|Home | About | Journals | Submit | Contact Us | Français|
Tumor necrosis factor (TNF) elicits its biological activities by stimulation of two receptors, TNFR1 and TNFR2, both belonging to the TNF receptor superfamily. Whereas TNFR1-mediated signal transduction has been intensively studied and is understood in detail, especially with respect to activation of the classical NFκB pathway, cell death induction, and MAP kinase signaling, TNFR2-associated signal transduction is poorly defined. Here, we demonstrate in various tumor cell lines and primary T-cells that TNFR2, but not TNFR1, induces activation of the alternative NFκB pathway. In accord with earlier findings demonstrating that only membrane TNF, but not soluble TNF, properly activates TNFR2, we further show by use of TNFR1- and TNFR2-specific mutants of soluble TNF and membrane TNF that soluble ligand trimers fail to activate the alternative NFκB pathway. In accord with the known inhibitory role of TRAF2 in the alternative NFκB pathway, TNFR2-, but not TNFR1-specific TNF induced depletion of cytosolic TRAF2. Thus, we identified activation of the alternative NFκB pathway as a TNF signaling effect that can be specifically assigned to TNFR2 and membrane TNF.
Tumor necrosis factor (TNF)2 is a highly pleiotropic cytokine and the prototypic member of the phylogenetically conserved TNF ligand family (1, 2). TNF, as well as other members of this cytokine family, is a type II transmembrane protein, which self-assembles into noncovalently bound trimers (1, 2). TNF occurs also as a soluble trimeric protein, which is derived from the transmembrane form by limited proteolysis (1, 2). TNF is mainly released from activated macrophages and T-cells, but it can also be produced by a variety of other cell types, especially after contact with bacterial products. TNF interacts with two receptors, TNFR1 and TNFR2, which both belong to the TNF receptor superfamily. Whereas TNFR1 is constitutively expressed in most cell types, TNFR2 is typically found on immune and endothelial cells (3). Remarkably, soluble and transmembrane TNF differ in their capability to stimulate signaling via TNFR1 and TNFR2. Whereas transmembrane TNF (memTNF) triggers signaling potently via both TNF receptors, soluble TNF trimers (sTNF) only activate TNFR1 robustly and show none or only limited activity on TNFR2 (4).
TNFR1 contains a death domain (DD) and utilizes this protein-protein interaction domain to recruit intracellular signaling proteins involved in the activation of proinflammatory pathways, but also in cell death induction. For example, activation of the classical NFκB pathway and the various MAP kinases by TNFR1 rely on recruitment of the DD-containing serine-threonine kinase RIP, the DD-containing adaptor protein TRADD, and a complex of the TRADD-interacting TRAF2 protein with cIAP1 and cIAP2 (3, 5, 6). Notably, RIP has also been implicated in TNFR1-induced necrosis, and TRADD, together with FADD and caspase-8, is crucially involved in TNFR1-mediated apoptosis (3). In general, the signaling mechanisms utilized by TNFR1 are biochemically well understood and the importance of the various signaling intermediates of TNFR1 have been verified in vivo in corresponding knock-out mice. In contrast to the detailed picture we have of TNFR1 signaling, the signaling mechanisms of TNFR2 are poorly defined. In fact, the mechanistically best investigated aspect of TNFR2 signaling is the capability of this receptor to modulate TNFR1 signaling. So, TNFR2 can specifically enhance TNFR1-mediated apoptosis by depletion of the NFκB-promoting/caspase-8 inhibitory TRAF2-cIAP1/2 complex from TNFR1 via competitive recruitment to TNFR2 and subsequent proteasomal degradation (7,–15). Noteworthy, a number of studies demonstrated apoptosis induction after selective stimulation of TNFR2. In most of these reports, apoptosis was not directly triggered by TNFR2, but was mediated indirectly by up-regulation of transmembrane TNF, which then secondarily stimulated TNFR1 (16, 17). There is further evidence from the rat-mouse cytotoxic T-cell hybridoma PC60 that TNFR2 can also induce cell death independently from TNFR1, but the underlying signaling mechanisms are unknown (18). It is clearly evident from analyses of TNFR1 knock-out mice that TNFR2 activation alone is sufficient to stimulate most of the signaling pathways activated by TNFR1, including those leading to activation of NFκB, ERK, JNK, p38, and Akt. Accordingly, TNFR2 elicits a variety of non-apoptotic cellular responses in TNFR1-deficient T-cells and TNFR1-deficient endothelial cells. For example, TNFR2 is necessary for antigen-driven differentiation and survival of T-cells (19, 20). TNFR2 further mediates up-regulation of ICAM-1, E-selectin, and MCP-1/JE in endothelial cells (21) and triggers the migration of intestinal epithelial cells and Langerhans cells (22, 23) as well as proliferation of myofibroblasts (24) and angiogenesis (25). There is also evidence for a neuroprotective role of TNFR2 (26,–28). With regard to TNFR2-induced migration and angiogenesis of endothelial cells a crucial role has been demonstrated for the tyrosine kinase BMX and the phosphatidylinositol 3-kinase/Akt pathway (29). Activation of the latter has also been implicated in the neuroprotective effects of TNFR2 (28).
