Tumor Necrosis Factor Receptor-associated Factor (TRAF) 1 Regulates CD40-induced TRAF2-mediated NF-κB Activation*

To investigate CD40 signaling complex formation in living cells, we used green fluorescent protein (GFP)-tagged CD40 signaling intermediates and confocal life imaging. The majority of cytoplasmic TRAF2-GFP and, to a lesser extent, TRAF3-GFP, but not TRAF1-GFP or TRAF4-GFP, translocated into CD40 signaling complexes within a few minutes after CD40 triggering with the CD40 ligand. The inhibitor of apoptosis proteins cIAP1 and cIAP2 were also recruited by TRAF2 to sites of CD40 signaling. An excess of TRAF2 allowed recruitment of TRAF1-GFP to sites of CD40 signaling, whereas an excess of TRAF1 abrogated the interaction of TRAF2 and CD40. Overexpression of TRAF1, however, had no effect on the interaction of TRADD and TRAF2, known to be important for tumor necrosis factor receptor 1 (TNF-R1)-mediated NF-κB activation. Accordingly, TRAF1 inhibited CD40-dependent but not TNF-R1-dependent NF-κB activation. Moreover, down-regulation of TRAF1 with small interfering RNAs enhanced CD40/CD40 ligand-induced NF-κB activation but showed no effect on TNF signaling. Because of the trimeric organization of TRAF proteins, we propose that the stoichiometry of TRAF1-TRAF2 heteromeric complexes ((TRAF2)2-TRAF1 versus TRAF2-(TRAF1)2) determines their capability to mediate CD40 signaling but has no major effect on TNF signaling.

CD40 and its ligand CD40L 1 /CD154 are members of the tumor necrosis factor (TNF) receptor and TNF ligand family and represent major regulators of lymphocyte function (1). Aside from T-and B-cells, CD40 and CD40L are expressed in a variety of non-lymphocytic cell types including monocytes, dendritic cells, fibroblasts, smooth muscle, and endothelial cells (1). The CD40/CD40L system plays a critical role in the regu-lation of thymus-dependent humoral immune responses but also contributes to chronic inflammatory processes in autoimmune diseases, neurodegenerative disorders, graft-versus-host disease, cancer, and atherosclerosis (1).
Engagement of CD40 results in the recruitment of members of the TNF receptor-associated factor (TRAF) adaptor protein family (1,2). In addition, triggering of CD40 leads to Janus family kinase 3 (Jak3)-dependent activation of signal transducers and activators of transcription (STAT) proteins and to activation of the Src-related tyrosine kinase Lyn (3)(4)(5)(6). TRAF proteins couple TNF receptors and Toll/interleukin-1 receptor family members to pathways leading to the activation of the inhibitor of I-B kinases and kinases of the mitogen-activated protein kinase (MAPK) family (2). All members of the TRAF family share a conserved C-terminal homology domain of ϳ180 amino acids (TRAF domain), which mediates interactions with the above mentioned receptors and the majority of cytosolic factors known for their TRAF binding capacity, including kinases, inhibitor of apoptosis proteins, and death domain adaptor proteins (2). With the exception of TRAF1, the N-terminal domain of all six known mammalian TRAFs comprise a single RING finger followed by several zinc finger motifs (2) that are important for downstream signaling events, e.g. via interaction with mitogen-activated protein kinase/extracellular signal-regulated kinase kinase kinase 1 (7), IB kinase 1 (IKK1), or IKK2 (8). All known TRAF proteins except TRAF4 have been reported to associate directly or indirectly with CD40 (9 -15). The cytoplasmic domain of human CD40 consists of only 62 amino acids and comprises two distinct TRAF binding sites located between amino acids 231 and 238 and amino acids 250 and 266. Whereas the membrane-proximal TRAF-interacting segment of CD40 mediates association with TRAF6, the receptor-distal TRAF-interacting segment allows recruitment of TRAF2 and TRAF3 (13,15,16). TRAF5 can also be indirectly recruited