Molecular Characterization of CD40 Signaling Intermediates*

Signal transduction through the CD40 receptor is initiated by binding of its trimeric ligand and propagated by interactions of tumor necrosis factor receptor-associated factor (TRAF) proteins with the multimerized CD40 cytoplasmic domain. Using defined multimeric constructs of the CD40 cytoplasmic domain expressed as either soluble or myristoylated proteins, we have addressed the extent of receptor multimerization needed to initiate signal transduction and identified components of CD40 signaling complexes. Signal transduction in human embryonic kidney 293 cells, measured by nuclear factor κB activation, was observed in cells expressing soluble trimeric CD40 cytoplasmic domain and to a lesser extent in cells expressing dimeric CD40 cytoplasmic domain. Nuclear factor κB activation was strongest in cells expressing myristoylated trimeric CD40 cytoplasmic domain. Signal transduction through trimeric CD40 cytoplasmic domains with various point mutations in the TRAF binding sites was similar to signal transduction through analogous full-length receptors. Transiently expressed soluble trimeric CD40 cytoplasmic domain was isolated as complexes that contained TRAF2, TRAF3, TRAF5, TRAF6, and the inhibitor of apoptosis protein (c-IAP1). Association of c-IAP1 with the CD40 cytoplasmic domain complex was indirect and dependent on the presence of an intact TRAF1/2/3 binding site. These results suggest that extracellular ligation of CD40 can be bypassed and that soluble trimerized CD40 complexes can be isolated and used to identify components that link CD40 with signaling pathways.

Signal transduction through the CD40 receptor is initiated by binding of its trimeric ligand and propagated by interactions of tumor necrosis factor receptor-associated factor (TRAF) proteins with the multimerized CD40 cytoplasmic domain. Using defined multimeric constructs of the CD40 cytoplasmic domain expressed as either soluble or myristoylated proteins, we have addressed the extent of receptor multimerization needed to initiate signal transduction and identified components of CD40 signaling complexes. Signal transduction in human embryonic kidney 293 cells, measured by nuclear factor B activation, was observed in cells expressing soluble trimeric CD40 cytoplasmic domain and to a lesser extent in cells expressing dimeric CD40 cytoplasmic domain. Nuclear factor B activation was strongest in cells expressing myristoylated trimeric CD40 cytoplasmic domain. Signal transduction through trimeric CD40 cytoplasmic domains with various point mutations in the TRAF binding sites was similar to signal transduction through analogous full-length receptors. Transiently expressed soluble trimeric CD40 cytoplasmic domain was isolated as complexes that contained TRAF2, TRAF3, TRAF5, TRAF6, and the inhibitor of apoptosis protein (c-IAP1). Association of c-IAP1 with the CD40 cytoplasmic domain complex was indirect and dependent on the presence of an intact TRAF1/2/3 binding site. These results suggest that extracellular ligation of CD40 can be bypassed and that soluble trimerized CD40 complexes can be isolated and used to identify components that link CD40 with signaling pathways.
The tumor necrosis factor (TNF) 1 receptor superfamily member CD40 is expressed primarily by professional antigen-presenting cells and plays a critical role in costimulation and antigen-presenting cell activation in T cell-dependent immune responses (1,2). Signals generated through CD40 in B cells are antiapoptotic and are required for T cell-dependent B cell ac-tivation and proliferation, isotype switching, up-regulation of costimulatory receptors, germinal center formation, and memory generation (1,3,4). Signal transduction through CD40 is initiated by binding trimeric CD40 ligand on the surface of activated T cells (1,5). Signaling pathways highly activated by CD40 engagement in B cells include NF-B and the mitogenactivated protein kinases p38 and c-Jun N-terminal kinase (6 -10).
Activation of CD40-dependent signaling pathways is thought to be mediated primarily by recruitment of several TRAF protein family members to the multimerized CD40 cytoplasmic domain (for review, see Ref. 11). The 62-amino acid human CD40 cytoplasmic domain (CD40c) contains two linear TRAF binding sites, a membrane proximal site that binds TRAF6 and a membrane distal site that directly binds TRAF1, TRAF2, and TRAF3 (12)(13)(14)(15). The conserved C-terminal half of the TRAF proteins, designated the TRAF domain, mediates interactions with multimerized receptor cytoplasmic domains (11,16,17).
