A novel cytoplasmic domain of the p55 tumor necrosis factor receptor initiates the neutral sphingomyelinase pathway.

The human p55 tumor necrosis factor (TNF) receptor (TR55) initiates at least two independent signaling cascades. The acidic sphingomyelinase (A-SMase) pathway involves a phosphatidylcholine-specific phospholipase C, an endosomal A-SMase, and controls expression of multiple TNF-responsive genes through induction of transcription factors such as NF-kappaB. The neutral sphingomyelinase (N-SMase) pathway comprises a membrane-bound N-SMase, proline-directed protein kinases, as well as phospholipase A2 and appears critical for the inflammatory responses induced by TNF. While the domain of TR55 that induces A-SMase is probably identical to the death domain, the exact location and extent of a putative N-SMase activation domain are still unknown. Structure-function analysis of TR55 deletion mutants revealed a novel region of 11 amino acids at position 309-319 that is both necessary and sufficient for activation of N-SMase. The N-SMase activation domain is distinct from the death domain and incapable of induction of A-SMase, NF-kappaB, and cytotoxicity. Taken together, our results suggest that a functionally independent region of TR55 is responsible for selectively initiating the N-SMase pathway that couples to an important inflammatory signaling cascade.

The human p55 tumor necrosis factor (TNF) receptor (TR55) initiates at least two independent signaling cascades. The acidic sphingomyelinase (A-SMase) pathway involves a phosphatidylcholine-specific phospholipase C, an endosomal A-SMase, and controls expression of multiple TNF-responsive genes through induction of transcription factors such as NF-B. The neutral sphingomyelinase (N-SMase) pathway comprises a membrane-bound N-SMase, proline-directed protein kinases, as well as phospholipase A 2

and appears critical for the inflammatory responses induced by TNF. While the domain of TR55 that induces A-SMase is probably identical to the death domain, the exact location and extent of a putative N-SMase activation domain are still unknown. Structure-function analysis of TR55 deletion mutants revealed a novel region of 11 amino acids at position 309 -319 that is both necessary and sufficient for activation of N-SMase. The N-SMase activation domain is distinct from the death domain and incapable of induction of A-SMase, NF-B, and cytotoxicity. Taken together, our results suggest that a functionally independent region of TR55 is responsible for selectively initiating the N-SMase pathway that couples to an important inflammatory signaling cascade.
TNF 1 is a pleiotropic cytokine that exerts a wide variety of cellular responses, including immunoregulatory activities, antiviral activity, cytotoxicity, and the transcriptional regulation of many genes (for reviews, see Refs. [1][2][3][4]. Induction of these responses by TNF is initiated by its interaction with two distinct cell surface receptors of 55 kDa (TR55) and 75 kDa (TR75) apparent molecular mass (reviewed in Refs. 5,6). It has become clear that a large majority of TNF activities, including cytotoxicity, antiviral activity, and stimulation of several second messenger pathways, is mediated by TR55 (7)(8)(9). Recent studies have demonstrated that signal transduction through TR55 involves the activation of a PC-PLC and, subsequently, of an A-SMase (10). Ceramide, generated by A-SMase, in turn leads to induction of NF-B (10 -12), possibly involving protein kinase C as an intermediate (13,14).
A second, equally important signaling pathway initiated by TR55 involves a membrane-bound N-SMase (for review, see Ref. 15). This enzyme hydrolyzes membrane sphingomyelin, leading to activation of proline-directed protein kinases, including CAP kinase (16,17) and members of the MAP kinase family (18,19). MAP kinases may be responsible for phosphorylation and activation of a cytosolic PLA 2 leading to the generation of proinflammatory metabolites (reviewed in Ref. 20). Finally, a cytosolic heterotrimeric protein phosphatase 2A (21), which is likely to play an important role in TR55-mediated down-regulation of c-myc expression (22), may be another member of the N-SMase pathway. Recently, it has been shown that N-SMase mediates TNF-induced activation of the protein kinase Raf (23), possibly involving CAP kinase (24) and thereby linking TR55 to the MAP kinase cascade.
