Interaction of Tumor Necrosis Factor Receptor-associated Factor 6 (TRAF6) and Vav3 in the Receptor Activator of Nuclear Factor κB (RANK) Signaling Complex Enhances Osteoclastogenesis*

The signaling pathway downstream of stimulation of receptor activator of nuclear factor κB (RANK) by RANK ligand is crucial for osteoclastogenesis. RANK recruits TNF receptor-associated factor 6 (TRAF6) to TRAF6-binding sites (T6BSs) in the RANK cytoplasmic tail (RANKcyto) to trigger downstream osteoclastogenic signaling cascades. RANKcyto harbors an additional highly conserved domain (HCR) that also activates crucial signaling during RANK-mediated osteoclastogenesis. However, the functional cross-talk between T6BSs and the HCR in the RANK signaling complex remains unclear. To characterize the cross-talk between T6BSs and the HCR, we screened TRAF6-interacting proteins using a proteomics approach. We identified Vav3 as a novel TRAF6 binding partner and evaluated the functional importance of the TRAF6-Vav3 interaction in the RANK signaling complex. We demonstrated that the coiled-coil domain of TRAF6 interacts directly with the Dbl homology domain of Vav3 to form the RANK signaling complex independent of the TRAF6 ubiquitination pathway. TRAF6 is recruited to the RANKcyto mutant, which lacks T6BSs, via the Vav3 interaction; conversely, Vav3 is recruited to the RANKcyto mutant, which lacks the IVVY motif, via the TRAF6 interaction. Finally, we determined that the TRAF6-Vav3 interaction resulting from cross-talk between T6BSs and the IVVY motif in RANKcyto enhances downstream NF-κB, MAPK, and NFATc1 activation by further strengthening TRAF6 signaling, thereby inducing RANK-mediated osteoclastogenesis. Thus, Vav3 is a novel TRAF6 interaction partner that functions in the activation of cooperative signaling between T6BSs and the IVVY motif in the RANK signaling complex.

The signaling pathway downstream of stimulation of receptor activator of nuclear factor B (RANK) by RANK ligand is crucial for osteoclastogenesis. RANK recruits TNF receptor-associated factor 6 (TRAF6) to TRAF6-binding sites (T6BSs) in the RANK cytoplasmic tail (RANK cyto ) to trigger downstream osteoclastogenic signaling cascades. RANK cyto harbors an additional highly conserved domain (HCR) that also activates crucial signaling during RANK-mediated osteoclastogenesis. However, the functional cross-talk between T6BSs and the HCR in the RANK signaling complex remains unclear. To characterize the cross-talk between T6BSs and the HCR, we screened TRAF6-interacting proteins using a proteomics approach. We identified Vav3 as a novel TRAF6 binding partner and evaluated the functional importance of the TRAF6-Vav3 interaction in the RANK signaling complex. We demonstrated that the coiled-coil domain of TRAF6 interacts directly with the Dbl homology domain of Vav3 to form the RANK signaling complex independent of the TRAF6 ubiquitination pathway. TRAF6 is recruited to the RANK cyto mutant, which lacks T6BSs, via the Vav3 interaction; conversely, Vav3 is recruited to the RANK cyto mutant, which lacks the IVVY motif, via the TRAF6 interaction. Finally, we determined that the TRAF6-Vav3 interaction resulting from cross-talk between T6BSs and the IVVY motif in RANK cyto enhances downstream NF-B, MAPK, and NFATc1 activation by further strengthening TRAF6 signaling, thereby inducing RANK-mediated osteoclastogenesis. Thus, Vav3 is a novel TRAF6 interaction partner that functions in the activation of cooperative signaling between T6BSs and the IVVY motif in the RANK signaling complex.
The integrity of the skeletal system is maintained by the process of bone remodeling, in which old and damaged bone is continuously replaced with new bone through the balanced action of bone-resorbing osteoclasts (OCs) 3 and bone-forming osteoblasts (1)(2)(3). As the only cells with bone-resorbing activity, OCs are clinically important in bone-related diseases; accelerated bone destruction by OCs results in pathological bone loss associated with rheumatoid arthritis, multiple myeloma, metastatic cancer, and osteoporosis, whereas impaired function of OCs leads to osteopetrotic disorders (2,4).