to the receptor-distal TRAF-interacting segment by TRAF3 (15). There are inconsistent reports in respect to a direct interaction of CD40 and TRAF1 (15,17,18). As TRAF2, TRAF5, and TRAF6 have the capability to stimulate NF-B-and c-Jun NH 2 -terminal kinase (JNK) activation, the two TRAF binding sites of CD40 may, in principle, allow redundant activation of these pathways. For TRAF3, a negative regulatory role in CD40-induced differentiation of B-cells and an involvement in CD40-mediated activation of JNK and p38 (19) have been described. TRAF6 has to been shown to contribute to CD40dependent extracellular signal-regulated kinase via a Ras-independent pathway (20). Although there is ample evidence that CD40 can signal NF-B activation by at least two independent pathways mediated by either TRAF2 or TRAF6 (see above), splenocytes from both TRAF2 (21) and TRAF6 (22) knock-out mice showed complete abrogation of NF-B activation. CD40-dependent activation of NF-B or JNK is apparently normal in TRAF5 deficient mice; nevertheless B cells from these mice showed defects in proliferation, compromised in vitro immunoglobulin production, and reduced up-regulation of various surface molecules including CD23, CD54, CD80, CD86, and Fas (23). The contribution of TRAF3 to CD40 signaling in vivo is poorly understood. B cells from TRAF3-deficient mice proliferate normally after CD40 triggering but have a defect in isotype switching in response to thymus-dependent antigens (24). Bcells of TRAF1 knock-out mice are largely unaffected with respect to CD40 signaling (25) whereas CD40L-matured bone marrow-derived dendritic cells exert reduced CD40-mediated NF-B upon restimulation (26). In contrast, NF-B activation by the CD40-related TNF-R2 is enhanced in T-cells of TRAF1deficient mice (25). Interpretation of data obtained from the study of TRAF-deficient mice is partly confounded by the fact that this group of molecules is commonly used by several members of the TNF receptor family. However, two studies with mutated CD40 transgenes lacking the TRAF2/3 and/or TRAF6 binding site confirmed the above-mentioned pivotal role of TRAF2 for CD40-mediated activation of NF-B, JNK, and p38 kinase (27,28).
Here we show in living cells that TRAF2 secondarily recruits TRAF1 and the cellular inhibitor of apoptosis (cIAP) proteins cIAP1 and cIAP2 to sites of CD40 signaling. TRAF1 interfered with TRAF2 recruitment to sites of CD40 signaling and diminished CD40-mediated NF-B activation, whereas down-regulation of TRAF1 enhanced this response. These data suggest that CD40-mediated NF-B activation is regulated by the stoichiometry of TRAF1-TRAF2 heteromeric complexes.

MATERIALS AND METHODS
Cells and Reagents-The human cervical carcinoma cell line HeLa and HEK293 cells were obtained from the American Type Culture Collection (Manassas, VA) and maintained in RPMI medium (Biochrom, Berlin, Germany) supplemented with 5% FCS. HeLa transfectants stably expressing TRAF1 have been described previously elsewhere (29). A20-GFP was a gift from Rudi Beyaert and described previously elsewhere (30). IKK1-YFP was a kind gift by Johannes Schmid (University of Vienna, Vienna, Austria). All other chimeric proteins containing GFP, YFP, or CFP were generated by standard cloning techniques with proofreading polymerase and the pEGFP-N1 vector (Clontech). Thus, in all these chimeras the GFP part is localized C-terminally. In case of the TRAF proteins, chimeras with N-terminal GFP have also been investigated but showed no difference when compared with the variants mentioned above. FLAG-tagged CD40L comprises the extracellular domain of human CD40L (amino acids 116 -261) fused with an N-terminal FLAG tag by a linker of six amino acids. The small interfering RNA oligonucleotides (siRNAs) were a kind gift from Hans-Peter Vornlocher and Philipp Hadwiger (Ribopharma AG, Kulmbach, Germany).