The domain of TR55 activating the A-SMase pathway strikingly corresponds to the so-called death domain responsible for mediating the cytocidal effects of TNF (9,11,25), and proteins have been described that bind directly to TR55 within this region (TRADD, Ref. 26;TRAK,Ref. 27). Conversely, an activation domain for the N-SMase pathway has not yet been identified. So far, structure-function analysis of TR55 has revealed that the N-SMase pathway must be signaled for by a domain N-terminal of amino acid 345 (11). Intriguingly, N-SMase has been proposed to operate at the outer leaflet of the plasma membrane (28), leading to the hypothesis that activation of N-SMase may even require extracytoplasmic parts of TR55.
By analysis of a set of TR55 deletion mutants stably expressed in murine 70Z/3 pre-B cells, we show that the N-SMase activation domain (NSD) consists of a cytoplasmic region of the receptor that comprises only 11 amino acids directly adjacent to the N terminus of the death domain. The NSD extends from amino acids 309 -319 and is both necessary and sufficient to mediate TNF-dependent N-SMase activation. The identification of the NSD represents a prerequisite for the isolation of novel TR55-associated proteins that are components of the N-SMase pathway.

MATERIALS AND METHODS
Plasmids and Reagents-pEF-BOS-TR55 was obtained by cloning a SalI-XbaI fragment of pADB-TR55 (7) containing the complete coding region of the human p55 TNF receptor into the XbaI site of pEF-BOS (Ref. 29; kindly provided by Dr. S. Nagata) following treatment with Klenow polymerase. The TR55-specific mouse monoclonal antibody H398 (30) was obtained from Dr. Scheurich (Stuttgart). Highly purified murine and human recombinant TNF-␣ was provided by Dr. G. Adolf (Boehringer Ingelheim, Vienna). Human recombinant IL-1 was purchased from Genzyme.
Cell Culture and Transfections-COS-1 cells and the murine pre-B cell line 70Z/3 were originally obtained from ATCC. Cell lines were maintained in a mixture of Click's/RPMI 1640 (50/50 volume %) supplemented with 10% calf serum, 10 mM glutamine, and 50 g/ml each of streptomycin and penicillin in a humidified incubator containing 5% CO 2 .
For transient expression experiments, 1 ϫ 10 7 COS-1 cells were electroporated at 960 microfarads/280 V with 20 g of the appropriate pEF-BOS-TR55 plasmid. Cells were seeded onto 10-cm dishes, supplied with fresh medium after 24 h, and harvested after 72 h. Stably transfected 70Z/3 cells were obtained by cotransfection of the corresponding pEF-BOS-TR55 construct with BMGNeo (31) using electroporation and subsequent selection with 1 mg/ml Geneticin (Life Technologies, Inc.). 70Z/3 cells expressing the corresponding TR55 deletion mutant were isolated by magnetic staining using antibody H398 and Mini-MACS columns (Miltenyi Biotec).
