Tumor necrosis factor receptors (Tnfr) in mouse fibroblasts deficient in Tnfr1 or Tnfr2 are signaling competent and activate the mitogen-activated protein kinase pathway with differential kinetics.

To dissect tumor necrosis factor receptor (Tnfr)-1 (CD120a) and Tnfr2 (CD120b)-dependent signal transduction pathways, primary fibroblasts isolated from inguinal adipose tissue of wild type (wt), tnfr1o, tnfr2o, and tnfr1o/tnfr2o mice were studied. The mitogen-activated protein kinases Erk1 and Erk2 were found to be tyrosine-phosphorylated and activated by Tnf treatment in all wt, tnfr1o, and tnfr2o fibroblasts; the activation was down-regulated 60 min after the start of steady state Tnf treatment. Distinct kinetics of Erk1 and Erk2 activation were detected; the Tnfr1-mediated activation of Erk1 and Erk2 started more slowly and persisted for more prolonged times as compared with Tnfr2 activation. Raf-1, Raf-B, Mek-1, Mek kinase, and p90rsk kinases were also shown to be activated independently in a distinct time-dependent pattern through the two Tnf receptors. In addition, both Tnfr1 and Tnfr2 mediated independently the activation of the transcription factor Ap-1 albeit with parallel activation kinetics. In contrast, Tnfr1 exclusively mediated activation of NF-κB and fibroblast proliferation; however, Tnfr2 enhanced proliferation triggered through Tnfr1. These findings indicate distinct but also overlapping roles of Tnfr1 and Tnfr2 in primary mouse fibroblasts and suggest different regulation mechanisms of signal transduction pathways under the control of both Tnf receptors.

Tumor necrosis factor ␣ (Tnf␣) 1 is one of the most pleiotropic cytokines, which exerts cytotoxic as well as differentiation and growth modulatory activities on many different target cells (for review see Refs. [1][2][3]. Tnf␣ activities are elicited by binding to at least two distinct surface receptors of 55 kDa (Tnfr1, CD120a) and 75 kDa (Tnfr2, CD120b), which are ubiquitously coexpressed on almost all cell types in various proportions (reviewed in Refs. 1,4,5). Both Tnf receptors have similar extracellular domains, but their intracellular parts are entirely unrelated suggesting distinct functions by addressing different signal pathways (5)(6)(7)(8). Tnfr1 and Tnfr2 bind Tnf␣ and Tnf␤ with similar high equilibrium affinity, but they differ significantly in their Tnf␣ binding kinetics; Tnf␣ bound to Tnfr2 is exchanged at a significantly faster rate than when complexed with Tnfr1 (9 -12).
Most of the cellular responses to Tnf␣ such as cytotoxicity, cell growth, activation of NF-B, and up-regulation of adhesion molecules are triggered by Tnfr1 engagement (for review see Refs. [13][14][15][16], but a small subset of Tnf activities, mainly proliferation of lymphoid cells, is mediated by direct Tnfr2 signaling; in some cell lines Tnfr2-mediated activation of NF-B and cytotoxicity has also been reported (17)(18)(19)(20)(21)(22)(23). The predominant role of Tnfr2 has been proposed to be an accessory function in enhancing or synergizing Tnfr1 signaling, e.g. by ligand passing (5,10,24). Studies of Tnfr1-and Tnfr2-deficient mice revealed a decisive role in vivo of Tnfr1 in the host defense against intracellular pathogens, whereas Tnfr2 played a role in Tnf-induced necrosis (16,25,26).
More recently, new insight in Tnf postreceptor mechanisms has been gained with the identification of a number of presumable signal transducing molecules that bind to the intracellular domains of Tnfr1 and Tnfr2 (27)(28)(29)(30)(31)(32)(33)(34)(35). The role in the activation of the transcription factor NF-B of one of these molecules, Traf2, which associates with the Tnfr2 intracellular domain has been demonstrated in transfection studies and using a B element-driven reporter construct (29,35).
In fibroblasts, Tnf␣ has been reported to activate several kinases such as c-Jun kinase, a member of the Mapk family, the phosphorylation of cytosolic proteins, and the transcription factors Ap-1 and NF-B (36 -44). Furthermore, it induces c-jun and c-fos gene expression (45). Mitogen-activated protein kinases (Mapk) play a central role in the early signal transduction events after receptor engagement of a variety of growth factors, cytokines and hormones; Tnf␣ has been found to activate c-Raf-1 and the Map kinases Erk1 and Erk2 (46 -49).
