Regulation of NF- (cid:1) B2 p100 Processing by Its cis -Acting Domain*

Processing of NF- (cid:1) B2 precursor protein p100 to generate p52 is tightly regulated. However, this proteolytic event could be actively induced by the NF- (cid:1) B-inducing kinase and the human T-cell leukemia virus-encoded oncoprotein Tax or be constitutively turned on due to the loss of the C-terminal portion of p100. Whereas NF- (cid:1) B-inducing kinase-mediated p100 processing requires (cid:2) -transducin repeat-containing protein, constitutive processing of p100 is independent of this protein. On the other hand, Tax-induced processing of p100 appears to be both (cid:2) -transducin repeat-containing protein-depend-ent and -independent. We show here that, besides the C-terminal sequences, multiple functional regions, including the two (cid:3) -helices, dimerization domain, nuclear localization sequence, and glycine-rich region, located in the N terminus of p100, also play important roles in both constitutive and inducible processing, suggesting a common mechanism for p100 processing. We further demonstrate that with the help of the C-terminal death domain and I (cid:1) B kinase (cid:3) -targeting serines, the C-termi-nal ankyrin-repeat domain of p100 strongly interacts with its N-terminal dimerization domain and nuclear localization sequence, thereby bringing the C- and N-ter-minal sequences together to form a three-dimensional domain. This presumptive domain is not only responsible for suppression of constitutive processing but also required for inducible

sequestered in the cytoplasm as inactive complexes with a family of ankyrin repeat domain (ARD)-containing inhibitors called IB proteins, which interact with RHD of NF-B proteins (1)(2)(3). Accordingly, the major pathway leading to NF-B activation is based on inducible IB degradation. This canonical pathway could be rapidly and transiently activated by a large variety of substances, such as mitogens, cytokines, and microbial components (4), and depends on a specific IB kinase (IKK) complex composed of two catalytic subunits, IKK␣ (IKK1) and IKK␤ (IKK2), and a regulatory subunit, IKK␥ (NEMO) (5,6). Upon activation by those stimuli, IKK phosphorylates specific serines within the IB proteins, triggering their ubiquitination by the ␤-transducin repeat-containing protein (␤-TrCP)-containing ubiquitin ligase complex and degradation by proteasome, thus allowing the NF-B dimers to move to the nucleus to induce gene expression (5,6).
Unlike the three Rel proteins, which are directly synthesized as mature proteins, p50 and p52 are generated by proteolytic processing from their large precursors NF-B1 p105 and NF-B2 p100, respectively (7,8). Interestingly, both p105 and p100 contain ankyrin repeats at their C-terminal regions and function as IB-like inhibitors of NF-B (9,10). Different from complete degradation of other IB proteins, the proteasomemediated degradation of p105 and p100 only leads to loss of their C-terminal ankyrin repeat regions, leaving intact N termini, p50 and p52, respectively (5,11). Thus, the processing of p105 and p100 not only serves to generate p50 and p52 but also works as an alternative pathway for NF-B activation. Whereas the processing of p105 is constitutive (11), the processing of p100 is tightly regulated by both positive and negative mechanisms (12). Induction of p100 processing is mediated by the NF-B-inducing kinase (NIK) and its downstream kinase IKK␣ (12,13). Interestingly, neither IKK␤ nor IKK␥, two key components of the canonical NF-B signaling pathway, is required for this noncanonical NF-B pathway (13,14). Consistently, this novel pathway of NF-B activation cannot be stimulated by most of the classical NF-B inducers but rather responds to signals involved in B cell maturation and lymphoid organogenesis, including those triggered by lymphotoxin ␤ (12,15), B-cell-activating factor (16,17), and CD40 ligand (18). These stimuli are physiological stimulators of NIK to induce processing of p100.
Recent genetic evidence has demonstrated that disruption of p52 production causes defects in development, organization, and function of lymphoid organs (19,20). On the other hand, overexpression of p52 leads to lymphocyte hyperplasia and transformation (21). In humans, constitutive processing of p100, which is caused by the loss of its C-terminal processing inhibitory domain due to nfb2 gene rearrangements (12), is associated with the development of various lymphomas (22)(23)(24)(25)(26). Furthermore, those constitutive processing forms of p100 can oncogenically transform fibroblasts in vitro (27). Interestingly, overproduction of p52 is also found to be associated with T-cell transformation by the human T-cell leukemia virus type I (HTLV-I), an etiological agent of an acute and fatal T-cell malignancy named adult T cell leukemia (14,28). Actually, overproduction of p52 in T cells is a hallmark of HTLV-I infection and transformation, since p100 processing is very poor in T cells, including activated T cells (14).