Radioligand Binding Assays and Scatchard Analyses-Cells were detached by incubation in PBS, 2 mM EDTA at 37°C for 15 min if necessary, washed, and resuspended in cold PBS with 0.2% FCS and 0.02% sodium azide. Assays were set up in triplicate as follows. 1 ng of 125 I-radiolabeled human recombinant TNF-␣ (DuPont NEN, specific activity 1882 kBq/g) was added to 10 6 cells in a total volume of 300 l of PBS/FCS/azide with or without a 200-fold excess of unlabeled ligand. Cells were incubated at 0°C for 2 h, washed twice in cold PBS/FCS/ azide, and total as well as nonspecific binding was analyzed by ␥-counting. Specific binding was calculated by subtracting nonspecific from total binding. For Scatchard analyses, cells were incubated with serial dilutions of labeled ligand (0.125-5 ng per well) in a total volume of 300 l of PBS/FCS/azide with or without unlabeled ligand. Cells were washed twice in cold PBS/FCS/azide and analyzed by ␥-counting. K a values and receptor numbers per cell were calculated using the computer program "Enzfitter." Preparation of Nuclear Extracts and Electrophoretic Mobility Shift Assays-Nuclear extracts were prepared as described (32). Electrophoretic mobility shift assays were performed by incubating 6 g of nuclear extract with 4 g of poly(dI-dC) (Pharmacia Biotech Inc.) in binding buffer (5 mM HEPES (pH 7.8), 5 mM MgCl 2 , 50 mM KCl, 0.5 mM dithiothreitol, 10% glycerol) in a total volume of 20 l for 20 min at room temperature. Then end-labeled double-stranded oligonucleotide probe (NF-B-specific oligonucleotide containing two tandemly arranged NF-B binding sites of the HIV-1 long terminal repeat enhancer (5Ј-ATCAGGGACTTTCCGCTGGGGACTTTCCG-3Ј), 1 ϫ 10 4 to 5 ϫ 10 4 cpm) was added, and the reaction mixture was incubated for 7 min. The samples were separated on native 6% polyacrylamide gels in low ionic strength buffer (0.25 ϫ Tris borate-EDTA).
Assays for Neutral and Acidic SMase-The micellar SMase assay using exogenous radiolabeled sphingomyelin was performed according to Quintern and Sandhoff (33) with some modifications (11). Briefly, cells were treated in triplicates in 0.5 ml of medium with 100 ng/ml human recombinant TNF-␣ or IL-1 for the indicated times. To measure neutral SMase, cells were homogenized as described, except that 0.5% CHAPS was substituted for 0.2% Triton X-100 in the lysis buffer. Radioactive phosphocholine produced from [N-methyl-14 C]sphingomyelin (labeled in the choline moeity, Amersham CFA566) was identified by TLC and routinely determined in the aqueous phase by scintillation counting. To measure acidic SMase, cells treated with TNF-␣ were homogenized in 200 l of 0.2% Triton X-100. The amount of radioactive phosphocholine produced was measured as described (11).
Cytotoxicity Assays-10 4 cells were seeded in flat-bottom 96-well plates in medium containing 1-5 ng/ml actinomycin D, for optimal sensitization, as well as serial dilutions of human recombinant TNF-␣. After 24 h, 20 l of MTT (Sigma, 2.5 mg/ml in PBS) were added, and incubation was continued for 2 h to allow metabolization of MTT to MTT-formazan. MTT-formazan was solubilized with isopropyl alcohol-HCl (24:1) and colorimetrically determined at 570 nm in a microplate reader (Dynatech).

The N-SMase Activation Domain Is Located in the Cytoplasmic Portion of the p55 TNF Receptor-To clarify whether N-
SMase was possibly activated by the extracellular domain of TR55, we generated a C-terminal deletion mutant of TR55 (TR55⌬205, Fig. 1) that completely lacks the cytoplasmic domain. TR55⌬205 was transfected into the murine pre-B cell line 70Z/3 that does not express endogenous TR55 but is capable of displaying TNF-specific responses after transfection of cDNA encoding human TR55 (7). Geneticin-resistant cells were selected for expression of the mutant receptor by two rounds of staining using the monoclonal antibody H398 directed against the extracellular portion of hTR55 followed by selection of stained cells in a magnetic column. The selected pool of transfectants was analyzed for cell surface expression of the mutant receptor by radioligand binding assays and additional flow cytometry analyses (data not shown). Average receptor numbers per cell as well as ligand binding affinity of the truncated receptor was determined by Scatchard analysis. While the binding affinity of the truncated receptor for TNF was comparable with that of wild type TR55, the average receptor number per cell was distinctly larger than in cells expressing wild type receptors (Fig. 1). This is probably due to the absence of toxic effects that have been described for highly expressed intact cytoplasmic domains of TR55 (34).