Crystal structures of the CD40 ligand trimer (5) and of the LT␣⅐TNFR1 and TRAIL⅐DR5 receptor complexes (18,19) suggest that CD40 forms at least a trimeric complex upon binding its ligand. Recent x-ray crystallographic structures of TRAF2⅐CD40 and TRAF3⅐CD40 peptide complexes also demonstrate that the TRAF domains of TRAF2 and TRAF3 form trimers where the geometrical spacing of the CD40c peptides closely matches that of the extracellularly ligated CD40 receptor (20 -22). Biochemical experiments suggest that the requirement for CD40c trimerization in the recruitment of TRAF proteins is solely avidity-driven. As an alternative to dimerization, receptor trimerization may regulate initiation of CD40 signaling by providing a higher degree of discrimination between liganded and unliganded receptors (22)(23)(24). Differential TRAF recruitment and distinct cellular signaling outcomes depend on the extent of higher order CD40 multimerization (23,25,26). However, minimum multimerization requirements for CD40c to initiate signaling have not been defined in cells.
The TRAF domain of TRAF2 has been reported to associate with a number of signaling and apoptosis-regulating proteins including NIK, RIP, RIP2, GCK, ASK-1, and Casper (27)(28)(29)(30)(31)(32). Because the association of many of these signaling proteins in native receptor complexes with TRAFs has been confirmed in systems in which the TRAFs or signaling proteins are overexpressed, components in native receptor complexes have not been well characterized. Additionally, the inhibitor of apoptosis proteins, c-IAP1 and c-IAP2, were identified as components of TNFR2 complexes which bound to the receptor via TRAF1 and TRAF2 (33). In particular, the formation of native TRAF⅐CD40 complexes is well documented (34,35), but components that potentially link TRAF⅐CD40 complexes with antiapoptotic and other primary signaling pathways are only beginning to be elucidated (36).
We present here a system for defining the minimal multimerization requirements for initiating CD40 signal transduction and for isolating and characterizing components in CD40 * 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  signaling complexes. Soluble or myristoylated stable monomers, dimers, or trimers of CD40c were expressed in human embryonic kidney (HEK) 293 cells, and their ability to initiate signal transduction was assessed. Expression of soluble trimeric CD40c in HEK 293 cells was sufficient to activate NF-B. NF-B activation through soluble or myristoylated trimeric CD40c was similar to that found through full-length CD40 receptors. Transiently expressed, soluble, trimeric CD40c could be isolated as complexes with TRAF proteins and c-IAP1. This system establishes a means to study endogenous CD40 signaling intermediates which will provide insight into the mechanisms by which multimerized CD40 receptors propagate signals.
Transfection and Immunoblot Analyses-HEK 293 cells (3 ϫ 10 6 or 6 ϫ 10 6 ) were seeded in 75-cm 2 or 225-cm 2 tissue culture flasks, grown overnight, and transfected with Superfect-DNA complexes according to the manufacturer's protocol. The amount of each pcDNA3.1 plasmid DNA used for each transfection was adjusted to achieve equivalent levels of protein expression. The cells were harvested by scraping either 18 h or 36 h post-transfection, lysed on ice in 0.5 ml of 200 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.02 mg/ml pepstatin A, 0.02 mg/ml leupeptin, and 0.1 mM phenylmethylsulfonyl fluoride (lysis buffer) for 30 min, and centrifuged at 13,000 ϫ g. Lysates were precleared with 20 l of protein A/G-agarose (1 h on ice) prior to immunoprecipitation of the dimeric and trimeric CD40 molecules. The dimeric and trimeric coiled coils have identical amino acids in the solvent-exposed regions (39) and are both recognized by the anti-IZIP monoclonal antibody. Anti-IZIP IgG (1 g) and 40 l of protein A/Gagarose beads were added to the lysates and incubated overnight at 4°C with rotation. The beads were washed four times in lysis buffer and eluted in 30 l of SDS sample buffer containing 40 mM dithiothreitol at 95°C. The samples were resolved on 10% NuPage polyacrylamide Bis-Tris gels and electrotransferred to polyvinylidene difluoride membranes. Membranes were blocked with 1 ϫ TBST containing 1 ϫ Genosys blocking buffer and 5% w/v sucrose, and the CD40/2, MCD40/2, CD40/3, and MCD40/3 proteins were detected by incubation with anti-CD40 (C-20) (1/200), anti-rabbit IgG alkaline phosphatase (1/200), and Western Blue alkaline phosphatase substrate solution.