Immunofluorescence and Confocal Microscopy-Cells (10 6 cells/ml) were electroporated (4-mm cuvette; 250 V, 1800 F, maximal resist- . The next day, CD40 signaling was initiated using soluble FLAG-tagged CD40L (200 ng/ml) crosslinked with anti-FLAG mAb M2 (␣ Flag; 1 g/ml). The distribution of the GFP-tagged proteins was analyzed by online confocal microscopy using a conditioned chamber (37°C; 5% CO 2 ) both prior to stimulation and for a further 20 min after stimulation. The TRAF2-YFP emission was recorded in the GFP channel (510 -560 nm) and therefore appears green. B and D, HeLa-CD40 cells were transfected with expression plasmids encoding the indicated proteins. The following day, non-stimulated and sCD40L/M2stimulated cells (10 min) were analyzed by confocal microscopy. Endogenous CD40 expression was detected by staining FLAG-tagged sCD40L with M2 and PElabeled anti-murine IgG. In contrast to stimulated cells, non-stimulated cells were fixed not after but prior to incubation and staining with sCD40L. Emission of both the GFP and the YFP chimeras were recorded between 510 and 560 nm. Prime symbol (Ј), minutes. ance) in a medium with 5% FCS and seeded onto 13-mm glass coverslips in six-well dishes or in glass-bottomed dishes (MatTek Corporation, Ashland, MA). For an on-line analysis of CD40 signaling complex formation, cells expressing GFP/YFP/CFP chimeras were stimulated with FLAG-tagged soluble CD40L (sCD40L) crosslinked with 1 g/ml anti-FLAG mAb M2 (Sigma) and maintained in a conditioned chamber (37°C; 5% CO 2 ) for up to 2 h on the microscope stage. Fluorescent specimens were analyzed with a Leica SP2 confocal microscope and imaged using the Leica TCS software. For colocalization of endogenous CD40 and GFP fusion proteins, transfected cells were stimulated with crosslinked sCD40L. To detect CD40 in non-stimulated cells, samples were fixed (15 min, 3% paraformaldehyde), washed three times in phosphate-buffered saline, and blocked with 5% FCS (30 min at room temperature). Fixed cells were incubated with sCD40L (200 ng/ml) and anti-FLAG mAb M2 (1 g/ml) for 1 h at room temperature and, after six washes, M2 was detected with 1:500 diluted mouse IgG-specific antisera conjugated with Alexa Fluor 546 (A-11060) (1 h at room temperature). After nine washes, cells were finally mounted onto microscope slides. Cells already stimulated with sCD40L/M2 were prepared the same way but without additional sCD40L/M2 treatment after fixation. For counting experiments of the recruitment of YFP/CFP-labeled proteins to sCD40L-stimulated CD40 and TRADD filaments, HeLa cells were transiently transfected with plasmids encoding the indicated proteins. The next day, the portion of successfully transfected cells with clear recruitment of YFP/CFP-labeled proteins to CD40 or TRADD filaments, respectively, were counted in at least three independent samples.
Reporter Gene Assay-HEK293, HeLa, or HeLa cells stably transfected with TRAF1 (20 ϫ 10 3 ) were seeded in 96-well tissue culture plates. The following day cells were transfected with a 3ϫNF-B luciferase reporter plasmid (35 ng/well) and a SV40 promoter-driven ␤-galactosidase expression plasmid (15 ng/well) to normalize the transfection efficiency along with the indicated mixtures of empty vector and the constructs of interest (200 ng/well). Transfections were performed with SuperFect reagent according to the manufacturer's recommendations (Qiagen, Hilden, Germany). The next day, cell extracts were prepared by the addition of 50 l of luciferase lysis solution (Galactolight-Kit, Tropix, Bedford, MA) and one freeze-thaw cycle. A portion of the extracts (25 l) was mixed with 50 l of luciferase substrate (luciferase assay system from Promega), and the luminescence was determined in the single photon mode using an Anthos microplate luminometer (Lucy 2). In parallel, 25 l of each cell extract was incubated for 1 h with a 1:100 dilution of Galacton substrate in reaction buffer and mixed with 100 l of accelerator II solution to determine relative ␤-galactosidase activity (Galactolight-Kit, Tropix), again using the Lucy 2 luminometer. Luciferase activities were normalized based on the ␤-galactosidase activities.
Analysis of TRAF1/TRAF2 Heterocomplexes by Immunoprecipitation-8 ϫ 10 5 HEK293 cells were transfected with 5 g of DNA (ratios of plasmids are indicated in Fig. 2E), using the SuperFect reagent according to the manufacturer's recommendations (Qiagen). Cells were harvested 2 days post-transfection and lysed for 30 min on ice in 1 ml of lysis buffer containing 20 mM Tris (pH 7.4), 1% Triton X-100, 150 mM NaCl, 5 mM MgCl 2 , and a mixture of standard protease inhibitors. Lysates were cleared by centrifugation (15 min at 13,000 rpm), and 75 l of each lysate was kept for analysis of protein expression levels. Lysates were incubated with 2 g of a TRAF1-directed rabbit antiserum (H-186; Santa Cruz Biotechnology) under rotation for 1.5 h at 4 C o . Protein G-Sepharose (Amersham Biosciences; 3 l per sample) was then added to each sample, and the incubation was continued for another hour. Immunoprecipitates were washed three times in lysis buffer and analyzed by SDS-PAGE/Western blotting.

