient mutant) by measuring the conversion of [³H]-ATP to [³H]cAMP in cells preloaded with [³H]adenine. Pentobarbital enhanced cAMP accumulation in WT cells in the absence (basal) or presence of isoproterenol but had no effect on the EC₅₀ for isoproterenol. This effect was dose dependent with a 50–60% enhancement at 2 mM pentobarbital. Pentobarbital did not affect forskolin-stimulated cAMP accumulation in WT cells. In cyc^? cells, basal and forskolin-stimulated cAMP accumulation were stimulated only at the highest concentration of pentobarbital used (2 mM). Pentobarbital did not affect the inhibition of cAMP accumulation by somatostatin in WT cells, and pertussis toxin treatment of WT cells did not affect the action of pentobarbital on cAMP accumulation. Pentobarbital did not affect isoproterenol-stimulated adenylyl cyclase activity in whole-cell homogenates or membranes prepared from WT cells. The S-(?)-isomer of pentobarbital enhanced isoproterenol-stimulated cAMP accumulation more than the R-(+)-isomer. Phenobarbital and barbituric acid did not enhance isoproterenol-stimulated cAMP accumulation, whereas the anesthetic barbiturates hexobarbital, pentobarbital, and thiopental all enhanced activity. These results suggest that pentobarbital enhances cAMP accumulation in intact WT cells by a mechanism that is dependent on G_sα but independent of G_i. The properties of barbiturates that are responsible for the enhancement of cAMP accumulation may be related to the properties that are responsible for producing sedation and anesthesia.     

Accumulation of cAMP and calcium in S49 mouse lymphoma cells following hyposmotic swelling

Swelling of S49 "wild type" mouse lymphoma cells in hyposmolar medium was used to examine the effects of cellular deformation on cAMP metabolism. In S49 wild type mouse lymphoma cells incubated in a defined medium, progressive reductions in medium osmolarity of 5-50% resulted in proportionate expansion of cell volume. Increases in cell volume were accompanied by incremental increases in intracellular cAMP and calcium. These responses in S49 cells occurred rapidly, with increases in calcium concentration and cAMP content occurring within 1-2 min. Swelling of S49 cells in the absence of ions (hyposmolar versus normosmolar sorbitol) resulted in a significant accumulation of cAMP. Inclusion of papaverine or isobutyl methylxanthine amplified cAMP accumulation, and omission of calcium, sodium, or magnesium from the medium attenuated, but did not prevent accumulation of cAMP in S49 cells in response to swelling. Exposure to propranolol or nadolol attenuated the ability of swelling to increase cAMP concentration, while treatment with 2',5'-dideoxyadenosine or phentolamine had no effect on swelling-induced cAMP accumulation. It is concluded that cellular deformation of S49 wild type mouse lymphoma cells stimulates rapid accumulation of intracellular calcium and cAMP.     

Identification and characterization of novel PKA holoenzymes in human T lymphocytes

Cyclic AMP-dependent protein kinase (PKA) is a holoenzyme that consists of a regulatory (R) subunit dimer and two catalytic (C) subunits that are released upon stimulation by cAMP. Immunoblotting and immunoprecipitation of T-cell protein extracts, immunofluorescence of permeabilized T cells and RT/PCR of T-cell RNA using C subunit-specific primers revealed expression of two catalytically active PKA C subunits C alpha1 (40 kDa) and C beta2 (47 kDa) in these cells. Anti-RI alpha and Anti-RII alpha immunoprecipitations demonstrated that both C alpha1 and C beta2 associate with RI alpha and RII alpha to form PKAI and PKAII holoenzymes. Moreover, Anti-C beta2 immunoprecipitation revealed that C alpha1 coimmunoprecipitates with C beta2. Addition of 8-CPT-cAMP which disrupts the PKA holoenzyme, released C alpha1 but not C beta2 from the Anti-C beta2 precipitate, indicating that C beta2 and C alpha1 form part of the same holoenzyme. Our results demonstrate for the first time that various C subunits may colocate on the same PKA holoenzyme to form novel cAMP-responsive enzymes that may mediate specific effects of cAMP.     

Antifibrotic effects of curcumin are associated with overexpression of cathepsins K and L in bleomycin treated mice and human fibroblasts

Background

Lung fibrosis is characterized by fibroblast proliferation and the deposition of collagens. Curcumin, a polyphenol antioxidant from the spice tumeric, has been shown to effectively counteract fibroblast proliferation and reducing inflammation and fibrotic progression in animal models of bleomycin-induced lung injury. However, there is little mechanistic insight in the biological activity of curcumin. Here, we study the effects of curcumin on the expression and activity of cathepsins which have been implicated in the development of fibrotic lung diseases.