NF-B Luciferase Reporter Assay-HEK 293 cells (1 ϫ 10 6 ) were seeded in 60-mm dishes and grown overnight in Dulbecco's modified Eagle's medium and transfected with Superfect-DNA complexes. For each transfection, 0.5 g of pNF-B-Luc and the indicated amounts of pcDNA3.1/CD40 plasmids were used to achieve equal levels of CD40c protein expression. The total amount of DNA transfected was 2 g. Cells were lysed 24 h after transfection and assayed for luciferase activity with the Promega luciferase assay system.
Immunofluorescence-Purified anti-IZIP mouse monoclonal IgG1, 5D10, was biotinylated (40). HEK 293 cells (1 ϫ 10 5 ) were grown on gelatin-coated coverslips and transfected with pcDNA3.1, pcDNA3.1/CD40/3, or pcDNA3.1/MCD40/3 using FuGENE 6 transfection reagent. After 24 h the cells on the coverslips were fixed with 4% paraformaldehyde in PBS and permeabilized in PBS containing 1% bovine serum albumin and 0.5% Triton X-100 for 10 min at room temperature. Coverslips were incubated with 1 g/ml biotinylated anti-IZIP for 1 h at room temperature, washed three times with PBS, and incubated with 0.5 g/ml streptavidin-rhodamine for 45 min at room temperature in the dark. Coverslips were washed three times with PBS and mounted with GEL/MOUNT. Slides were examined under a Nikon E-800 microscope using oil immersion and either a 40ϫ or 100ϫ objective and photographed.
AntpCD40/3 Treatment-Overlapping oligonucleotides encoding helix 3 from the Drosophila melanogaster antennapedia protein which contained a proline substitution (MNRQIKIWFPNRRMKWKK) were fused to the 5Ј-end of CD40/3 at the NcoI site in pET23d (23). The resulting AntpCD40/3 protein was expressed in Escherichia coli strain BL21DE3pLysS, and lysates were prepared as described previously (12). Protein in clarified extracts (150,000 ϫ g for 10 min) was precipitated with 40% ammonium sulfate, suspended in 20 mM HEPES, pH 7.5, 150 mM NaCl, 1 mM dithiothreitol, 10% v/v glycerol, 0.1 mM EDTA, 0.1 mM EGTA, and 1 mM phenylmethylsulfonyl fluoride. AntpCD40/3 protein was purified (ϳ90%) by chromatography on a Resource S15 column. The identity of the protein was verified by tryptic digestion and mass spectrometry.

Expression of Multimerized CD40c in HEK 293
Cells-Engagement of CD40 with its trimeric ligand is expected to form intracellularly trimerized CD40 cytoplasmic domains. To ascertain the minimal multimeric state required for initiating CD40 signal transduction, we bypassed ligand binding by constructing monomeric and multimeric CD40c from the 62-amino acid cytoplasmic domain of human CD40. Our previous bio-chemical studies on purified CD40c monomers and trimers defined their stable multimeric states and demonstrated that trimeric CD40c has a significantly higher affinity than monomeric CD40c for TRAF2 (23). Dimeric and trimeric CD40c were constructed as N-terminal fusion proteins to stable dimeric and trimeric ␣-helical coiled coils derived from the yeast GCN4 transcription factor (23,39). A second set of monomeric, dimeric, and trimeric CD40c constructs was produced by fusing the 9-amino acid Src myristoylation sequence (37) to the N termini. Myristoylation of this sequence is predicted to localize the molecules to the plasma membrane. Multimeric CD40c constructs were transiently expressed in HEK 293 cells for 24 h, and cell lysates were immunoprecipitated with a monoclonal antibody that recognized both the dimeric and trimeric forms of the ␣-helical coiled coil (anti-IZIP; Fig. 1). Equivalent levels of protein expression were achieved by adjusting the amount of DNA used in each transfection (Fig. 1). The expression levels were not altered upon exchanging the Src myristoylation sequence with an alternative myristoylation signal, MGKS (data not shown).