CD40L-induced Recruitment of TRAF Proteins in Living
Cells-To investigate the recruitment of various cytoplasmic signaling intermediates to the CD40 signaling complex in intact living cells, we transiently transfected HeLa cells stably expressing CD40 with constructs encoding GFP/YFP-tagged forms of the proteins of interest in glass-bottomed dishes. The next day, cells were stimulated with anti-FLAG mAbcrosslinked recombinant soluble N-terminal FLAG-tagged CD40L and monitored on-line in a conditioned chamber (37°C; 5% CO 2 ) by confocal microscopy with respect to intracellular distribution of GFP fusion proteins. In unstimulated cells, GFP/YFP-tagged TRAF1, TRAF2, TRAF3, and TRAF4 were mainly localized in the cytoplasm and excluded from the nucleus. In cells stimulated with crosslinked sCD40L, TRAF2-YFP (Fig. 1A, middle row) and, to a lesser extent, TRAF3-GFP (Fig. 1A, bottom row), but not TRAF1-GFP or TRAF4-GFP, became compartmentalized in small membrane-localized aggregates (Fig. 1A, top row). sCD40L-induced aggregation of TRAF2-YFP and TRAF3-GFP occurred in all transfected cells independently of the individual expression levels of the chimeric proteins within 5-10 min (data not shown). Staining of sCD40L by detection of its FLAG tag showed extensive colocalization with TRAF2-GFP and TRAF3-GFP only in the TRAF2/3-GFP aggregates of CD40-stimulated cells (Fig. 1B). As crosslinked sCD40L activates CD40 and activated CD40, in turn, interacts strongly with TRAF2 and TRAF3 (13,15,16), these data suggest that the observed sCD40L-induced TRAF2-GFP and TRAF3-GFP aggregates represent CD40-TRAF2 and CD40-TRAF3 complexes (Fig. 1B). The marked preference of sCD40L/CD40 complexes to interact with TRAF2-GFP and TRAF3-GFP is in good agreement with data from the literature showing superior binding of in vitro translated TRAF2 and TRAF3 to recombinant CD40-GST as compared with other TRAFs. Moreover, in surface plasmon resonance analyses of the recombinant TRAF domains of TRAF2 and TRAF3 and the immobilized cytoplasmic domain of CD40, K d -values of 2.5 and 15 M have been reported, whereas no detectable interaction was found with TRAF1 and TRAF6 (17,18).
TRAF2 Recruits cIAP1 and 2 and TRAF1 to Sites of CD40 Signaling-To verify that TRAF2 and TRAF3 can act as adaptors for the recruitment of known TRAF-interacting proteins into the CD40 signaling complex, we cotransfected GFP/YFPtagged variants of TRAF1, TRAF4, cIAP1, cIAP2, A20, and IKK1 along with non-tagged TRAF2 and TRAF3 and analyzed cells as described above. To analyze cIAP1 and cIAP2, we have used, in most experiments, C-terminal deletion mutants (cIAP1/2⌬C) of these molecules that lack the RING finger domain but can still interact with TRAF1 and TRAF2 via their N-terminal BIR domains. These deletion mutants were more efficiently expressed than their parental counterparts, most likely due to the lack of an E3 ligase activity located in the C termini of cIAP1 and cIAP2, mediating their own degradation via the proteasome (32,33). In unstimulated cells, cIAP1-GFP and cIAP2-GFP (41) as well as cIAP1⌬C-GFP and cIAP2⌬C-GFP alone localize in the cytoplasm and the nucleus (Fig. 1C, top row). However, upon cotransfection with non-tagged TRAF2, both cIAP1⌬C-GFP and cIAP2⌬C-GFP were predominantly relocated to the cytoplasm (Fig. 1C, middle and bottom  row). This is in accordance with published data suggesting that cIAP1/2 and TRAF2 are parts of a preformed cytoplasmic complex (34). Transfected cIAP1/2-GFP (data not shown) or cIAP1/ 2⌬C-GFP molecules alone colocalized only marginally with the CD40 signaling complex after CD40 triggering. However, in the presence of cotransfected TRAF2, extensive colocalization of the cIAP1/2⌬C-GFP proteins and sCD40L was observed (Fig.  1D). Because of the known strong interaction between TRAF2 and cIAP1/2 (34) and TRAF2 and activated CD40 (13,15,16), colocalization of sCD40L with cIAP1⌬C-GFP and cIAP2⌬C-GFP confirmed that sCD40L-induced cIAP⌬C-GFP aggregates represent CD40-containing complexes (Fig. 1D). In contrast, TRAF3 was unable to redirect nuclear localized cIAP1/2⌬C-GFP to the cytoplasm and also failed to induce colocalization of cIAP1/2⌬C-GFP and sCD40L/CD40 complexes (data not shown). sCD40L-induced, TRAF2-mediated colocalization of cIAP1/2⌬C-GFP and sCD40L occurred in all transfected cells mainly independently of the individual expression levels of the chimeric proteins (data not shown). For TRAF1-GFP we observed TRAF2-dependent recruitment in sCD40L-induced aggregates also ( Fig. 2A, far right column). Compared with TRAF2-dependent colocalization of cIAP1/2-GFP and sCD40L, however, this was less efficient and only observed in a subset of cells. In accordance with a TRAF2-mediated recruitment of TRAF1 to sites of CD40 signaling, we found that TRAF2-CFP and TRAF1-YFP colocalized upon sCD40L stimulation ( Fig.  2A, left three columns). For IKK1-YFP (41) and A20-GFP (data not shown), no significant TRAF2-dependent recruitment into CD40 signaling complexes was found, although TRAF2 is able to efficiently recruit these molecules to TRADD, a signaling intermediate of the TNF-R1 signaling complex.