In the present study, primary fibroblasts isolated from mice deficient in Tnfr1 (tnfr1 o ), in Tnfr2 (tnfr2 o ), and in both Tnfr1 and Tnfr2 (tnfr1 o /tnfr2 o ) were used to further investigate signal pathways under Tnfr1 and Tnfr2 control (16,25,26). Tnfr1 and Tnfr2 were found to activate distinct but also overlapping signal pathways.

MATERIALS AND METHODS
Mice-The generation of homozygous Tnfr1-deficient (tnfr1 o ) or Tnfr2-deficient (tnfr2 o ) mice by gene targeting has been described else-* 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.
Reagents-Rabbit polyclonal anti-Map kinase R2 (Erk1-CT; recognizing Erk1 and Erk2 independent of phosphorylation state), rabbit polyclonal anti-Map kinase R3 (erk1-NT; recognizing Erk1), and mouse monoclonal anti-Map kinase antibodies (erk2; recognizing Erk2) were purchased from Upstate Biotechnology (Lake Placid, NY). Rabbit polyclonal anti-Mek-1, anti-Mekk, anti-Raf-1, anti-Raf-B, and anti-p90 rsk antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit polyclonal antibodies to tyrosine-phosphorylated Erk1 and Erk2 (phospho-Mapk) were purchased from New England BioLabs (Beverly, MA). Horseradish peroxidase (HRP)-labeled secondary antirabbit antibodies, [␥-32 P]ATP (Ͼ5000 Ci/mmol), [␥- 33  Isolation of Mouse Fibroblasts and Cell Culture-The preparation of mouse fibroblasts was performed as described by Mackay et al. (24). Briefly, mice were sacrificed by CO 2 asphyxia, and inguinal adipose pads were isolated and rinsed with DMEM (Life Technologies, Inc.) containing 100 units/ml penicillin and 100 g/ml streptomycin. Adipose tissue was cut into small pieces, washed once at room temperature with DMEM, 100 units/ml penicillin, 100 g/ml streptomycin, and resuspended in DMEM supplemented with 1% bovine serum albumin and 0.3% collagenase type I (Sigma). The suspension was incubated for 1 h at 37°C under agitation and finally homogenized using a loose-fitting Dounce homogenizer. The homogenate was filtered through a 100-m nylon mesh and centrifuged for 10 min at 1500 rpm, 4°C. Fibroblasts, localized to the pellet, were washed twice and then cultured in DMEM/ F12 medium (Life Technologies, Inc.) containing 20% human serum (SRK Blutspendezentrum, Bern, Switzerland), 5% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM glutamine. Cellular homogeneity of the cultures was controlled by flow cytometry analysis using a rat anti-mouse reticular fibroblast antibody (ER-TR7; Bachem, Torrance, CA). The primary fibroblasts were used at passages three to seven for assays.
Cell Proliferation Assay-Mouse fibroblasts were incubated in 96well microtiter plates in 100 l of medium at a density of 1 ϫ 10 4 cells/well with increasing amounts of recombinant mouse Tnf␣ (mTnf␣) (0.1-100 ng/ml). After 48 h of culture at 37°C, cells were pulsed for 12 h with 1 Ci/well [methyl-3 H]thymidine and harvested using an LKB cell harvester. Incorporated radioactivity was measured in a Betaplate liquid scintillation counter (Pharmacia LKB, Uppsala, Sweden). Each assay was performed in triplicate.
