An Atypical Tumor Necrosis Factor (TNF) Receptor-associated Factor-binding Motif of B Cell-activating Factor Belonging to the TNF Family (BAFF) Receptor Mediates Induction of the Noncanonical NF-κB Signaling Pathway*

BAFF receptor (BAFFR) is a member of the TNF receptor (TNFR) superfamily that regulates the survival and maturation of B cells. BAFFR exerts its signaling function by inducing activation of NF-κB, although the underlying mechanism has not been well defined. By using a chimeric BAFFR, we show that BAFFR preferentially induces the noncanonical NF-κB signaling pathway. This specific function of BAFFR is mediated by a sequence motif, PVPAT, which is homologous to the TRAF-binding site (PVQET) present in CD40, a TNFR known to induce both the canonical and noncanonical NF-κB pathways. Mutation of this putative TRAF-binding motif within BAFFR abolishes its interaction with TRAF3 as well as its ability to induce noncanonical NF-κB. Interestingly, modification of the PVPAT sequence to the typical TRAF-binding sequence, PVQET, is sufficient to render the BAFFR capable of inducing strong canonical NF-κB signaling. Further, this functional acquisition of the modified BAFFR is associated with its stronger and more rapid association with TRAF3. These findings suggest that the PVPAT sequence of BAFFR not only functions as a key signaling motif of BAFFR but also determines its signaling specificity in the induction of the noncanonical NF-κB pathway.

interacts with two TRAF members, TRAF2 and TRAF3 (23). This TRAF-binding motif, with the consensus sequence of PXQXT(S), has also been found in several other TNFR family members (24). CD40 is able to induce the activation of both noncanonical and canonical NF-B signaling pathways (8,19). Further, ligation of CD40 is sufficient for triggering the expression of various target genes, including those encoding the apoptosis inhibitors Bfl-1/A1 and Bcl-XL (25). Unlike CD40, BAFFR does not contain a conserved TRAF-binding motif. The signaling determinant of BAFFR has not been well defined, although truncation at its C-terminal region abolishes its signaling function (2,3,21). The precise role of BAFFR in mediating different pathways of NF-B activation also remains unclear. Whereas BAFFR clearly mediates the induction of the noncanonical NF-B signaling pathway (18,21), its role in activating the canonical pathway remains ambiguous (8,26). One complexity that may contribute to the discrepancy of the prior studies is the involvement of different BAFF-responding receptors: TACI, BCMA, and BAFFR (27).
In the present study, we examined the specific signaling function of BAFFR using a chimeric receptor containing the ligand binding domain of CD40 and the intracellular signaling domain of BAFFR. We show that the BAFFR predominantly targets the activation of noncanonical NF-B. Moreover, we have identified a signaling motif of BAFFR that is critical for its signaling function in NF-B activation. This motif is homologous to the TRAF-binding site of CD40, although it does not contain all the conserved residues of the typical TRAF-binding motif. Substitution of the conserved residues with alanines completely abolishes the function of BAFFR in the induction of p100 processing. Interestingly, mutation of the BAFFR signaling motif to the typical TRAF-binding sequence rendered the BAFFR competent in activation of both the noncanonical and canonical NF-B pathways. These findings suggest that the atypical TRAF-binding sequence of BAFFR not only functions as a key signaling motif but also determines the unique signaling property of BAFFR in mediating noncanonical NF-B activation.

MATERIALS AND METHODS
Plasmid Constructs-The hCD40-BAFFR encodes the extracellular and transmembrane domains of human CD40 (amino acids 1-215) and the cytoplasmic domain of murine BAFFR (amino acids 98 -175). The corresponding cDNA fragments were isolated by reverse transcription-PCR using human BJAB and murine M12 B cell lines, respectively, and cloned into the pCLXSN retroviral vector (provided by Dr. Inder M. Verma (28)). To create the M1 and M2 mutants, site-directed mutagenesis was performed, using hCD40-BAFFR as template, to introduce the indicated amino acid substitutions in the cytoplasmic domain of BAFFR. pcDNA-HA-TRAF2 was described previously (29). The oligonucleotide primers for PCR and site-directed mutagenesis (sequences are available upon request) were synthesized at the Macromolecular Core of Penn State University College of Medicine. All the plasmids used in this work were sequenced at the Molecular Genetics Core of Pennsylvania State University College of Medicine.