TRAF1 Regulates the CD40-TRAF2 Interplay-As discussed above, TRAF2 is capable to recruit TRAF1 to sites of CD40 signaling. However, in our experiments TRAF2-dependent recruitment of TRAF1 was paralleled by an overall reduction in colocalization of CD40 and TRAF2-GFP. We therefore analyzed the effect of the TRAF1 to TRAF2 ratio on the colocalization efficiency of TRAF2-YFP and sCD40L/CD40 complexes. The percentage of cells that showed colocalization of TRAF2-YFP and sCD40L successively dropped down from 100%, when TRAF2-YFP was cotransfected with empty vector, to ϳ50%, when the TRAF2-YFP expression plasmid was cotransfected along with a 5-fold excess of the TRAF1 encoding plasmid (Fig.  2B). Moreover, in the residual TRAF2-YFP aggregate-positive cells, the density and the number of aggregates were significantly reduced (data not shown). Complementary to these data, we found that TRAF1-YFP most efficiently colocalized with sCD40L/CD40 complexes in the presence of an excess of TRAF2 (Fig. 2C). The inhibitory effect of TRAF1 expression on the recruitment of TRAF2 into sites of CD40 signaling was also apparent in neighboring cells expressing similar levels of TRAF2-CFP and varying amounts of TRAF1-YFP, as is evident in Fig. 2A, for example. Notably, the cell showing the highest TRAF1-YFP expression shows almost no sCD40L-induced clustering of TRAF2-CFP, whereas the cell with the lowest TRAF1-YFP expression displays a high number of such aggregates ( Fig. 2A, bottom row on the left). In contrast to TRAF1, TRAF3 did not interfere with colocalization of TRAF2-YFP and sCD40L even at high TRAF3 to TRAF2 ratios (Fig. 2D). Thus, we have the puzzling observation of a TRAF2-dependent recruitment of TRAF1-YFP to sites of CD40 signaling despite an were detected by Western blotting (WB) using TRAF1-or TRAF2-specific antisera and, alternatively, a GFP-specific monoclonal antibody (third section from top). Bottom section, TRAF1-YFP was immunoprecipitated (IP) from these lysates using the TRAF1 specific antiserum. TRAF1-YFP and co-immunoprecipitated TRAF2-YFP were simultaneously detected using the GFP-specific monoclonal antibody. F, model of CD40 interaction with TRAF1, TRAF2, and TRAF1-TRAF2 heteromeric complexes.

FIG. 3. TRAF1 differentially affects CD40 and TNF-R1-induced activation of NF-B and does not interfere with TRADD-TRAF2 interaction.
