Specificity of TRAF3 in Its Negative Regulation of the Noncanonical NF-κB Pathway*

Tumor necrosis factor (TNF) receptor-associated factors (TRAFs) are critical signaling adaptors downstream of many receptors in the TNF receptor and interleukin-1 receptor/Toll-like receptor superfamilies. Whereas TRAF2, 5, and 6 are activators of the canonical NF-κB signaling pathway, TRAF3 is an inhibitor of the noncanonical NF-κB pathway. The contribution of the different domains in TRAFs to their respective functions remains unclear. To elucidate the structural and functional specificities of TRAF3, we reconstituted TRAF3-deficient cells with a series of TRAF3 mutants and assessed their abilities to restore TRAF3-mediated inhibition of the noncanonical NF-κB pathway as measured by NF-κB-inducing kinase (NIK) protein levels and processing of p100 to p52. We found that a structurally intact RING finger domain of TRAF3 is required for inhibition of the noncanonical NF-κB pathway. In addition, the three N-terminal domains, but not the C-terminal TRAF domain, of the highly homologous TRAF5 can functionally replace the corresponding domains of TRAF3 in suppression of the noncanonical NF-κB pathway. This functional specificity correlates with the specific binding of TRAF3, but not TRAF5, to the previously reported TRAF3 binding motif in NIK. Our studies suggest that both the RING finger domain activity and the specific binding of the TRAF domain to NIK are two critical components of TRAF3 suppression of NIK protein levels and the processing of p100 to p52.

Tumor necrosis factor (TNF) receptor-associated factors (TRAFs) are critical signaling adaptors downstream of many receptors in the TNF receptor and interleukin-1 receptor/Tolllike receptor superfamilies. Whereas TRAF2, 5, and 6 are activators of the canonical NF-B signaling pathway, TRAF3 is an inhibitor of the noncanonical NF-B pathway. The contribution of the different domains in TRAFs to their respective functions remains unclear. To elucidate the structural and functional specificities of TRAF3, we reconstituted TRAF3-deficient cells with a series of TRAF3 mutants and assessed their abilities to restore TRAF3-mediated inhibition of the noncanonical NF-B pathway as measured by NF-B-inducing kinase (NIK) protein levels and processing of p100 to p52. We found that a structurally intact RING finger domain of TRAF3 is required for inhibition of the noncanonical NF-B pathway. In addition, the three N-terminal domains, but not the C-terminal TRAF domain, of the highly homologous TRAF5 can functionally replace the corresponding domains of TRAF3 in suppression of the noncanonical NF-B pathway. This functional specificity correlates with the specific binding of TRAF3, but not TRAF5, to the previously reported TRAF3 binding motif in NIK. Our studies suggest that both the RING finger domain activity and the specific binding of the TRAF domain to NIK are two critical components of TRAF3 suppression of NIK protein levels and the processing of p100 to p52.
Tumor necrosis factor (TNF) 5 receptor-associated factors (TRAFs) are evolutionarily conserved proteins found in mam-mals as well as other organisms, including Drosophila melanogaster and Caenorhabditis elegans (1,2). In mammalian cells, TRAFs 1-6 associate either directly or indirectly with many members of the TNF receptor and the interleukin 1/Toll-like receptor superfamilies (3). TRAFs are thought to modulate a number of biological processes mediated by TNF receptors, ranging from cell survival and bone metabolism to immune responses through activation of intracellular signaling pathways such as NF-B and the c-Jun N-terminal kinase pathways (4,5). In addition, TRAF proteins have been implicated in interferon response against viral infections (6 -8).
Among the TRAF family of proteins, TRAFs 2, 5, and 6 are activators of the canonical NF-B pathway (9 -12). The canonical NF-B pathway involves activation of the IB kinase complex, leading to the degradation of IB␣, nuclear translocation of predominately the p50 -p65 transcription complex, and activation of genes with pleiotropic roles in cell survival and immune and inflammatory responses (13,14). Various domains in TRAF proteins have been characterized as important for activation of the canonical NF-B pathway. The conserved TRAF domain at the C terminus of all TRAF members is critical for association with cytoplasmic tails of receptors, intracellular signaling proteins, and the formation of homo-or heterodimers (15,16). The RING finger domain at the N terminus of TRAF2 and 6 has been shown to act as an E3 ubiquitin ligase that is critical for induction of the canonical NF-B signaling pathway (16 -18). Interestingly, TRAF1, the only TRAF member lacking the RING domain, functions as a feedback inhibitor of the canonical NF-B pathway activated by tumor necrosis factor receptor 2 (19). Moreover, the zinc fingers of TRAF2 and 5 have also been suggested to contribute to the activation of the canonical NF-B signaling pathway (16,20).