The mammalian transcription factors of the NFκB family are homo- or heterodimers of p65/RelA, RelB, cRel, NFκB1/p50, and NFκB2/p52. The latter are released by limited proteasomal proteolysis from the precursor proteins p105 (p50) and p100 (p52) (30,–32). In nonstimulated cells, NFκB dimers are retained in the cytoplasm by binding of inhibitors of the IκB family. The structural hallmark of IκBs is a repeat containing six or seven ankyrin domains that mediate interaction and inhibition of the NFκB proteins. Notably, in addition to their N-terminal Rel homology domain (RHD), p105 and p100 also contain an IκB domain in their C terminus (30,–32). These precursor proteins can therefore act as IκBs. Activation of NFκBs, thus their translocation into the nucleus, occurs after IκB degradation or in the case of p105 and p100 after destruction of the IκB domain by limited processing. Degradation/processing of IκB proteins and p105 are triggered via the classical NFκB pathway (30,–32). In this pathway, treatment with appropriate inducers, such as TNF, IL-1, or LPS results in stimulation of the kinase activity of the IκB kinase (IKK) complex composed of the scaffolding protein NEMO and the serine/threonine kinases IKK1 and IKK2. Phosphorylation of IκB proteins by the IKK complex then marks these inhibitory proteins for ubiquitination and proteasomal degradation (30,–32). In contrast, the alternative or noncanonical NFκB pathway, which triggers p100 processing and activation of p52-containing NFκBs, relies on phosphorylation of p100 by IKK1, independent from NEMO and IKK2 (30,–32). Notably, TRAF2 is required by many stimuli to induce the classical NFκB pathway, whereas it elicits together with TRAF3 an inhibitory function in the noncanonical pathway (32). In nonstimulated cells, TRAF3 recruits the TRAF2-cIAP1/2 complex to NIK, a MAP3K crucially involved in the activation of the noncanonical NFκB pathway (32). In the resulting complex, the TRAF2-associated cIAPs ubiquitinate NIK and thereby trigger the proteasomal degradation of this protein. Receptors that activate the noncanonical NFκB pathway deflect the TRAF2-cIAP1/2 complex from the cytoplasm and thereby prevent degradation of newly synthesized NIK, which now stimulates IKK1 and p100 processing.
Here, we show that membrane TNF, but not soluble TNF trimers, induces p100 processing by activation of TNFR2 in several cell lines and primary immune cells. This identifies the noncanonical NFκB pathway as a cellular TNF response that can be specifically assigned to TNFR2 and membrane TNF.