A, 20 ϫ 10 3 HeLa cells stably expressing TRAF1-GFP were seeded in 96-well cell culture plates. The next day, cells were transiently transfected as indicated with expression vectors (200 ng/well) encoding CD40, TNF-R1, or empty vector along with a 3ϫNF-B-luciferase reporter plasmid (35 ng/well) and a SV40 promoter-driven ␤-galactosidase expression plasmid (15 ng/well). After a further day, cells, except controls, were incubated for 9 h with TNF (10 ng/ml) or interleukin 1 (IL1; 10 ng/ml) and assayed for luciferase and galactosidase activity. B, 20 ϫ 10 3 HeLa cells were cultured in 96-well cell culture plates. The next day, cells were transiently transfected as indicated with 120 ng/well empty vector or an expression vector encoding TRAF1-GFP along with a 3ϫNF-B-luciferase reporter plasmid (35 ng/well) and a SV40 promoter-driven ␤-galactosidase expression plasmid (15 ng/well). NF-B activation was triggered by cotransfection of an 80 ng/well, 1:1 mixture of CD40 and CD40L expression plasmids or by cotransfection with 80 ng of a TNF-R1 expression plasmid. One day post-transfection, cells were lysed and assayed for luciferase and galactosidase activity. C, quantification of endogenous TRAF1 and TRAF2 expression. HeLa, HT1080 (4 ϫ 10 5 ), and SV80 (1 ϫ 10 5 ) cells were stimulated with TNF (20 ng/ml) for 6 h or remained untreated. Subsequently, total cell lysates were prepared and analyzed by Western blotting (WB) together with TRAF1-GFP and TRAF-GFP as mass standards. Using TRAF1 (top section) and TRAF2-specific antibodies (bottom section), the concentrations of endogenously expressed TRAF1 and TRAF2 were estimated by comparison with the corresponding mass standards. TRAF1-GFP and TRAF2-GFP mass standards were obtained by transient expression and calibration with a commercial GFP standard by fluorescence spectroscopy. D, quantification of ectopically expressed TRAF1-GFP and TRAF2-GFP. FACS analyses of enhanced GFP FACS calibration beads with 0, 4,500, 15,000, 44,000, and 116,000 molecules of equivalent soluble fluorochrome (thin solid line) and HEK293 cells 24 h post-electroporation with 10 g of expression plasmids of TRAF1-GFP (thick dotted line) and TRAF2-GFP (thick solid line) or empty vector (gray shading) are shown. The localization of the calibration beads was corrected according to the different auto-fluorescence of vector-transfected cells and beads. E, the TRAF1-GFP expression plasmid was electroporated in triplicate along with 150 nM siRNAs specific for TRAF1 (TRAF1-1 and TRAF1-2), neomycin (K4), or GRP (S11). The next day, cells were analyzed by FACS, and relative expression levels (product of the percentage of positive cells and the mean intensity of positive cells) were calculated. F, to determine the effect of endogenously induced TRAF1 on NF-B activation induced by various stimuli, the appropriate reporter genes (see panel A) were cotransfected along with the indicated siRNAs. Bcl2-5 is a low efficacy, Bcl2-specific siRNA included as an additional negative control. The next day, cells were stimulated for 6 h with 20 ng/ml TNF or remained without further treatment. Finally, cells were assayed for luciferase and ␤-galactosidase activity. G and H, HeLa cells were transiently transfected with expression plasmids encoding the indicated proteins and empty vector. To inhibit TRADD-induced apoptosis, cells were treated with benzyloxycarbonyl-VAD-fluoromethyl ketone (20 M) immediately after transfection. The next day, transfected cells were analyzed by confocal microscopy with regard to the recruitment of TRAF1-GFP and TRAF2-YFP into TRADD filaments by calculating the percentage of transfected cells showing fluorescent filaments (G), and representative cells were selected for photography (H). overall inhibitory effect of TRAF1 on sCD40L-induced colocalization of TRAF2-YFP and CD40. To further analyze the interaction of TRAF1 and TRAF2, immunoprecipitation experiments were performed using lysates of cells that express equal levels of TRAF1-YFP but increasing amounts of TRAF2-YFP (Fig. 2E, top two sections). Using a GFP/YFP specific monoclonal antibody that recognizes both proteins, we have found that, with increasing amounts of the TRAF2-YFP plasmid, comparable expressions of both proteins had been reached (Fig.  2E, third section from the top). Using a polyclonal TRAF1directed antiserum, TRAF1-YFP was immunoprecipitated. To estimate the relative amounts of TRAF1-YFP and TRAF1bound TRAF2-YFP, immunoprecipitates were analyzed by Western blotting using the GFP-specific monoclonal antibody (Fig. 2E, bottom section). Even at the lowest TRAF1/TRAF2 expression ratio, TRAF2-YFP was detectable in association with TRAF1-YFP. Increasing amounts of TRAF2-YFP in the lysate led to immunoprecipitation of complexes that contain equal amounts of TRAF1-YFP and TRAF2-YFP. In this experiment, the TRAF2-YFP content of the immunoprecipitates could even