Cell Culture, Transfection, and Retroviral Infection-Murine B cell line M12.4.1 (called M12 in this report) (30) and its derivative stably transfected with hCD40 (M12-hCD40) (31) were kindly provided by Dr. Gail A. Bishop. The M12-hCD40-BAFFR, M12-hCD40-BAFFR M1, and M12-hCD40-BAFFR M2 cells were created by infecting the M12 cells with the pCLXSN retroviruses encoding wild type and mutant forms of the hCD40-BAFFR chimeras as previously described (32). After drug selection (G418), bulk infected cells were used in the experiments to avoid clonal variations. All of these B-cell lines were cultured in RPMI medium supplemented with 10% fetal bovine serum, 2 mM L-glutamine, antibiotics, and 10 M 2-mercaptoethanol. The kidney carcinoma cell line 293T was cultured in Dulbecco's medium with the same supplements except the lack of 2-mercaptoethanol. Transient transfection of 293 cells was carried out using Lipofectamine 2000 (Invitrogen). For retrovirus production (32), the cells were transfected with FuGENE 6 (Roche Applied Science).
Fluorescence-activated Cell Sorting-About 1 ϫ 10 6 cells were resuspended in 100 l of FACS buffer (2% fetal bovine serum in phosphatebuffered saline). The cells were stained with fluorescein isothiocyanateconjugated anti-hCD40 antibody and subjected to FACS analysis using a BD Biosciences FACScan at the Pennsylvania State College of Medicine Cell Science/Flow Cytometry Core Facility.
RNase Protection Assays-Total cellular RNA was isolated from M12 and derivative cells using the TRI reagent (Molecular Research Center, Inc., Cincinnati, OH). RNase protection assays were performed using the BD RiboQuant Reagents and mouse APO-2 template set according to the manufacturer's instruction (BD Biosciences).
Immunoblotting and Electrophoresis Mobility Shift Assays-Whole cell lysates were prepared in radioimmune precipitation assay buffer (33), and nuclear extracts were prepared as previous described (34). The proteins were fractionated in SDS-polyacrylamide gels, electrophoretically transferred onto nitrocellulose membranes, and then subjected to immunoblotting using the indicated primary antibodies and HRP-conjugated secondary antibodies (35). Nuclear extracts were subjected to EMSA using a 32 P-radiolabeled B probe (34).
Coimmunoprecipitation Assays to Detect BAFFR/TRAF3 Binding-The coimmunoprecipitation assays were essentially according to Hostager et al. (36). Briefly, M12 cells (2 ϫ 10 7 ) stably expressing hCD40-BAFFR and derivatives or full-length hCD40 were resuspended in 1 ml of fresh growth medium and stimulated for the indicated times with anti-hCD40. The cells were washed twice with PBS and then lysed in 600 l of a lysis buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 0.1% SDS, 1% Triton X-100, 20 mM ␤-glycerophosphate, 5 mg/ml p-nitrophenyl phosphate, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, and 1/1000 volume of a protease inhibitor mixture, Sigma). The protein lysates were incubated in at 4°C for 2 h with 30 l of protein G-agarose to isolate the anti-hCD40/receptor immune complexes. After washing the agarose beads with lysis buffer for three times, the coprecipitated TRAF3 was eluted from the beads with SDS loading buffer and detected by IB using HRP-conjugated anti-TRAF3.

BAFFR Preferentially Induces the Noncanonical NF-B Sig-
naling Pathway-To study the signaling function of BAFFR, we generated a chimeric receptor containing the cytoplasmic domain of murine BAFFR and the extracellular ligand-binding domain of human CD40 (hCD40) (Fig. 1A). Within this chimera, the signaling activity of BAFFR could be stimulated with an agonistic anti-hCD40 antibody without the interference of TACI and BCMA or the endogenous BAFFR. Further, this chimeric receptor allows direct comparison of the signaling functions of BAFFR and CD40 using the same stimulus. By retroviral infection, we stably expressed the hCD40-BAFFR chimera in a murine B cell line, M12 (30), which has been used as a model for studying the signaling function of both CD40 (36 -38) and BAFFR (29). Because close to 100% of the cells expressed hCD40-BAFFR (Fig. 1B), we used the bulk-infected cells for the signaling studies. To compare the signaling activity of BAFFR with that of CD40, we also included M12 cells stably expressing the wild type (intact) human CD40 (M12-hCD40).
