Phosphorylation of the tumor necrosis factor receptor CD120a (p55) recruits Bcl-2 and protects against apoptosis.

Ligation of the tumor necrosis factor alpha receptor CD120a initiates responses as diverse as apoptosis and the expression of NF-kappaB-dependent pro-survival genes. How these opposing responses are controlled remains poorly understood. Here we demonstrate that phosphorylation by p42(mapk/erk2) inhibits the apoptotic activity of CD120a while preserving its ability to activate NF-kappaB. Phosphorylated CD120a is re-localized from the Golgi complex to tubular structures of the endoplasmic reticulum wherein it recruits Bcl-2. Antisense-mediated down-regulation of Bcl-2 antagonized the localization of CD120a to tubular structures and reversed the protection from apoptosis conferred by receptor phosphorylation. We propose that phosphorylation of CD120a represents a novel, Bcl-2-dependent mechanism by which the apoptotic activity of the receptor may be regulated. Thus, oncogenic activation of p42(mapk/erk2) may serve to inhibit the apoptotic activity of this death receptor while preserving NF-kappaB-dependent responses and may thus indirectly contribute to a failure to eliminate cells bearing oncogenes of the Ras-Raf-MEK-p42(mapk/erk2) pathway.

Engagement of the tumor necrosis factor-␣ (TNF␣) 1 receptor CD120a (p55) sets in motion the activation of signaling pathways that lead to apoptosis in certain cells (1). Interaction with TNF␣ leads to trimerization of the receptor and to the binding of the adapter molecule TRADD through death domain⅐death domain interactions (2). The complex then recruits several signaling molecules, including TRAF2 and RIP, which stimulate MAPK and NF-B activation pathways (3)(4)(5), and FADD, which mediates activation of apoptosis by recruitment of caspase-8 (3). The requirement of the FADD-caspase-8 pathway in CD120a-mediated apoptosis has been elegantly demonstrated by the resistance of cells derived from FADD-knock-out mouse embryos to TNF␣-induced apoptosis (6). Activation of caspase-8 leads to the activation of caspase-3 and the cleavage of Bid to generate a truncated, pro-apoptotic fragment (tBid) capable of inducing the release of cytochrome c from mitochondria (7) and thereby promoting apoptosis.
Fas ligand (Fas-L) and to a lesser extent TNF␣-mediated apoptosis have been shown to be negatively regulated by several molecules, including Casper/FLICE-inhibitory protein (FLIP) (8,9), Toso (10), the inhibitor of apoptosis family of proteins (IAP), including c-IAP1, c-IAP2, XIAP, and survivin/ TIAP (11), as well as several members of the Bcl-2 family (12). Casper/FLIP, a homolog of caspase-8, interacts with FADD (8) and prevents the recruitment and activation of caspase-8, thereby inhibiting the signaling cascade initiated by death receptors (9), whereas Toso inhibits Fas-dependent apoptosis through the induction of Casper/FLIP expression (10). In contrast, IAP proteins have been shown to interact with, and inhibit, caspases 3, 7, and 9. In addition, c-IAP1 and c-IAP2 are capable of binding to TRAF1 and TRAF2, although the mechanism by which they promote survival is poorly understood (11).
The family of Bcl-2-related proteins comprises both deathinducing (Bax, Bak, Bcl-xS, Bad, Bid, Bik, and Hrk) and deathinhibiting (Bcl-2, Bcl-xL, Bcl-w, Bfl-1, Mcl-1, and Boo) members (13), and, although the mechanism of action of death-inducing Bcl-2 family members is rapidly emerging, the mechanism underlying the death-inhibiting properties of Bcl-2 family members is only partly understood. The ratio of death antagonists to agonists has been proposed to regulate the death-life rheostat within the cell (13). In addition, several Bcl-2 family members are regulated by subcellular localization (7, 14 -16). Although Bcl-2 and Bcl-xL are essential components of the general apoptosis regulatory machinery, they are relatively poor inhibitors of the death signals induced by TNF␣ and Fas-L in cells that efficiently activate caspase-8 at the death-inducing signaling complex (DISC) (so-called "type I" cells), but effectively block apoptosis in cells that poorly activate caspase-8 at the DISC ("type II" cells) (13,(17)(18)(19)(20). When present, the survival influence of Bcl-2 family proteins in Fas-L or TNF␣induced apoptosis generally occurs downstream of the activation of caspase-8 through the inhibition of cytochrome c release from mitochondria (12,(21)(22)(23).
The MAPKs comprise three major sub-families in mamma-* This work was supported in part by United States Public Health Services Grants HL55549 and SCOR HL 56556 from the National Institutes of Health. 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 U.S.C. Section 1734 solely to indicate this fact.