be underestimated, because a TRAF2-YFP degradation product occurred that co-migrated with TRAF1-YFP (Fig. 2E). Thus, TRAF1 and TRAF2 have a very strong tend-ency to form heteromeric complexes when expressed at comparable concentrations. Such expression levels can indeed occur upon up-regulation of endogenous TRAF1 by NF-B inducers, e.g. TNF (see also Fig. 3C). The most likely explanation for the TRAF1-mediated inhibition of TRAF2 recruitment to sites of CD40 signaling is that TRAF2 homotrimers and TRAF2-TRAF1 heteromeric complexes with distinct subunit compositions ((TRAF2) 2 -TRAF1 versus TRAF2-(TRAF1) 2 ) interact differentially with CD40 (Fig. 2F). It has been reported that contacts between receptor peptides derived from TRAF-interacting receptors like CD40 or TNF-R2 and TRAF2 or TRAF3 buried only a small area (35)(36)(37). Consistent with these findings, an avidity model of receptor-TRAF interaction has been suggested in which three low affinity binding sites, each built up by a single receptor molecule and one subunit of a TRAF trimer, act together to form a receptor-TRAF complex of reasonable stability. Hence, it seems conceivable that the total affinity of a (TRAF2) 2 -TRAF1 heteromer, which based on the TRAF2-CD40 interaction and the avidity effect of the two TRAF2 subunits, is high enough to form stable CD40-TRAF2-TRAF1 complexes, whereas the single low affinity interaction between CD40 and a TRAF2-(TRAF1) 2 heteromer is too low to form a stable complex with CD40 (Fig. 2F). A recent study has  shown that TRAF1 redistributes TRAF2 and CD40 from low density raft fractions to the soluble fraction (26). These data are in good agreement with the inhibitory effect of TRAF1 on CD40-TRAF2 colocalization, as previous reports suggested that, upon stimulation, CD40 translocates to lipid rafts leading to the binding of TRAF2 (38,39). TRAF1 Inhibits CD40-induced but Not TNF-R1-induced NF-B Activation-Because TRAF2 is critically involved in CD40-mediated NF-B activation (21), we next analyzed the impact of TRAF1 on CD40/CD40L-induced activation of a NF-B-driven reporter gene. Ligand-independent NF-B activation by transient overexpression of CD40 was significantly reduced in HeLa cells stably expressing TRAF1. In contrast, NF-B activation induced by TNF, interleukin 1, or overexpressed TNF-R1 was unaffected by TRAF1 (Fig. 3A). Similarly, NF-B activation induced by cotransfection of a mixture of a CD40 expression plasmid and a construct encoding full-length, membrane-bound CD40L showed a marked reduction when transfected along with a TRAF1 expression plasmid, whereas activation of the NF-B reporter gene by a TNF-R1 expression plasmid remained unaffected by TRAF1 cotransfection (Fig. 3,  A and B). Thus, the capability of TRAF1 to weaken sCD40Linduced colocalization of TRAF2 and CD40 (Fig. 2) correlates with the inhibitory effect of TRAF1 on CD40-mediated NF-B activation.
To judge whether the expression of endogenous TRAF1 can be high enough to block TRAF2-CD40 interaction and CD40mediated NF-B activation, we determined and compared the expression levels of endogenous TRAF1 and TRAF2 with that of TRAF1-GFP. To obtain TRAF1 and TRAF2 mass standards, we prepared lysates of TRAF1-GFP-expressing and TRAF2-GFP-expressing cells and determined the concentration of the respective GFP fusion protein by the help of a fluorescence spectrophotometer and a commercially available GFP standard of known concentration. With the TRAF-GFP mass standards and TRAF-specific antibodies, we quantified the endogenous expression of TRAF1 and TRAF2 by comparative Western blot analysis (Fig. 3C). Because TRAF1 is not or only modestly expressed in most cells but can typically be up-regulated by NF-B-inducing stimuli (2), we analyzed TNF-stimulated cells. Depending on the cell type, we found expression of endogenous TRAF1 corresponding to 30,000 -600,000 molecules per cell in TNF-treated cells (Fig. 3C). In contrast, TRAF2 expression was independent from TNF treatment and rather similar in all investigated cell types with 50,000 -200,000 molecules per cell (Fig. 3C). Thus, endogenous TRAF1 to TRAF2 ratios of 10:1 to 1:6 can be reached in NF-B-activated cells. Because the TRAF1-GFP and TRAF2-GFP expression plasmids used in this study show comparable protein expression, this range of TRAF1 to TRAF2 ratios covers those of the TRAF fusion proteins in Fig. 2. Next, we determined the distribution of TRAF1-GFP expression by the use of GFP fluorescence standard beads by FACS analyses. These analyses revealed that ϳ30% of the HeLa-TRAF1-GFP cells (data not shown) and 50% of the transiently transfected cells (Fig. 3D) express Ͼ200,000 molecules per cell. As CD40-mediated NF-B activation was reduced 50 and 66% in these cells, this implies that in a significant portion (20 -30%) of the cells the NF-B inhibitory effect is the result of endogenously occurring expression levels of TRAF1.