We examined the BAFFR-mediated activation of NF-B signaling pathways by detecting the processing of p100 and degradation of IB␣, indicators of noncanonical and canonical NF-B pathways, respectively (17). Consistent with the lack of cross-reactivity of the anti-hCD40 (G28 -5) with murine CD40 (31), the parental murine M12 B cells did not respond to stimulation by the agonistic anti-hCD40 antibody (Fig. 1C, top panel, lane 2). On the other hand, the hCD40-BAFFR cells efficiently responded to the anti-hCD40, as demonstrated by the inducible processing of p100 to p52 (lane 4). A similar result was obtained with the M12-CD40 cells (lane 6). Because the induction of p100 processing by recombinant BAFF is associated with degradation of TRAF3 (29), we examined whether the BAFFR chimera also induces TRAF3 degradation. Indeed, TRAF3 was lost concomitant with the induction of p100 processing by both CD40-BAFFR and the intact CD40 (Fig. 1C, middle panel). The loss of TRAF3 was not due to its translocation to the nucleus (data not shown) and could be detected at early time points of cell stimulation (see Fig. 2F).
Using the CD40-BAFFR chimera system, we next examined whether BAFFR targets the canonical NF-B signaling pathway by analyzing degradation of IB␣. Consistent with the ability of CD40 to activate canonical NF-B, stimulation of M12-hCD40 cells resulted in efficient degradation of IB␣ (Fig.  1D, upper panel, lanes [5][6][7][8]. In contrast, however, stimulation of the M12-hCD40-BAFFR cells only led to a weak loss of IB␣ (lanes 1-4). This result was not due to the lower expression of hCD40-BAFFR than hCD40 (Fig. 1B). Further, because the p100 processing was even more strongly induced by the hCD40-BAFFR than hCD40 (Fig. 1C), it is unlikely that the ineffectiveness of hCD40-BAFFR in inducing IB␣ degradation was due to its intrinsically weaker signaling activity. Thus, unlike CD40, BAFFR is a TNFR member that preferentially stimulates the noncanonical NF-B signaling pathway.
An Atypical TRAF-binding Sequence Motif Is Critical for BAFFR Signaling-The signaling function of TNFRs requires TRAF molecules. A specific TRAF-binding motif has been iden-tified in a number of TNFR members (39), with those in CD40 being the most extensively characterized (40). To understand the biochemical mechanism mediating the unique signaling function of BAFFR, we compared the amino acid sequences of the cytoplasmic domains of BAFFR and CD40. Interestingly, although BAFFR lacks a typical TRAF-binding motif, it possesses a sequence element (PVPAT) that shares significant homology, in both location and sequence, with the TRAF-binding motif of CD40 ( Fig. 2A, bolded and underlined). To examine the role of this sequence element in BAFFR signaling, we performed site-directed mutagenesis to substitute the conserved amino acids within the hCD40-BAFFR chimera with alanines (Fig. 2B, M1). In parallel, we also mutated the BAFFR motif to the typical TRAF-binding sequence, PVQET, of CD40 (Fig. 2B, M2). FACS analyses revealed a comparable expression level between the wild type and mutant forms of hCD40-BAFFR (Fig. 2C). Importantly, the M1 mutant (harboring PVPAT to AVAAA mutations) completely lost the ability to induce p100 processing (Fig. 2D, top panel, lane 4). On the other hand, the M2 mutant remained competent in inducing p100 processing (lane 6). The mutations introduced into M1, but not M2, also abolished the induction of TRAF3 degradation (Fig. 2D, middle panel). A separate experiment using shorter stimulation time points revealed that the TRAF3 degradation by wild type BAFFR and M2 mutant occurred around 2 h following receptor cross-linking, but this response was not detected in cells expressing the M1 mutant (Fig. 2F).