§ lian cells: 1) the p38 mapk subfamily; 2) the extracellular signalregulated kinases (ERK) p44 mapk/erk1 and p42 mapk/erk2 ; and 3) the c-Jun NH 2 -terminal kinases (JNK). In work previously reported from this laboratory, we have shown that specific members of each MAPK sub-family are rapidly and transiently activated in response to stimulation of mouse macrophages with TNF␣ (24 -26). Moreover, whereas the targets of these kinases include transcription factors involved in the inflammatory response, recent studies have also shown these targets to include CD120a itself (27). In addition, we have shown that activation of p42 mapk/erk2 by other receptors, e.g. growth factor receptors, also promotes phosphorylation of CD120a. (28). As a result of phosphorylation, CD120a expression, especially in the Golgi complex, is down-regulated and the receptor is redistributed to tubular structures associated with the endoplasmic reticulum (27). Thus, the phosphorylation of CD120a by p42 mapk/erk2 may represent a broad mechanism by which TNF␣, growth factors, and other stimuli may regulate the activity of CD120a.
Recent studies have suggested that the activation of p42 mapk/erk2 in the context of other pro-apoptotic signals confers a dominant pro-survival advantage (29,30). For example, activation of MEK1, an upstream activator of p42 mapk/erk2 , has been shown to antagonize Fas-triggered cell death through the up-regulation of Casper/FLIP (29). Activation of p42 mapk/erk2 has also been shown to be protective against TNF␣-induced apoptosis of L929 cells (31), serum deprivation-induced apoptosis in PC12 cells (32), and UV-induced apoptosis in human primary neutrophils (33). However, the p42 mapk/erk2 substrates involved in the survival responses are not known. Given the finding that CD120a itself is phosphorylated upon activation of p42 mapk/erk2 (27), we have investigated the influence of CD120a phosphorylation on TNF␣-induced apoptosis. We report herein that the phosphorylation of CD120a by p42 mapk/erk2 inhibits TNF␣-induced apoptosis in HeLa cells through a Bcl-2-dependent mechanism. In contrast, the ability of the receptor to activate NF-B is unaffected by phosphorylation. As we will show, these findings shed new light on how phosphorylation of CD120a may contribute to the regulation of the multiple functions of this death receptor.

EXPERIMENTAL PROCEDURES
Materials-Rabbit polyclonal anti-Bcl-xL (Ab 1690), goat polyclonal anti-CD120a (Ab 1069), anti-TRADD (Ab 1165), and mouse monoclonal anti-Bcl-2 (Ab 7382) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The hamster monoclonal agonistic (Ab 80-4004-01) and antagonistic anti-CD120a (Ab 80-4005-01) antibodies and the monoclonal anti-CD120b antibody (Ab 80-4009-01) were purchased from R&D Systems (Minneapolis, MN). Anti-myc and antitubulin mouse monoclonal antibodies were from Sigma Chemical Co. (St. Louis, MO). Mouse monoclonal anti-caspase-8 (Ab 66231A) and anti-FADD (Ab 65751A) were from PharMingen (San Diego, CA). The mouse monoclonal anti-FLAG antibody (M2) was from Kodak-Scientific Imaging Systems (Rochester, NY). The mouse monoclonal phosphospecific anti-ERK antibody (E10) was from New England BioLabs (Beverly, MA). Fluorescent secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). The expression vector encoding CD120a has been described previously (27). All deletion and point mutants were constructed using overlapping polymerase chain reaction (34) and verified by restriction enzyme analysis and nucleotide sequencing. The expression vectors for p42 mapk/erk2 and MEK1.ca were a generous gift from Dr. Lynn Heasley, University of Colorado Health Sciences Center, Denver, CO. The vector expressing Bcl-xL was kindly provided by Dr. Hong Zhou, Boston, MA. The vector expressing Bcl-2 in pcDNA3 was a gift from Dr. Gary Johnson, National Jewish Medical and Research Center, Denver, CO. The vectors expressing myc-TRADD and FLAG-FADD were provided by Dr. Hong-Bing Shu, National Jewish Medical and Research Center, Denver, CO.
Transfections and Confocal Immunofluorescence Microscopy-HeLa cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin G, 100 g/ml strep-tomycin, and 2 mM glutamine. Approximately 6 ϫ 10 4 cells/well were seeded in 12-well plates containing 18-mm glass coverslips and grown in 5% CO 2 . Cells were transfected with 250 ng of DNA the following day using LipofectAMINE reagent (Life Technologies). 14 h after transfection, the cells were washed with PBS, fixed for 15 min at room temperature in a solution containing 3% (w/v) paraformaldehyde and 3% (w/v) sucrose in PBS (pH 7.5), washed again, and permeabilized with 0.2% (v/v) Triton X-100 for 10 min. The cells were then washed, blocked for 30 min in Hank's balanced salt solution (without Mg 2ϩ , Ca 2ϩ , or phenol red, pH 7.2) containing 5% normal donkey serum and then incubated with the primary antibodies (1:100) in blocking solution for 2 h. After washing with PBS, the cells were incubated for 1 h with Cy3-and/or fluorescein-conjugated donkey secondary antibodies (1:200). Staining for CD120a was performed using a primary hamster monoclonal antibody (1:200), and a Cy3-conjugated F(abЈ) 2 goat anti-hamster IgG in blocking solution containing 5% normal goat serum. The coverslips were incubated overnight in PBS supplemented with 0.02% sodium azide and mounted in a solution containing 90% glycerol, 10% Tris-HCl 0.1 M, pH 8.5, and 20 mg/ml o-phenylenediamine as an anti-fading agent. To visualize the nuclei, cells were incubated with 10 g/ml Hoechst 33342 together with the secondary antibodies. Cells were observed with a Leica DMR/XA fluorescence microscope using a 100ϫ plan objective. Digital images were captured using a SensiCam camera, deconvolved using the software Slidebook 2.6 (Intelligent Imaging Innovations, Inc., Denver, CO) to remove fluorescence that was not in focus, and processed using Adobe Photoshop 5.0 (Adobe Systems, Inc.).