NF-B Activation Requires CD40c Multimerization-To determine whether monomeric, dimeric, or trimeric CD40c was capable of activating signaling pathways, each construct was cotransfected into HEK 293 cells with an NF-B luciferase reporter plasmid, and luciferase activity was quantified after 24 h. Expression of equivalent levels of dimeric and trimeric CD40c forms was confirmed by analyses as in Fig. 1. The highest amount of plasmid DNA used was also chosen as the amount of monomeric CD40c plasmid DNA to use for each transfection. Expression of monomeric soluble CD40c did not result in an increase in NF-B activation. In contrast, expression of the soluble trimeric CD40c resulted in an ϳ8-fold increase in NF-B activation over expression of empty vector controls, whereas soluble dimeric CD40c afforded a ϳ4-fold increase in NF-B activation over background ( Fig. 2A). Expression of myristoylated trimeric CD40c resulted in a larger increase in NF-B activation compared with soluble trimeric CD40c. Myristoylated trimeric CD40c containing the MGKS myristoylation sequence produced a 46-fold increase in NF-B activation over empty vector control (Fig. 2B). A 250-fold increase in NF-B activation was observed upon expression of myristoylated trimeric CD40c containing the Src myristoylation sequence (Fig. 2C). NF-B activation observed in cells expressing myristoylated monomeric or dimeric CD40c was weak (Fig. 2, B and C). In two independent time course experiments, the soluble and myristoylated CD40c constructs each demonstrated maximal signaling between 18 and 42 h after transfection (data not shown). These results suggest that in HEK 293 cells, trimerization of the CD40 cytoplasmic domain is sufficient to activate the NF-B signaling pathway and that membrane-proximal CD40c trimerization provides maximal NF-B activation.
Localization of Soluble and Myristoylated Trimeric CD40c-To localize the expressed trimeric CD40c proteins, HEK 293 cells were transfected with either soluble or myristoylated (Src) CD40c trimers, permeabilized, stained with biotin-anti-IZIP and streptavidin-rhodamine, and analyzed for fluorescence. HEK 293 cells expressing soluble trimeric CD40 displayed fluorescence distributed evenly throughout the cytoplasm with increased uniform fluorescence in the cell nucleus (Fig. 3A). In contrast, HEK 293 cells expressing myristoylated trimeric CD40c exhibited intense fluorescence in the cytoplasm and finely defined plasma membrane processes, with minimal expression in the nucleus (Fig. 3B). Although a fraction of myristoylated trimeric CD40c appeared to be in the cell cytoplasm, a significant portion was also localized to the plasma membrane. This dual distribution of CD40c containing the Src myristoylation sequence may result from the high expression levels obtained in the transient expression system coupled with a limitation in the efficiency of myristoylation of overexpressed proteins (37). Cells transfected with empty expression vector and stained with biotin-anti-IZIP exhibited background fluorescence (Fig. 3C). These results suggest that myristoylation enhances signaling by CD40c trimers (Fig. 2) by promoting their localization to the cell membrane.