Our previous studies have demonstrated that Tax functions as a pathogenic inducer of p100 processing in HTLV-I-infected T cells (14). Unlike NIK, the physiological inducer of p100 processing, which can directly interact with IKK␣ (29,30), Tax associates with IKK␣ via binding to IKK␥ (31)(32)(33). Once forming a complex with NIK or Tax, IKK␣ is activated and recruited to p100 via two amino acid residues, serine 866 and 870, of p100, to phosphorylate the specific serines located in both N and C termini of p100, thereby triggering p100 ubiquitination mediated by ␤-TrCP and subsequent processing by 26 S proteasome (34 -36). However, a ␤-TrCP-independent mechanism may also contribute to Tax-induced p100 processing (35). Interestingly, constitutive processing of p100 does not require ␤-TrCP either (36) but rather is regulated by the nuclear translocation of p100 (37). Although much progress on the regulation of p100 processing has been made recently, it is still unknown whether there is a general mechanism for both inducible and constitutive processing of p100 and whether the ␤-TrCP-independent mechanism of Tax-induced p100 processing also involves p100 nuclear shuttling. Additionally, it is also unknown why, unlike the IKK targeting serines within other proteins such as IB␣, the IKK␣-targeting serines in p100 are located at distant positions in the amino acid sequence (34,35).
In the present study, we show that, unlike the C-terminal ARD and death domain (DD), which suppress p100 processing, multiple regions located in the N terminus of p100, such as the two ␣-helices, dimerization domain, nuclear localization sequences (NLS), and glycine-rich region (GRR), are important for both constitutive and inducible processing of p100. We also demonstrate that the C-terminal ARD of p100 strongly interacts with its N-terminal dimerization domain and NLS, and the C-terminal DD and IKK␣-targeting serines also contribute to this interaction. Through this N-and C-terminal interaction, these important regions within p100 located at distant positions in the amino acid sequence appear to come close to form a functional ternary domain required for regulation of p100 processing. Additionally, we find that NIK does not alter the cytoplasmic expression of p100, whereas Tax induces a significant nuclear localization of p100, suggesting that nuclear shuttling may be the ␤-TrCP-independent mechanism for Taxinduced p100 processing. These studies provide evidence for common and different mechanisms of p100 processing under physiological and pathogenic conditions.

MATERIALS AND METHODS
Expression Vectors and Antibodies-Expression vectors encoding p100 and its S866A/S870A and helix disruption mutants (labeled as p100 ⌬S and p100 ⌬␣ in the figures, respectively), Tax, HA-tagged NIK, IKK␣, ␤-TrCP, and ubiquitin have been described (12,14,29,34,35). Other p100 mutants, including HA-or Myc-tagged p100 mutants, were generated by PCR amplification or site-directed mutagenesis using human p100 cDNA (gift from Dr. W. C. Greene) or its corresponding mutants as templates. The anti-HA monoclonal antibody (anti-HA, 12CA5) or the horseradish peroxidase (HRP)-conjugated HA antibodies (anti-HA-HRP, 3F10) were purchased from Roche Applied Science. The antibodies for c-Myc (9E10), NIK (H248), and IKK␣ (H744) were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). The fluorescein isothiocyanate-conjugated anti-rabbit secondary antibodies were from Amersham Biosciences. The antibodies recognizing the N and C termini of p100 (anti-p100N and anti-p100C) were kindly provided by Drs. W. C. Greene (38) and N. Rice and M. Ernst (9), respectively. The anti-Tax monoclonal antibody was prepared from a hybridoma (168B17-46-49) provided by the AIDS Research and Reference Program (NIAID, National Institutes of Health).
Inducible Processing of p100 in MEF-MEFs derived from wild type mice (gift from Dr. Michael Karin) were transfected with p100 or its mutants. After 24 h post-transfection, the cells were treated with mouse LT-␣ 1 ␤ 2 recombinant proteins (500 ng/ml; R & D Systems) or control solution (0.2% bovine serum albumin in phosphate-buffered saline) for another 24 h, followed by in vivo processing assays as described previously (12)(13)(14).
In Vivo Ubiquitin Conjugation Assays-293 cells were transfected with HA-tagged ubiquitin and p100 or its mutants in the presence or absence of NIK or Tax. After 36 -48 h post-transfection, the cells were lysed in RIPA buffer and immediately subjected to immunoprecipitation (IP) using anti-p100N. The agarose beads were washed three times with RIPA buffer followed by two additional washes with RIPA buffer supplemented with 1 M urea, and the bound proteins were eluted with SDS loading buffer (62.5 mM Tris-HCl, pH 6.8, 4% SDS, 0.004% bromphenol blue, 8 M urea, 10% glycerol). The eluted ubiquitin-conjugated p100 or its mutants were fractionated by SDS-PAGE and detected by IB using anti-HA-HRP (12,34,35).