Electrophoretic mobility shift assays (EMSA) were performed by incubating nuclear extracts containing 10 g of total nuclear protein with 2 g of poly(dI-dC) (Pharmacia, Uppsala, Sweden) and 30 g of bovine serum albumin in binding buffer (10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM DTT, 1 mM EDTA, 5% glycerol) in a total volume of 15 l for 20 min at room temperature. As NF-B probe, a double-stranded oligonucleotide containing two tandemly arranged B elements of the human immunodeficiency virus long terminal repeat (5Ј-ATCAGG-GACTTTCCGCTGGGGACTTTCCG-3Ј) was used. A double-stranded oligomer containing a single consensus binding site (5Ј-AGCTTGAT-GAGTCAGCCG-3Ј) served as a specific Ap-1 probe. About 6 ϫ 10 4 cpm of the 32 P-end-labeled oligonucleotides were added to the preincubated nuclear extracts, and the reaction mixtures were further incubated for 15 min at room temperature. An unrelated oligonucleotide (5Ј-AGGAT-GGGAAGTGTGTGATATATCCTTGAT-3Ј) and an oligonucleotide with a single nucleotide deletion (5Ј-AGCTTGATGGTCAGCCG-3Ј) were included as controls in the NF-B and Ap-1 assays, respectively. The samples were then analyzed by electrophoresis through a 4% polyacrylamide gel in 1 ϫ TAE buffer (40 mM Tris acetate, 1 mM EDTA, pH 8.0), and the gels were dried and exposed to x-ray films under standard conditions. Autoradiograms were analyzed with a laser densitometer (Molecular Dynamics, Sunnyvale, CA). Cell Lysates-Mouse primary fibroblasts were seeded in 100 ϫ 20-mm polystyrene tissue culture dishes (Falcon) and were allowed to grow to 80% confluency (about 1 ϫ 10 6 cells total). Cells were then washed twice with PBS and incubated for 3 days with medium containing 0.25% fetal calf serum (DMEM/F12, 0.25% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM glutamine) to render cells quiescent before being treated as indicated. After stimulation, the cells were washed twice with ice-cold PBS and lysed in 300 l of lysis buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 2 mM EDTA, 10% glycerol, 1.0% Nonidet P-40, 14 M pepstatin A, 100 M leupeptin, 3 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium pyrophosphate, 10 mM sodium orthovanadate, 100 units/ml aprotinin, and 100 mM sodium fluoride) for 30 min at 4°C. The cell lysates were centrifuged (10 min, 70000 rpm, 4°C) in a Beckman Microfuge, and supernatants were recovered. The protein concentration of lysates was determined using a BCA protein assay (Pierce) with bovine serum albumin as standard.
Immunoblotting-Equal amounts of cell lysate protein (typically 25 g total protein) were separated by SDS-PAGE in Mini-Protean Electrophoresis cells (Bio-Rad) using the Laemmli buffer system (51). Proteins were transferred electrophoretically to PVDF membranes (Immobilon-P membranes, Millipore, Bedford, MA). After blocking of free binding sites with 5% defatted milk powder in blocking buffer (TBS, 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 0.1% Tween 20), the membranes were incubated for 1 h at room temperature with appropriate concentrations of first antibodies in TBS followed by a 1-2-h incubation with HRP-labeled secondary anti-rabbit antibodies at room temperature. The membranes were then washed and subjected to enhanced chemiluminescence detection (Amersham Corp.) under standard conditions. Detection of Kinase Activity Using Mbp-containing SDS-Polyacrylamide Gels-The Mbp-SDS-PAGE kinase assays were performed as described by Vietor et al. (49) with minor modifications. Cell lysates containing aliquots of 20 g of total protein were electrophoretically separated in 10.5% SDS-polyacrylamide gels containing 0.1 mg/ml Mbp (Sigma) co-polymerized in the running gel. Gels were washed for 1 h in two changes of buffer A (50 mM HEPES, pH 7.4, 5 mM 2-mercaptoethanol) supplemented with 20% isopropyl alcohol, followed by an incubation in buffer A for 1 h. The proteins in the gels were denatured for 1 h with two changes of buffer A containing 6 M guanidine HCl and finally renatured for 12-16 h at 4°C in buffer A containing 0.04% Tween 40. To perform kinase assays, the gels were preincubated in kinase reaction buffer (buffer B, 25 mM HEPES, pH 7.4, 10 mM MgCl 2 , 90 M sodium orthovanadate, 5 mM 2-mercaptoethanol) for 30 min at 30°C and incubated for 1 h at 30°C in buffer B containing 100 Ci of [␥-32 P]ATP. The gels were washed extensively with a buffer containing 5% trichloroacetic acid and 1% sodium pyrophosphate, dried, and analyzed by a Phos-phorImager (Molecular Dynamics, Sunnyvale, CA).
Immunoprecipitations and Kinase Assays-Equal amounts of cell lysates used for immunoprecipitation were precleared with protein A-Sepharose beads (3 ϫ 20 min on ice). Aliquots of 30 g of total protein of the precleared extracts were incubated with 1 g of the respective antibodies in a final volume of 100 l for 20 min at 4°C under agitation.

Predominant Role of Tnfr1 in Tnf-stimulated Mouse Fibro-
blast Proliferation-To dissect the activity of the two Tnf receptors in proliferation, primary mouse fibroblasts from wild type (wt) and homozygous tnfr1 o or tnfr2 o mice were stimulated for 48 h with various amounts of mTnf␣, and thymidine incorporation was measured. Confirming previous studies (24,52), proliferation of wt fibroblasts was stimulated by mTnf␣, reaching a maximum at about 5-10 ng/ml mTnf␣ (Fig. 1). tnfr2 o fibroblasts similarly showed enhanced proliferation with mTnf␣ stimulation, but the maximal proliferative response achieved was reproducibly lower than with wt fibroblasts. In contrast, mTnf␣ had no effect on tnfr1 o fibroblast proliferation (Fig. 1). These findings confirm that Tnfr1 is essential in mediating proliferative signals, whereas Tnfr2 engagement merely enhances proliferation triggered through Tnfr1 (24).