Altered Signal Transduction through Soluble and Myristoylated Trimeric CD40c Containing Mutations in TRAF Binding
Sites-Point mutations in the two TRAF binding sites in CD40c have been characterized previously for their effects on CD40 signaling (35,38,42,43). Mutations in CD40c which greatly reduce interactions with either TRAF1/2/3(T254A) or TRAF6(P233G/E235A) significantly reduced but did not eliminate NF-B activation by full-length CD40 receptor in stable cell lines (35,38,43). To test whether the signaling responses of the trimeric CD40c were similar to full-length CD40 receptor, the above point mutations were introduced into the soluble or myristoylated trimeric CD40c constructs to produce TRAF1/ 2/3 Ϫ , TRAF6 Ϫ , or TRAF1/2/3 Ϫ TRAF6 Ϫ proteins. The mutant trimeric CD40c constructs were transiently expressed in HEK 293 cells cotransfected with the NF-B luciferase reporter plasmid. Cells expressing each trimeric CD40c mutation showed significantly reduced ability to activate NF-B relative to cells expressing the wild type sequence (Fig. 4). Expression of the trimeric CD40c containing mutations in both TRAF binding sites (T254A/P233G/E235A) resulted in NF-B activation lower than that obtained with each mutation individually. Cells expressing the T254A/P233G/E235A mutations still showed NF-B activity slightly above background levels of cells transfected with empty vector (Fig. 4). Relative NF-B activation in cells expressing the soluble CD40c (Fig. 4A) and myristoylated CD40c (Fig. 4C) mutant trimers was similar, with the exception of the TRAF6 Ϫ mutants. Eliminating TRAF6 binding resulted in a more profound reduction in NF-B activation by the myristoylated trimeric CD40c (Fig. 4C). Equivalent levels of wild type and mutant protein expression were confirmed by immunoprecipitation and immunoblotting of transiently transfected HEK 293 cells (Fig. 4, B and D).
Another mutation (N237D) that has been shown to confer increased affinity for TRAF6 (38) was also introduced into soluble and myristoylated trimeric CD40c. Cells expressing intact CD40(N237D) have been shown to exhibit enhanced ligand-dependent NF-B and c-Jun N-terminal kinase activation (38). As expected from these data, HEK 293 cells transiently transfected with soluble or myristoylated trimeric CD40c(N237D) and the NF-B luciferase reporter plasmid exhibited enhanced NF-B activation relative to cells expressing the wild type trimeric CD40c (Fig. 4).
The relative levels of NF-B activation mediated by the mutant trimeric CD40c constructs in transiently transfected HEK 293 cells correlated well with the ligand-dependent NF-B activation exhibited by mutant full-length CD40 receptors in stable cell lines (38). These results suggest that the trimerized CD40c molecules activate NF-B using the normal intracellular pathway utilized by ligated, full-length CD40.
Soluble Trimeric CD40c Is Complexed with Endogenous TRAFs and c-IAP1-To characterize endogenous proteins that participate in activation of CD40-dependent signaling pathways, soluble trimeric CD40c was transiently expressed in HEK 293 cells for 36 h and isolated from cell extracts. Soluble trimeric CD40c was used for these studies because the protein could be extracted in soluble form by Dounce homogenization of transfected cells in the absence of detergent. 2 In contrast, extraction of the majority of myristoylated trimeric CD40c from transfected cells required detergent, 2 which might also disrupt protein binding to trimeric CD40c. After Dounce homogenization of transfected HEK 293 cells, soluble trimeric CD40c in clarified extracts was immunoprecipitated with anti-IZIP and subjected to immunoblot analysis with antisera that recognized either TRAF1, TRAF2, TRAF3, TRAF5, or TRAF6. Endogenous TRAF2, TRAF3, TRAF5, and TRAF6 were detected in immunoprecipitates from cells expressing soluble trimeric CD40c but not in immunoprecipitates from cells transfected with empty vector (Fig. 5). TRAF1 protein was not detected in the immunoprecipitates but was detected at very low levels in whole cell extracts of HEK 293 cells (data not shown). Analogous experiments with detergent extracts of cells expressing myristoylated (Src) trimeric CD40c showed the presence of endogenous TRAF3 coprecipitating with myristoylated trimeric CD40c (data not shown). Anti-IZIP immunoprecipitates of soluble trimeric CD40c were immunoblotted additionally with antisera to c-IAP1 and c-IAP2. Endogenous c-IAP1, but not c-IAP2, was found associated with soluble trimeric CD40c (Fig. 5). c-IAP1 and c-IAP2 protein expression was readily detected in whole cell lysates of HEK 293 cells (data not shown). Additionally, immunoprecipitates were immunoblotted with antibodies to ASK1, GCK, RIP, or RIP2, molecules that have been demonstrated by several methods to interact with TRAF2 (27)(28)(29)(30)(31)33). No evidence for the presence of these endogenous proteins in the soluble trimeric CD40c immunoprecipitates was obtained with the antisera. ASK1, GCK, RIP, and RIP2 were readily detectable in whole cell lysates of HEK 293 cells (data not shown). These results demonstrate that soluble trimeric CD40c expressed in HEK 293 cells is complexed with endogenous TRAF proteins and c-IAP1.