NCTC, HEK293, HeLa, HeLa-TNFR2, EM-2, Kym1, and Jurkat-TNFR2 cells were maintained in RPMI1640 medium (PAA, Pasching, Germany), containing 10% heat-inactivated fetal calf serum. Jurkat-TNFR2, HeLa-TNFR2, and HeLa-TNFR2-TRAF1 transfectants have been described elsewhere (8, 33, 34). TNF mutations conferring specificity for TNFR1 and TNFR2, respectively, have been described by Lötscher et al. (35) for soluble TNF. The pan-caspase inhibitor z-VAD-fmk was purchased from Bachem (Weil am Rhein, Germany). The various Flag-tagged variants of soluble TNF used in this study were produced in HEK293 cells and purified by affinity chromatography on anti-Flag mAb M2-agarose. The TRAF2-specific antibody, TRAF1, anti-p65, anti-p50, anti-RelB, anti-IκBα, and anti-lamin B were purchased form Santa Cruz Biotechnology (Santa Cruz, CA). Anti-tubulin was from Dunn Labortechnik (Asbach, Germany), anti-p100 from Upstate (Temecula, CA), anti-β-actin from Sigma, and anti-cRel and anti-NIK and anti-pIκBα from Cell Signaling Technologies (Danvers, MA). TNFR2-Fc was purchased from Wyeth Pharma (Münster, Germany). TNFR1-Fc was a kind gift from Pascal Schneider (University of Lausanne, Switzerland). The IKK2 inhibitor TPCA-1 was from Tocris Bioscience, (Ellisville, MO).
NCTC cells were transiently transfected with expression plasmids encoding N-terminal YFP fusion proteins of non-cleavable membrane TNF variants (36) binding both TNF receptors (YFP-TNFΔ1–12) or only TNFR1 (YFP-TNFΔ1–12(32W/86T)) or TNFR2 (YFP-TNFΔ1–12(143N/145R)) with LipofectamineTM 2000 (Invitrogen, Karlsruhe, Germany) following the supplier's protocol. An expression plasmid encoding soluble TNF was used as a control. After 48 h, expression of membrane TNF was verified by FACS analysis using a TNF-specific PE-conjugated antibody (eBioscience, San Diego, CA) recognizing the various TNF variants irrespective of their TNF receptor specificity and a corresponding isotype control (mouse IgG1, eBioscience, San Diego, CA). Transfected cells used for experiments were regularly >60% positive for expression of the various membrane TNF variant.
Cells (2 × 104) were seeded in triplicate in 96-well tissue plates and were then either stimulated with TNF-expressing NCTC transfectants (2 × 104 per well) or with the indicated recombinant TNF variants. After 18 h of cell culture, supernatants were collected, cleared by centrifugation, and analyzed for IL-8 with a commercially available ELISA kit (BD Biosciences, Franklin Lakes, NJ) according to the manufacturer's instructions.
TNFR2-expressing cells were either challenged in a 4:1 ratio with NCTC transfectants expressing the indicated TNF variant or with the recombinant purified forms of TNF. After the indicated time points, cells were scraped into ice-cold phosphate-buffered saline and collected by centrifugation. For preparation of total cell lysates, cells were lysed in 4× Laemmli sample buffer supplemented with phosphatase inhibitors I and II (Sigma-Aldrich, Munich, Germany) and a protease inhibitor mixture (Roche, Mannheim, Germany). After sonification (10 pulses), samples were boiled for 5 min at 96 °C. Cytoplasmic and nuclear protein extracts were prepared using NE-PER nuclear and cytoplasmic extraction reagents following the manufacturer's instructions (Thermo Fisher Scientific, Rockford, IL). Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes. Protein binding sites on the membrane were further saturated by incubation with PBS containing 0.1% Tween 20 and 5% dry milk. Proteins of interest were finally detected by sequential incubation with primary antibodies of the indicated specificity and corresponding horseradish peroxidase-conjugated secondary antibodies (Dako, Hamburg, Germany) and visualization of antigen-antibody complexes using the ECL Western blotting detection system (Amersham Biosciences, Freiburg, Germany). Alternatively, IRDye-conjugated secondary antibodies were used for detection with the LI-COR Odyssey infrared imager (LI-COR Biosciences, Lincoln, NE).
Cells retained in a leukoreduction system chamber were subjected to Ficoll (PAA, Pasching, Germany) density gradient centrifugation to gain peripheral blood mononuclear cells (PBMCs). T-cells were isolated from PBMCs using a MACS separator and CD3-specific micro beads (Milteny Biotech, Auburn, CA). T-cells were controlled for purity by FACS analysis with a CD3-specific antibody (Milteny Biotech) and activated by stimulation with PHA (5 μg/ml, Sigma-Aldrich) and IL-2 (20 units/ml, Chiron, München, Germany) for 7 days.