Despite sharing structural domains similar to TRAFs 2, 5, and 6, TRAF3 does not activate the canonical NF-B pathway (9,10,21). Recent studies revealed that, in fact, TRAF3 acts as an inhibitor of the noncanonical NF-B pathway (22)(23)(24). TRAF3 has been shown to associate with TNF receptors (e.g. BAFFR, CD40, LT␤R, RANK, CD30, and Fn14) that are activators of the noncanonical NF-B pathway (23,(25)(26)(27). Overexpression of TRAF3 inhibits p100 processing activated by these receptors (23). In agreement with these findings, TRAF3 Ϫ/Ϫ cells exhibit constitutive p100 processing to p52, and more importantly the early postnatal lethality of TRAF3 Ϫ/Ϫ mice was rescued by the compound deletion of the p100 gene (22). Thus, TRAF3 plays an essential role in regulating the noncanonical NF-B activity in vivo. * This work was supported in part by National Institutes of Health Grants RO1 AI056154, RO1 CA87924, and R01 GM57559. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Activation of the noncanonical NF-B pathway requires NF-B-inducing kinase (NIK) to induce the processing of p100 (or NF-B2) to p52 (28). Then p52-associated NF-B complexes, predominantly p52-RelB heterodimers, are translocated into the nucleus and activate transcription of genes with preferential functions in lymphoid organogenesis (14,28). The mechanism by which TRAF3 negatively regulates the noncanonical NF-B pathway is unclear. Several studies have suggested that TRAF3 suppresses p100 processing by inducing NIK degradation. TNF receptor activation of the noncanonical NF-B pathway may involve inhibition of this TRAF3 function, allowing for an increase in NIK protein levels and resulting in processing of p100 to p52 via the proteasome (22,24).
In this study, we explored the structural and functional specificities between TRAF3 and its close homologue TRAF5 in suppressing NIK protein levels and p100 processing to p52. We reconstituted TRAF3 Ϫ/Ϫ mouse embryonic fibroblasts (MEFs) with a series of TRAF3 mutants including TRAF3/5 chimeric molecules and then examined TRAF3-mediated suppression of p100 processing. We report that both the RING finger and TRAF domains of TRAF3 are essential for its negative regulation of NIK protein levels and p100 processing. In addition, the TRAF domain of TRAF3 confers its specificity in suppression of the noncanonical NF-B pathway by its association with the TRAF3 binding motif (T3BM) within NIK.
Constructs-GST-NIK(T3BM) was constructed as previously described (29). Briefly, synthetic DNA oligos corresponding to amino acid residues 73-94 of human NIK (GenBank TM accession number Y10256) containing the TRAF3 binding motif were annealed and ligated into the BamH1 and EcoRI sites of pGEX-2T. Murine TRAF3 (GenBank TM accession number U21050), TRAF5 (GenBank TM accession number D78141), and mutant constructs were generated as described previously (8). Briefly, TRAF3 deletion mutants were cloned into pBABE-puromycin (pBABEpuro) or pBABE-puro-TAP (Tandem Affinity Purification tag) retroviral vector with an N-terminal FLAG tag. TRAF3/TRAF5 chimeras were constructed using standard overlapping PCR techniques and cloned into pBABEpuro-TAP. Double point mutants of TRAF3 were constructed using a QuikChange kit (Stratagene) and the pBABE-puro-TAP-TRAF3 vector as a template.
GST Pulldown Assay-GST pulldown assays were performed as previously described (30). Glutathione beads (Sigma Aldrich) were incubated with Escherichia coli-expressed GST alone or FIGURE 1. The N terminus of TRAF3 is required for negative regulation of p100 processing. A, diagram of full-length TRAF3 and N terminus truncation mutant constructs. The numbering below each domain indicates the predicted amino acid residues for the RING finger, zinc fingers, isoleucine zipper, and TRAF domains. B, TRAF3 Ϫ/Ϫ MEFs were reconstituted with the pBABEpuro-TAP-FLAG or pBABEpuro-TAP-FLAG-TRAF3 constructs described in panel A. Basal p100 processing to p52 and expression of each TRAF3 construct in whole cell extracts were detected by immunoblotting. * indicates expression of each FLAG-tagged TRAF3 construct. Total ␤ actin is shown as a loading control.
GST-NIK(T3BM) for 2 h. Glutathione beads were then washed and incubated for 1.5 h with lysates of 293T cells expressing the indicated TRAF3 or TRAF5 constructs. After washing, proteins were eluted from the beads and separated by 10% SDS-PAGE.