TUNEL Assay-HeLa cells grown on coverslips and transfected 18 -24 h earlier with the appropriate expression vectors were washed with PBS, fixed and permeabilized as described above, and incubated with terminal transferase reaction solution containing fluorescein-conjugated dUTP for 1 h at 37°C as recommended by the manufacturer (Roche Diagnostics Corp., Indianapolis, IN). The cells were washed three times with 0.03 M sodium citrate, pH 7.4, containing 0.3 M sodium chloride, to remove unbound nucleotides, then washed with PBS. Cells were then blocked and incubated with antibodies as above. The percentage of TUNEL-positive cells among transfected cells was determined by counting at least 200 cells with a confocal microscope.
Caspase-8 Cleavage-2 ϫ 10 5 cells grown in 6-well plates were transfected with 500 ng of DNA using the LipofectAMINE reagent. 18 h after transfection, the cells were washed and lysed in Nonidet P-40 lysis buffer (50 mM Tris buffer, pH 8.0, containing 1% Nonidet P-40, 137 mM NaCl, and protease inhibitors). Post-nuclear supernatants were subjected to SDS-PAGE and transfer to nitrocellulose membranes. Procaspase-8 was detected by Western blotting, and quantified by scanning densitometry using the IMAGE 1.6 software (National Institutes of Health).
NF-B Reporter Gene Assays-Activation of NF-B was determined by a reporter gene assay. Briefly, 2.1 ϫ 10 5 HEK293GT cells/well were seeded in 6-well cluster plates. The following day, the media were replaced 1 h prior to transfection with 0.5 g/ml NF-B-luciferase reporter gene plasmid and varying amounts of CD120a constructs using the calcium phosphate method as previously described (35). The cells were then maintained for 18 -24 h prior to stimulation in fresh media with 0.5 g/ml agonist hamster anti-CD120a monoclonal antibody or an irrelevant IgG as control for 6 h prior to lysis and determination of luciferase activity (36).
Co-immunoprecipitation and Western Analysis-Approximately 1.25 ϫ 10 6 HEK293GT cells/well were seeded on 100-mm plates and grown in 5% CO 2 . Cells were transfected the following day by the calcium phosphate precipitation method. Transfected cells were lysed in 1 ml of immunoprecipitation buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 5 g/ml aprotinin, and 1 mM Na 3 VO 4 ). Post-nuclear supernatants were precleared for 1 h with 25 l of protein A/Gϩ beads (Santa Cruz Biotechnology), then incubated for 1 h at 4°C with 3 g of monoclonal hamster anti-CD120a, or 3 g of monoclonal hamster anti-CD120b or non-immune hamster antibody as controls. Immune complexes were precipitated with 50 l of protein A/Gϩ beads for 1 h at 4°C. The Sepharose beads were washed four times with immunoprecipitation buffer. The precipitates were fractionated on SDS-PAGE under non-reducing conditions, and subsequent Western blotting analysis was performed as described previously (27).
Phosphorothioate Oligonucleotides-Antisense DNA oligonucleotides with a phosphorothioate backbone were synthesized and purified by high pressure liquid chromatography (BIOSOURCE International, Camarillo, CA). The oligonucleotides were 3Ј-biotinylated to allow staining and to prevent them from interfering with the TUNEL assay. The 20-mer antisense oligonucleotide (2009) targeted against the coding region of the bcl-2 gene, and the scrambled-sequence oligonucleotide with a similar nucleotide content (sc48), were described previously (37). A BLASTN search of the NCBI DNA data base revealed no homology of the oligonucleotides to other human genes. The sequences were as follows: antisense, 5Ј-AATCCTCCCCCAGTTCACCC-3Ј; scrambled, 5Ј-CTCATTCCTACCGACACCCC-3Ј. Oligonucleotides were transfected for 24 h using LipofectAMINE Plus reagent as recommended by the manufacturer (Life Technologies). Medium was changed 7 h after the transfection.
Results shown for all experiments are representative of at least three separate experiments.