Tnfr1 Exclusively Activates Transcription Factor NF-B in Mouse Fibroblasts-Several previous studies had attributed NF-B activation by Tnf to Tnfr1 signaling, but in some cell lines NF-B activation was also reported to be under Tnfr2 control (16,20,21,28,(53)(54)(55). To investigate the role of Tnfr1 and Tnfr2 in NF-B activation in the present cell systems, primary wt and Tnfr-deficient fibroblasts were stimulated for 15 and 30 min with predetermined optimal doses of mTnf␣, and equal amounts of nuclear protein extracts were tested for activated and translocated NF-B in electrophoretic mobility shift assays (EMSA). NF-B was found strongly activated by mTnf␣ treatment in wt fibroblasts ( Fig. 2A). In contrast, mTnf␣ completely failed to activate NF-B in tnfr1 o fibroblasts (Fig. 2B), confirming similar results with other cells from tnfr1 o mice (16). However, NF-B could be activated in tnfr1 o fibroblasts by hIL-1␣ treatment, demonstrating that NF-B activation in general was not affected. Tnf␣ stimulation of tnfr2 o fibroblasts resulted in NF-B activation (Fig. 2C); for a more quantitative evaluation, the integrated relative peak intensities of the Bspecific bands were determined by densitometry of the autoradiograms of three independent series of experiments, each using all four cell types in parallel. Despite non-negligible variation among experiments, the data support the conclusion that NF-B activation in wt fibroblasts increases from the 15-to the 30-min time point, whereas it has decreased after 30 min, when compared with 15 min, of Tnf␣ stimulation of tnfr2 o fibroblasts, suggesting that Tnfr2 signals are required to achieve a more sustained activation of NF-B. In tnfr1 o /tnfr2 o fibroblasts, no B binding activity was detected in the nuclear extracts stimulated with mTnf␣, whereas hIL-1␣ treatment still elicited NF-B activation (Fig. 2D).
Both Tnf Receptors Signal in Ap-1 Activation-Tnf has previously been reported to be a potent activator of transcription factor Ap-1 through prolonged activation of the c-Jun kinase (44,45). The role of the two Tnf receptors in Ap-1 activation in wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblasts was therefore investigated by EMSA. Interestingly, Ap-1 was found to be activated by mTnf␣ in all wt, tnfr1 o , and tnfr2 o fibroblasts, showing that Ap-1 was independently activated through both Tnf receptors (Fig. 3, A-C). Increased Ap-1 probe binding activity in nuclear extracts from wt, tnfr1 o , and tnfr2 o fibroblasts was already detected after 30 min and was still apparent 3 h after mTnf␣ stimulation (Fig. 3, A-C). The intensity of the shifted bands appeared to be comparable in the nuclear extracts of all these cell types, since the integrated relative intensities of the Ap-1 bands determined by densitometry of autoradiograms of three independent series of parallel experiments with all four cell types were not significantly different. As expected, treatment of tnfr1 o /tnfr2 o fibroblasts with hIL-1␣, but not with mTnf␣, resulted in Ap-1 activation (Fig. 3D).
Increased Phosphorylation of Erk1 and Erk2 after Tnf Treatment-To investigate the control of the Map kinases Erk1 and Erk2 through Tnfr1 and Tnfr2 in the primary mouse fibroblasts, Western blot analyses were performed. Equal amounts of total cell lysate protein of wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblasts treated for various times with mTnf␣ and hIL-1␣ were separated by SDS-PAGE, transferred to PVDF membranes, and stained with erk1-CT antibody. Two distinct bands in the range of 40 -45 kDa, assigned to Erk1 and Erk2, were visualized with unstimulated, mTnf␣, and hIL-1␣-stimulated cells of all four mouse types (Fig. 4A). The amount of Erk1 and of Erk2 protein staining was approximately constant in all fibroblast types and did not significantly change with the mTnf␣ or hIL-1␣ treatment (Fig. 4A). Furthermore, to test for Erk1 and Erk2 phosphorylation in the TEY activation site (56,57), Western blots were performed using an antibody exclusively recognizing tyrosine-phosphorylated forms of Erk1 and Erk2. As shown in Fig. 4B, no phosphorylated Erk1 and Erk2 was detected in any of the unstimulated fibroblasts, but increased tyrosine phosphorylation of Erk2 and, to a lesser extent, of Erk1 was apparent already after 5 min of mTnf␣ treatment in wt, tnfr2 o , and interestingly also in tnfr1 o fibroblasts, whereas tnfr1 o /tnfr2 o fibroblasts did not respond. In contrast, phosphorylation of Erk1 and Erk2 was induced by hIL-1␣ in all wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblasts, demonstrating that the Mapk cascade was operating in all these cells. Erk1 and Erk2 tyrosine phosphorylation was transient, and almost no phosphorylated forms of these proteins were found 60 min after the start of mTnf␣ treatment despite the continued presence of the ligand.