Intact RING Finger and TRAF Domains of TRAF3 Are
Required for Suppression of p100 Processing-Recently, TRAF3 has been shown to be an essential negative regulator of p100 processing to p52 (22). To map the domain(s) in TRAF3 that are required for this function, we generated TRAF3 mutants with successive truncations at the N terminus (Fig. 1A). These TRAF3 mutant constructs were cloned into the low expression pBABE retroviral vector and were transduced into TRAF3 Ϫ/Ϫ MEFs. We then assessed the ability of TRAF3 mutants to suppress noncanonical NF-B activity by immunoblotting for p100 and p52 protein in the reconstituted TRAF3 Ϫ/Ϫ MEFs. As seen in Fig. 1B, whereas wild-type MEFs showed low basal levels of p52, TRAF3 Ϫ/Ϫ MEFs showed constitutive, high levels of p100 processing to p52. Reconstitution of TRAF3 Ϫ/Ϫ MEFs with the full-length TRAF3 restored basal p100 processing levels. However, reconstitution with the N-terminal deletion mutants of TRAF3 failed to reduce the high basal p52 levels in TRAF3 Ϫ/Ϫ MEFs to wild-type levels. All these truncation mutants lack the N-terminal sequence (amino acid residues 1-107) of TRAF3, indicating that this region is required for the negative regulation of p100 processing.
Amino acid residues 1-107 of TRAF3 contain a RING finger domain that is thought to form a conformation necessary for zinc chelation. To determine whether an intact RING finger structure is required for TRAF3 to suppress p100 processing, we mutated the highly conserved cysteines and a histidine to alanines (C52A/C55A and C67A/H69A) to generate two point mutations in TRAF3 that disrupt the zinc-chelating positions of the RING finger ( Fig. 2A). As shown in Fig. 2B, TRAF3 Ϫ/Ϫ MEFs reconstituted with either TRAF3 (C52A/C55A) or TRAF3 (C67A/H69A) showed high basal p52 levels, like the TRAF3 Ϫ/Ϫ MEFs reconstituted with the vector control. Thus, these two RING finger mutants could not restore the TRAF3mediated suppression of basal p100 processing to p52 to wildtype levels. Collectively, these data suggest that proper folding of the RING finger domain is critical for TRAF3-mediated suppression of the noncanonical NF-B pathway.  Previously, TRAF3 association with NIK via the TRAF domain was shown to be critical for suppression of p100 processing in a 293T cell line under conditions of overexpression (24). To further determine the role of the TRAF domain of TRAF3 in suppression of p100 processing, we reconstituted TRAF3 Ϫ/Ϫ MEFs with a TRAF3 construct lacking the TRAF domain (Fig. 3A, T3(1-300)). As seen in Fig. 3B, TRAF3 Ϫ/Ϫ MEFs reconstituted with the T3(1-300) mutant are unable to reduce the high basal p52 levels to the wildtype levels, indicating that this region containing the NIK binding site is indeed required for TRAF3mediated suppression of p52 activity.
Specificity of TRAF3 in Negative Regulation of p100 Processing-Among all the TRAF family members, TRAF5 shares the same domain organization as TRAF3 (1). However, TRAF5 cannot functionally replace TRAF3 in suppressing p100 processing using our reconstitution assay (Fig. 4A), implying that different functions exist for the domains in these two TRAF members. To determine which domains of TRAF3 and TRAF5 are functionally interchangeable, we reconstituted TRAF3 Ϫ/Ϫ MEFs with a series of TRAF3/5 chimeric molecules where individual domains of TRAF3 are substituted with the corresponding domain of TRAF5 (Fig.  4B). As shown in Fig. 4C, reconstitution of TRAF3 Ϫ/Ϫ MEFs with the TRAF3 chimeras containing the RING finger, zinc fingers, or isoleucine zipper of TRAF5 could reduce the high basal p100 processing to p52. Interestingly, the TRAF3/5TD chimera containing the TRAF domain of TRAF5 could not suppress the high basal p52 levels in the TRAF3 Ϫ/Ϫ MEFs, indicating that the TRAF domain of TRAF3 is functionally distinct from that of TRAF5 for its ability to negatively regulate p100 processing. To confirm this observation, we replaced the TRAF domain of TRAF5 with the corresponding region of TRAF3 and found that this mutant TRAF5/3TD could now suppress constitutive p100 processing in TRAF3 Ϫ/Ϫ MEFs (Fig. 4C). Thus, our studies have demonstrated that the difference in the functional specificity between TRAF3 and TRAF5 in negatively regulating the noncanonical NF-B pathway is conferred by the C-terminal TRAF domain of TRAF3.