Because a primary consequence of p100 processing is nuclear translocation of RelB and p52 NF-B heterodimer, we analyzed the expression of these two NF-B members in the nucleus of Thus, these data establish the PVPAT sequence as a key signaling motif of BAFFR that mediates the induction of p100 processing and nuclear expression of the noncanonical NF-B members.
The PVPAT Motif Also Determines the Signaling Specificity of BAFFR-An important question regarding BAFFR signaling is how its signaling specificity is regulated. As shown in Fig. 1  (C and D), BAFFR preferentially stimulates the noncanonical NF-B signaling pathway, whereas CD40 stimulates both the canonical and noncanonical NF-B pathways. One hypothesis we were considering was that the unique signaling function of BAFFR was likely due to its possession of an atypical TRAFbinding motif. To test this hypothesis, we examined the effect of PVPAT-to-PVQET conversion within the cytoplasmic domain of BAFFR on its ability to induce IB␣ degradation. As expected, cross-linking of the wild type hCD40-BAFFR only led to a low level of IB␣ degradation (Fig. 3A, upper panel, lanes  1-4), which was not detected in the M1 mutant of the receptor (lanes 5-8). Remarkably, however, conversion of the PVPAT sequence to PVQET rendered the BAFFR capable of stimulating efficient degradation of IB␣ (lanes 9 -12). Thus, a twoamino acid substitution in the putative TRAF-binding site of BAFFR is sufficient to convert it into a receptor capable of stimulating the canonical pathway of NF-B.
One major difference between the canonical and noncanonical NF-B pathways resides in their signaling kinetics. Whereas the canonical pathway is rapid and transient, the noncanonical pathway is slow and persistent. We performed EMSA to examine the acute and delayed phases of NF-B activation by the wild type and mutant forms of BAFFR. During the early phase of cell stimulation, the wild type BAFFR only mediated a weak activation of NF-B (Fig. 3B, lanes 2 and  3), but the M2 mutant of BAFFR caused a much stronger activation of NF-B under the same conditions (lanes 6 and 7). This result was consistent with the elevated induction of IB␣ degradation by the M2 mutant (Fig. 3A). In contrast to the acute NF-B activation, no appreciable difference was detected in the late-phase NF-B activation mediated by the wild type and M2 mutant of BAFFR (compare lanes 4 and 8). Further, the M1 mutant of BAFFR was defective in activation of both the acute and delayed phases of NF-B (lanes 10 -12). Thus, the PVPAT motif of BAFFR specifically mediates the delayed activation of NF-B, which in turn is associated with the processing of p100 and nuclear translocation of RelB/p52 (Fig. 2, D  and E).

FIG. 2. A putative TRAF-binding motif is required for BAFFR-mediated induction of the noncanonical NF-B signaling pathway.
A, sequence alignment of the C-terminal portions of murine (m) and human (h) BAFFR and CD40. The TRAF-binding motifs in CD40 and a homologous sequence in BAFFR are shown in bolded and underlined letters. B, schematic picture of hCD40-BAFFR chimeras. All the chimeras contain the N-terminal region of human CD40 and C-terminal region of murine BAFFR, as described in Fig. 1A. The wild type PVPAT motif and the altered sequences in two mutants (M1 and M2) are indicated. C, FACS analysis of receptor expression. M12 cells stably expressing the indicated hCD40-BAFFR chimeras were subjected to FACS analyses as described in Fig. 1B. D, processing of p100 and degradation of TRAF3. The indicated cells were stimulated for 24 h with anti-hCD40 followed by isolating total-cell lysates and IB to determine p100 processing (top panel), TRAF3 degradation (middle panel), and tubulin level (bottom panel). E, M12 cells stably expressing the indicated chimeric receptors were stimulated as in D. Nuclear extracts were isolated and subjected to IB using anti-RelB (top panel), anti-p52 (middle panel), and anti-Oct-1 antibodies. Oct-1 serves as a loading control, because it is constitutively expressed in the nucleus. F, M12 cells stably expressing the indicated chimeric receptors were stimulated with anti-hCD40 for the indicated times. Whole cell lysates were prepared and subjected to IB analyses using anti-TRAF3 (upper panel) and anti-tubulin (lower panel), respectively.