Subcellular Fraction Preparation-293 cells were transfected with vectors encoding p100 or its mutants in the presence or absence of NIK or Tax. After 36 -48 h post-transfection, the cells were lysed in Buffer B (10 mM Hepes, pH 7.9, 10 mM KCl, 0.4% Nonidet P-40, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride) and swelled on ice for 15 min, followed by centrifugation for 5 min at 4°C and 12,000 ϫ g. The supernatant was cytoplasmic extract. The pellet (nucleus) was further washed three times with Buffer B, suspended in Buffer C (20 mM Hepes, pH 7.9, 0.4 M NaCl, 0.1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride), and rotated at 4°C for 15 min, followed by centrifugation for 5 min at 4°C and 12,000 ϫ g. The resultant supernatant was nuclear extract (14). Both cytoplasmic and nuclear extracts were subjected to IB using anti-p100N.
Immunofluorescence Assays-COS cells were transfected with p100 in the presence or absence of NIK or Tax using Fugene 6 reagent. After 24 h, the recipient cells were directly fixed, permeabilized, and sequentially incubated with anti-p100C, followed by fluorescein isothiocyanate-conjugated anti-rabbit secondary antibodies. To block protein nuclear export, the transfected cells were incubated with 10 ng/ml leptomycin B for 5 h prior to the immunofluorescence staining. The subcellular localization of stained proteins was detected using an inverted fluorescence microscope. The cells were also counterstained with Hoechst 33258 (Sigma) for nuclear staining by detecting DNA (37).

RESULTS
Multiple Domains within the p100 N Terminus Are Involved in p100 Processing Induced by NIK or Tax-Although it is known that p100 C-terminal mutants, including ARD and DD deletion mutants, fail to respond to NIK-or Tax-induced processing (12,14,34,35) (also see Fig. 2), the functional significance of the N-terminal domains of p100 in its inducible processing remains to be investigated. For these studies, we created full-length p100 mutants harboring mutations at the following domains: two ␣-helices, dimerization domain, NLS, or GRR (Fig. 1A). The processing of these mutants to generate p52 in the absence or presence of NIK or Tax was then examined. As shown in Fig. 1B, disruption of the ␣-helices or dimerization domain efficiently prevented p100 processing induced by both NIK and Tax (top, lanes 4 -9), whereas mutation of the NLS also partially blocked these events (compare lanes 11 and 12 with lanes 2 and 3, respectively). In agreement with previous studies indicating that GRR provides the stop signal for processing of p105 and p100 (42,43), both NIK and Tax could induce p100 GRR deletion mutant degradation instead of processing, as indicated by a loss of the precursors but no processing products when NIK or Tax was co-expressed (lanes 13-15). As a control, a C-terminal IKK␣-targeting serine mutant of p100, which shows a defect in p100 processing, was also included (lanes 16 -18). To further verify the physiological significance of these domains in inducible p100 processing, we used the purified lymphotoxin proteins (LT-␣ 1 ␤ 2 ) to treat MEFs expressing p100 or its N-terminal mutants. Consistently, disruption of these important domains also prevented p100 processing induced by lymphotoxin ␤ receptor ligation (Fig. 1C, compare lanes 3-10 with lane 2). Thus, multiple domains within the p100 N terminus are important for inducible p100 processing under both physiological and pathogenic conditions. FIG. 1. N-terminal domains within p100 required for p100 processing induced by NIK or Tax. A, schematic representation of p100 and its mutants. ␣, ␣ helices; D, dimerization domain; N, NLS; G, GRR; S, the C-terminal IKK␣-targeting serines. The processing site is also indicated. Note that p100-(1-405) is actually p52, the processing product of p100. B, 293 cells were transfected with the indicated p100 constructs either alone (Ϫ) or together with (ϩ) NIK or Tax, followed by IB analysis of p100 and p52 (p100 processing) (top), NIK (middle), and Tax (bottom). C, MEFs were transfected with p100 or its mutants, followed by LT-␣ 1 ␤ 2 treatment. After 24 h post-treatment, the p100 processing assays were performed as described in B.
Both the C-and N-terminal Functional Domains of p100 Are Essential for Its Inducible ␤-TrCP Binding and Subsequent Ubiquitination-The significance of these functional domains in ␤-TrCP recruitment and ubiquitination of p100 induced by NIK and Tax was also investigated, since the E3 ligase ␤-TrCP plays an important role in inducible processing of p100 (34 -36). Consistent with our previous studies (12,14,34,35), both NIK and Tax could efficiently induce ␤-TrCP binding to and polyubiquitination of p100 (Fig. 2, lanes 3). Importantly, the helices, dimerization domain, NLS, ARD, and DD were required or partially required for inducible ␤-TrCP recruitment to p100 and subsequent p100 ubiquitination (lanes 4 -9 and 12-15), whereas GRR was dispensable for these events (lanes 10 and 11). In agreement with our previous studies (12,14,34,35), the IKK␣-targeting serines were also involved (lanes 16 and 17). Taken together, the results are highly consistent with the above studies showing the roles of these regions in the inducible p100 processing (Fig. 1).