Interaction of the TRAF domain of TRAF3 with the TRAF3 binding motif (amino residues 78 -84) in NIK has previously been demonstrated to be essential for TRAF3 suppression of NIK protein stability (24). Thus, one potential explanation for Basal p100 processing to p52 and expression of TRAF3 or TRAF5 constructs in whole cell extracts were detected by immunoblotting. Total ␤ actin is shown as a loading control. B, diagram of full-length TRAF3, TRAF5, and chimeric molecules generated by domain swapping between TRAF3 and 5. The overall amino acid residue identity between each domain is indicated. T35 RD, TRAF3 residues (1-108) replaced with TRAF5 (residues 1-102); T35 ZF, TRAF3 (residues 108 -279) replaced with TRAF5 (residues 102-270); T35 IZ, TRAF3 (residues 279 -345) replaced with TRAF5 (residues 270 -335); T35 TD, TRAF3 (residues 345-end) replaced with TRAF5 (residues 335-end); and T53 TD, TRAF5 (residues 335-end) replaced with TRAF3 (residues 345-end). C, TRAF3 Ϫ/Ϫ MEFs were reconstituted with pBABEpuro-TAP-FLAG, pBABEpuro-TAP-FLAG-TRAF3, -TRAF5, or TRAF3/5 chimeric constructs described in panel B. Basal p100 processing to p52 and expression each reconstituted TRAF molecule in whole cell extracts were detected by immunoblotting. Total ␤ actin is shown as a loading control. Blots of p100 and p52 were quantitated, and the band intensities were normalized to background. The resulting relative p100 and p52 protein levels were calculated as p52/p100 ratios and presented in the bar graph. the functional specificity between the TRAF domain of TRAF3 and TRAF5 is that TRAF5 cannot associate with NIK at the T3BM. To test this hypothesis, we constructed GST fusion proteins with a 22-amino acid peptide encompassing the T3BM sequence (Fig. 5). We then performed a GST pulldown assay on TRAF3 and TRAF5. As shown in Fig. 5, unlike TRAF3, TRAF5 cannot associate with NIK(T3BM).
The RING Finger and TRAF Domains of TRAF3 Are Required for Suppression of NIK Protein Levels-We have previously demonstrated that TRAF3 Ϫ/Ϫ cells have an accumulation of NIK protein and that NIK is responsible for the constitutive p100 processing in TRAF3 Ϫ/Ϫ MEFs (22). In the current study, we have concluded that both the RING and TRAF domains of TRAF3 are required for suppression of p100 processing to p52. To test whether these domains are also important for suppression of NIK protein levels, we compared NIK expression in TRAF3 Ϫ/Ϫ MEFs reconstituted with the TRAF3 RING finger mutants or the TRAF3/5 chimeras. As shown in Fig. 6A, reconstitution with full-length TRAF3, but not the vector control, suppressed the high NIK levels in TRAF3 Ϫ/Ϫ MEFs. However, the TRAF3 RING finger mutants (C52A/C55A and C67A/ H69A) failed to suppress NIK levels to the same extent as the full-length TRAF3, suggesting that an intact RING finger domain is important for inducing NIK degradation. Similarly, reconstitution of the TRAF3 Ϫ/Ϫ MEFs with the TRAF3/5TD mutant or TRAF5 showed high NIK protein levels comparable with those of the vector control. However, TRAF3 Ϫ/Ϫ MEFs reconstituted with TRAF3/5 chimeras T3/5RD, ZF, IZ, and T5/3TD showed greatly reduced NIK protein levels comparable with the cells reconstituted with the full-length TRAF3 (Fig. 6B). Moreover, NIK expression levels in these reconstituted cells correlated well with the amount of p100 processing to p52 (Fig. 6, A and B, top two panels). Therefore, the RING finger and TRAF domains of TRAF3, which are critical for inhibiting p100 processing, are also important for suppressing NIK protein levels.

DISCUSSION
The structural similarities between TRAF proteins prompt the interesting question of the specificity of each TRAF protein in regulation of downstream signal transduction pathways. A specific example is that TRAF5 activates the canonical NF-B pathway, whereas its closest structural homologue, TRAF3, functions as a negative regulator of the nonca-  . Basal NIK and p100/p52 protein levels and expression of each reconstituted TRAF molecule in whole cell extracts were detected by immunoblotting. Whole cell extracts from NIK ϩ/ϩ and NIK Ϫ/Ϫ MEFs incubated with 25 M MG132 for 2 h were served as a control for NIK expression. Total ␤ actin is shown as a loading control. Blots of p100 and p52 were quantitated, and the band intensities were normalized to background. The resulting relative p100 and p52 protein levels were calculated as p52/ p100 ratios and presented in the bar graph.
nonical NF-B pathway. In this study, we determined that the N-terminal domains of TRAF5, but not the C-terminal TRAF domain, are functionally interchangeable with those of TRAF3 in suppression of NIK protein levels and p100 processing. Furthermore, we clearly showed that an intact RING finger domain of TRAF3 is required for the suppression of the noncanonical NF-B pathway. Therefore, the TRAF and RING finger domains are two important determinants for the role and the specificity of TRAF3 in its negative regulation of the noncanonical NF-B pathway.
By reconstituting TRAF3 Ϫ/Ϫ MEFs with two TRAF3 mutants (C52A/C55A and C67A/H69A), we observed that an intact RING finger domain is required for TRAF3 to suppress p100 processing to p52. The RING finger domain is a common motif found in many E3 ubiquitin ligases, including TRAF2 and TRAF6 (18,31). Thus, it is tempting to speculate that TRAF3 is also an E3 ubiquitin ligase, particularly considering that TRAF3 has been hypothesized to trigger NIK ubiquitination and degradation (24). However, thus far, we and other groups have been unable to demonstrate by in vitro assays that TRAF3 acts as an E3 ubiquitin ligase for NIK (24). Based on the previous work in isolating the TRAF6 ubiquitin ligase complex, there may be additional components that are required for NIK ubiquitination (18,32). In support of this, Grech et al. (33) have reported that TRAF2 Ϫ/Ϫ B cells exhibit constitutive p100 processing to p52, suggesting that TRAF2 is also a negative regulator of the noncanonical NF-B pathway. We have examined whether TRAF2 can functionally replace TRAF3 in the suppression of p100 processing by retrovirally transducing TRAF2 into TRAF3 Ϫ/Ϫ MEFs. As shown in supplemental Fig S1, TRAF3 Ϫ/Ϫ MEFs with overexpression of TRAF2 still display high NIK protein levels and constitutive processing of p100 to p52. Hence, TRAF2 and TRAF3 perform distinct roles in suppressing the noncanonical NF-B pathway. Furthermore, Act1 and TNAP, two other intracellular proteins known to associate with TRAF2 and TRAF3, have been shown to suppress CD40-and NIK-mediated p100 processing (34,35). How TRAF2, Act1, or TNAP negatively regulate the noncanonical NF-B pathway and whether any of these molecules participate in TRAF3-mediated suppression of p100 processing has yet to be elucidated. Moreover, the mechanism by which TNF receptors such as CD40, BAFFR, and LT␤R lead to activation of NIK remains controversial. Some reports suggest that CD40 and BAFF stimulation of B cell lines induces TRAF3 degradation, thereby allowing NIK accumulation and activation of p100 processing (24). However, this observation is inconsistent in the case of LT␤R stimulation of MEFs or CD40 and BAFF stimulation of primary B cells where TRAF3 levels are not significantly reduced upon receptor ligation and induction of p100 processing to p52 (33,36). These discrepancies point to the need for further experimentation to decipher the mechanism by which TRAF3 negatively regulates the noncanonical NF-B pathway during homeostasis as well as during cell activation by TNF receptors.
By reconstituting TRAF3 Ϫ/Ϫ MEFs with TRAF3/5 chimeras, we have determined that the TRAF domain is the structural feature that differentiates the functionality between TRAF3 and TRAF5. In support of the significance and specificity of the TRAF domain of TRAF3, we observed that TRAF3, but not TRAF5, associates with the TRAF3 binding motif in NIK. These data correspond to the study by Liao et al. (24) that showed that the binding of NIK and TRAF3 at the TRAF domain is required for TRAF3 to induce NIK degradation. Together, this evidence suggests that a specific TRAF3 interaction with NIK is critical for regulating basal protein levels of NIK. Moreover, recent studies indicate that TRAF3 is also a positive regulator of the interferon pathway (6 -8). Thus, TRAF3 functions not only as an essential regulator of basal noncanonical NF-B activity but also as a critical mediator of the cellular response to viral infections. Interestingly, similar structural requirements and specificity were observed for TRAF3 in both of these functions (8). Thus, it is unclear what mechanism directs TRAF3 to act as an activator of the type 1 interferon pathway or an inhibitor of the noncanonical NF-B pathway. Future characterization of endogenous TRAF3 complexes involved in these two distinct pathways will provide us with a better understanding of the role of TRAF3 in the antiviral response and could possibly identify novel targets to treat immune disorders and cancers caused by constitutive activation of the noncanonical NF-B pathway (28,37,38).