OCs are multinucleated giant cells that are formed by the fusion of monocyte/macrophage lineage precursors through the process of OC differentiation or osteoclastogenesis (2). The signaling pathways of receptor activator of nuclear factor B ligand (RANKL) and macrophage colony-stimulating factor (M-CSF) play key roles in osteoclastogenesis (2,5). Mice lacking either RANK or RANKL exhibit severe osteopetrotic phenotypes, highlighting the importance of RANK-RANKL signaling as a crucial regulator of OC differentiation, activation, and survival (6,7). RANK-RANKL signaling is initiated primarily by the recruitment of cytoplasmic TNF receptor-associated factors (TRAFs) that trigger the activation of signaling cascades downstream of adaptors/kinases, such as nuclear factor B (NF-B) essential modulator (NEMO), inhibitor of IB kinases (IKKs), Akt kinase, and MAPKs, including p38, JNK, and ERK (1,2). The downstream consequence of RANK-RANKL signaling is the activation of osteoclastogenic transcription factors, including NF-B, activator protein 1 (AP-1), cAMP response element-binding protein (CREB), and nuclear factor of activated T cell c1 (NFATc1), which induce osteoclastogenic genes, such as Trap, Dcstamp, Atp6v0d2, and Oscar; these genes code for the proteins tartrate-resistant acid phosphatase (TRAP), dendritic cell-specific transmembrane protein (DC-STAMP), V-ATPase subunit d2 (Atp6v0d2), and OC-associated receptor (OSCAR), respectively (1)(2)(3).
TRAF6 is a TRAF adaptor molecule recruited to the RANK cytoplasmic tail (RANK cyto ) signaling complexes and is critical for activating downstream targets. Similar to RANKL-and RANK-deficient mice, TRAF6-deficient mice exhibit severe osteopetrosis caused by a defect in OC activity (8). TRAF6 contains an N-terminal RING domain and zinc finger motifs, a central coiled-coil (CC) domain, and a conserved C-terminal TRAF domain. The RING domain has E3 ubiquitin (Ub) ligase activity in conjunction with the E2 Ub-conjugating enzyme Ubc13/Uev1A, which promotes Lys 63 -linked autoubiquitination (9). Lys 63 -linked TRAF6 autoubiquitination induced by RANKL stimulation leads to the recruitment and activation of downstream adaptors/kinases, including TGF-␤-activated kinase 1 (TAK1), TAK1-binding proteins, NEMO, IKKs, and MAPKs (10,11). Upon RANKL stimulation, the C-terminal TRAF domain of TRAF6 directly interacts with three TRAF6 binding sites (T6BS-I, -II, and -III), which are defined by a PXEXX[Ar/Ac] consensus sequence, where Ar is an aromatic residue, and Ac is an acidic residue in RANK cyto (2). Triplemutant analyses have revealed a crucial role of the T6BSs in RANK-mediated osteoclastogenesis (12,13). T6BS-I has the highest binding affinity for TRAF6, and T6BS-II and -III exhibit 10-fold lower binding affinity (12,14). Despite different binding affinities for TRAF6, all T6BSs in RANK cyto have the potential to induce osteoclastogenesis via signaling (13). However, osteoclastogenesis is not completely disrupted in TRAF6deficient mice (8,15), suggesting that RANK cyto has additional motifs that are functionally independent of T6BS activity and that recruit specific adaptor/kinase molecules to promote osteoclastogenesis.
Studies of deletion mutations of RANK cyto have revealed that RANK cyto harbors a unique domain named the highly conserved domain in RANK (HCR, amino acids 487-546) that functions as a crucial regulatory motif in RANK-mediated osteoclastogenesis (16,17). Upon RANKL stimulation, HCR recruits additional adaptor/kinase molecules, such as Vav3, Grb-2-associated binder 2 (Gab2), and phospholipase C ␥2 (PLC␥2), to induce the late phase of RANK-mediated signaling (16,18,19). Fine deletion analyses of HCR further demonstrated that the IVVY motif (amino acids 535-538) in HCR plays a pivotal role in late phase signaling of RANK-mediated osteoclastogenesis by recruiting Vav3 to the RANK signaling complex (16,18,20,21). Vav3 is a Rho-guanine nucleotide exchange factor belonging to the Vav family, which shares a common domain structure comprising a calponin homology domain, an acidic domain (AD), a Dbl homology (DH) domain, a pleckstrin homology (PH) domain, a cysteine-rich (C1) domain, a Src homology 2 domain (SH2), and a Src homology 3 (SH3) domain (22,23). Interestingly, a direct link between TRAF6 and HCR-recruited molecules, such as Vav3, Gab2, and PLC␥2, in the RANK signaling complex has been proposed (18,19). However, how T6BSs and HCR function in concert to regulate cooperative signaling networks between TRAF6 and HCR-recruited Vav3, Gab2, and PLC␥2 or unidentified additional adaptors/kinases during RANK-mediated osteoclastogenesis remains to be elucidated at the molecular level.
In this study, we attempted to identify novel TRAF6-interacting proteins using a proteomics approach to investigate the regulation of RANK-mediated osteoclastogenesis by TRAF6. We determined that TRAF6 interacts directly with Vav3 in the RANK signaling complex, thereby enhancing osteoclastogenesis. The DH domain of Vav3 is required for the interaction with the CC domain of TRAF6. TRAF6 recruits Vav3 and vice versa by interacting directly with the RANK signaling complex independent of its binding motifs in RANK cyto . Finally, we demonstrated that the TRAF6-Vav3 interaction in the RANK signaling complex enhances OC differentiation and function. Thus, we have discovered a new role of Vav3 as a novel TRAF6interacting partner in RANK-mediated osteoclastogenesis.