p100 N-terminal Domains Involved in Inducible Processing Are Also Required for Its Constitutive Processing-As we have shown before (12) (also see Fig. 2), the C-terminal ARD and DD are important for suppressing the constitutive processing of p100. However, it remains to be investigated whether the Nterminal domains, which are essential for inducible p100 processing, are also involved in constitutive processing of p100. For these studies, a p100 C-terminal truncation mutant (p100N) was utilized, and these important domains were disrupted from this mutant (Fig. 3A). Consistent with our previous studies (12), p100N showed very strong constitutive processing, due to the loss of both ARD and DD (Fig. 3B, lane  1). In sharp contrast to the C-terminal ARD and DD, disruption of either the two ␣ helices or the dimerization domain blocked p100N constitutive processing, whereas the mutant lacking NLS also exhibited a significant defect in constitutive processing (lanes 2-4). Again, consistent with its role in providing a stop signal of processing/degradation (Fig. 2), the GRR domain was also required for p52 production from constitutive processing of p100N (lane 5). Collectively, these results suggest a common mechanism for both constitutive and inducible processing of p100.
Identification of the Reciprocal Binding Domains within p100 That Mediate Interaction between the N and C Termini of p100 -The fact that both inducible and constitutive processing of p100 involve the same regions alludes to that these regions may form a three-dimensional domain required for general regulation of its processing. In this regard, the previous structural studies of co-crystals of NF-B proteins in association with IB␣ or IB␤ revealed that the ARD of IBs interacts with NF-B proteins via the RHD (44). Although it is still unknown whether the p100 C terminus binds to its N terminus, it is interesting to note that p100 contains RHD and ARD at its N and C termini, respectively (Fig. 4A). Additionally, the C terminus of p100 also contains a DD, a well known protein-protein interaction domain (Fig. 4A), although its role in protein-protein interaction remains undetermined. To first test the interaction between p100 C and N termini, we performed Co-IP assays using p52 and p100C, the p100 C-and N-terminal deletion mutants, respectively (Fig. 4A). As expected, the interaction between the N-and C-terminal parts of p100 could be easily detected (Fig. 4B, top, lane 2). This result suggests that the N-and C-terminal sequences are actually close in three dimensions, further explaining why the IKK␣-targeting serines in p100 are located at distant positions in the amino acid sequence (34,35).
Next, the roles of ARD and DD in the C-and N-terminal interaction of p100 were investigated. As shown in Fig. 4C, deletion of ARD blocked the interaction between p100 C and N termini under these conditions (top, lane 3), whereas deletion of DD moderately prevented this interaction (lane 4), suggesting that both ARD and DD contribute to the C-and N-terminal interaction of p100, although to different extents. Interestingly, the downstream sequences of p100 DD are also partially involved in this C-and N-terminal interaction of p100 (lane 5). It seems that the contribution of the downstream sequences of DD to this interaction is attributed to serines 866, 870, and 872 of p100, the C-terminal IKK␣ targets (compare lane 6 with lane 5), which are required for p100 processing induced by both NIK and Tax (12)(13)(14).
To define N-terminal domains involved in this C-and Nterminal interaction of p100, we repeated Co-IP assays using p100C and various mutants of p52 (Fig. 4A). Not surprisingly, the interaction of p100 N terminus and its C terminus involves the dimerization domain and NLS (Fig. 4D, top,  lanes 3 and 4). However, the proximate GRR is not required, since deletion of GRR did not affect the ability of p52 to bind to the C terminus of p100 (lane 5). Similarly, the two ␣-helices are not involved in this C-and N-terminal interaction of p100 (lane 2).
To further define reciprocal interaction domains of p100, multiple co-IP assays were performed by using different p100 C-and N-terminal mutants (Fig. 4A). As expected, the ARD alone could efficiently associate with p52 ( Fig. 4E, top, lane 2). Disruption of helices or GRR from p52 failed to prevent the interaction between p52 and the ARD of p100 (lanes 3 and 6). However, disruption of the dimerization domain or NLS from p52 completely blocked the binding of p100 ARD to p100 N terminus (lanes 4 and 5), suggesting that both dimerization domain and NLS of p100 are required for p100 ARD binding to its N terminus. These results further confirmed the co-IP studies above. Since the interaction between p52 and p100C ⌬ARD, which contains DD and C-terminal IKK␣ targeting serines, is very weak (Fig. 4C, lane 3, and data not shown), it is difficult to directly identify the reciprocal binding site for DD or C-terminal IKK␣ targeting serines. Thus, we examined whether DD or C-terminal IKK␣ targeting serines of p100 could compensate ARD for its binding to p52 dimerization domain or NLS disruption mutant. As shown in Fig. 4, F and G (top, lanes 4 and  5), the DD or serines at the p100 C terminus significantly, although only partially, restored the ability of ARD to bind p52 dimerization domain or NLS disruption mutant, indicating that both DD-and IKK␣-targeting serines at the p100 C terminus indeed enhance the interaction between p100 ARD and its dimerization domain/NLS. These results are also consistent with the data showing that disruption of the dimerization domain or NLS only partially blocked the N-and C-terminal interaction of p100 (Fig. 4D), although both of them are required for the binding of ARD of p100 to its N terminus (Fig. 4E).