HEK293 cells (25 × 106) were electroporated (4-mm cuvette, 250 V, 1800 μF, maximum resistance) in medium containing 40 μg of plasmid DNA encoding the different membrane TNF variants or the corresponding empty vector using an Easyject Plus electroporator (PeqLab, Erlangen, Germany). Two days post-transfection, cells were harvested and analyzed for TNF receptor specificity by FACS. In brief, transfectants were incubated on ice with Fc fusion proteins (4 μg/ml) of the extracellular domain of TNFR1 and TNFR2 for 30 min. TRAILR3-Fc (kind gift of Pascal Schneider, University of Lausanne, Switzerland) served as a negative control. Cells were washed three times and were then incubated (ice, 1 h) with 1 μg/ml anti-Fc-PE (Alexis, Grünberg, Germany). After removal of unbound anti-Fc-PE by washing cells three times with PBS, cell-bound receptor-Fc/anti-Fc-PE complexes were measured by FACS analysis (BD Callibur, BD, Heidelberg, Germany).
Purified ligands (100 μl, 80–100 μg/ml) were analyzed by size exclusion chromatography on a BioSep-SEC-S3000 (300 × 7.8) gel filtration column (Phenomenex, Aschaffenburg, Germany) equilibrated with PBS at a flow rate of 1 ml/min. Calibration of the column was carried out with the column performance check standard aqueous SEC 1 solution (Phenomenex) containing bovine thyroglobulin (670 kDa), IgG (150 kDa), ovalbumin (44 kDa), and myoglobin (17 kDa).
HeLa-TNFR2 cells (~105) were seeded on glass coverslips and cultured overnight at 37 °C. After incubation either with medium, Flag-TNF(32W/86T) for 45 min, or Flag-TNC-scTNF(143N/145R) for 18 h, cells were fixed with 2% formalin solution in PBS (10 min, room temperature), permeabilized with 0.1% Triton X-100 in PBS, and treated with 10% normal goat serum and 1% bovine serum albumin (30 min, room temperature) to block nonspecific protein binding. Anti-p65 (Santa Cruz Biotechnology, 1:100 in PBS) anti-RelB (Santa Cruz Biotechnology, 1:50 in PBS), and anti-p100 (Upstate, 1:100 in PBS) antibodies were then applied overnight at 4 °C. After rinsing three times for 5 min with PBS, cells were incubated with Cy3-labeled secondary antibodies (Dianova, Hamburg Germany) in a 1:600 dilution in PBS (1 h, room temperature). Following another three rounds of PBS rinsing, cells were mounted on glass slides with Fluoromount-G (Southern Biotech, Birmingham, AL) as antifading compound. Coverslips were imaged with a laser confocal setup (MRC-1024, Bio-Rad, Munich, Germany) via a Plan-Apochromat 63×/1.40 oil objective attached to a Axiovert 135TV microscope (both Carl Zeiss Microimaging, Göttingen, Germany). Mean nuclear and cytosolic fluorescence intensities were determined within a cell using ImageJ software. Ratios of nuclear to cytoplasmic fluorescence were calculated from >120 cells for every condition in three independent experiments. Statistical analysis was performed with Kruskal-Wallis ANOVA on ranks with Tukey posthoc test by the use of SigmaStat 3.5 software (Systat Software GmbH, Erkrath, Germany). Significance was assumed for p < 0.05.