Phosphorylation of CD120a
Protects from Apoptosis-Given the striking changes in the subcellular distribution of CD120a that result from its phosphorylation by p42 mapk/erk2 (27), we investigated the effect of phosphorylation on receptor function and signaling, including CD120a-mediated apoptosis and NF-B activation. To address this question, we used HeLa cells, because previous studies have shown these cells to undergo apoptosis through the endogenous receptor (38) and hence they express the necessary signaling pathways to mediate this response. In previous studies we induced receptor phosphorylation by co-transfection with constitutively active MEK1 and p42 mapk/erk2 . However, these kinases have been shown to affect the balance between apoptosis and survival (29 -32) and thus the activation of p42 mapk/erk2 , per se, may affect the ability of CD120a to induce apoptosis in a CD120a phosphorylation-independent fashion in cells co-transfected with p42 mapk/erk2 and MEK1.ca. Therefore, to address the role of phosphorylation of CD120a on TNF␣-induced apoptosis in the absence of potential artifacts introduced by activated p42 mapk/erk2 , HeLa cells were transfected with expression vectors encoding wild type CD120a and CD120a.4D/E, a mutant receptor that we have previously shown to mimic the phosphorylated receptor by being redistributed to the endoplasmic reticulum in the absence of p42 mapk/erk2 (27). 14 h later, the cells FIG. 1. Phosphorylation of CD120a protects from the induction of apoptosis. HeLa cells were transfected with wild-type CD120a or mutants of CD120a as indicated. 14 h after transfection, the cells were stimulated with TNF␣ and cycloheximide (CHX) (50 ng/ml and 10 g/ml, respectively, for 4 h), subjected to the TUNEL assay, and stained for CD120a. Nucleic acids were stained with the Hoechst 33342 dye. The percentage of apoptotic cells among CD120a-transfected cells was quantified using a confocal fluorescence microscope, on the basis of characteristic changes in the morphology of the nucleus (a) or by the TUNEL assay (b). c, transfected cells were lysed, and equal amounts of proteins were subjected to Western blotting for uncleaved procaspase-8, or ␣-tubulin as a control. Signals were quantified by scanning densitometry.
FIG. 2. Activation of NF-B is unaffected by the state of phosphorylation of CD120a. a, HEK293 cells were co-transfected with either wild type CD120a or CD120 4D/E (200 ng of each construct) and the NF-B luciferase reporter gene construct. 18 h after transfection, the cells were lysed and assayed for luciferase activity. b, wild type CD120a or CD120a.4D/E were expressed at low level in HEK293 cells by co-transfection with 2.5 ng of each construct with the NF-B luciferase reporter gene construct. 18 h after transfection, the cells were stimulated with 0.5 g of either agonistic hamster anti-mouse CD120a, hamster anti-CD120b as a control, or medium alone. Luciferase activity was quantified in cell lysates after 6 h of stimulation with the indicated antibodies. c, HEK293 cells were transfected as in b but were stimulated with mouse TNF␣ (10 ng/ml for 6 h) before determining luciferase activity.
were stimulated with TNF␣ and cycloheximide (50 ng/ml and 10 g/ml, respectively) for 4 h, fixed, permeabilized, and stained for CD120a under conditions that did not permit the detection of the endogenous receptor. Apoptotic cells were quantified on the basis of characteristic changes in the morphology of the nucleus stained with the Hoechst dye 33342 (Fig.  1a). As expected, transfection of wild type CD120a induced a robust increase in the proportion of apoptotic cells, both in the presence and in the absence of TNF␣ and cycloheximide. In contrast, transfection of the CD120a.4D/E mutant receptor completely abolished apoptosis as compared with that induced by the transfected wild type receptor. To confirm these results, apoptotic cells were also identified by TUNEL staining. Similar results were obtained when the percentage of TUNEL-positive cells among CD120a-transfected cells was quantified by confocal fluorescence microscopy (Fig. 1b). In addition, the inability of the transfected CD120a.4D/E mutant to induce apoptosis was similar to that of CD120a.L351N, a receptor bearing a point mutation in the death domain known to abolish the apoptotic activity of the receptor (39). In contrast, the CD120a.4A (T236A, S240A, S244A, S270A) mutant receptor, in which the four p42 mapk/erk2 consensus sites had been mutated to Ala residues, did not protect the cells from the induction of apoptosis (data not shown). To further confirm the effect of phosphorylation of CD120a on the activation of the apoptotic process, transfected HeLa cells were lysed and equal amounts of proteins from post-nuclear lysates were resolved by SDS-PAGE and subjected to Western blotting for procaspase-8 using an antibody that specifically recognizes uncleaved and inactive procaspase-8. As shown in Fig. 1c and as expected, expression of wild type CD120a induced the processing of procaspase-8 (as did TRADD, which was used as an alternative positive control). In contrast, expression of the CD120a.4D/E mutant receptor did not induce the processing of procaspase-8.