TR55⌬205 transfectants were assayed for activation of N-SMase after treatment of cells with TNF or IL-1, which was used as a positive control for N-SMase activation because 70Z/3 cells express endogenous IL-1 receptors. While exposure to IL-1 elicited elevated N-SMase activity in 70Z/3TR55⌬205 cells, no changes of N-SMase activity were seen when the cells were treated with TNF (Fig. 2), suggesting that activation of the N-SMase pathway requires sequences from the cytoplasmic rather than the extracellular portion of TR55.
The N-SMase Activation Domain Consists of a Stretch of 11 Amino Acids between Positions 309 and 319 and Is Distinct from the Death Domain-The above results strongly suggested that the domain through which TR55 interacts with N-SMase is located in the cytoplasmic region between amino acids 205 and 344. To obtain a more precise estimate, we first generated two stable transfectant lines (70Z/3TR55⌬244 and 70Z/ 3TR55⌬308 -340) expressing TR55 mutants that contain suc-FIG. 2. TNF-mediated activation of N-SMase requires amino acids 309 -319 of the p55 TNF receptor. 70Z/3 cells expressing wild type TR55 or deletion mutants of TR55 were stimulated with 100 ng/ml TNF for the indicated times (f). Extracts containing the cytosolic/membrane fraction were prepared and assayed for N-SMase activity using the substrate [ 14 C]sphingomyelin as described. Treatment of cells with 100 ng/ml IL-1 (Ⅺ) served as an internal positive control. N-SMase activities are expressed as percent of control. Basal levels of phosphocholine production were 96 -165 pmol⅐mg Ϫ1 ⅐h Ϫ1 . The values shown represent the means from triplicate determinations; error bars indicate the respective standard deviations. One representative from multiple experiments (n Ն 3 for each deletion mutant) is shown. cessively smaller deletions of the TR55 cytoplasmic domain (Fig. 1). When the transfectants were tested for N-SMase activity, neither 70Z/3TR55⌬244 nor 70Z/3TR55⌬308 -340 cells showed an increase after treatment with TNF (Fig. 2). Therefore, amino acids 308 -340 of TR55 must contain determinants required for successful activation of the N-SMase pathway.
Two additional deletion mutants proved that a domain between amino acids 308 and 340 was not only necessary but also sufficient to activate the N-SMase pathway. Mutant TR55⌬-212-308/346 contains exactly the fragment from amino acid 309 to 345 but lacks almost all of the remaining cytoplasmic domain. The mutant TR55⌬320 was generated to examine whether the death domain (which extends C-terminal of position 326, Ref. 25) was involved in N-SMase activation. As shown in Fig. 2, both TR55⌬212-308/346 and TR55⌬320 mediated TNF-dependent induction of N-SMase with the same kinetics as observed with IL-1. In summary, these results indicate that amino acids 309 -319 are sufficient for activation of N-SMase and that the death domain is not involved in activation of this enzyme.
The N-SMase Activation Domain Does Not Contribute to Induction of A-SMase, Cytotoxicity, or NF-B-The death domain, located in close vicinity of the NSD between amino acids 326 and 413, is responsible for induction of cell death by TR55 (9, 25). The same region is also required for activation of A-SMase and NF-B (11). When the generated 70Z/3 transfec-tants (all devoid of an intact death domain) were tested for their ability to signal for any of the above-mentioned responses through the NSD, neither transfectant was able to mediate TNF functions (Figs. [3][4][5]. These data indicate that the NSD by itself does not mediate TNF-induced activation of A-SMase, NF-B, or cytotoxic response. DISCUSSION N-and A-SMase are activated independently and by different cytoplasmic domains of TR55. Each SMase couples to a select signaling pathway with a distinct set of downstream enzymes (11). A similar functional dichotomy of SMase signaling has recently been described for the Fas/APO-1 receptor, another member of the TNF/NGF receptor family (35). For Fas/APO-1 mutants, it has been shown that the inability to activate PC-PLC and A-SMase correlates to ablation of the cytotoxic signal (35), suggesting that the A-SMase pathway is initiated by the death domain. The recent discovery of proteins binding directly or indirectly to the death domain (26,27,36,37) promises a rapid advance in elucidation of the A-SMase/ apoptosis pathway.