Binding of soluble TNF results in strong activation of TNFR1, but fails to stimulate robust TNFR2 signaling (3). For other members of the TNF receptor superfamily that remained inactive/poorly active upon binding of their soluble trimeric ligands (e.g. CD95L and CD95, OX40L and OX40, APRIL and TACI, and TRAIL and TRAILR2), it has been observed that receptor activation becomes possible upon oligomerization of the soluble ligand molecules. In fact, bacterially produced preparations of His-tagged soluble TNF mutants recognizing TNFR2, which contains significant amounts of aggregated ligand trimers, activates TNFR2 (data not shown). As expected the corresponding eukaryotically produced secretable Flag-tagged trimeric variant of TNF(143N/145R) was unable to activate TNFR2 (Fig. 1D). Notably, Flag-tagged TNF(143N/145R) remained inactive despite oligomerization with anti-Flag antibody (data not shown), a treatment enabling Flag-tagged variants of CD95L (37), OX40L (38), APRIL (39), and TRAIL (40) to gain high activity on CD95, OX40, TACI, and TRAILR2. We recently observed that some soluble ligands of the TNF family benefit from covalent linkage of their subunits in a way resulting in strongly enhanced activity without overcoming, however, the need for oligomerization (38, 41). We therefore reasoned that not only oligomerization, but also “stabilization” of the trimeric structure of TNF(143N/145R) is necessary to gain a highly active TNFR2-specific agonist. To achieve “stabilization” of the TNF(143N/145R) trimer, we used a single chain cassette, in which three TNF(143N/145R) domains were genetically linked by two short peptide linkers. To achieve self-assembly of the scTNF(143N/145R) cassette into oligomers of defined stoichiometry, we further linked the scTNF(143N/145R) cassette N-terminally with the Flag-tagged trimerization domain of chicken tenascin-C (TNC). The latter forms tightly packed trimers because of strong non-covalent interactions and disulfide bonding (42). Accordingly, the huge majority (>95%) of Flag-TNC-scTNF(143N/145R) purified from cell culture supernatants by anti-Flag affinity chromatography eluted in gel filtration analysis with a molecular mass of 120–150 kDa, which is consistent with the expected self-assembly of three scTNF(143N/145R) cassettes in the formation of a defined species of molecules containing nine TNF domains (Fig. 1, A–C). To prove that the increased avidity of Flag-TNC-scTNF(143N/145R) compared with conventional Flag-TNF(143N/145R) does not compromise TNFR2 specificity conferred by the Flag-TNC-143N/145R mutation, we analyzed TNF-induced IL-8 production in HeLa cells, which express no TNFR2, and corresponding TNFR2 transfectants. Whereas conventional TNFR1-specific Flag-TNF(32W/86T) readily induced IL-8 production in both cell types, Flag-TNC-scTNF(143N/145R) induced IL-8 production only in the TNFR2-expressing HeLa variant (Fig. 1D). Flag-TNC-scTNF(143N/145R)-induced IL-8 production was significantly lower than those induced via TNFR1, but this was in good accord with earlier observations, indicating that TNFR2 is only a weak activator of the classical NFκB pathway in HeLa cells (8). We have recently found that TRAF1 enhances the capability of TNFR2 to activate the classical NFκB pathway and to induce IL-8 production (33). We therefore also analyzed the TNFR2 specificity of Flag-TNC-scTNF(143N/145R) in this more sensitive system. As expected, Flag-TNC-scTNF(143N/145R) induced in this HeLa type showed much stronger IL-8 production than in conventional HeLa-TNFR2 cells (Fig. 1D). Production of IL-8 was inhibited by adding TNFR2-Fc, but not by TNFR1-Fc, again emphasizing the specificity of Flag-TNC-scTNF(143N/145R) (Fig. 1E). Flag-TNF(32W/86T)-induced IL-8 production was blocked by TNFR1-Fc, but not by TNFR2-Fc (Fig. 1E).