Clear distinctions in the kinetics of Erk1 and Erk2 phosphorylation between wt, tnfr1 o , and tnfr2 o fibroblasts were detected (Fig. 4B). In wt fibroblasts, both Erk1 and Erk2 showed significant phosphorylation already 5 and 15 min after mTnf␣ stimulation; only a minor fraction of phosphorylated Erk2, and no phosphorylated Erk1, was seen after 60 min. In contrast, phosphorylation of Erk1 and Erk2 was more short-lived in tnfr1 o fibroblasts when compared with wt fibroblasts, since phosphorylated forms of Erk1 or Erk2 were evident 5 min, but no longer detected 15 min, after mTnf␣ stimulation (Fig. 4B). In tnfr2 o fibroblasts, a relatively weak Erk2 phosphorylation was apparent 5 min after mTnf␣ stimulation, but it became stronger, and Erk1 phosphorylation became evident 15 min after mTnf␣ stimulation. Similar to wt fibroblasts, a weak Erk2, but no Erk1, phosphorylation persisted 60 min after mTnf␣ stimulation in tnfr2 o fibroblasts. hIL-1␣, but not mTnf␣, elicited phosphorylation of Erk1 and Erk2 in tnfr1 o /tnfr2 o fibroblasts (Fig. 4B).
Tnfr1 and Tnfr2 Control Erk1 and Erk2 Kinase Activity-To corroborate the above findings, the Tnf-stimulated kinase activity of Erk1 and Erk2 in lysates of wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblasts was analyzed in Mbp-SDS-PAGE ki-nase assays (Fig. 5), and by in-solution kinase assays using specific Erk1 and Erk2 immunoprecipitates (Fig. 6). First, Mbp-SDS-PAGE kinase assays of lysates of mTnf␣ or hIL-1␣treated fibroblasts revealed strongly inducible kinase activities of two proteins in the typical locations of Erk1 and Erk2 of about 40 -45 kDa (Fig. 5). The mTnf␣ concentration chosen in these studies was 10 ng/ml; preliminary studies had shown that the Mapk responses saturated at Tnf concentrations of 0.5 ng/ml. Furthermore, no significant decrease in the active Tnf concentration in the culture medium could be measured at the end of the experiment. The two kinases, tentatively assigned to Erk1 and Erk2, in wt fibroblasts showed similar activity 5 and 15 min after Tnf stimulation and some residual activity after 60 min. Two further kinases of about 90 and 140 kDa were also found strongly activated by mTnf␣ treatment, following a similar activation time dependence. Both presumptive Erk1 and Erk2 were triggered through Tnfr1 and Tnfr2 as demonstrated with the tnfr1 o and tnfr2 o fibroblasts (Fig. 5). In tnfr1 o fibroblasts, the activity of these kinases was strongly stimulated after 5 min and had decayed to a large extent after 15 min of Tnf treatment. In contrast, in tnfr2 o fibroblasts the time dependence of kinase activation was reversed, the signal intensity increasing from a relatively low level at 5 min to a maximum at 15 min. The significance of the differential Tnfr1-and Tnfr2-mediated time dependence of kinase activity was further supported by the 90-and 140-kDa kinases following similar activation kinetics in the tnfr1 o and tnfr2 o fibroblasts. As expected, tnfr1 o /tnfr2 o fibroblasts showed kinase activation after hIL-1␣ but not after mTnf␣ stimulation.