To study whether the induction of endogenous TRAF1 counteracted CD40-induced NF-B signaling, we used siRNAs of 21 nucleotides that are complementary to nucleotides 406 -426 and 1034 -1054 of the TRAF1 cDNA. Transfection with the TRAF1-specific siRNAs enhanced CD40/CD40L-induced NF-B activation but showed no effect on NF-B activation induced by overexpressed p65 or TNF-R1 or by treatment with TNF (Fig. 3, E and F). Control siRNAs specific for neomycin or Bcl2 did not, in any case, show an effect. We conclude from these data that NF-B-induced TRAF1 expression selectively counteracts CD40mediated NF-B activation.
Because TNF-R1-induced NF-B activation, which is also mediated by TRAF2 (40), was not regulated by TRAF1 (Fig. 3,  A, B and F), we next checked whether TRAF1 interferes with TRAF2 recruitment to TRADD, an adaptor protein bridging TNF-R1 and TRAF2. Ligand-induced assembly of TNF-R1/ TRADD signaling complexes is difficult to analyze by the method used above for CD40 because of the very low expression level of endogenous TNF-R1 and the strong tendency of overexpressed TNF-R1 to auto-aggregate. To analyze TRAF2-TRADD interaction, we therefore employed the capability of overexpressed TRADD to form distinct filaments (41)(42)(43). Upon cotransfection with TRADD, TRAF2-YFP was recruited very efficiently to TRADD filaments (Fig. 3, G and H) because of the interaction of its TRAF domain with the N-terminal TRAF-binding domain of TRADD (50). In contrast to the CD40-TRAF2 colocalization, the TRADD-TRAF2 interaction was not affected by TRAF1 (Fig. 3G). Indeed, TRAF1 itself was able to interact with TRADD filaments (Fig. 3, G and H). Thus, the capability of TRAF2-utilizing receptors or adaptor proteins to interact or not interact with TRAF1 and (TRAF1) 2 -TRAF2, might determine the impact of TRAF1 on the signaling capabilities of these molecules. At first glance, TRAF1-mediated inhibition of CD40L-induced NF-B activation is in contrast to data obtained from CD40L-maturated bone marrow dendritic cells of TRAF-deficient mice showing reduced CD40-mediated NF-B activation (26). However, TRAF2 expression in these traf1 Ϫ/Ϫ cells was strongly reduced compared with that of the corresponding cells derived from wild type mice (26). Thus, the observed reduction of CD40-mediated NF-B activation in this case most likely reflects the limited availability of TRAF2 rather than a NF-B stimulatory action of TRAF1. Like CD40, TNF-R2 did not directly interact with TRAF1 but can recruit this molecule by the help of TRAF2 (44). It is therefore tempting to speculate that the inhibitory effect of TRAF1 on TNF-R2 signaling deduced from the phenotype of TRAF1-deficient mice (25) is also related to the mechanisms discussed above for CD40. However, one should note in this regard that the molecular mechanistic interpretation of data derived from T-cells of TRAF1-deficient mice is difficult and has the following limitation. Early proliferation of T-cells following their activation is dependent on the non-apoptotic activation of caspases (45)(46)(47). Because TRAF1 can be cleaved by caspases under generation of a fragment that is inhibitory for TNF-induced NF-B activation (48 -50), one cannot decide whether the reported increase of TNF-induced proliferation of TRAF1-deficient T-cells is due to the absence of the NF-B inhibitory TRAF1 fragment or to the absence of full-length TRAF1. As TRAF1 itself is a target gene of the NF-B pathway (29,(51)(52)(53), it might serve as feedback regulator of this pathway that interferes with NF-B activation of a selected range of NF-B-inducing receptors.