To determine if the phosphorylation of CD120a simply rendered the receptor non-functional we investigated the effect of phosphorylation on the ability of CD120a to activate NF-B using an NF-B-dependent luciferase reporter gene assay. Three approaches were taken. First, HEK293 cells were transfected with expression constructs for wild type CD120a and CD120a.4D/E under conditions that led to constitutive activation of NF-B as a result of receptor overexpression. As can be seen in Fig. 2a, transfection of equal amounts of both wild type CD120a and the phosphorylation mimic CD120a.4D/E led to approximately equivalent levels of NF-B-dependent luciferase reporter gene expression. Second, we expressed low levels of both receptors in HEK293 and specifically activated the transfected receptors by cross-linking with an agonistic hamster anti-mouse CD120a monoclonal antibody. As can be seen in Fig. 2b, cross-linking of the transfected receptors also resulted in approximately equivalent increases in luciferase reporter gene activity, whereas a control hamster monoclonal antibody (anti-mouse CD120b) was without effect. Third, HEK293 cells were co-transfected with the NF-B-luciferase reporter construct and equivalent amounts of either wild type CD120a, CD120a.4D/E, or empty vector (2.5 ng/ml) for 18 h prior to stimulation with mouse TNF␣ (10 ng/ml) for 6 h. As can be seen in Fig. 2c, although transfection with the reporter gene and empty vector led to a low level of luciferase expression, transfection with wild type CD120a or CD120a.4D/E led to a similar increase in basal luciferase expression. However, stimulation with mouse TNF␣ resulted in a ligand-dependent increase in reporter gene expression that was equivalent for both receptors. Thus, although the induction of apoptotic activity was lost as a consequence of CD120a phosphorylation, the ability to activate NF-B was preserved. Furthermore, specific ligation of the transfected receptors using both agonistic antibodies and TNF␣ resulted in the activation of NF-B thus indicating that both the wild type and the phosphorylated receptors are available for signaling this response.
TRADD, FADD, TRAF2, Casper/FLIP, and Caspase-8 Are Not Recruited by CD120a.4D/E-Following the binding of TNF␣ to CD120a, the adapter molecule TRADD binds through its death domain to the death domain of CD120a (2) thereby forming a platform for the assembly of other signaling molecules of the apoptosis cascade, including FADD (3) and caspase-8 (3). Other molecules with pro-survival activity such as TRAF2 and Casper/FLIP are also recruited to the receptor complex. Thus, protection from apoptosis by phosphorylated CD120a may result from sequestration of the pro-apoptotic molecules TRADD and FADD, thereby removing them from their defined sites of action, or from the recruitment of prosurvival molecules to phosphorylated CD120a-associated tubules. To address this question, HeLa cells were transfected with the CD120a.4D/E mutant receptor and were co-stained for CD120a, and TRADD or FADD. As shown in Fig. 3, both endogenous TRADD and FADD showed a homogenous intracellular staining pattern, which was not affected by expression of the CD120a.4D/E mutant receptor (Fig. 3, rows a and b). TRAF2, was also absent from the tubular structures containing CD120a.4D/E (Fig. 3, row c), rendering the involvement of cIAP proteins in the survival effect of the phosphorylated receptor unlikely. Similarly, co-expression of myc-tagged Casper/FLIP and CD120a.4D/E resulted in a lack of co-localization (Fig. 3,  row d), consistent with the failure of FADD to be recruited by CD120a.4D/E and with the conclusion that Casper/FLIP is unlikely to be involved in the pro-survival effect of CD120a.4D/E. Thus, the adapters TRADD, FADD, and TRAF2, and the pro-survival molecule Casper/FLIP, are not recruited to the tubular structures formed by phosphorylated CD120a.
To verify that we could detect the interaction of these proteins by confocal microscopy, we tested whether caspase-8 was recruited to "filaments" formed by the overexpression of FADD and TRADD as previously reported (40) as well as to the CD120a-associated tubular structures. Cells were transfected with the CD120a.4D/E mutant receptor, and stained for endogenous caspase-8. As a control, cells were transfected with FLAG-tagged FADD or myc-tagged TRADD, and then costained for caspase-8 and either the FLAG or myc epitopes. As shown in Fig. 3 (row e), caspase-8 exhibited a diffuse staining pattern in untransfected cells and this was not affected by transfection of the CD120a.4D/E mutant receptor. Consistent with previous reports (21,40), a proportion of cells transfected with FADD showed thick and extensive filaments in the cytoplasm as well as in (or across) the nucleus, the latter imprinting the surface of the nuclear envelope as seen by Nomarski imaging or Hoechst staining (Fig. 3, row g). Cells transfected with TRADD also showed cytoplasmic filaments (Fig. 3, row f). However, in contrast to what was seen with the phosphorylation mimic CD120a.4D/E, caspase-8 was co-localized with the filaments formed by FADD and TRADD (Fig. 3, rows f and g). Thus, caspase-8 is recruited to FADD and TRADD filaments but not to CD120a-associated tubular structures. Collectively, these results suggest that the down-regulation of apoptosis associated with phosphorylation of CD120a does not involve the recruitment of TRADD, FADD, TRAF2, Casper/FLIP, or caspase-8.