Antennapedia-mediated Delivery of Soluble Trimeric CD40c-To examine signaling and protein associations of soluble trimeric CD40c in a relevant cell type, such as a B lymphocyte, peptide-mediated delivery was used. 16 amino acids from the Drosophila antennapedia homeodomain are sufficient to mediate protein uptake into many cell types including B cells (44 -47). An antennapedia homeodomain-related peptide was fused N-terminally to soluble CD40c trimers to produce Ant-pCD40/3. Purified AntpCD40/3 was introduced efficiently in a dose-and time-dependent fashion into DND39 cells, an Epstein-Barr virus negative B lymphocyte cell line. Punctate fluorescent patches were observed in cells treated for 2 h with AntpCD40/3 but not in cells treated with soluble trimeric CD40c (Fig. 7A). The punctate staining of AntpCD40/3-treated cells differed from the even distribution of fluorescence observed in Antp peptide-treated cells (Fig. 7A). Immunoprecipitation of AntpCD40/3 with anti-IZIP from soluble extracts of DND39 cells demonstrated coprecipitation of endogenous TRAF3 (Fig. 7B). Similarly, TRAF3 was detected in immuno- precipitates of the AntpCD40/3-treated monocytic cell line, THP1. 3 Although TRAFs 2 and 6 were not detected by immunoblot analysis (data not shown), a reproducible increase in expression of the CD40-inducible intercellular adhesion molecule 1 protein was detected in AntpCD40/3-treated DND39 B cells. 3,4 Together these results imply that soluble trimeric Ant-pCD40/3 was introduced into B cells and formed functional signaling complexes. DISCUSSION We have designed a cellular system to delineate the biochemical requirements for CD40 receptor multimerization in initiating signal transduction. The system facilitates isolation and characterization of functional CD40 signaling complexes containing endogenous cellular proteins. Defined monomeric, di- meric, and trimeric CD40 cytoplasmic domains that bypass ligand-receptor interaction were produced and evaluated for their effects on NF-B activation and their ability to recruit downstream activators of TNF receptor family signaling pathways. The minimal soluble CD40 multimeric state found to be necessary and sufficient to initiate significant NF-B activation in HEK 293 cells was a trimeric CD40 cytoplasmic domain.
Introduction of myristoylation sequences to facilitate membrane localization of the proteins significantly enhanced the ability of trimeric CD40c to activate NF-B and conveyed a low NF-B activation to dimeric CD40c. Expressed levels of soluble and myristoylated CD40c protein were similar, suggesting that recruitment of downstream signaling mediators to CD40 may be more efficient at the membrane because of enhanced proximity or diffusion of the receptor in the membrane. These findings are consistent with previous studies that have noted an increase in basal NF-B activation in stable cell lines overexpressing full-length CD40 receptor (38). This suggests that a high density of unliganded CD40 may form transient dimers or trimers in the plasma membrane which results in an increased basal activation of the NF-B signaling pathway. These results also imply that extracellular ligand binding may function only to bring three cytoplasmic domains together and is not required for inducing or propagating a conformational change in CD40c.