Results
TRAF6 Interacts with Vav3-To identify novel protein partners that interact with TRAF6, lysates of FLAG-TRAF6-expressing 293T cells were immunoprecipitated with FLAG beads, and the TRAF6-bound protein band at ϳ90 kDa was identified by nanoscale liquid chromatography coupled with tandem mass spectrometry ( Table 1). The interaction of TRAF6 and Vav3 was confirmed by a GST bead pulldown (PD) assay and anti-Myc immunoprecipitation (IP) assay (Fig. 1A). To further confirm the TRAF6-Vav3 interaction at the endogenous protein level, we conducted an endogenous TRAF6 IP assay with an anti-TRAF6 antibody in bone marrow-derived macrophages (BMMs) and observed the interaction between endogenous TRAF6 and Vav3 (Fig. 1B). Consistent with these results, TRAF6 co-localized with Vav3 in subcellular localization analysis (Fig. 1C). Next, we examined the effect of the TRAF6-Vav3 interaction on the RANK signaling complex. The GST-PD assay with GST-RANK cyto revealed that Vav3 recruitment to GST-RANK cyto was increased by TRAF6 expression in a dose-dependent manner (Fig. 1D). Because RANK can recruit TRAFs 1, 2, 3, 5, and 6 (2), we also examined whether Vav3 interacts with other TRAF family members. We detected interaction bands for TRAFs 1, 3, and 6 but not TRAFs 2 or 5 in an anti-FLAG IP assay (Fig. 1E).
The DH Domain of Vav3 Physically Interacts with the CC Domain of TRAF6 -To identify the specific domain of Vav3 that interacts with TRAF6, we constructed expression plasmids for Vav3 deletion mutants lacking the N-and C-terminal regions ( Fig. 2A). The GST-TRAF6 PD assay of the C-terminal deletion mutants of Vav3 revealed that TRAF6 bound to the Vav3  and Vav3  mutants but not to Vav3  , indicating that the region between residues 160 and 847 in Vav3 is involved in the TRAF6 interaction (Fig. 2B). Among the N-terminal deletion mutants of Vav3, TRAF6 bound to the Vav3 [160 -847] and Vav3 [196 -847] mutants but not to Vav3  or Vav3  , indicating that the region between residues 1 and 400 in Vav3 is required for the TRAF6 interaction (Fig. 2C). Collectively, these results indicate that the region between residues 160 and 400, including the AD and DH domains of Vav3, may be involved in the TRAF6 interaction. To further determine the precise domain of Vav3 that interacts with TRAF6, we constructed internal deletion mutants of Vav3 ( Fig. 2A) (Fig. 2D). To further confirm the specific domain of Vav3 interacting with TRAF6 in BMMs, we conducted an endogenous TRAF6 IP assay with an anti-TRAF6 antibody in BMMs, and we observed the specific interaction between the DH domain of Vav3 and endogenous TRAF6 (Fig. 2E). As summarized in Fig. 2A, these results indicate that the DH domain of Vav3 containing residues 196 -370 is required for the TRAF6 interaction.
To identify the specific domain in TRAF6 that interacts with Vav3, we generated a series of TRAF6 deletion mutants lacking the N-terminal, C-terminal, or internal regions (Fig. 3A). In an anti-Myc IP assay, Vav3 interacted only with TRAF6  , TRAF6 [109 -530] , and TRAF6  , indicating that the region between residues 301 and 355 containing the CC domain of TRAF6 is required for the Vav3 interaction (Fig. 3B). Moreover, in the internal deletion mutant assay, we did not observe any protein-protein interaction of the TRAF6 ⌬[301-355] mutant, which harbors an internal deletion of the CC domain of TRAF6 (Fig. 3C). To further confirm the specific domain of TRAF6 interacting with Vav3 in BMMs, we conducted an endogenous Vav3 IP assay with an anti-Vav3 antibody in BMMs, and we observed the specific interaction between the CC domain of TRAF6 and endogenous Vav3 (Fig. 3D). As summarized in Fig.  3A, these results indicate that residues 301-355 of TRAF6 containing the CC domain are required for the Vav3 interaction.
The CC domain of TRAF6 is essential for autoubiquitination and NF-B activation (24,25). Thus, we next examined whether the TRAF6-Vav3 interaction is involved in a Ub-dependent pathway. The binding assay results for Lys-deficient TRAF6 (TRAF6⌬K) demonstrated that the TRAF6-Vav3 interaction is not dependent on TRAF6 ubiquitination (Fig. 3E). Moreover, the E3 Ub ligase activity of TRAF6 was not affected by Vav3 expression (Fig. 3F). Thus, we conclude that the TRAF6-Vav3 interaction is not dependent on the TRAF6 ubiquitination pathway.