Recruitment of Bcl-2 but Not of Bcl-xL by Phosphorylated CD120a-Because the mechanism underlying the protection from apoptosis by phosphorylated CD120a did not appear to involve sequestration of the pro-apoptotic adapter molecules TRADD, FADD, or caspase-8, we next explored the hypothesis that protection may be mediated by the recruitment of prosurvival factors of the Bcl-2 family, especially because Bcl-2 is associated with intracellular membranes especially mitochondria and the endoplasmic reticulum (41). HeLa cells were transfected with the CD120a.4D/E mutant receptor and stained for endogenous Bcl-2 and CD120a. Endogenous Bcl-2 was detected in the cytoplasm and was associated with small granular structures (Fig. 4, a and b), consistent with previous reports (42). This pattern was not significantly affected by transfection of wild type CD120a (Fig. 4, e-h). However, transfection of the CD120a.4D/E mutant receptor induced a dramatic redistribution of Bcl-2 to cytoplasmic tubular structures (Fig. 4, c and d). Double-staining experiments showed precise co-localization of the phosphorylation mimic CD120a.4D/E and Bcl-2, indicating that endogenous Bcl-2 is recruited to the tubules formed by phosphorylated CD120a (Fig. 4, i-l). Cotransfection of vectors encoding Bcl-2 and CD120a.4D/E also resulted in the recruitment of Bcl-2 to CD120a-containing tu-bules (data not shown). Bcl-2 was also recruited to the tubular structures formed when wild type CD120a was phosphorylated by cotransfection with MEK1.ca and p42 mapk/erk2 (Fig. 4, m-p). Colocalization of Bcl-2 and CD120a.4D/E was also observed in cells fixed and permeabilized with methanol instead of paraformaldehyde and Triton X-100, respectively, excluding the possibility that this association could be the result of the use of detergents (43).
We next tested whether the interaction between Bcl-2 and phosphorylated CD120a could be detected as a protein complex by co-immunoprecipitation. CD120a or the CD120a phosphorylation mutants were co-expressed with Bcl-2 in HEK293 cells. As can be seen in Fig. 4q, Bcl-2 was co-immunoprecipitated with the CD120a.4D/E mutant receptor but not with wild type CD120a.wt. Bcl-2 was also co-immunoprecipitated with CD120a when phosphorylated by co-expression with MEK1.ca and p42 mapk/erk2 , a point confirmed by the gel shift of the phosphorylated receptor upon SDS-PAGE (Fig. 4q). No interaction was observed between Bcl-2 and the death domaininactive CD120a.L351N, suggesting that different mechanisms account for the inability of this receptor mutant and the phosphorylated CD120a to protect against the induction of apoptosis.
To determine if the recruitment of Bcl-2 by phosphorylated CD120a was specific, we also stained for Bcl-xL. Bcl-xL was mainly expressed in cytoplasmic structures (Fig. 5) (mostly mitochondria, as assessed by co-localization with Mitotracker orange, data not shown), but was not recruited to the tubular structures containing CD120a.4D/E (Fig. 5). Bcl-xL was also absent from CD120a-containing tubules in HeLa cells that had been cotransfected with CD120a.wt, MEK1.ca, and p42 mapk/erk2 (Fig. 5). These findings thus indicate that phosphorylated CD120a is targeted to the endoplasmic reticulum where it preferentially interacts with Bcl-2.
Bcl-2 Is a Critical Determinant in the Regulation of Apoptosis by Phosphorylated CD120a-We next examined whether the protection against apoptosis afforded by phosphorylated CD120a was mediated by Bcl-2. HeLa cells were transfected for 24 h with a 3Ј-biotinylated Bcl-2 antisense phosphorothioate oligonucleotide, or an oligonucleotide of the same nucleotide content but with a scrambled sequence as a control. Staining with Cy5-streptavidin confirmed the efficient nuclear delivery of the oligonucleotides in 100% of the cells (Fig. 6a). Western blot analyses showed decreased Bcl-2 protein levels in cells treated with the Bcl-2 antisense oligonucleotide at concentrations of at least 300 nM, whereas treatment with the control oligonucleotide had little effect (Fig. 6b). Protein levels of Bcl-xL and ␣-tubulin were unchanged, indicating that the ef-FIG. 5. Bcl-xL is not recruited by phosphorylated CD120a. HeLa cells were cotransfected with Bcl-xL and CD120a.4D/E, or CD120a.wt, p42 mapk/erk2 , constitutively active MEK1 and Bcl-xL. The cells were stained for CD120a and Bcl-xL and observed by confocal fluorescence microscopy. Text in italics refers to staining; roman text refers to transfection. fect of the antisense oligonucleotide was specific for Bcl-2 (Fig.  6b). To examine whether Bcl-2 accounts for some of the prosurvival effect of phosphorylated CD120a, HeLa cells were cotransfected with the CD120a.4D/E mutant receptor together with the Bcl-2 antisense or control oligonucleotides. HeLa cells were also transfected with wild type CD120a as a control. The conditions for both receptors were carefully selected to yield a low level of apoptosis in the absence of antisense Bcl-2 oligonucleotide thereby ensuring that we would be able to detect any increase in the degree of apoptosis in the presence of the oligonucleotide. It was also for this reason that we elected to use modest receptor overexpression to induce the necessary low level of apoptosis in the controls as opposed to inducing apoptosis with ligand exposure. 24 h after transfection, the cells were fixed, subjected to TUNEL assay, and stained for CD120a. In cells transfected with the antisense Bcl-2 oligonucleotide, the level of expression of CD120a was greatly reduced in apoptotic cells and the localization of CD120a.4D/E to cytoplasmic tubules was lost. Instead, the receptor exhibited a dim intracytoplasmic punctate staining (Fig. 6c), suggesting that Bcl-2 may be required for the localization of the phosphorylated receptor to the tubular structures associated with the endoplasmic reticulum. Importantly, treatment with the Bcl-2 antisense oligonucleotide (300 -450 nM) also induced a significant increase in the level of apoptosis induced by CD120a.4D/E, as shown in Fig. 6d. The Bcl-2 antisense oligonucleotide also potentiated apoptosis in cells transfected with the wild type receptor, although to a lesser extent than that seen with CD120a.4D/E. The apoptosis observed was CD120a-specific, because the antisense oligonucleotide did not promote cell death in cells transfected with the empty vector. In contrast, treatment with the control oligonucleotide induced little increase in apoptosis, indicating that most of the effect of the Bcl-2 antisense oligonucleotide was sequence specific. However, some sequence-independent effects of the oligonucleotides may also have occurred to a much lower extent, as reported (44). Taken together, these data demonstrate that Bcl-2 forms a complex with phosphorylated CD120a and participates in the protection against apoptosis conferred by these events.