To further assess the signaling specificity of BAFFR, we examined the expression of downstream genes. Previous studies suggest that the BAFFR signal alone is insufficient for triggering significant induction of anti-apoptotic genes, such as Bfl-1/A1 (7). In contrast, the Bfl-1/A1 gene can be potently stimulated by the CD40 signal (25,41). We thus examined the mRNA levels of this downstream gene in cells expressing the chimeric BAFFR receptors. Consistent with the prior studies, stimulation of the wild type BAFFR did not appreciably induce the expression of Bfl-1/A1 mRNA (Fig. 3C, lanes 1-5). Interestingly, however, stimulation of the BAFFR harboring PVPAT-to-PVQET modifications led to potent induction of Bfl-1/A1 (lanes 11-15). Together, these data suggest that the atypical TRAF-binding motif of BAFFR not only serves as a key signaling motif but also regulates its signaling specificity.
The PVPAT Motif Regulates Recruitment of TRAF3 to BAFFR-A prior study demonstrated that BAFFR specifically interacts with TRAF3 (42). To directly determine the role of the PVPAT motif in TRAF binding, we examined the recruitment of TRAF3 to the hCD40-BAFFR receptor in cells stimulated with anti-hCD40. In agreement with the previous report, we found that TRAF3 was recruited to the wild type BAFFR upon ligation by anti-hCD40, although significant BAFFR/TRAF3 association was not detected until 1 h following anti- hCD40   FIG. 3. Induction of canonical NF-B  by a modified BAFFR. A, IB analyses to determine the degradation of IB␣. M12 B cells stably expressing the indicated chimeric receptors were stimulated with anti-hCD40 antibody, and the cell lysates were subjected to IB using anti-IB␣ or anti-tubulin. B, EMSA to detect the activation of NF-B. The different cells were stimulated with anti-hCD40 for the indicated times and then subjected to nuclear extract isolation. NF-B DNA binding activity was examined by EMSA using a 32 P-radiolabeled B probe. The NF-B/DNA complexes are indicated, and the free probe is not shown. C, RNase protection assays to detect the induction of Bfl-1/A1 gene expression by BAFFR variants. M12 cells expressing the indicated chimeric BAFFRs were stimulated with anti-hCD40. Total RNA was isolated and subjected to RPA analyses to detect Bfl-1/A1 expression. Two housekeeping genes, L32 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), were included in the assay to monitor the consistency of RNA amounts.
FIG. 4. Coimmunoprecipitation to detect the binding of BAFFR with TRAF2 and TRAF3. A, M12 cells stably expressing the different BAFFR chimeras or the intact hCD40 were stimulated with anti-hCD40 for the indicated times. The receptor complexes were isolated by IP, and the coprecipitated TRAF3 was detected by IB using an HRP-conjugated anti-TRAF3 antibody (upper panel). The level of TRAF3 in the cell lysates were monitored by direct IB (lower panel). B, 293 cells were transfected with HAtagged TRAF2 (in pcDNA vector) together with either vector control (V) or the indicated hCD40-BAFFR chimeric receptors. The receptor complexes were isolated by IP using anti-hCD40 followed by detecting the associated TRAF2 by IB using HRP-conjugated anti-HA (upper panel). The TRAF2 expression level was monitored by direct IB using HRP-anti-HA (lower panel). treatment (Fig. 4A, upper panel, lane 3). More importantly, this physical interaction critically requires the PVPAT motif, because the M1 mutant (carrying the AVAAA mutation) failed to recruit TRAF3 (lanes 4 -6). Interestingly, the M2 mutant (carrying the PVQET mutation) interacted with TRAF3 more strongly and more rapidly than the wild type BAFFR (lanes [7][8][9]. This property of the BAFFR M2 mutant was reminiscent of the CD40 molecule, which potently and rapidly interacted with TRAF3 in response to the anti-hCD40 treatment (lanes 10 -12). These biochemical results support the functional studies described above.