Based on the trimeric structure of CD40 ligand (5) and the co-crystal structures of LT␣⅐NFR1 and TRAIL⅐DR5 receptor (18,19,48), CD40 trimerization is predicted to result from ligand binding. The geometry of spacing between CD40c peptides in the trimeric structures of TRAF2⅐CD40 and TRAF3⅐CD40 complexes matches the geometry of extracellularly ligated receptor (20 -22), suggesting a minimum requirement for a CD40c trimer in initiating activation of NF-B. However, it is not clear from the static crystal structures or the in vitro biochemistry whether trimerized receptor cytoplasmic domains are sufficient for initiation of cellular responses. Several cellular studies suggest the need for higher order multimerization in initiation of CD40 signaling responses (23,25,26). Interestingly, recent studies using fluorescence resonance energy transfer techniques have proposed a dimeric extracellular domain interaction for TNFR2, Fas, and CD40 which was independent of receptor ligation and was essential for mediating ligand-dependent signaling (49,50). This finding may provide a mechanism by which overexpressed full-length CD40 could increase basal NF-B activity (38). Taken together with our findings that cells expressing CD40c dimers have the ability to activate low levels of NF-B, these results suggest that ligand-independent CD40 dimers form and may serve a specific signaling function. In previous studies, the proportion of CD40 receptors found as dimers was not known but appeared to be a small percentage of total CD40 (49,50). Dynamic formation of CD40 dimers in cells may provide low levels of basal NF-B signaling which could be antiapoptotic or maintain low levels of expression of costimulatory or adhesion molecules. Alternatively, preformed receptor dimers could provide a mechanism for initiating ligand-dependent formation of higher order multimers of trimerized CD40 receptors.
The relative signaling abilities of full-length CD40 receptors with mutations in the two TRAF binding sites have been well studied (35,38,42,43). Point mutations that eliminated TRAF1/2/3 or TRAF6 binding in full-length CD40 receptor reduced but did not eliminate NF-B activation. Cells expressing soluble or myristoylated CD40c trimers containing the same point mutations in the TRAF1/2/3 or TRAF6 binding sites showed comparable reductions in NF-B activation. Additionally, a point mutation that increased TRAF6 binding significantly increased NF-B activation through both intact CD40 receptor and trimeric CD40c. These results suggest that activation of the NF-B signaling pathway in cells expressing engineered CD40c multimers recapitulates signaling through full-length CD40 receptor. It is not clear why the severe defects in CD40 signaling in cells from TRAF6 knockout mice (51) are not reproduced in HEK 293 cells expressing CD40 molecules defective in TRAF6 binding. It is possible that some TRAF6 recruitment to mutant CD40 could occur indirectly through a complex containing TRAF6 and other TRAF molecules.
Soluble trimeric CD40c could be isolated by immunoprecipitation of cellular homogenates made in the absence of detergents. Previous studies examining association of intracellular proteins with the CD40 cytoplasmic domain have required the receptor to be solubilized with detergents and have demonstrated recruitment of endogenous TRAF2 and TRAF3 to crosslinked, full-length CD40 (34,52). In HEK 293 cells, we also found endogenous TRAF2 and TRAF3 associated with soluble trimeric CD40c but not with soluble trimeric CD40c containing mutations in the TRAF1/2/3 binding site. This suggests that trimeric CD40c forms a direct complex with these TRAFs. In addition, endogenous TRAF5 and TRAF6 were associated with soluble trimeric CD40c in HEK 293 cells.
The finding that TRAF5 is associated with soluble trimeric CD40c is particularly interesting because human TRAF5 does not associate directly with human CD40 (12,53). Previous coexpression and biochemical experiments have suggested that TRAF3 and TRAF5 associate in cells through their extended coiled coil domains, and the complex is capable of binding to CD40 through TRAF3 (12). Our results are consistent with TRAF5 recruitment to soluble trimeric CD40c in HEK 293 cells through an interaction with TRAF3. TRAF6 has been shown to play an important role in signaling through CD40 as assessed by significant defects in CD40 signaling in TRAF6 knockout mice (51). Although endogenous TRAF6-CD40 association has not been demonstrated previously in cells either because of low intrinsic affinity or low levels of TRAF6 expression, we readily detected interaction of TRAF6 with soluble trimeric CD40c. Thus, it is likely that the repertoire of TRAF proteins associated with the soluble trimeric CD40c in HEK 293 cells reflects TRAF-CD40c affinity as well as the level of TRAF protein expression. Together these results indicate that an endogenous repertoire of TRAF proteins associates with soluble trimeric CD40c as a population of stable CD40 signaling intermediates.