In contrast, no accurate knowledge is currently available of how TR55 initiates the N-SMase pathway. This pathway includes important cellular responses such as activation of PLA 2 followed by production of proinflammatory metabolites. Equally important, the N-SMase pathway is potentially linked to the network of receptor tyrosine kinase signaling. It has recently been shown that TNF-dependent phosphorylation and activation of the c-raf-1 kinase is still functional in 70Z/ 3TR55⌬345 transfectants but not in 70Z/3TR55⌬205 cells that lack the NSD, implicating N-SMase in c-raf-1 kinase activation (23). The link of the N-SMase pathway to the Ras/Raf pathway is most likely accomplished by CAP kinase, as TNF-dependent phosphorylation of c-raf-1 kinase by CAP kinase as well as complex formation of c-raf-1 kinase and CAP kinase have been reported followed by increased activity of c-raf-1 kinase toward MAP kinase/extracellular signal-regulated kinase kinase (MEK) (24). In line with the lack of information on the N-SMase activation mechanism, no receptor-associated protein has been isolated so far that could be attributed to the N-SMase pathway.
In an effort to characterize the region(s) of TR55 that initiates the N-SMase pathway, we have generated, expressed, and analyzed a set of deletion mutants that all affect receptor sequences N-terminal of the death domain. The initial possibility that TR55 might contribute to N-SMase activation through its extracellular domain was ruled out by the observation that a receptor that completely lacks the cytoplasmic domain is unable to signal through N-SMase. This clearly indicates that the cytoplasmic portion of TR55 carries information critical for induction of the N-SMase pathway. By analysis of further deletion mutants of TR55, a region of 11 amino acids (309 -319) could be defined that apparently is both required and sufficient for the N-SMase pathway.
The delineation of the N-SMase activation domain to positions 309 -319 excludes a contribution by the adjacent death domain, whose N terminus is located between position 326 and 340 (25). This fits nicely into a model where different domains of TR55 initiate independent signaling pathways, most likely by binding different sets of associated proteins. The functional independence of the N-SMase pathway is further underscored by the fact that mutants with an inactive death domain were also unable to signal through A-SMase and NF-B regardless whether a functional NSD was present or not. In addition, no effect of the NSD on cytotoxicity was seen. This provides evidence that N-SMase and supposedly also enzymes secondary to N-SMase like MAP kinases and PLA 2 (reviewed in Ref. 20) by themselves are not sufficient for induction of cell death in 70/Z3 cells. At this point, however, we cannot rule out a possible contribution of the N-SMase pathway to the induction of apoptosis. In Jurkat cells, Fas/APO-1 ligation has been shown to induce Ras via the sphingomyelin (N-SMase) pathway, leading to subsequent apoptosis (38), which would imply a participation of N-SMase in the induction of programmed cell death. However, the activation of Ras via the action of ceramide by itself appeared not to be sufficient for Fas/APO-1-mediated apoptosis (38), which is in line with our findings that the NSD on its own is unable to signal cell death. The possibility remains that the NSD may be required for an apoptotic response in concert with the death domain.
Although both TR55 and Fas/APO-1 are capable of inducing the N-SMase pathway, there are no obvious homologies between the NSD and Fas/APO-1 sequence, indicating that both receptors have their own specific activation domains for N-SMase.
In summary, the results of our study indicate that the N-SMase pathway is activated by a domain both functionally and spatially distinct from the death domain. It is thus plausible to assume that a distinct, yet unknown factor may bind to the NSD initiating this signaling cascade. So far, two proteins have been described that bind to TR55 outside the death domain; however, one of them has a binding site that does not match the NSD (protein 55.11, binding site between residues 243 and 308; Ref. 39) while the amino acid sequences that mediate binding of the other protein (TRAP-1; Ref. 40) are diffusely distributed outside the death domain and therefore do not seem to be specific for the NSD. The identification of proteins that bind specifically to the newly defined NSD is a current subject of investigation.