␣-Helices of p100 Are Required for the Suppression of Its Constitutive IKK␣ Recruitment-Although the two ␣-helices, like the dimerization domain, NLS, ARD, and DD, are required for p100 processing, the mechanism of how these two helices are involved remains largely unknown. In this regard, our recent study revealed that deletion of the N-terminal region of p100 leads to constitutive p100/IKK␣ interaction, although the relevant sequences remain unknown (35). We thus examined the role of the helices in suppressing IKK␣ constitutive recruitment of p100. Consistent with our previous studies (14,34,35), no significant binding between IKK␣ and p100 was detected when these two proteins were coexpressed (Fig. 5, top, lane 2), whereas NIK or Tax could significantly promote their interaction (lanes 3 and 4). Interestingly, disruption of these two FIG. 2. Both the C-and N-terminal domains within p100 are required for ␤-TrCP recruitment into p100 and subsequent ubiquitination of p100 induced by NIK or Tax. A, multiple domains within p100 are required for NIK-mediated ␤-TrCP recruitment into p100. 293 cells were transfected with expression vectors encoding HA-tagged ␤-TrCP and p100 wild type or its mutants together with either an empty vector (Ϫ) or NIK (ϩ). Cell lysates were subjected to IP using anti-p100N, and the co-precipitated ␤-TrCP was detected by IB using anti-HA-HRP (top). The cell lysates were also directly subjected to IB to detect the expression levels of ␤-TrCP (middle) and of p100 and its processing product p52 (bottom) by using anti-HA-HRP and anti-p100N, respectively. The N-and C-terminal p100 mutants were described in the legend to Fig. 1 and before (12), respectively. B, multiple domains within p100 are required for Tax-mediated ␤-TrCP recruitment into p100. 293 cells were transfected with expression vectors encoding HA-tagged ␤-TrCP and p100 wild type or its mutants in the presence (ϩ) or absence (Ϫ) of Tax, followed by IP as described in A. The co-precipitated ␤-TrCP (top) and the expression levels of ␤-TrCP (middle) and of p100 and its processing product p52 (bottom) were indicated. C, multiple domains within p100 are required for NIK-induced polyubiquitination of p100. 293 cells were transfected with HA-tagged ubiquitin together with indicated constructs. The p100 wild type and its mutants were isolated by IP with anti-p100N followed by IB using anti-HA-HRP to detect the ubiquitin-conjugated p100. D, multiple domains within p100 are required for Tax-induced polyubiquitination of p100. 293 cells were transfected with expression vectors encoding HA-tagged ubiquitin and p100 wild type or its mutants in the presence (ϩ) or absence (Ϫ) of Tax, followed by ubiquitin conjugation assays as described in C. ␣-helices from p100 resulted in stable association of IKK␣/p100 even without coexpression of NIK or Tax (lane 5). These results clearly indicate that the two helices within p100 normally function as a doorkeeper in this three-dimensional domain to block the access of p100 to IKK␣.