Methods

We investigated the effects of curcumin administration to bleomycin stimulated C57BL/6 mice and human fetal lung fibroblasts (HFL-1) on the expression of cathepsins K and L which have been implicated in matrix degradation, TGF-β1 modulation, and apoptosis. Lung tissues were evaluated for their contents of cathepsins K and L, collagen, and TGF-β1. HFL-1 cells were used to investigate the effects of curcumin and cathepsin inhibition on cell proliferation, migration, apoptosis, and the expression of cathepsins K and L and TGF-β1.

Results

Collagen deposition in lungs was decreased by 17-28% after curcumin treatment which was accompanied by increased expression levels of cathepsins L (25%-39%) and K (41%-76%) and a 30% decrease in TGF-β1 expression. Moreover, Tunel staining of lung tissue revealed a 33-41% increase in apoptotic cells after curcumin treatment. These in vivo data correlated well with data obtained from the human fibroblast line, HFL-1. Here, cathepsin K and L expression increased 190% and 240%, respectively, in the presence of curcumin and the expression of TGF-β1 decreased by 34%. Furthermore, curcumin significantly decreased cell proliferation and migration and increased the expression of surrogate markers of apoptosis. In contrast, these curcumin effects were partly reversed by a potent cathepsin inhibitor.

Conclusion

This study demonstrates that curcumin increases the expression of cathepsins K and L in lung which an effect on lung fibroblast cell behavior such as proliferation, migration and apoptosis rates and on the expression of TGF-β1 in mouse lung and HFL-1 cells. These results suggest that cathepsin-inducing drugs such as curcumin may be beneficial in the treatment of lung fibrosis.     

Involvement of NADPH-dependent and cAMP-PKA sensitive H+ channels in the chorda tympani nerve responses to strong acids

To investigate if chorda tympani (CT) taste nerve responses to strong (HCl) and weak (CO(2) and acetic acid) acidic stimuli are dependent upon NADPH oxidase-linked and cAMP-sensitive proton conductances in taste cell membranes, CT responses were monitored in rats, wild-type (WT) mice, and gp91(phox) knockout (KO) mice in the absence and presence of blockers (Zn(2+) and diethyl pyrocarbonate [DEPC]) or activators (8-(4-chlorophenylthio)-cAMP; 8-CPT-cAMP) of proton channels and activators of the NADPH oxidase enzyme (phorbol 12-myristate 13-acetate [PMA], H(2)O(2), and nitrazepam). Zn(2+) and DEPC inhibited and 8-CPT-cAMP, PMA, H(2)O(2), and nitrazepam enhanced the tonic CT responses to HCl without altering responses to CO(2) and acetic acid. In KO mice, the tonic HCl CT response was reduced by 64% relative to WT mice. The residual CT response was insensitive to H(2)O(2) but was blocked by Zn(2+). Its magnitude was further enhanced by 8-CPT-cAMP treatment, and the enhancement was blocked by 8-CPT-adenosine-3'-5'-cyclic monophospho-rothioate, a protein kinase A (PKA) inhibitor. Under voltage-clamp conditions, before cAMP treatment, rat tonic HCl CT responses demonstrated voltage-dependence only at ±90 mV, suggesting the presence of H(+) channels with voltage-dependent conductances. After cAMP treatment, the tonic HCl CT response had a quasi-linear dependence on voltage, suggesting that the cAMP-dependent part of the HCl CT response has a quasi-linear voltage dependence between +60 and -60 mV, only becoming sigmoidal when approaching +90 and -90 mV. The results suggest that CT responses to HCl involve 2 proton entry pathways, an NADPH oxidase-dependent proton channel, and a cAMP-PKA sensitive proton channel.     