DISCUSSION
CD120a, a death receptor of the TNF receptor superfamily, induces apoptosis in certain cell types upon ligation by TNF␣.
In the present study we show that phosphorylation of CD120a by p42 mapk/erk2 differentially regulates its function as a death receptor. When phosphorylated, CD120a loses its ability to induce apoptosis and thus confers a survival advantage to cells expressing the phosphorylated receptor. However, the receptor retains its ability to stimulate the activation of NF-B either by high level overexpression or by agonist antibody-or TNF␣mediated ligation of the transfected receptor following low level expression. In addition, we show that Bcl-2 is specifically recruited by phosphorylated CD120a to tubular structures associated with elements of the endoplasmic reticulum and contrib- utes to the survival activity of the phosphorylated receptor. Thus, the phosphorylated CD120a⅐Bcl-2 complex localized within endoplasmic reticulum tubules may represent an alternative signaling complex that is protective against apoptosis in contrast to the previously described pro-apoptotic CD120a⅐ TRADD complex.
Recently reported studies have established the concept that activation of p42 mapk/erk2 confers survival advantages to cells in the face of activation of apoptotic pathways. For example, Xia et al. (32) showed that apoptosis induced by growth factor withdrawal from PC12 cells could be overcome by dominant activation of p42 mapk/erk2 . Similarly, the activation of T-cells by concanavalin A was shown by Yeh et al. (29) to antagonize Fas-triggered cell death through the MEK1-dependent up-regulation of Casper/FLIP, indicating that cross-talk between p42 mapk/erk2 and the pro-apoptotic signals induced by Fas ligation results in a dominant survival response. Therefore, to distinguish between (i) direct inhibition of CD120a-induced apoptosis by the phosphorylated receptor and (ii) potential secondary (or additional) CD120a-independent pro-survival effects of activated p42 mapk/erk2 we studied the mechanism of protection against apoptosis by phosphorylated CD120a with a mutant receptor in which the p42 mapk/erk2 phosphorylation sites were mutated to Glu and Asp residues (CD120a.4D/E) to mimic the effects of phosphorylation on receptor function (27). The findings from the present study reveal a novel mechanism by which p42 mapk/erk2 protects against CD120a-mediated apoptosis, namely, redistribution of the phosphorylated receptor to tubular structures associated with the endoplasmic reticulum wherein it serves as a platform (or signal) for the co-recruitment of Bcl-2.
The mechanism underlying the protective effect conferred by the phosphorylated receptor may be attributable to the loss of the dephosphorylated receptor from its primary site of apoptotic signaling, and/or to a pro-survival effect mediated by Bcl-2. We propose that both mechanisms contribute a role in the protection from apoptosis by the phosphorylated receptor. First, work reported by Schutze et al. (45) has shown that the endocytic inhibitor monodansylcadaverine inhibits the induction of death following cross-linking of CD120a, and studies initially reported by Bradley et al. (46) and confirmed by other groups including our own have conclusively shown that an extensive pool of CD120a exists in the trans-Golgi network (27,47). Likewise, Bennett and colleagues (48) have reported that Fas is localized within the Golgi complex and that disruption of the Golgi complex with brefeldin A inhibits the ability of Fas to induce apoptosis. Thus, although ligation of the cell surface receptor would intuitively seem likely to be necessary for the initiation of apoptosis, these recent studies support the contention that the initiation of apoptotic signaling by death receptors may involve assembly of the signaling complex within the broad spatial confines of the Golgi complex. In contrast, other functions of CD120a, such as the activation of proline-directed protein kinases, occur independently of receptor internalization (45). Because phosphorylation of CD120a by p42 mapk/erk2 promotes its translocation from the Golgi complex to the endoplasmic reticulum while preserving a level of receptor expression at the cell surface, we speculate that the inability of the phosphorylated receptor to signal apoptosis may result in part from selective receptor loss from the Golgi complex while the fraction preserved at the cell surface may be responsible for signaling NF-B activation.