NIK Fails to Alter Subcellular Localization of p100, whereas Tax Triggers p100 Nuclear Shuttling-Our previous studies indicate that Tax-induced processing of p100 is both ␤-TrCPdependent and -independent (35). However, the ␤-TrCP-independent mechanism remains unknown. A recent study suggested that the constitutive processing of p100 C-terminal deletion mutants, which is also ␤-TrCP-independent (12,36), is regulated by the nuclear shuttling of p100 (37). We thus examined whether Tax induces p100 nuclear translocation by performing both subcellular IB and immunofluorescence assays. Consistent with a previous immunofluorescence study (37), our IB studies also indicated that p100 was primarily located in the cytoplasm (Fig. 6A, lanes 1 and 2). NIK expression did not alter the pattern of p100 subcellular localization (lanes 3 and 4). However, co-expression of Tax resulted in dramatic nuclear localization of p100 (lanes 5 and 6). These data are correlated with high nuclear/cytoplasmic shuttling activity of Tax (Fig.  6A, third panel from top, lanes 5 and 6) (45). In support of the results of subcellular immunoblot assays, our immunofluorescence assays showed that, in the presence of Tax, p100 exhibited a whole-cell expression pattern (Fig. 6B, upper right). More importantly, p100 accumulated in the nucleus upon treatment of leptomycin B, a potent nuclear export inhibitor (Fig. 6B,  lower right). In contrast, p100 was still retained in the cytoplasm even in the presence of NIK and leptomycin B treatment (Fig. 6B, left). These results indicate that NIK-mediated p100 processing solely occurs in the cytoplasm, whereas Tax-induced processing of p100 may happen in both cytoplasm and nucleus. These results further suggest that nuclear shuttling, which is ␤-TrCP-independent, may be one of the mechanisms of Taxinduced p100 processing. DISCUSSION The data presented in this study revealed that the optimal interaction of the N-and C-terminal regions of p100 involves its multiple domains, including the N-terminal dimerization domain, NLS, and C-terminal ARDs, DD, and IKK␣-targeting serines. Whereas the binding of p100 C-terminal ARD to its N-terminal dimerization domain and NLS plays a major role in its C-/N-terminal interaction, C-terminal DD and serines also contribute to this interaction. Interestingly, besides the Cterminal region of p100, regulation of p100 processing by NIK and Tax also involves its N-terminal regions, including the two ␣-helices, dimerization domain, and NLS. Through this N-and C-terminal interaction, these important domains located at distant positions in the amino acid sequence of p100 thus come close to form a functional three-dimensional domain required for general regulation of p100 processing. Indeed, like NLS, the two ␣-helices and dimerization domain are also involved in constitutive processing of C-terminal deletion forms of p100. The data presented here thus provide the first evidence for a common mechanism for both inducible and constitutive processing of p100. In addition, we observe that Tax but not NIK can trigger p100 nuclear translocation, an event associated with constitutive processing of p100 (37). These data thus further suggest that nuclear shuttling may account for the ␤-TrCP-independent mechanism of Tax-induced p100 processing we discovered before (35).
In contrast to the canonical NF-B pathway, which is activated by a plethora of stimuli (4), p100 processing has been found so far to be triggered only by lymphotoxin ␤ (15, 18), B-cell-activating factor (19,20), CD40 ligand (21), tumor necrosis factor-like weak inducer of apoptosis (46), and RANKL (47). Interestingly, induction of p100 processing by all of these five inducers relies on NIK and IKK␣ but is independent of IKK␤ or IKK␥. Since activation of the canonical NF-B signaling pathway is predominantly dependent on IKK␤ and IKK␥ (6,26), the NIK/IKK␣-dependent p100 processing serves as a novel mechanism for NF-B activation. Currently, the mechanisms of how NIK activity is suppressed in most cells and activated by these noncanonical NF-B stimuli are still largely unknown. It seems, however, that NIK activity is largely regulated by its protein level. In support of this idea, the amount of NIK is extremely low in most cell types, whereas expression of transfected NIK readily induces p100 processing without any further stimuli (12) (data not shown). The low level of NIK protein seems to involve TRAF3-mediated ubiquitination and subsequent degradation (39). Importantly, NIK protein level is upregulated by the noncanonical NF-B stimuli (39) (data not shown), which may induce de novo protein synthesis of NIK FIG. 3. Multiple domains located in the N terminus of p100 are involved in constitutive processing of p100 Cterminal truncation mutant. A, schematic representation of a constitutive processing form of p100 and its mutants. Full-length p100 and its processing product p52 are also represented. The abbreviations used are the same as those in Fig.  1. B, deletion of N-terminal domains prevents constitutive processing of p100. 293 cells were transfected with p100N or its mutants, followed by IB analysis using anti-p100N to detect p100N and p52 (p100 processing).

FIG. 4. Domains involved in p100 C-and N-terminal interaction.