In the second approach, Erk1/Erk2 in-solution kinase assays were performed with erk1-CT immunoprecipitates from lysates of unstimulated and mTnf␣ or hIL-1␣-treated fibroblasts, using Mbp as substrate. Erk1/Erk2 activity was found stimulated by mTnf␣ and hIL-1␣ treatment in wt, tnfr1 o , and tnfr2 o fibroblasts, whereas in tnfr1 o /tnfr2 o fibroblasts only IL-1␣ treatment enhanced Erk1/Erk2 activity (Fig. 6). Consistent with the findings in Fig. 5, the precipitation kinase assays revealed delayed and more prolonged, as compared with early and more short-lived, Erk1/Erk2 activation with exclusive Tnfr1 and Tnfr2 signaling, respectively. In wt and tnfr1 o fibroblasts, strong activation of Erk1/Erk2 was reached 5 and 15 min after Tnf stimulation; this activity decayed to a low level after 60 min, especially in tnfr1 o fibroblasts. In contrast, with tnfr2 o fibroblasts Erk1/Erk2 activity peaked only 15 min after Tnf stimulation, and significant activity was still obvious after 60 min. A more quantitative analysis, using PhosphorImager counts, confirmed the longer persistence of the Erk1/Erk2 activities in tnfr2 o fibroblasts (Table I).
Finally, Erk1 and Erk2 were assayed individually in insolution kinase assays with immunoprecipitates using Erk1and Erk2-selective antibodies (Table I). These assays confirmed the differential time dependence of Erk1 and Erk2 activation shown in Fig. 5 and Fig. 6.
Tnf-dependent Activation of p90 rsk Kinase Activity-A further confirmation for a distinct time course of intracellular Tnf signal progression mediated by Tnfr1 and Tnfr2 was presented by p90 rsk activation. Several reports had provided evidence that members of the 90-kDa S6 kinase family (collectively termed rsk) can be phosphorylated and thereby stimulated by Map kinases (56, 58 -60). Since the control of p90 rsk activity by Tnf has not previously been reported, p90 rsk activation was studied by in-solution kinase assays with anti-p90 rsk immunoprecipitates from wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblast cell lysates using Mbp as substrate. Tnf was found to stimulate p90 rsk activity by triggering either Tnfr1 or Tnfr2, but the kinetics of the activation were distinct (Fig. 7). A more quantitative PhosphorImager analysis with wt fibroblasts showed an increase of about 3-3.5-fold of p90 rsk activity 5 and 15 min after mTnf␣ stimulation (Table II); thereafter, p90 rsk activity declined and reached background levels after 60 min. In tnfr1 o fibroblasts, a similar activation time dependence was found (Fig. 7, Table II). In contrast, with tnfr2 o fibroblasts, p90 rsk activation was 1.6-fold 5 min after Tnf stimulation, followed by a 6-fold increase after 15 min and persisting with a 2-fold higher activity after 60 min, when compared with unstimulated fibroblasts. tnfr1 o /tnfr2 o fibroblasts showed p90 rsk activation by hIL-1␣ but not by mTnf␣ stimulation.
Upstream Kinases of Mapk Signal Cascade-Raf-1 kinase and Mapk kinase (Mek) function as upstream transducing elements of the Mapk (Erk) pathway (61)(62)(63). To explore pathways leading to Erk activation, Raf-1, Raf-B, Mekk, and Mek-1 activities were investigated in immunoprecipitation kinase assays with mTnf␣-stimulated wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o fibroblasts. PhosphorImager analyses of the labeled 33 P-Mbp substrate separated by SDS-PAGE are presented in Table II. It was found that Raf-1, Mek-1, and Mekk responded with enhanced kinase activity to Tnf treatment in all wt, tnfr1 o , and tnfr2 o fibroblasts, supporting again that both Tnf receptors individually are signaling-competent. The activation factors were about 1.5-4 for Raf-1, Mekk, and Mek-1 in wt, tnfr1 o , and tnfr2 o fibroblasts (Table II). Interestingly, Raf-B, an isoform of c-Raf-1 reported to be mainly localized to the brain (64 -66), also responded to Tnf treatment of the fibroblasts.
The time dependence of Raf-1, Raf-B, Mekk, and Mek-1 activation under Tnfr1 and Tnfr2 control in general was similar to that seen above with Erk1, Erk2, and p90 rsk . In tnfr1 o fibroblasts, the strongest activation of Raf-1, Raf-B, Mekk, and Mek-1 was found 5 min after Tnf stimulation, and the activities of these kinases decreased dramatically thereafter, reaching background levels after 60 min. In contrast, in tnfr2 o fibroblasts strong activation of Raf-1, Raf-B, Mekk, and Mek-1 was more persistent, a significant activation remaining even 60 min after Tnf stimulation (Table II). In tnfr1 o /tnfr2 o fibroblasts, hIL-1␣, but not mTnf␣, activated Raf-1, Raf-B, Mekk, and Mek-1.