To examine endogenous proteins associated with soluble CD40 signaling intermediates in a more relevant cell type, we introduced soluble CD40 trimers directly into a B cell line. The very low transient transfection efficiency of B cell lines precluded transient expression of soluble CD40 constructs. Cellular uptake of soluble CD40 trimers into DND39 human B cells was demonstrated to be mediated by a previously characterized sequence from D. melanogaster Antennapedia protein (44 -46). As indicated by the punctate pattern of AntpCD40/3 staining, these molecules were found in clusters after cellular uptake. A functional significance of the AntpCD40/3 clusters and possible colocalization with other subcellular structures remains to be characterized. Peptide-delivered AntpCD40/3 was shown to be complexed with TRAF3, indicating that soluble CD40 trimers also formed signaling intermediates in B cells. The inability to detect TRAF2 or TRAF6 association with AntpCD40/3 in DND39 cells may reflect low intracellular expression of these TRAF proteins. Previous studies have suggested that IL-4 significantly enhances TRAF2 association with CD40 in DND39 cells and may in addition be required for signaling through TRAF2 (34).
An endogenous protein not previously found associated with intact CD40 was the cellular inhibitor of apoptosis protein, c-IAP1. c-IAP1 was initially identified, along with c-IAP2, as a component of TNFR2 complexes which bound to the receptor via TRAF1 and TRAF2 (33) and has subsequently been demonstrated to be capable of directly inhibiting activation of caspases (54 -56). Our studies demonstrated that c-IAP1 associates with soluble, trimeric CD40c. This complex requires the presence of an intact TRAF1/2/3 binding site in CD40, implying that c-IAP1 association with CD40c is also indirect. In contrast to c-IAP association with TNFR2, the presence of TRAF1 was not required for endogenous c-IAP1 association with soluble CD40c signaling complexes. Engagement of CD40 on B cells has been demonstrated to induce antiapoptotic signals that include activation of NF-B and up-regulation of Bcl-x L (57)(58)(59). In addition, c-IAP1 and c-IAP2 are induced by NF-B and suppress apoptosis (54). The current findings suggest that endogenous c-IAP1 may contribute directly to antiapoptotic effects of CD40 signaling.
A number of studies in cells have characterized interactions of signaling molecules with TRAF proteins that could link TRAF complexes to downstream signaling pathways. These include the kinases ASK1 (31), GCK (30), RIP (28), and RIP2 (29). We failed to detect these endogenous proteins associated with soluble trimeric CD40c in HEK 293 cells. Because we have verified by immunoblot analyses that HEK 293 cells express each of these proteins, 2 a trivial explanation is that levels of endogenous kinases associated with CD40 signaling intermediates were below the level of detection for each antiserum. Alternative possibilities include competition between the various signaling or regulatory proteins (e.g. c-IAP1) and TRAF⅐receptor complexes, association of kinases with TRAF⅐receptor complexes may be transient, or these components may be down-regulated following signaling. The composition of the TRAF⅐CD40 complexes may also differ in different cell types, depending on expression levels of the various signaling proteins.
Signaling through different TNF receptors produces distinct outcomes, and it is thought that this may reflect a different hierarchy of affinities of the TRAF proteins for different receptors (23,24) resulting in a receptor-specific repertoire of associated TRAFs. It is also possible that some specificity in the association of signaling proteins with TRAF⅐receptor complexes could be provided by the receptor cytoplasmic domain. The recently determined structure of a TRAF3⅐CD40 peptide complex suggests that binding to different TRAFs can result in structuring of parts of CD40c (22). Supporting this idea is the finding that specific kinases have been demonstrated to associate with TRAF proteins in the context of particular receptors. For example, RIP has been shown to link TNFR1 with TRAF2 to activate the NF-B pathway (60, 61) and ASK1 has been found associated with TRAF2 following TNF treatment of cells (31). Thus, there is no a priori reason to expect these components to be recruited to TRAFs following CD40 ligation of cells if the associations are specific to TNFR1 signaling.
It is possible that TRAF proteins and links to downstream signaling pathways could be utilized in a variety of combinations with different receptor cytoplasmic domains. Depending on the cellular context, the composition and function of a CD40 signaling complex may depend on the expression levels of TRAF proteins, the expression and activation states of other TRAF-interacting receptors, and the repertoire of expression of TRAF-binding proteins in a given cell type. Future directions are focused on utilizing this system to assess the cellular contexts of signaling intermediates of TNF receptors in different cell types.