A, schematic representation of p100 C-and N-terminal truncation mutants. The important domains within p100, including RHD, are indicated. The abbreviations used are the same as those in Fig. 1. B, the N terminus of p100 interacts with its C terminus. 293 cells were transfected with expression vectors encoding HA-tagged p52 either in the absence (Ϫ) or presence (ϩ) of Myc-tagged p100-(406 -900) or full-length p100. The cell lysates were subjected to immunoprecipitation using anti-Myc, and the co-precipitated p52 (top) was detected by immunoblotting using anti-HA-HRP. Some of the cell lysates were directly subjected to immunoblotting to monitor the expression levels of p52 (middle), p100-(406 -900), and p100 wild type (bottom) using anti-HA-HRP and anti-Myc, respectively. C, C-terminal sequences of p100 involved in its C-and N-terminal interaction. 293 cells were transfected with HA-tagged p52 together with Myc-p100-(406 -900) or its mutants. Cell lysates were subjected to IP using anti-Myc for examining the p52/p100-(406 -900) interaction (top) or to IB for detecting the expression levels of p52 (middle), p100-(406 -900), and its mutants (bottom) using the indicated antibodies. D, N-terminal sequences of p100 involved in its C-and N-terminal interaction. Myc-p100-(406 -900), together with HA-p52 or its mutants, was transfected into 293 cells, followed by IP using anti-Myc antibody. The co-precipitated p52 and its mutants (top) were detected by immunoblotting using anti-HA-HRP. The expression levels of p100-(406 -900) (bottom), p52, and its mutants (middle) were monitored by direct IB using cell lysates. E, the N-terminal sequences of p100 involved in binding to its ARD. Myc-p100C ⌬C2 (p100-(406 -753)), together with HA-p52 or its mutants, was transfected into 293 cells, followed by IP using anti-HA antibody. The co-precipitated p100-(406 -753) (top) was detected by immunoblotting using and/or TRAF3 degradation indirectly via other members of the TRAF family. Indeed, TRAF2 and TRAF6 can induce p100 processing via NIK (data not shown). Thus, it seems that the function of NIK is positively or negatively regulated by different TRAF family members. Whereas the regulation of NIK remains to be further investigated, the mechanisms of NIKinduced p100 processing have been well addressed. We have demonstrated that, by direct interaction with IKK␣, NIK activates and recruits IKK␣ to p100 via two specific amino acid residues, serine 866 and serine 870, of p100. Once recruited into the p100 complex, activated IKK␣ phosphorylates both N-and C-terminal serines of p100 (serines 99, 108, 115, 123, and 872), thereby triggering p100 ubiquitination mediated by ␤-TrCP and subsequent processing by the proteasome (12,13,34,35).
Although p100 processing is hardly detected in most cell types, including activated T cells, p52 is overproduced in HTLV-I-infected T cells due to the aberrantly persistent processing of p100 (14). Since constitutive processing of p100 is associated with various human lymphomas and leukemia (22)(23)(24)(25)(26), and constitutive processing forms of p100 show transforming activity in vitro (27), deregulated processing of p100 may be part of the oncogenic mechanism of HTLV-I. We have demonstrated that HTLV-I-encoded oncoprotein Tax is a potent inducer of p100 processing (14). This finding made Tax the first pathogenic agent found to deregulate the p100 processing pathway. Interestingly, Tax and NIK utilize similar strategies to trigger p100 processing, although Tax-induced p100 processing does not require NIK (14). Like NIK-mediated processing, Taxinduced p100 processing also involves IKK␣, which phosphorylates serines located in both N and C termini of p100 to trigger ubiquitination of p100 by the E3 ligase ␤-TrCP (34,35). Although this general mechanism plays a central role in inducible processing of p100 under both physiological and pathogenic FIG. 6. Different effects of NIK and Tax on p100 nuclear shuttling. A, Tax but not NIK induces nuclear translocation of p100. 293 cells were transfected with p100 either in the presence or absence of NIK or Tax, followed by the use of cytoplasmic or nuclear extracts for IB to detect p100, p52 (p100 processing) (top panel), NIK (second panel), and Tax (third panel). The subcellular expression of endogenous IKK␣ (fourth panel) and c-Myc (bottom panel) were also detected for cytoplasmic and nuclear protein control, respectively. B, immunofluorescence analysis to determine the effect of NIK or Tax on subcellular localization of p100. COS cells were transfected with p100 either in the presence or absence of NIK or Tax. The subcellular localization of p100 was detected by immunofluorescence assays using the C terminus-specific anti-p100 (anti-p100C) primary antibody followed by fluorescein isothiocyanate-conjugated anti-rabbit Ig secondary antibody (upper panels). The nuclei were counterstained with Hoechst 33258 and visualized with a UV filter (lower panels). As indicated, some of the cells were treated with leptomycin B (LMB) for 5 h before staining.
anti-Myc. The expression levels of p100-(406 -753) (bottom) and p52 and its mutants (middle) were monitored by direct IB using cell lysates. F, the C-terminal IKK␣-targeting serines of p100 rescued its ARD binding to dimerization domain or NLS disruption mutant of p52. Myc-p100C ⌬DD, together with HA-p52 or its mutants, was transfected into 293 cells, followed by IP using anti-HA antibody. The co-precipitated p100C ⌬DD (top) was detected by immunoblotting using anti-Myc. The expression levels of p100C ⌬DD (bottom) and p52 and its mutants (middle) were monitored by direct IB using cell lysates. G, the DD of p100 rescues its ARD binding to the dimerization domain or NLS disruption mutant of p52. Myc-p100C ⌬S, together with HA-p52 or its mutants, was transfected into 293 cells, followed by IP using anti-HA antibody. The co-precipitated p100C ⌬S (top) was detected by immunoblotting using anti-Myc. The expression levels of p100C ⌬S (bottom) and p52 and its mutants (middle) were monitored by direct IB using cell lysates.