Toll-like receptor 4-mediated cAMP production up-regulates B-cell activating factor expression in Raw264.7 macrophages

B-cell activating factor (BAFF) plays a role in the generation and the maintenance of mature B cells. Lipopolysaccharide (LPS) increased BAFF expression through the activation of toll-like receptor 4 (TLR4)-dependent signal transduction. Here, we investigated the mechanism of action on mouse BAFF (mBAFF) expression by cAMP production in Raw264.7 mouse macrophages. mBAFF expression was increased by the treatment with a cAMP analogue, dibutyryl-cAMP which is the activator of protein kinase A (PKA), cAMP effector protein. PKA activation was measured by the phosphorylation of cAMP-response element binding protein (CREB) on serine 133 (S133). cAMP production and CREB (S133) phosphorylation were augmented by LPS-stimulation. While mBAFF promoter activity was enhanced by the co-transfection with pS6-RSV-CREB, it was reduced by siRNA-CREB. PKA inhibitor, H-89, reduced CREB (S133) phosphorylation and mBAFF expression in control and LPS-stimulated macrophages. Another principal cAMP effector protein is cAMP-responsive guanine nucleotide exchange factor (Epac), a Rap GDP exchange factor. Epac was activated by the treatment with 8-(4-chloro-phenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate (CPT), Epac activator, as judged by the measurement of Rap1 activation. Basal level of mBAFF expression was increased by CPT treatment. LPS-stimulated mBAFF expression was also slightly enhanced by co-treatment with CPT. In addition, dibutyryl-cAMP and CPT enhanced mBAFF expression in bone marrow-derived macrophages (BMDM). With these data, it suggests that the activation of PKA and cAMP/Epac1/Rap1 pathways could be required for basal mBAFF expression, as well as being up-regulated in the TLR4-induced mBAFF expression.     

Transfection of a mutant regulatory subunit gene of cAMP-dependent protein kinase causes increased drug sensitivity and decreased expression of P-glycoprotein

Wild-type Chinese hamster ovary (CHO) cells were transfected with a DNA clone (MT-REV, site A) carrying a mouse gene for a dominant mutant regulatory subunit (RI) gene of cAMP-dependent protein kinase (PKA) from S49 cells along with a marker for G418 resistance. G418-resistant transfectant clone R-2D1 was resistant to 8-Br-cAMP-induced growth inhibition and morphological changes. The cells also did not phosphorylate a 50-kDa protein after cAMP stimulation and had decreased PKA activity, both characteristics of PKA mutants. Northern blot analysis indicated that clone R-2D1 was actively transcribing the MT-REV (site A)-specific RNA. We also tested clone R-2D1 for sensitivity to certain natural product hydrophobic drugs and found increased sensitivity to several drugs including adriamycin. Hypersensitivity to these drugs has previously been shown by us to be a characteristic of a CHO PKA mutant cell line. Expression of the mutant RI gene is also associated with a decrease in expression of the multidrug resistance associated P-glycoprotein (gp170) mRNA and protein. These results show that the PKA mutant RI gene from S49 cells acts as a dominant mutation to reduce the total PKA activity in the CHO transfectants as it does in mouse S49 cells. This study also confirms that reduced PKA activity modulates the basal multidrug resistance of these cells, apparently by causing decreased expression of the mdr gene at the protein and mRNA level.     

Regulation of death complexes formation in tumor necrosis factor receptor signaling

TNFα stimulation triggers both cell death and survival programs. Since dysregulated apoptosis or cell growth can cause inflammatory diseases, cancer, or autoimmune disorders, it is important to understand the molecular mechanism of controlling cell death and survival by TNFR downstream signaling molecules. In this study, we used normal diploid cells, mouse embryonic fibroblasts (MEFs), to mimic the general TNFα-resistant phenomenon seen under physiological conditions. We elucidated the TNFα-induced death signaling complexes in TNF α-resistant WT MEFs and TNFα-sensitive MEFs that were cFLIP-, RelA-, TRAF2- or RIP1-deficient. Consistent with TNFα-mediated killing, we detected TNFα-induced high molecular weight complexes containing caspase-8 and FADD by gel filtration in the deficient MEFs, especially in those devoid of cFLIP. In addition to the presence of caspase-8-FADD in the TNFα-induced-death complex in the deficient MEFs, we also detected an intermediate protein complex containing RIP1, TRAF2 and caspase-8. Moreover, we demonstrated a correlation between TNFα-sensitivity and death-inducing complex ability in two transformed cell lines, E1A- and Ras- transformed MEFs and PDGF-B-transformed NIH-3T3 cells with PDGF-B signaling inhibited by the tyrosine kinase inhibitor STI571. Taken together, our results suggest the involvement of cFLIP-, RelA-, RIP1-, or TRAF2-related mechanisms for preventing FADD-caspase-8 interaction in wild-type MEFs.