TRAF2, TRAF3, and cIAP1/2 act together in triggering the proteasomal degradation of NIK and thus inhibit activation of the noncanonical NFκB pathway (32). As we and others (7, 12–14) have previously shown that TNFR2 stimulation results in the translocation of TRAF2 into a detergent-insoluble fraction and subsequent proteasomal degradation, we wondered whether TNFR2 activation also triggers p100 processing. To investigate the effect of TNFR1 and TNFR2 stimulation on the activity of the alternative NFκB pathway, we challenged Kym1 and EM-2 cells as well as HeLa and Jurkat cells stably transfected with TNFR2 with Flag-TNF(32W/86T) and Flag-TNC-scTNF(143N/145R). In agreement with the known fact that p100 is regulated by the classical NFκB pathway (31), this protein was up-regulated in Kym1, EM-2, Jurkat-TNFR2, and HeLa-TNFR2 cells upon TNFR1 activation. However, the ratio of p52 to p100 did not increase after Flag-TNF(32W/86T) stimulation, arguing against activation of the alternative NFκB pathway (Fig. 2A). In contrast, TNFR2 activation had no major impact on p100/p52 expression, but shifted the ratio of p100 and p52 toward the latter (Fig. 2A). Analysis of phosphorylation and degradation of IκBα, two hallmarks of activation of the classical NFκB pathway, confirmed that TNFR1, but not TNFR2 is a strong activator of this pathway (Fig. 2B). Thus, the classical NFκB pathway was activated via TNFR1 and the alternative NFκB pathway was activated by TNFR2. In agreement with the requirement of TNFR2 for oligomerized soluble TNF to become activated, conventional Flag-TNF(143N/145R) failed to trigger p100 processing (Fig. 2C). Stimulation of TNFR2 was accompanied by depletion of TRAF2 (Fig. 2D). We were unable to identify TRAF3 and cIAPs undoubtedly by Western blotting because of the notoriously low expression of these proteins (data not shown), but we observed that TNFR2 activation results also in NIK accumulation (Fig. 2E). Thus, TNFR2 most likely utilized the same TRAF2-, cIAP-, TRAF3-, and NIK-dependent pathway to trigger IKK1 activation and p100 processing that has been described for other receptors activating alternative NFκB signaling (32). Furthermore, in agreement with the established independence of the alternative NFκB pathway from IKK2, an inhibitor of the latter almost completely blocked TNFR1-mediated activation of the classical pathway, but showed no major effect on Flag-TNC-scTNF(143N/145R)-induced p100 processing (Fig. 2F). TRAF1 expression strongly enhances TNFR2-induced activation of the classical NFκB pathway (Ref. 33 and Fig. 1D). We, therefore, also analyzed TNFR2-induced processing of p100 in TRAF1-expressing HeLa-TNFR2 cells. Flag-TNC-scTNF(143N/145R) induced p100 processing with similar efficacy and to the same extent in HeLa-TNFR2-TRAF1 cells and the corresponding vector-transfected HeLa-TNFR2 cells (Fig. 2G). Thus, TRAF1 enhances activation of the classical but not the alternative NFκB pathway in response to TNFR2 activation.
Cytosolic and nuclear fractions were prepared from Jurkat-TNFR2 and HeLa-TNFR2 cells. Increased amounts of nuclear p52 and RelB were observed after stimulation of TNFR2, whereas there was none or only a minor increase upon TNFR1 stimulation (Fig. 3, A and B). The latter correlated with the abovementioned observation that total p100 synthesis is induced by TNFR1 via the classical pathway resulting in more p52 without changing the p100/p52 ratio. TNFR2 stimulation resulted in barely detectable nuclear translocation of p65 and p50 (Fig. 3, A and B). In contrast, the TNFR1-specific TNF mutant triggered robust nuclear translocation of both proteins (Fig. 3, A and B), which is in good agreement with the fact that p65-p50 NFκB heterodimers are preferentially regulated by the classical NFκB pathway. To confirm the differential effect of TNFR1 and TNFR2 stimulation on nuclear translocation of different NFκB subunits by an independent method, we analyzed the subcellular localization of these proteins by immunofluorescence. In nonstimulated HeLa-TNFR2 cells, p65 was readily detectable in the cytoplasm, but barely visible in the nucleus (Fig. 4A). RelB and p100/p52, however, were present in both cellular compartments. The latter matches to the previous biochemical analysis showing significant p100 to p52 processing (Fig. 2A) and p52 nuclear localization (Fig. 3A) in nonstimulated HeLa-TNFR2 cells. More importantly, stimulation of TNFR2 shifted the ratio of nuclear to cytoplasmic RelB and p100/p52 toward nuclear localization whereas there was no effect on the localization of p65 (Fig. 4, A and B). TNFR1 stimulation resulted in strong redistribution of cytoplasmic p65 into the nucleus, but showed no (RelB) or only a very minor (p100/p52) effect on the nuclear to cytoplasmic intensity ratio of RelB and p100/p52 (Fig. 4, A and B).