FIG. 5. Two ␣-helices of p100 function as a suppressor of constitutive recruitment of IKK␣ to p100. 293 cells were transfected with the indicated constructs, followed by immunoprecipitation using anti-p100N. The co-precipitated IKK␣ was detected by immunoblotting using anti-HA-HRP (top panel). Some of the cell lysates were directly subjected to immunoblotting to monitor the expression levels of HAtagged IKK␣ (second panel), Tax (third panel), and p100 and its processing product p52 (bottom panel).
conditions, different mechanisms may also exist for Tax induction of this novel NF-B pathway. First, NIK-mediated p100 processing is independent of IKK␥, whereas IKK␥ is essential for Tax induction of p100 processing by serving as an adaptor between Tax and IKK␣ (14,33,41). Second, NIK-mediated processing of p100 relies on ␤-TrCP (34,36), whereas a ␤-TrCPindependent mechanism also contributes to Tax-induced p100 processing (35). Our current studies indicate that this ␤-TrCPindependent mechanism of p100 processing induced by Tax may be attributed to p100 nuclear shuttling specifically triggered by Tax (37), since Tax but not NIK induces significant nuclear translocation of p100 (Fig. 6) (see below for detailed discussion).
Another important mechanism for p100 processing involves its sequence. It appears that p100 processing is both positively and negatively regulated by its sequence (12). Whereas the C-terminal ARD and DD suppress constitutive processing of p100 (Fig. 2), the multiple regions located in both C and N termini of p100 are required for inducible and constitutive processing of p100 (Figs. 1 and 3), suggesting a common mechanism for p100 processing. Although so many regions of p100 are positively or negatively involved in its processing, they work collaboratively and delicately. Since the optimal C-and N-terminal interaction of p100 involves multiple domains located in both its N and C termini (Fig. 4), it seems plausible that this C-and N-terminal interaction could serve as a platform to form a potential three-dimensional domain by bringing its C-and N-terminal sequences together. This functional three-dimensional domain is essential not only for suppression of constitutive processing but also for inducible processing of p100. Usually, the C-terminal IKK␣ recruiting site of p100 is hidden within the three-dimensional domain by the N-terminal two ␣-helices (Fig. 5). In the presence of NIK or Tax, the conformation of this three-dimensional domain will be changed, which promotes the recruitment of IKK␣ to p100 and subsequent phosphorylation of p100 (Fig. 5). Once interacting with p100, activated IKK␣ by NIK or Tax phosphorylates the serines located in both N and C termini of p100, which triggers ␤-TrCP recruitment into p100 and subsequent p100 ubiquitination ( Fig. 2) (34,35). On the other hand, disruption of this functional domain by deleting its ARD and DD makes the N-terminal NLS exposed ( Fig. 2 and 3), which results in p100 nuclear translocation and subsequent constitutive processing (37). This model also provides an explanation for why IKK␣targeting serines (serines 99, 108, 115, 123, 868, 870, and 872) in p100 are located at distant positions in amino acid sequences (34,35).
Whereas the N-terminal sequences of the presumptive threedimensional domain are required for p100 processing (Figs. 1  and 3), deletion of the C-terminal ARD and DD of p100 paradoxically leads to its constitutive processing (12,37) (also see Fig. 2). Although currently the mechanism of constitutive processing of p100 is largely unclear, a recent study suggested that constitutive processing of p100 is associated with its nuclear expression (37). We also found that deletion of ARD and DD leads to a significant level of nuclear translocation of p100 (data not shown). These data are also consistent with our studies showing that NLS of p100 may be masked by its Nand C-terminal interaction involving ARD, DD, and NLS (Fig. 4). Interestingly, Tax also induces a significant nuclear localization of p100, whereas NIK fails to alter the subcellular distribution of p100 (Fig. 6). Thus, it appears that NIKinduced p100 processing only occurs in the cytoplasm, whereas Tax-deregulated p100 processing may happen in both cytoplasm and nucleus. Given that the constitutive processing of p100 C-terminal deletion mutants is ␤-TrCP-independent and regulated by nuclear translocation of p100, these results suggest that p100 nuclear shuttling may be the ␤-TrCP-independent mechanism of Tax-induced p100 processing.
In summary, we have found a common mechanism for both inducible and constitutive processing of p100. We show that p100 processing is tightly and delicately regulated by its sequences, which may form a tertiary domain. However, different mechanisms also exist. Whereas NIK-mediated p100 processing is ␤-TrCP-dependent and occurs in the cytoplasm, constitutive processing of p100 is ␤-TrCP-independent and nuclear shuttling-associated. On the other hand, Tax takes advantage of both mechanisms for its deregulation of p100 processing.