DISCUSSION
The ubiquitous and simultaneous expression of two distinct Tnf receptors on almost all cells is intriguing (6,(67)(68)(69)(70). While their extracellular domains have sequence similarity, their intracellular domains are unrelated, suggesting distinct modes of activation of signal transduction pathways (6,71). Studies using Tnfr1-and Tnfr2-selective agonists, such as monoclonal antibodies and Tnf muteins, demonstrated that the majority of known Tnf activities is mediated by Tnfr1 and can be elicited by exclusive Tnfr1 engagement. In contrast, a direct signaling activity of Tnfr2 was found only in a subset of Tnf activities, including the enhancement of T cell proliferation and the activation of NF-B and cytotoxicity in cell lines and transfected cells (17)(18)(19)(20)(21)(22)(23). These findings prompted the proposition that Tnfr2 mainly has an accessory function to Tnfr1 signaling, either by enhancement and modulation of the Tnfr1 signals in intracellular transduction pathways or by ligand passing (5,10,24).
The view of distinct modes of Tnfr1 and Tnfr2 signaling has been further confirmed in studies on receptor-associated intracellular molecules. Traf1 and Traf2, two members of a newly defined family of this class of proteins, were initially found to interact with a specific intracellular region of Tnfr2 but not with Tnfr1 (28). In contrast, Tradd, a member of a second family of such presumptive signal transducers, was identified as a protein specifically associating with the intracellular domain of Tnfr1 (31). More recently, it was shown that Tradd-Traf2 and Tradd-Fas-associated death domain protein interactions define two distinct Tnfr1 signal transduction pathways leading to a bifurcation between Tnfr1-dependent NF-B activation and apoptosis, respectively (35). The discovery of additional proteins that may engage directly or in complexes cytoplasmic domains of Tnfr1 and Tnfr2, such as Trap-1, receptor interacting protein, LMP1-associated protein, cellular inhibitor Cell lysates (calibrated to 30 g of total protein) were immunoprecipitated with 1 g of erk1-CT antibody recognizing Erk1 and Erk2. Insolution kinase assays were performed with immunoprecipitates, using Mbp and [␥-33 P]ATP. Phosphorylated Mbp substrate was visualized by SDS-PAGE and PhosphorImager analyses under standard conditions. Dried gels were exposed to PhosphorImager screens, and 33 P incorporation was quantified by volume integration (Imagequant software package, Molecular Dynamics). of apoptosis proteins 1 and 2, reveals the complexity of these presumably early steps after Tnf stimulation (8,34).
To further dissect Tnfr1 and Tnfr2 functions and signal transduction pathways at the level of transcription factors and the Mapk cascade, presumably downstream of the proteins engaging the cytoplasmic receptor domains such as Traf and Tradd and of the sphingomyelinase pathways, we have studied Tnf responses in primary fibroblast cultures from wt, tnfr1 o , tnfr2 o , and tnfr1 o /tnfr2 o mice. Tnfr1 was found to exclusively mediate early and transient NF-B activation and fibroblast proliferation. The exclusive control of NF-B activation through Tnfr1 in this system is consistent with previous findings in human umbilical cord vein endothelial cells and HL60 cells, whereas both Tnf receptors were found to control NF-B activation in other cell lines and transfection studies, pointing to cell specificity in this response (14,29,72). In contrast to NF-B, both Tnfr1 and Tnfr2 independently and synchronously activated Ap-1 from 30 min to 3 h after Tnf stimulation. The Ap-1 activation demonstrated that both Tnfr1 and Tnfr2 independently are competent to accede major signal pathways. The independent signaling competence of Tnfr1 and Tnfr2 was further confirmed in the studies of various kinases of the Erk/ Mapk signal transduction cascade; Erk1, Erk2, Raf-1, Raf-B, Mek-1, Mekk, and p90 rsk all were found to be activated independently by Tnfr1 as well as Tnfr2. This clearly demonstrates that the activation of the Mapk pathway alone does not suffice to elicit those Tnf functions that are under exclusive Tnfr1 control. The activation of the Mapk cascade through both Tnfr1 and Tnfr2 was transient, the stimulated enzyme activities in general being again down-regulated on a time scale of less than 1 h, whereas other responses such as Ap-1 activation were more long lasting. It is noted that the down-regulation of these kinase activities occurred in the continued presence of active Tnf concentrations, since no significant decrease of Tnf activity within the first few hours could be detected in the culture media in control studies. Furthermore, the Tnf concentration chosen was found in preliminary control studies to be saturating with regard to the Mapk response.