Thus, the effects of TNFR1 and TNFR2 stimulation on nuclear translocation of the various NFκB subunits correspond to the concept that the two TNF receptors complementary regulate the classical and alternative NFκB pathway.
TNFR2 has been implicated in a variety of immune cell-related functions especially in T-cell activation and survival (19, 20, 43, 44). We, therefore, analyzed activated primary human T-cells with respect to TNF-induced p100 processing. Although TNFR2 was only expressed in a subset of activated T-cells and although TNFR2 expression of T-cells derived from different donors was variable, in ~50% (5 of 11) samples analyzed, TNFR2 stimulation with Flag-TNC-scTNF(143N/145R) resulted in significant p100 processing, whereas TNFR1 activation showed no effect in all cases (Fig. 3C).
To directly test the effect of membrane TNF stimulation on the activity of the two NFκB signaling pathways, we introduced the mutations conferring TNFR1 and TNFR2 specificity into an expression plasmid-encoding membrane TNF (Fig. 5A). To minimize the potential interfering of soluble TNF molecules released from membrane TNF, we also deleted the TACE-processing site of the molecule. To further relieve analysis of the molecules, we used membrane TNF fusion proteins with an intracellular YFP domain (Fig. 5A). To prove the TNF receptor specificity of the various YFP-membrane TNF expression constructs, transiently transfected cells were analyzed with respect to cell surface expression of the molecules by FACS analysis using TNFR1-Fc and TNFR2-Fc (Fig. 5B). Whereas expression of YFP-membrane TNF was easily detectable with both Fc proteins, YFP-membrane TNF(32W/86T) was only stained with TNFR1-Fc, but not with TNFR2-Fc. In contrast, YFP-membrane TNF(143N/145R) was stained with TNFR2-Fc, but not with TNFR1-Fc. Thus, the 32W/86T and 143N/145R mutations are sufficient to make membrane TNF specific for TNFR1 and TNFR2, respectively. As observed before with Flag-TNC-scTNF(143N/145R), YFP-memTNF(143N/145R) induced no IL-8 production in HeLa cells, modest production in HeLa-TNFR2 cells, and showed the best IL-8 up-regulation in HeLa-TNFR2-TRAF1 cells. In contrast, YFP-memTNF(32W/86T) and YFP-memTNF induced in all three cell lines a robust IL-8 production (Fig. 5C). In accord with the specificity of the membrane TNF variants, TNFR1-Fc efficiently blocked IL-8 production by YFP-memTNF(32W/86T) and YFP-memTNF, but had no effect on YFP-memTNF(143N/145R)-induced up-regulation of IL-8 (Fig. 5C).
To analyze the effect of membrane TNF on NFκB activity, we performed coculture assays between murine NCTC cells transfected with the various TNF variants and human TNFR2-expressing cells. IL-8 production was measured as an indicator of the activity of the classical NFκB pathway, and p100 processing was monitored to determine the activity of the noncanonical NFκB pathway. Transfection of the TNF molecules into murine cells facilitates the analysis of IL-8 production and p100 processing in the human target cells. NCTC transfectants expressing soluble TNF or TNFR1-specific or wild-type membrane TNF strongly induced IL-8 production in both HeLa and HeLa-TNFR2-TRAF1 cells. In contrast, TNFR2-specific membrane TNF induced IL-8 only in the HeLa-TNFR2-TRAF1 cells (Fig. 6A). Processing of p100 however was only induced in cocultures with NCTC cells expressing TNFR2-specific membrane TNF, but not by TNFR1-specific and wild-type membrane TNF or soluble TNF (Fig. 6B). At the first glance, it is confusing that wild-type membrane TNF failed in this experiment to induce p100 processing, but a closer inspection of the corresponding group revealed that the target cells were killed by apoptosis. In fact, earlier studies have already shown that TNFR1 and TNFR2 costimulation, in contrast to individual TNF receptor stimulation, leads to apoptosis induction in the Jurkat-TNFR2 cells used in our experiment (11, 12). Together, these experiments again argue for a differential capability of the two TNF receptors to activate the classical and alternative NFκB pathway.