Second, although initial reports suggested that Bcl-2 was incapable of protecting against apoptosis induced by death receptors, recent studies have clarified the mechanisms and conditions under which Bcl-2 is protective against death recep-tors ligation (17). In particular, in type II cells, of which HeLa cells are a representative (49,50), Bcl-2 inhibits apoptosis by blocking the release of cytochrome c from mitochondria (51,52). Our findings also show that Bcl-2 is involved in the protection against apoptosis that results from CD120a phosphorylation, because antisense oligonucleotide-mediated down-regulation of Bcl-2 protein reversed the protection against apoptosis afforded by the CD120a.4D/E expression. In addition, bulk cleavage of pro-caspase-8 in type II cells has been shown to occur as a consequence of cytochrome c release from mitochondria and the ensuing activation of caspase-9 (17). We also observed an absence of bulk pro-caspase-8 processing in response to CD120a.4D/E expression, a finding that is consistent with the conclusion that Bcl-2 serves to inhibit the activation of the mitochondrial amplification pathway in these cells. Thus, Bcl-2 is involved in the protection against apoptosis conferred by CD120a phosphorylation, and this likely is mediated by the prevention of cytochrome c release from mitochondria.
The mechanism by which the phosphorylated CD120a⅐Bcl-2 complex is formed upon CD120a phosphorylation and how it specifically protects against CD120a-induced apoptosis remains to be investigated. We speculate that the recruitment of Bcl-2 to the CD120a-associated endoplasmic reticulum tubules serves to recruit other pro-apoptotic Bcl-2 family members thereby preventing these molecules from interacting with mitochondria in initiating cytochrome c release. A similar mechanism has recently been proposed to explain the pro-survival effect of the adenoviral protein E1B 19K, a distantly related Bcl-2 homolog that has been shown to inhibit FADD-induced death upstream of caspase-8 (21) and to interact with tBidactivated Bax, thereby preventing Bax from promoting cytochrome c release from mitochondria (50). Other pro-apoptotic molecules such as Bak, Bim, and Bid are also potential candidates that could potentially be sequestered by the phosphorylated CD120a⅐Bcl-2 complex. Although an attractive hypothesis, we consider it less likely that the phosphorylated CD120a⅐Bcl-2 complex would recruit non-Bcl-2 family pro-apoptotic proteins such as Apaf-1 or effector caspases, because it has been quite difficult to detect interactions between these proteins in mammalian cells (53,54). Our findings also raise the question of the possible role of the endoplasmic reticulum in the protection against apoptosis conferred by CD120a phosphorylation. Previous work has shown the endoplasmic reticulum to be a major site of Bcl-2 localization (41) and protection against apoptosis. Deletion of the C-terminal 20-residue hydrophobic membrane insertion domain abrogates or diminishes the death-inhibitory effects of Bcl-2 (42), indicating that the subcellular localization of Bcl-2 may contribute to its anti-apoptotic function. Similarly, the pro-survival effect of Bcl-2 is affected by its localization to either the endoplasmic reticulum or mitochondria in a cell-dependent fashion (55). Furthermore, Hacki and colleagues (56) have recently shown that the endoplasmic reticulum-targeted expression of Bcl-2 also prevented the release of cytochrome c from mitochondria thus emphasizing the potential of cross-talk between these two organelles in the regulation of apoptosis. Unexpectedly, although Bcl-2 was recruited to phosphorylated CD120a, Bcl-xL was not. The significance of this finding is currently unclear. However, while in many situations Bcl-2 and Bcl-xL exhibit similar activities, differences in their activities have previously been noted. For example, Bcl-xL, but not Bcl-2, has been found to redistribute from the cytosol to intracellular membranes upon induction of thymocyte apoptosis in response to dexamethasone (57), whereas FK506 and cyclosporin-induced apoptosis in WEHI-231.7 cells is prevented by enforced expression of Bcl-xL but not by Bcl-2 (58).
In conclusion, we have shown that phosphorylated CD120a recruits Bcl-2 to tubular structures associated with the endoplasmic reticulum and that Bcl-2 contributes to the protection against apoptosis that arises as a consequence of CD120a phosphorylation. Phosphorylation of CD120a may thus contribute to the promotion of resistance to death receptor-induced apoptosis in tumor cells while at the same time preserving the ability of the receptor to activate NF-B, a transcription factor that regulates the expression of both pro-survival molecules, including c-IAPs as well as pro-inflammatory cytokines (59). Whether similar events regulate the apoptotic pathway induced by other death receptors such as Fas, or whether this mechanism is specific for CD120a signaling, remains to be elucidated. However, these findings suggest that post-translational modification of the cytoplasmic domain of CD120a may play an important role in the regulation of receptor function and may thus have significant implications in cell survival and function in cancer and in chronic inflammatory diseases, which are associated with persistent activation of the Ras-Raf-MEK-p42 mapk/erk2 pathway by activated oncogenes and cytokines, respectively.