Interestingly, the time dependence of kinase activation after Tnfr1 and Tnfr2 stimulation was distinct. The kinase activations, when triggered through Tnfr1, were shifted to a later onset and persisted for more prolonged times when compared with Tnfr2 stimulation. To interpret these findings the different binding kinetics of Tnf␣ to Tnfr1 and Tnfr2 must first be considered. The half-life times of Tnf binding from the measurement of exchange rates at the cell surface have been determined as t1 ⁄2 Ͼ 3 h and t1 ⁄2 ϭ 10 min for Tnfr1 and Tnfr2, respectively (10). These values agree well with exchange rates measured with the respective human dimeric recombinant TNFR1-and TNFR2-IgG heavy chain fusion constructs (i.e. t1 ⁄2 ϭ 7 h and t1 ⁄2 ϭ 5-10 min, respectively) which due to their dimeric structure with two receptor moieties binding one Tnf trimer may be thought to approximate the Tnf-Tnf receptor interaction at the cell surface (9 -12). These data are also consistent with the association kinetics of TNF to the two receptors measured with U937 cells where rate constants for the observed initial association of 0.002 and 0.037 min Ϫ1 at 0°C were measured for TNFR1 and TNFR2, respectively, the steady state of TNFR2 association not being reached before 2 h (10). There is an obvious parallel between the kinetics of Tnf binding to the Tnf receptors and the onset of the activation of the Mapk pathway, and it may be argued that the more rapid onset of the Mapk pathway activation under Tnfr2 control reflects the faster on-rate of Tnf and Tnfr2 binding although there may be further rate-determining steps in the pathway that couples the receptor-proximal events and the Mapk pathway. However, it appears that the persistence of the activated state of the Mapk pathway is independent of the time of receptor occupancy. This is obviously true for Tnfr1 where the time scale of Tnf-Tnfr1 exchange far exceeds that of the Mapk pathway down-regulation. Furthermore, assuming that the intracellular signal transduction machine experiences the integrated input of all activated surface receptors of one specificity, even the short exchange rate of Tnf-Tnfr2 binding cannot suffice to explain the observed down-regulation of the Tnfr2-mediated Mapk pathway activation, because the cells were kept under steady state saturating Tnf concentrations. It may be proposed that the down-regulation and the distinct time dependence of the Mapk pathway activation result from mechanisms such as Tnfr1-and Tnfr2-specific counter-regulatory intracellular signal pathways with negative feedback effect or by the activation of distinct parallel, yet unknown pathways regulating the kinase activities. Among these, kinase-specific phosphatases such as 3CH134/CL100 specifically inactivating Mapk or phosphatase 2A might have an important role in FIG. 7. Time-dependent activation of p90 rsk in mTnf␣-and hIL-1␣-treated primary mouse fibroblasts. Wt, tnfr1 o , tnfr2 o , and tnfr1 o / tnfr2 o fibroblasts in parallel cultures were treated with mTnf␣ (10 ng/ml) for various times or with rhIL-1␣ (10 ng/ml, 15 min). Cell lysates (calibrated to 30 g total protein) were immunoprecipitated with 1 g of p90 rsk -specific polyclonal antibody, and in-solution kinase assays were performed with immunoprecipitates using Mbp and [␥-33 P]ATP. Phosphorylated Mbp substrate was visualized by SDS-PAGE and Phospho-rImager analyses under standard conditions. Dried gels were exposed to PhosphorImager screens, and 33 P incorporation was quantified by volume integration (Imagequant software package, Molecular Dynamics).  5  1 5  6 0  -5  1 5  6 0  -5  1 5  6 0  -5  1 5  6 0  ----1 5  ---1 5  ---1 5  ---1 5 Erk1/Erk2 6.1 Ϯ 0.8 7.5 2.2 8. regulating signal progression (73)(74)(75). The cells used in the present studies have been rendered quiescent by serum deprivation in order to increase the signal-to-noise ratio of activated kinase responses. While it cannot be ruled out entirely, it is unlikely that quiescence generates a kinase reactivity pattern unrelated to the normal cell condition. It may be envisioned rather that in a cell under standard conditions the signals leading to Mapk activation as documented in the present studies are superimposed as transients on a background of ongoing normal cellular activity in the Mapk pathway.  5  1 5  6 0  -5  1 5  6 0  -5  1 5  6 0  -5  1 5  6 0  ----1 5  ---1 5  ---1 5  ---1 5 p90 rsk 2.9 Ϯ 0.