Activation of the classical NFκB pathway has been reported for all members of the TNF receptor superfamily having a cytoplasmic domain. In contrast, activation of the alternative NFκB pathway has only been described for a subset of TNF receptors including LTβR, FN14, CD27, and the B-cell regulatory receptors Baff and CD40 (30–32, 45). Particularly, TNF has repeatedly been reported to fail to stimulate the activity of the alternative NFκB pathway (45, 46), although it can regulate this pathway indirectly by up-regulation of p100 via the classical NFκB pathway. In this study, we showed with the help of oligomerized and membrane-bound TNF receptor-specific mutants that TNFR2 activation results in a variety of cell lines in increased p100 processing and nuclear translocation of “alternative pathway”-associated NFκB proteins (Figs. 2A, ,3,3, A and B, and and4).4). Soluble TNF, in contrast to membrane TNF, is not or only poorly active on TNFR2, if it is not oligomerized or aggregated. The difference in our findings to earlier reports failing to observe TNF-induced activation of the alternative pathway may therefore lie in the exclusive use of soluble TNF in those studies. In fact, we observed also that soluble TNF, despite stimulating strong activation of the classical NFκB pathway, fails to activate robust p100 processing and nuclear translocation of p52 and RelB (Figs. 22–4 and and6).6). The differential activation of the two TNF receptors by soluble and membrane TNF together with the complementary regulation of the two NFκB pathways by TNFR1 and TNFR2 described here, obviously facilitate the diversification of cellular TNF response and might therefore contribute to the known high pleiotropic action of TNF.
In contrast to the ubiquitous expression of TNFR1, TNFR2 expression is confined to immune cells and endothelial cells. In accord with the known intensive cross-talk between the two TNF receptors, analysis of TNFR1 and TNFR2 knock-out mice revealed that both receptors cooperate in the regulation of TNF-dependent cell death in ConA-activated T-cells, dendritic cells (DCs), and activated macrophages (47, 48). However, there is also clear evidence for TNFR2-specific functions, especially in CD8+ T-cells and CD4+ CD25+ regulatory T-cells. So, TNFR2 promotes expansion of CD4+ CD25+ regulatory T-cells and has been identified as a marker for regulatory T-cells with high suppressive capacity (43, 44). It also mediates a strong costimulatory signal in CD8+ T-cells, contributes to the up-regulation of anti-apoptotic proteins such as Bcl2, BclxL, and survivin and is required for priming of tumor-specific T-cells (19, 20, 50–51). The relevance of alternative NFκB signaling for T-cell biology has been poorly addressed so far. However, the fact that the T-cell-transforming Tax protein of human T-cell leukemia virus type is a potent activator of this pathway points to a role of alternative NFκB signaling in proliferation and survival of T-cells, which would be consistent with the aforementioned findings for TNFR2. RelB has been implicated in the development of myeloid-related DCs as well as in the maturation/differentiation of bone marrow-derived DCs and osteoclasts (52,–54). As p100 interacts with RelB and inhibits its nuclear translocation, activation of the alternative NFκB pathway and p100 processing are a crucial step in the aforementioned processes (55, 56). In view of the observations that TNF-induced maturation is impaired in TNFR2-deficient bone marrow-derived DCs and osteoclasts, it is tempting to speculate that TNFR2-dependent activation of the alternative NFκB pathway plays a role (48, 49).
Thus, our finding that TNFR2 in contrast to TNFR1 is a receptor that triggers p100 processing will support the future functional characterization of the alternative NFκB pathway in T-cells and other “TNF-controlled” immune cells such as macrophages and dendritic cells.
*This work was supported by Wyeth BioPharma (Forschungspreis Rheumatologie 2007), Deutsche Forschungsgemeinschaft (Sonderforschungsbereich 487, Project B7, Clinical Research Unit 216, Project 8, and Grant Wa 1025/18-1), and IZKF Würzburg (Project A-49).
2The abbreviations used are: