Regulation of the NF-κB-inducing Kinase by Tumor Necrosis Factor Receptor-associated Factor 3-induced Degradation*

The NF-κB family of transcription factors plays a pivotal role in regulation of diverse biological processes, including immune responses, cell growth, and apoptosis. Activation of NF-κB is mediated by both canonical and noncanonical signaling pathways. Although the canonical pathway has been extensively studied, the mechanism mediating the noncanonical pathway is still poorly understood. Recent studies have identified the NF-κB-inducing kinase (NIK) as a key component of the noncanonical pathway of NF-κB activation; however, how the signaling function of NIK is regulated remains unknown. We report here that one important mechanism of NIK regulation is through its dynamic interaction with the tumor necrosis factor receptor-associated factor 3 (TRAF3). TRAF3 physically associates with NIK via a specific sequence motif located in the N-terminal region of NIK; this molecular interaction appears to target NIK for degradation by the proteasome. Interestingly, induction of noncanonical NF-κB signaling by extracellular signals involves degradation of TRAF3 and the concomitant enhancement of NIK expression. These results suggest that induction of noncanonical NF-κB signaling may involve the rescue of NIK from TRAF3-mediated negative regulation.

The NF-B family of transcription factors plays a pivotal role in regulation of diverse biological processes, including immune responses, cell growth, and apoptosis. Activation of NF-B is mediated by both canonical and noncanonical signaling pathways. Although the canonical pathway has been extensively studied, the mechanism mediating the noncanonical pathway is still poorly understood. Recent studies have identified the NF-B-inducing kinase (NIK) as a key component of the noncanonical pathway of NF-B activation; however, how the signaling function of NIK is regulated remains unknown. We report here that one important mechanism of NIK regulation is through its dynamic interaction with the tumor necrosis factor receptor-associated factor 3 (TRAF3). TRAF3 physically associates with NIK via a specific sequence motif located in the N-terminal region of NIK; this molecular interaction appears to target NIK for degradation by the proteasome. Interestingly, induction of noncanonical NF-B signaling by extracellular signals involves degradation of TRAF3 and the concomitant enhancement of NIK expression. These results suggest that induction of noncanonical NF-B signaling may involve the rescue of NIK from TRAF3mediated negative regulation.
The transcription factor NF-B regulates genes involved in diverse biological processes, such as immune and inflammatory responses, cell growth, and apoptosis (1)(2)(3). NF-B represents a family of dimeric DNA-binding proteins, composed of RelA, RelB, c-Rel, NF-B1 (or p50), and NF-B2 (or p52) (4). NF-B1 and NF-B2 are produced as inactive precursors, p105 and p100, which undergo proteasome-mediated processing to generate the mature proteins, p50 and p52, respectively (4). Although various NF-B complexes have been identified, the predominant forms of NF-B are the heterodimers formed between the Rel proteins and p50 or p52. The NF-B dimers are normally sequestered in the cytoplasm as inactive complexes with a family of ankyrin repeat-containing inhibitors, including IB␣ and related proteins (5). Activation of NF-B can be stimulated by a large variety of cellular stimuli, such as cytokines, mitogens, DNA-damaging agents, and microbial components (6). The canonical pathway of NF-B activation is mediated by a large IB kinase (IKK) 1 complex, composed of two catalytic subunits, IKK␣ and IKK␤, and a regulatory subunit termed IKK␥ (also named NEMO, IKKAP1, or FIP-3) (6). Upon activation by the various NF-B stimuli, IKK phosphorylates IB␣, triggering its ubiquitination and subsequent degradation by the proteasome, which results in rapid and transient nuclear translocation of the active NF-B dimers.
Another mechanism of NF-B regulation is through proteolytic processing of p105 and p100. In addition to serving as the precursors of p50 and p52, p105 and p100 function as IB-like molecules (7)(8)(9). The processing of these precursor proteins not only leads to generation of p50 and p52 but also results in disruption of the IB-like function of these precursor proteins. Recent studies suggest that although the processing of p105 is largely constitutive and cotranslational (10), the processing of p100 is tightly regulated by both positive and negative mechanisms (11). Induction of p100 processing is mediated by a noncanonical NF-B signaling pathway that relies on the NF-B-inducing kinase (NIK) (11) as well as its downstream kinase IKK␣ (12). This novel pathway of NF-B activation does not require IKK␤ and IKK␥ (12)(13)(14), key components of the canonical NF-B signaling pathway (15). In addition to producing p52, the processing of p100 is required for liberating RelB, allowing this specific NF-B member to move to the nucleus and exert its gene regulation function (16 -18). Genetic evidence suggests that the noncanonical NF-B signaling pathway plays a key regulatory role in maturation of B cells and development of lymphoid organs (12, 19 -21). Consistently, this novel pathway of NF-B activation is regulated by a subset of tumor necrosis factor receptor (TNFR) family members known to be important for B-cell function and lymphoid organogenesis, including Baff receptor (BaffR), CD40, and lymphotoxin ␤ receptor (14,17,22,23).
How the different receptors mediate activation of the noncanonical NF-B signaling pathway remains unclear. The TNFRassociated factor (TRAF) proteins may play an important role, since they are known to be recruited to various TNFR family members (24). Interestingly, a common TRAF molecule recruited by the different noncanonical NF-B-stimulating receptors (BaffR, CD40, and lymphotoxin ␤ receptor) is TRAF3 (25)(26)(27), although the role of this adaptor protein in regulation of the noncanonical NF-B pathway remains unclear. In the present study, we show that TRAF3 is a negative regulator of NIK, a key signaling molecule involved in noncanonical NF-B activation. TRAF3 physically interacts with NIK and targets NIK for degradation by the proteasome. Remarkably, the receptormediated activation of noncanonical NF-B signaling is associated with persistent degradation of TRAF3 and the marked elevation of NIK expression. These findings suggest that induction of the noncanonical NF-B signaling involves rescue of NIK from TRAF3-mediated destruction.

MATERIALS AND METHODS
Plasmid Constructs-HA-ubiquitin has been described previously (11,13). HA-NIK was created by cloning the human NIK cDNA, together with a C-terminal HA tag, into the pCMV4 expression vector (28). Truncation mutants of NIK were generated by PCR and designated by the specific amino acid residues retained in the mutant proteins. For example, NIK-(152-947) contains the region from amino acid 152 to 947. NIK mutants harboring internal deletions were produced by site-directed mutagenesis (Stratagene) using wild type NIK expression vector as template. To generate the HA-tagged TRAFs, human TRAF1 and TRAF3 and murine TRAF2 cDNAs were cloned by reverse transcriptase-PCR and inserted into the pcDNA-HA vector (29) downstream of an HA tag. Truncation mutants of TRAF3 were created by PCR. The retroviral expression vectors encoding NIK, NIK⌬78 -84 (NIK harboring deletion of amino acids 78 -84), and TRAF3 were constructed by inserting the corresponding cDNA into the pCLXSN retroviral vector (provided by Dr. Inder M. Verma (30)). The expression vector encoding human p100 was provided by Dr. Warner C. Greene.
Antibodies and Other Reagents-The antibodies for NIK (H248), TRAF3 (H122), horseradish peroxidase-conjugated TRAF3, TRAF2 (C-20), tubulin (TU-02), and RelB (C-19) were purchased from Santa Cruz Biotechnology, Inc. Horseradish peroxidase-conjugated HA monoclonal antibody (3F10) was from Roche Applied Science. Anti-mouse CD40 antibody was purchased from PharMingen. The anti-p100 antibody was kindly provided by Dr. Warner C. Greene. Recombinant IKK␣ was a gift from Dr. Michael Karin. Human recombinant Baff was purchased from BIOSOURCE. The proteasome inhibitor MG132 was purchased from Calbiochem and used at a dose of 25 M.
Yeast Two-hybrid Screening-Yeast two-hybrid screening was performed using the MATCHMAKER LexA two-hybrid system (Clontech). A cDNA fragment encoding the N-terminal 710 amino acids of NIK was cloned into the pLexA vector downstream of the DNA binding domain of LexA to generate the pLexA-NIK-(1-710) bait. This bait was used to screen a Jurkat T cell library constructed using pB42AD vector (human leukemia MATCHMAKER LexA cDNA library; Clontech) following the manufacturer's instructions. Putative positive clones were retransformed to the yeast strain with pLexA-NIK-(1-710) or a nonspecific bait, pLexA-lamin. After this step of false positive elimination, specific clones were subjected to DNA sequencing and subsequent BLAST search analyses.
Cell Culture, Transfection, and Retroviral Infection-M12.4.1 B cell line (31) and its derivative stably transfected with hCD40 (M12-hCD40) (32) were kindly provided by Dr. Gail A. Bishop. The M12-NIK cells were created by infecting the M12-hCD40 cells with the pCLXSN-NIK retroviral vector as previously described (33), and bulk-infected cells were used in the experiments. All of these B-cell lines were cultured in RPMI medium supplemented with 10% fetal bovine serum, 2 mM Lglutamine, antibiotics, and 10 M 2-mercaptoethanol. The kidney carcinoma cell line 293 was cultured in Dulbecco's medium with the same supplements. The 293-NIK, 293-NIK⌬78 -84, and 293-TRAF3 cells were constructed by infecting 293 cells with retroviruses carrying the corresponding cDNAs. Sequential infections were used to generate cells expressing two or more exogenous genes. Transient transfection of 293 cells was carried out in 6-well tissue culture plates using DEAE-dextran (34). The total amount of DNA (expression vectors plus carrier DNA) was 2 g/well, with the amounts of individual expression vectors being adjusted to achieve similar levels of protein expression.
In Vitro Translation-In vitro translation was performed using the TNT coupled reticulocyte lysate system (Promega).
For co-IP assays, whole-cell lysates (ϳ250 g for transfected 293 cells and ϳ500 g for M12 B cells) were diluted to 500 l using radioimmune precipitation buffer and precleared for 30 min with a preimmune serum and protein A-agarose. The precleared cell lysates were then incubated with the indicated antibodies (6 l) for 1 h and with protein A-agarose (30 l) for another 2 h. The agarose beads were washed three times with radioimmune precipitation buffer, and the bound proteins were eluted in 2ϫ SDS loading buffer and subjected to SDS-PAGE and IB analyses.
In Vitro Kinase Assays-In vitro kinase assays were performed as previously described (36). Briefly, NIK and NIK⌬78 -84 were isolated by IP from transiently transfected 293 cells and then incubated with its substrate, recombinant IKK␣ (25 ng), in a kinase buffer containing [␥-32 P]ATP. Phosphorylated IKK␣ as well as autophosphorylated NIK proteins were fractionated by SDS-PAGE and detected by autoradiography.
Cell Labeling and Pulse-Chase Assays-293 cells were transfected using DEAE-dextran. At 30 h posttransfection, the cells were starved for 1 h in Dulbecco's modified Eagle's medium lacking methionine and cysteine and then metabolically labeled for 30 min with 350 Ci/ml [ 35 S]methionine/cysteine. The pulse-labeled cells were either immediately lysed or chased for the indicated times in regular Dulbecco's modified Eagle's medium supplemented with 10 mM cold methionine. The cells were then lysed in radioimmune precipitation buffer and subjected to IP using anti-NIK antibody. The radiolabeled NIK proteins were fractionated by SDS-PAGE and visualized by autoradiography.
For siRNA delivery, 293 cells were seeded into 6-well plates to around 70% confluence. After 12-16 h, the cells were transfected with 140 pmol of siRNA using Oligofectamine (Invitrogen). To maximize the efficiency of RNAi, the transfection was repeated at 24 h following the first transfection. The highest efficiency of RNAi was achieved when the second transfection was carried out using LipofectAMINE 2000 (Invitrogen), in which the siRNA was mixed with 300 ng of carrier DNA (pcDNA). Accordingly, this strategy (an Oligofectamine transfection followed by a LipofectAMINE transfection) was used in this study. At about 30 h after the second transfection, the cells were collected for preparation of protein lysates. The efficiency of RNAi-mediated gene suppression was monitored by IB.
Reverse Transcriptase-PCR-RNA was prepared from the indicated cells and subjected to reverse transcriptase-PCR analysis as described (33) using the following primers: NIK, CAT GCA GAC AGG CAC CCC AC (forward) and CGA ACA ATA TTG AAG TAG CCA GC (reverse); glyceraldehyde-3-phosphate dehydrogenase, CTA CTG GCG CTG CCA AGG C (forward) and GTG GGT GTC GCT GTT GAA GTC (reverse).

Identification of TRAF3 as a Major NIK-associated Protein-
We have previously shown that the N-terminal region of NIK contains regulatory sequences that appear to inhibit the function of NIK (28). This finding prompted us to investigate whether this negative regulatory region associates with any cellular factors involved in NIK regulation. We performed yeast two-hybrid screening using the N-terminal portion of NIK (NIK1-710) as bait. A clone that was frequently isolated encoded the C-terminal portion of TRAF3. Although several other TRAF members have been shown to interact with NIK (37,38), none of them were identified using this bait. This result was consistent with the previous finding that the common TRAFbinding site of NIK is located at its C terminus (37, 38), which was missing in the bait used in our screening.
To confirm the NIK/TRAF3 physical interaction in mamma-lian cells, co-IP was performed using transfected 293 cells. Indeed, NIK and TRAF3 formed a stable complex, which was readily precipitated from the cell lysates by the anti-NIK antibody ( Fig. 1A, top panel, lane 3). Interestingly, parallel IP assays revealed that the interaction of NIK with TRAF3 was markedly stronger than that with TRAF1 and TRAF2 (lanes 1 and 2). This result was not due to the variation in protein expression, since the cell lysates contained even higher amounts of TRAF1 and TRAF2 than TRAF3 (bottom panel). We also performed binding assays using in vitro translated NIK and TRAF3 and demonstrated their strong interaction under such in vitro conditions (Fig. 1B). Together with the yeast two-hybrid results, these findings suggest that TRAF3 is a major partner protein of NIK and that the N-terminal region of NIK may contain a domain specifically mediating its strong interaction with TRAF3.
To map the domain of NIK mediating its interaction with TRAF3, a panel of deletion mutants of NIK was subjected to TRAF3-binding assays. Consistent with the yeast two-hybrid assays, two NIK mutants containing the N-terminal portion (1-736 and 1-710) were capable of TRAF3 association (Fig. 1C,  top panel, lanes 4 and 5). On the other hand, a NIK mutant lacking the N-terminal 151 amino acids (positions 152-947) was largely defective in binding to TRAF3 (lane 3). A NIK mutant harboring an internal deletion of amino acids 30 -120 (⌬30 -120) also exhibited a severe defect in TRAF3 association (lane 2). Additional truncation analyses revealed that the Nterminal 29 amino acids of NIK were dispensable for its TRAF3-binding function (Fig. 1C, upper panel, lanes 8 -10). These results, which are summarized in Fig. 1D, suggest that the N-terminal TRAF3-binding domain of NIK is located between amino acids 30 and 120.
A Specific Sequence Motif of NIK Is Required for Its Strong Interaction with TRAF3-Prior studies suggest that TRAF3 binds to target proteins via a conserved sequence motif, PX-QX(S/T), which is also recognized by TRAF2 and certain other TRAF members (39). However, the TRAF3-binding domain of NIK does not contain such a motif. The preferential binding of NIK to a specific TRAF member, TRAF3, also suggests the involvement of a novel interaction mechanism.
To systematically define the TRAF3-binding motif of NIK, we generated a large panel of NIK mutants harboring various internal deletions in the TRAF3-binding domain ( Fig. 2A). Interestingly, most of the sequences in this domain were dispensable for TRAF3 binding; these include those covering amino acids 31-50 (Fig. 2B, top panel, lane 3), 51-70 (lane 4), and 91-120 (lane 6). However, a region covering amino acids 71-90 was essential for the TRAF3-binding activity of NIK, since removal of this region largely abolished the NIK/TRAF3 interaction (lane 5). More detailed deletion analyses within this region revealed a sequence motif ISIIAQA (located between amino acids 78 and 84; see Fig. 2A) that was critical for the TRAF3-binding function of NIK. Deletion of this sequence motif generated a NIK mutant, NIK⌬78 -84, which was largely defective in TRAF3 binding (Fig. 2B, top panel, lane 9). On the other hand, removal of the flanking sequence downstream of this TRAF3-binding motif (amino acids 85-90) had no effect on the NIK/TRAF3 interaction (lane 10), although removal of the upstream flanking sequence (amino acids 71-77) caused a partial reduction (lane 8). These biochemical analyses, summarized in Fig. 2A, suggest that the strong interaction of NIK with TRAF3 requires a specific sequence motif (ISIIAQA) located in the N-terminal region of NIK.
We then determined the domain within TRAF3 that is required for its binding to NIK. As depicted in Fig. 2C, TRAF3 contains three major domains: a ring finger (RF) domain, a zinc finger (ZF) domain, and a TRAF domain. Co-IP assays were performed to detect the interaction between NIK and TRAF3 mutants lacking the different domains. Deletion of the ring finger domain of TRAF3 had no effect on its interaction with NIK (Fig. 2D, top panel, lane 2; also see summary in Fig. 2C). Further deletion of the zinc finger domain also did not abolish the NIK-binding function of TRAF3 (lane 3). On the other hand, the removal of 120 amino acids from the C terminus of TRAF3, which disrupted the TRAF domain, completely abolished its interaction with NIK (lane 4). Thus, the C-terminal TRAF domain of TRAF3 is required for its interaction with NIK.
Transfected TRAF3 Inhibits the Signaling Function of NIK-To investigate the functional significance of NIK/TRAF3 interaction, we examined the effect of TRAF3 on NIK-induced p100 processing. As expected (11), expression of NIK in 293 cells resulted in the induction of p100 processing, as demonstrated by generation of the processing product p52 (Fig. 3A,  top panel, lane 2). Interestingly, this signaling function of NIK was inhibited by TRAF3 in a dose-dependent manner (lanes 3 FIG. 1. N-terminal region of NIK mediates stable association with TRAF3. A, NIK strongly interacts with TRAF3 but only weakly interacts with TRAF1 and TRAF2. HA-tagged NIK was transfected into 293 cells together with HA-tagged TRAF1, TRAF2, or TRAF3. The NIK protein complex was isolated by IP using anti-NIK followed by detecting the coprecipitated TRAFs by IB using anti-HA antibody (top panel). The amounts of NIK and TRAFs in the cell lysates were monitored by direct IB using anti-HA (middle and lower panels). B, in vitro interaction between NIK and TRAF3. In vitro translated HA-NIK and HA-TRAF3 were incubated for 10 min on ice either alone (lanes 1 and 2) or together (lane 3) followed by IP using anti-NIK. The precipitated NIK and associated TRAF3 were detected by IB using horseradish peroxidaseanti-HA (upper panel). The NIK and TRAF3 proteins in the input materials were analyzed by direct IB (lower panel). C, the N-terminal region of NIK is required for its interaction with TRAF3. HA-tagged TRAF3 was transfected into 293 cells together with the indicated NIK mutants. The physical interaction of TRAF3 with the different NIK mutants was determined by co-IP assays (top panel) as described for A. The expression of NIK mutants and TRAF3 was monitored by direct IB using anti-HA (middle and bottom panels). D, schematic diagram of wild type (WT) form and deletion mutants of NIK. Their TRAF3 binding ability (ϩ) or deficiency (Ϫ) is summarized based on the results presented in C. and 4). A C-terminal truncation mutant of TRAF3, TRAF-(1-423), failed to inhibit the NIK-mediated induction of p100 processing (lanes 6 and 7). Since the TRAF3-(1-423) is defective in NIK binding (see Fig. 2D), it raised the possibility that the TRAF3/NIK interaction is required for the negative regulation of NIK by TRAF3.
To further confirm this idea, we examined the inhibitory effect of TRAF3 on NIK⌬78 -84, a NIK mutant lacking the TRAF3-binding motif (see Fig. 2, A and B). This NIK mutant retained its function in inducing p100 processing (Fig. 3B,  upper panel, lane 5). In fact, it exhibited elevated function compared with the wild type NIK (compare lanes 2 and 5). Importantly, the signaling function of NIK⌬78 -84 was no longer inhibited by TRAF3 (lanes 6 and 7). This result further suggests that the TRAF3/NIK physical interaction is required for TRAF3-mediated negative regulation of NIK. Parallel ki-

FIG. 2. NIK/TRAF3 interaction requires the C-terminal TRAF domain of TRAF3 and a specific N-terminal sequence motif of NIK.
A, schematic diagram of wild type (WT) NIK and its deletion mutants. Their TRAF3 binding activity (ϩ) or deficiency (Ϫ) is summarized based on the results presented in B. A sequence motif (located between amino acids 78 and 84) critical for TRAF3 binding is indicated. B, an N-terminal sequence motif of NIK is critical for its interaction with TRAF3. HA-TRAF3 was transfected into 293 cells together with the indicated NIK expression vectors. In lane 1, the cells were transfected with NIK wild type alone to serve as a control. Whole-cell lysates were prepared and subjected to co-IP to detect the interaction between TRAF3 and NIK mutants (top panel). The expression of NIK mutants and TRAF3 was monitored by direct IB using anti-HA (middle and bottom panels). C, schematic diagram of wild type TRAF3 and its deletion mutants showing the ring finger (RF), zinc finger (ZF), and TRAF domains. The NIK-binding activity (ϩ, active; Ϫ, inactive) of the different mutants is summarized based on the results presented in D. D, the C-terminal region of TRAF3 is required for its binding to NIK. HA-tagged NIK was transfected into 293 cells together with the wild type or mutant forms of TRAF3. In lane 5, the cells were transfected with TRAF3 in the absence of NIK. NIK protein complexes were isolated by IP using anti-NIK followed by detection of associated TRAF3 proteins by IB using anti-HA (top panel). The expression of NIK and TRAF3 in the cell lysates was analyzed by IB using anti-NIK (middle panel) and anti-HA (bottom panel), respectively.

FIG. 3. Inhibition of NIK-induced p100 processing by TRAF3.
A, the C-terminal region of TRAF3 is required for its NIK-inhibiting function. 293 cells were transfected with NIK (0.2 g) together with increasing amounts of HA-tagged TRAF3 or its C-terminal deletion mutant (residues 1-423). The cells were also transfected with the p100 expression vector (0.4 g). In lane 1, the cells were transfected with p100 only. The processing of p100 (top panel) and the expression of NIK (middle panel) and TRAF3 (bottom panel) proteins were analyzed by IB using the indicated antibodies. B, a NIK mutant defective in TRAF3 binding is insensitive to TRAF3-mediated inhibition. 293 cells were transfected with NIK (0.2 g) or NIK⌬78 -84 (0.12 g) together with increasing amounts of HA-TRAF3. The cells were also transfected with p100. The processing of p100 (top panel) and expression of NIK and TRAF3 proteins were detected by IB. C, deletion of the TRAF3-binding site in NIK has no effect on its catalytic activity. 293 cells were transfected with an empty vector (lane 1) or cDNA expression vectors encoding wild type NIK (lane 2) or NIK⌬78 -84 (lane 3). NIK proteins were isolated by IP using anti-NIK and subjected to in vitro kinase assays using recombinant IKK␣ as substrate (upper panel). The autophosphorylated NIK (P-NIK) and phosphorylated substrate (P-IKK␣) are indicated. Following autoradiography, the kinase assay membrane was subjected to IB to monitor the amount of the NIK protein (lower panel). nase assays revealed that deletion of the TRAF3-binding sequence of NIK did not alter its catalytic activity. Both NIK and NIK⌬78 -84 exhibited strong activity in autophosphorylation and phosphorylation of the target IKK␣ (Fig. 3C).
TRAF3/NIK Interaction Targets NIK Degradation through the Proteasome-In the transient transfection studies, we noticed that the level of NIK was reduced in cells cotransfected with TRAF3 (see Fig. 3, A and B, middle panel, lanes 3 and 4). This result was unlikely to be due to transfection variation, since it was observed in multiple experiments (data not shown). Further, the effect of TRAF3 on NIK expression appeared to require their physical interaction, since the C-terminal truncation mutant of TRAF3-(1-423) defective in NIK binding did not alter the level of NIK expression (Fig. 3A, middle panel, lanes  6 and 7). Similarly, the expression of NIK⌬78 -84, a NIK mutant incapable of TRAF3 binding, was not affected by TRAF3 (Fig. 3B, middle panel, lanes 6 and 7). These results indicated that the TRAF3/NIK interaction might also serve as a mechanism to control the steady expression of NIK. To confirm this notion, we performed pulse chase assays to determine the stability of NIK. Indeed, when NIK was coexpressed with TRAF3 in 293 cells, the half-life of this kinase was greatly reduced (Fig. 4A).
We also noticed that the expression efficiency of NIK⌬78 -84 was higher than that of the wild type NIK (data not shown) and had to adjust the amounts of expression vectors in the transfection experiments to achieve similar protein expression levels (see the legend to Fig. 3B). Although this difference could be due to the variation in transfection efficiency of different expression vectors, it also raised the possibility that the steady expression of NIK is affected by endogenous TRAF3. To minimize the effect of protein overexpression associated with transient transfection, we generated 293 cells stably expressing NIK or NIK⌬78 -84 using a retroviral gene delivery system. Due to the low copy number of retroviral insertion, the stably infected cells express lower levels of NIK than the transiently transfected cells (data not shown). Notably, the steady level of wild type NIK protein was significantly lower than that of NIK⌬78 -84 (Fig. 4B, top panel, compare lanes 1 and 5). Since bulk-infected cells were used in these studies, such a difference could not be due to clonal variations. Further, the mRNA for the wild type and mutant NIK constructs was expressed at comparable levels (Fig. 4B, panel 4, compare lanes 1 and 5), suggesting a regulation at the protein level.
To examine the effect of TRAF3 on NIK expression, the NIK-expressing cells were reinfected with a retroviral vector The differential mobility of the exogenous and endogenous TRAF3 proteins was due to the presence of the HA tag in the exogenous TRAF3. RNA was extracted from the same cells and subjected to reverse transcriptase-PCR analysis to determine the NIK expression at the mRNA level. C, knockdown of endogenous TRAF3 by siRNA enhances NIK expression and p100 processing. The 293-NIK and 293-NIK⌬78 -84 cells were transfected with siRNA for either the control GFP or TRAF3. Cell lysates were subjected to IB using the indicated antibodies to monitor expression of NIK (top panel), TRAF3 (panel 2), control tubulin (panel 3), or p100 and its processing product p52 (bottom panel). D, inhibition of NIK degradation by MG132 results in accumulation of NIK⅐TRAF3 complex. The indicated 293 infectants were either not treated (Ϫ) or incubated for 2 h with the proteasome inhibitor MG132. The interaction of NIK with endogenous TRAF3 was detected by co-IP (top panel), and the expression of NIK (middle panel) and TRAF3 (bottom panel) was detected by IB. E, polyubiquitination of wild type NIK but not NIK⌬78 -84. 293 cells stably infected with NIK or NIK⌬78 -84 were transiently transfected with HA-ubiquitin. The cells were incubated with MG132 (25 M) for the indicated times followed by isolating NIK by IP (using anti-NIK). The NIK/ubiquitin conjugates were detected by IB using anti-HA. encoding TRAF3. Indeed, the steady level of NIK was further reduced in cells co-infected with TRAF3 (Fig. 4B, top panel,  compare lanes 1 and 3). In contrast, the exogenous TRAF3 did not affect the expression level of the NIK mutant (⌬78 -84) defective in TRAF3 binding (compare lanes 5 and 7). These results further suggested that the steady level of NIK is negatively regulated by its interaction with TRAF3. To test whether the low steady level of NIK resulted from its degradation, we treated the cells with a proteasome inhibitor, MG132. Remarkably, the steady level of wild type NIK was greatly elevated following MG132 treatment (Fig. 4B, top  panel, lanes 2 and 4). In contrast, the expression of NIK⌬78 -84 was not significantly affected by MG132 (lanes 6  and 8). Taken together with the pulse-chase results (Fig. 4A), these data suggest the possibility that association of TRAF3 with NIK targets NIK for rapid degradation by the proteasome. Since NIK undergoes rapid turnover even in the absence of exogenous TRAF3, it is conceivable that the fate of NIK is also regulated by endogenous TRAF3.
To directly demonstrate that the instability of NIK is mediated by endogenous TRAF3, we took the RNAi approach to suppress the expression of endogenous TRAF3. siRNA for TRAF3 or the control GFP was transfected into cells expressing NIK or NIK⌬78 -84. The TRAF3 siRNA dramatically suppressed the expression of TRAF3 (Fig. 4C, panel 2, lanes 2 and 4) but did not alter the expression level of tubulin (panel 3, lanes 2 and 4). Further, as expected, the GFP siRNA did not suppress the expression of TRAF3 (panel 2, lanes 1 and 3). Importantly, the suppression of TRAF3 expression by TRAF3 siRNA was associated with marked elevation in the level of wild type NIK (top panel, lane 2). In contrast, the TRAF3 suppression did not significantly alter the expression of NIK⌬78 -84 (lane 4). Thus, binding of NIK to endogenous TRAF3 appears to target this kinase for degradation. Parallel IB assays revealed that the elevation of NIK expression in TRAF3 siRNA-treated cells was associated with enhanced processing of p100 (bottom panel, lane 2). On the other hand, the TRAF3 siRNA had no effect on p100 processing in cells expressing NIK⌬78 -84, which already exhibited high levels of p100 processing (bottom panel) and NIK expression (top panel).
If the TRAF3/NIK physical interaction indeed serves as a trigger for NIK degradation, we should detect accumulation of the TRAF3⅐NIK complexes in cells incubated with proteasome inhibitors. We examined this idea by co-IP assays using 293 cells stably expressing NIK or NIK⌬78 -84. In untreated cells, a low amount of endogenous TRAF3 was co-precipitated with NIK (Fig. 4D, top panel, lane 1). In contrast, no interaction was detected between TRAF3 and NIK⌬78 -84 (lane 3), although this NIK mutant was expressed at higher levels than the wild type NIK (middle panel, compare lanes 1 and 3). Importantly, a considerably higher amount of TRAF3 was assembled into the NIK complex in cells incubated with the proteasome inhibitor MG132 (Fig. 4D, top panel, lane 2). The enhanced NIK/ TRAF3 association in MG132-treated cells was apparently due to the accumulation of NIK (middle panel, lane 2). On the other hand, the MG132 treatment did not enhance the expression of TRAF3 (bottom panel, lane 2). These results have several implications: 1) newly synthesized NIK rapidly associates with TRAF3; 2) NIK, but not TRAF3, is targeted for degradation upon their physical association; and 3) the intracellular amount of TRAF3 is in excess compared with NIK.
To examine whether the NIK degradation is mediated through its ubiquitination, we analyzed the polyubiquitination of wild type NIK and NIK⌬78 -84 in cells treated with MG132. Indeed, when the degradation of NIK was inhibited by MG132, its ubiquitinated forms could be readily detected (Fig. 4E, lanes   1-3). Consistent with its inability to associate with TRAF3, the NIK⌬78 -84 was not significantly ubiquitinated (lanes 4 -6).
Noncanonical NF-B Stimuli Induce Degradation of TRAF3 and Elevation of NIK Expression-Recent studies have shown that the noncanonical NF-B signaling pathway can be activated by ligands for a subset of TNFR family members, including BaffR, CD40, and lymphotoxin ␤ receptor (14,17,22,23), although the underlying mechanisms remain unclear. We examined whether the noncanonical NF-B signaling involves modulation of the TRAF3/NIK molecular interplay. For these studies, we employed a murine B-cell line, M12.4.1 (hereafter referred to as M12), which had been frequently used to study CD40 signaling (31,32,40,41). Since both CD40 and BaffR are expressed on B cells, we reasoned that the M12 cells might serve as a convenient model for studying noncanonical NF-B signaling. Indeed, stimulation of the M12 cells with either anti-CD40 or Baff resulted in potent processing of p100 (Fig.  5A, upper panel) as well as nuclear translocation of RelB (data not shown) (Fig. 5B, bottom panel). Interestingly, the noncanonical NF-B signaling events were associated with persistent degradation of TRAF3 (Fig. 5A, lower panel). Time course studies showed that TRAF3 was largely depleted around 2 h after cellular stimulation by Baff or anti-CD40, although partial loss of TRAF3 was detectable at an earlier time point (1 h) (Fig. 5B, top panel). Parallel analysis of TRAF2 revealed that, as previously reported (42), this TRAF member was also degraded by the anti-CD40 signal (second panel, lanes 6 -8). In sharp contrast, however, the fate of TRAF2 was not appreciably affected in Baff-stimulated cells (second panel, lanes 2-4). On the other hand, both Baff and anti-CD40 efficiently stimulated the nuclear translocation of RelB (bottom panel). Thus, degradation of TRAF3, but not that of TRAF2, is associated with noncanonical NF-B activation. Since TRAF3 promotes NIK degradation, it is tempting to speculate that the CD40 and BaffR signals may rescue NIK from TRAF3-induced degradation.
Our initial IB analyses indicated that the level of endogenous NIK is extremely low, since no specific NIK protein was detected by various commercial anti-NIK antibodies (data not shown). To overcome this technical difficulty, we concentrated NIK by IP from the cell lysates before performing IB analyses. With this IP-coupled IB approach, we were able to detect a weak NIK band (Fig. 5C, top panel, lane 1). Interestingly, the level of NIK expression was significantly enhanced upon stimulation of the cells with anti-CD40 (lane 2) or Baff (lane 3) but not the canonical NF-B inducer tumor necrosis factor-␣ (lane 4). To further confirm that the steady level of NIK could be induced by noncanonical NF-B stimuli, we generated M12 B cells stably expressing exogenous NIK by retrovirus-based gene transfer. The steady level of exogenous NIK was also low, which required IP-coupled IB for detection (Fig. 5C, top panel,  lane 5). More importantly, the exogenous NIK was also induced by the anti-CD40 and Baff signals and was readily precipitated by the anti-NIK antibody (lanes 6 and 7) but not a preimmune serum (lane 8). Further, the induction of NIK expression was associated with degradation of TRAF3 (Fig. 5C, middle panel).
The studies presented above strongly suggest that the endogenous NIK may be subject to the negative regulation by TRAF3. To examine this possibility, we incubated the M12 B cells with the proteasome inhibitor MG132. Indeed, significantly more endogenous NIK was detected in the MG132treated cells (Fig. 5D, top panel, lane 2). We then investigated whether NIK and TRAF3 are associated under endogenous conditions and whether this molecular interaction precedes NIK degradation. Consistent with the low expression level of NIK, little TRAF3 was detected in the NIK immune complex in untreated cells (Fig. 5D, middle panel, lane 1). However, inhibition of NIK degradation by MG132 resulted in accumulation of the NIK⅐TRAF3 complex, as demonstrated by the coprecipitation of TRAF3 with NIK (lane 2). Our attempt to perform pulse-chase assays proved to be technically difficult, since the level of endogenous NIK or stably infected NIK was too low to obtain efficient pulse labeling (data not shown). Nevertheless, these findings indicate that binding of TRAF3 to NIK targets NIK for degradation by the proteasome and that the noncanonical NF-B stimuli may prevent NIK degradation by eliminating TRAF3. DISCUSSION NIK serves as a key component in a noncanonical NF-B signaling pathway that targets the processing of NF-B2/p100 and, in turn, the nuclear translocation of RelB-containing NF-B dimers (43). How the signaling function of NIK is regulated has remained unclear. The results presented in this paper establish TRAF3 as a negative regulator of NIK. TRAF3 physically interacts with NIK via a novel sequence motif located at the N-terminal region of NIK. This molecular interaction may serve to prevent uncontrolled function of NIK. At least in transfected cells, TRAF3 inhibits NIK-mediated induction of p100 processing. More importantly, the TRAF3/NIK interaction targets NIK for continuous degradation by the proteasome, a mechanism that may contribute to the low intracellular level of NIK. We have demonstrated that the TRAF3mediated NIK destruction can be prevented by the noncanonical NF-B stimuli, including anti-CD40 and Baff. This step of signaling seems to be achieved through signalinduced degradation of TRAF3. These findings suggest a novel mechanism mediating NF-B signaling, which involves the rescue of NIK from its TRAF3-programmed destruction.
It remains to be further investigated how the catalytic activity of NIK is activated in response to receptor signals. It is possible that the activation of NIK involves both its modifications by upstream signaling components (e.g. kinases) and its protection from TRAF3-induced proteolysis. However, since the transfected NIK exhibits remarkably high constitutive activity in mammalian cells (11,37,44), it is also conceivable that the function of NIK may be primarily controlled by its level of expression. We have found that the preexisting level of NIK is extremely low and is only weakly detected by IP-coupled IB assays. When stably expressed in M12 B cells and 293 cells, the exogenous NIK is also maintained at a low level. At least in 293 cells, in which RNAi could be effectively performed, the level of NIK can be markedly enhanced by RNAi-mediated suppression of endogenous TRAF3 (Fig. 4C). This genetic evidence strongly indicates that the low intracellular level of NIK is due to its degradation targeted by its partner protein TRAF3. In agreement with this model, induction of TRAF3 degradation by noncanonical NF-B stimuli is also associated with elevation of NIK expression (Fig. 5C).
The TRAF3/NIK interplay seems to be reminiscent of the p53 regulation by its inhibitor Mdm-2 (45). It has been shown that newly synthesized p53 is quickly bound by Mdm-2 and targeted to the proteasome for degradation. In this case, Mdm-2 is known to function as a ubiquitin ligase (E3) that stimulates polyubiquitination of p53 (46). Mdm-2 belongs to a family of ubiquitin ligases that contain ring finger domains (47,48). Notably, TRAF3, as well as most other TRAF members, also contain a ring finger. It is thus possible that TRAF3 may induce NIK degradation through the ubiquitination pathway. We have found that NIK indeed undergoes polyubiquitination in MG132-treated cells (Fig. 4E). This modification of NIK requires its association with TRAF3, since a NIK mutant lacking the TRAF3-binding site (⌬78 -84) is not ubiquitinated (Fig.  4E). However, whereas these results suggest that the TRAF3/ NIK association triggers NIK ubiquitination, our preliminary FIG. 5. Noncanonical NF-B activation by Baff and anti-CD40 is associated with TRAF3 degradation and induction of NIK expression. A, induction of p100 processing by anti-CD40 and Baff is associated with persistent degradation of TRAF3. M12 cells were either not treated (NT) or stimulated with anti-CD40 or Baff for the indicated times. Whole-cell lysates were subjected to IB to detect the processing of p100 (upper panel) and expression of TRAF3 (lower panel). B, degradation of TRAF3, but not that of TRAF2, is associated with nuclear translocation of RelB. M12 cells were treated with Baff or anti-CD40 for the indicated times. Whole-cell lysates were subjected to IB to detect expression of TRAF3 (top panel), TRAF2 (panel 2), and tubulin (panel 3); nuclear extracts were subjected to IB to detect the nuclear translocation of RelB (bottom panel). C, signal-induced TRAF3 degradation is associated with induction of NIK expression. M12 cells and M12-NIK cells (M12 cells stably infected with retroviruses encoding NIK) were either not treated (NT) or incubated for 2 h with anti-CD40, Baff, or tumor necrosis factor-␣. Whole-cell lysates were subjected to IP using anti-NIK (lanes 1-7 and 9) or a preimmune serum (lane 8) followed by detection of the enriched NIK protein by IB (top panel). The expression of TRAF3 (middle panel) and control tubulin (bottom panel) was detected by direct IB. D, endogenous NIK⅐TRAF3 complex can be detected in MG132-treated cells. M12 B cells were either not treated (NT) or incubated for 2 h with MG132. The NIK protein complex was isolated by IP followed by detection of NIK (top panel) and TRAF3 (middle panel) using anti-NIK and anti-TRAF3. Horseradish peroxidase-conjugated anti-TRAF3 was used to avoid interference of the immunoglobulin heavy chain. The TRAF3 expression level was detected by direct IB (bottom panel).
in vitro ubiquitination studies using purified TRAF3 did not reveal significant E3 activity (data not shown). Nevertheless, although this result suggests that TRAF3 itself may not be sufficient to act as a ubiquitin ligase, it does not exclude the possibility that this NIK-binding protein functions as a critical component of a ubiquitin ligase mediating NIK ubiquitination in vivo. Notwithstanding, the results presented in this paper establish TRAF3 as a negative regulator of NIK, mediating degradation of NIK through the proteasome pathway.