We next examined whether the PVPAT to PVQET conversion was sufficient to render the modified BAFFR capable of binding to TRAF2. These studies were performed using transfected 293 cells. Consistent with a previous study, the wild type BAFFR did not interact with TRAF2 even under the overexpression conditions (Fig. 4B, upper panel, lane 2). Similarly, the M1 mutant was inactive in this physical interaction (lane 3). Interestingly, however, the M2 mutant, which carries the PVQET motif, exhibited a significant TRAF2-binding activity (lane 4). Together, these results suggest that the atypical TRAF-binding motif of BAFFR specifically interacts with TRAF3 but not with TRAF2, which likely contributes to the signaling specificity of BAFFR. Further, conversion of the atypical TRAF-binding motif to a conserved one is sufficient to alter the specificity of BAFFR in TRAF binding and NF-B signaling.

DISCUSSION
The results of this study provide an insight into the signaling mechanism of BAFFR. We have identified a signaling motif, PVPAT, within the cytoplasmic domain of BAFFR that shares significant sequence homology with the TRAF-binding motif, PVQET, of CD40. However, the BAFFR motif does not fall into the consensus sequence of the TRAF-binding sequence (PXQXT) present in several TNFR superfamily members. We propose that this atypical TRAF-binding motif is responsible for the unique signaling property of BAFFR that preferentially targets the noncanonical NF-B pathway. We have obtained several lines of evidence that support this hypothesis. First, alanine substitutions of the conserved amino acids within the PVPAT motif abolishes the ability of BAFFR to induce p100 processing and nuclear translocation of RelB and p52 (Fig. 2), key steps in noncanonical NF-B signaling (17). Second, a two-amino acid substitution that converts the PVPAT sequence to the typical TRAF-binding sequence (PVQET) renders the modified BAFFR competent in activating the canonical NF-B pathway (Fig. 3). Third, the PVPAT motif is essential for the suboptimal association of BAFFR with TRAF3, and the PVPATto-PVQET mutation significantly enhances the TRAF3-binding activity of BAFFR (Fig. 4A). Additionally, we have also shown that mutating PVPAT to PVQET has no obvious effect on BAFFR-mediated p100 processing or the delayed activation of NF-B (Fig. 2), but this manipulation markedly enhances the acute-phase NF-B activation known to be associated with canonical NF-B signaling. Of note, the PVPAT to PVQET sequence change allows the modified BAFFR to interact with TRAF2, which may contribute to the gain-of-function in the canonical NF-B signaling.
We have recently shown that TRAF3 functions as a negative regulator of p100 processing induced by CD40 and BAFFR (29). Our current study suggests that recruitment of TRAF3 to BAFFR is critical for BAFFR-mediated induction of p100 processing. The M1 mutant of BAFFR, which is deficient in recruiting TRAF3, also fails to induce p100 processing. It is unclear how the TRAF3 recruitment to BAFFR contributes to signal transduction leading to p100 processing. One possibility is that the receptor recruitment serves as a trigger for targeting the degradation of TRAF3, which appears to be an important step in the noncanonical NF-B signaling pathway (29). In support of this idea, the functionally inactive BAFFR mutant (M1) fails to induce TRAF3 degradation (Fig. 2D). However, it also remains possible that the receptor recruitment of TRAF3 plays an active role in the initiation of the noncanonical NF-B signaling. Examination of this latter hypothesis will need genetically manipulated B cells lacking TRAF3.
TRAF2 and TRAF3 are generally thought to bind to the same sequence present in the cytoplasmic tails of TNFR family members (23). However, detailed mutagenesis analysis using the cytoplasmic domain of CD40 reveals that TRAF2 and TRAF3 exhibit subtle differences in target sequence requirement (40). Our data suggest that the atypical TRAF-binding sequence within BAFFR favors binding to TRAF3 but does not interact with TRAF2. This biochemical property may in turn contribute to the signaling specificity of BAFFR in the induction of noncanonical NF-B activation. In support of this hypothesis, conversion of the atypical TRAF-binding motif to a conserved TRAF-binding motif renders the modified BAFFR competent to interact with both TRAF3 and TRAF2 and to target both the noncanonical and canonical NF-B signaling pathways.