Because earlier studies suggested that Vav3 is recruited to RANK cyto (18,20), we next attempted to identify the specific domain of Vav3 that is recruited to RANK cyto . However, no specific domain of Vav3 interacted with the RANK cyto in the GST-RANK cyto PD assay, indicating that the entire protein or a broad range of Vav3 domains is required for the interaction with RANK cyto (data not shown). TRAF6 interacts directly with consensus T6BSs (PXEXX[Ar/Ac]) in RANK cyto through its C-terminal TRAF domain (2). We identified one similar sequence with a consensus T6BS at residues 348 -353 in the DH domain of Vav3. To assess the involvement of this putative T6BS sequence in the TRAF6-Vav3 interaction, we constructed a Vav3 E350A point mutant (mutation of the Glu residue at position 350 in Vav3 to Ala). The Vav3-TRAF6 interaction was not affected by the Vav3 E350A point mutation in the GST-PD assay (Fig. 3G). Moreover, the TRAF6 [356 -530] mutant harboring the C-terminal TRAF domain that binds to T6BSs in RANK cyto did not interact with Vav3 (Fig. 3B). Taken together, these results indicate that the DH domain of Vav3 physically interacts with the CC domain of TRAF6 and that the interaction domain between TRAF6 and Vav3 does not overlap with the RANK cytorecruited domains of the proteins.
Vav3 Enhances Osteoclastogenesis through the Activation of TRAF6 Signaling-Because TRAF6 functions as a crucial adaptor molecule in the RANK signaling complex that activates downstream targets of NF-B and MAPKs (2), we determined whether the TRAF6-Vav3 interaction enhances the RANK-mediated signaling pathway. In a reporter assay, we observed that the transcription activities of TRAF6-induced NF-B and AP-1 were significantly enhanced by Vav3 expression in a dose-dependent manner (Fig. 4, A and B). Consistent with these results, upon RANKL stimulation, the phosphorylation of NF-B p65 and MAPKs (p38, ERK, and JNK) was induced by Vav3 expression (Fig. 4C).
The activation of NF-B and MAPKs is a key signaling pathway in osteoclastogenesis (2,3). Thus, we next examined whether Vav3 expression is involved in osteoclastogenesis using a retroviral gene transfer system. Vav3-transduced  5D). Next, we investigated whether Vav3 knockdown (Vav3 KD ) is involved in osteoclastogenesis. As a preliminary experiment, BMMs transduced by retroviral supernatant harboring Vav3 shRNA expression cassettes were selected with puromycin and further tested for the expression of Vav3. Vav3 expression was reduced significantly by the combination of shRNAs 1 and 2 (sh1ϩ2) compared with shRNA 1 (sh 1) or shRNA 2 (sh 2) alone ( Fig. 6A). To examine the effects of Vav3 KD , puromycin-resistant BMMs transduced by sh1ϩ2 were differentiated into OCs with M-CSF and RANKL for 4 days. TRAP activity and the number of TRAP ϩ MNCs in Vav3 KD OCs were reduced significantly (0.36 Ϯ 0.08 and 42.7 Ϯ 11.6, respectively) compared with the mock control (0.73 Ϯ 0.14 and 86.3 Ϯ 15.5, respectively) (Fig. 6B). Similar results were obtained in a bone resorption assay, with a 3.1-fold decrease in the resorption area after Vav3 KD (13.2 Ϯ 1.8) compared with the mock control (41.3 Ϯ 8.8) (Fig. 6C). Consistent with these results, Vav3 KD reduced the expression or activation of the osteoclastogenic transcription factors (Fig. 6D) and markers (Fig. 6E). Collectively, these data indicate that Vav3 can induce the signaling pathway downstream of TRAF6, thereby enhancing osteoclastogenesis.
Physical Interaction of TRAF6 and Vav3 Forms a Receptor Complex with RANK cyto Lacking Either T6BSs or the IVVY Motif-RANK has distinct cytoplasmic T6BSs, and an IVVY motif that can recruit TRAF6 and Vav3, respectively, to form the RANK signaling complex during osteoclastogenesis (12,14,18,20). Thus, to confirm that TRAF6-Vav3 interaction can lead to RANK signaling complex formation independent of the binding motifs in RANK cyto , we constructed GST-RANK cyto mutants lacking either T6BSs or the IVVY motif (Fig. 7A). In a GST-RANK cyto -PD assay, TRAF6 was clearly recruited to the GST-RANK cyto -AAA mutant lacking T6BSs by Vav3, although a large fraction of the TRAF6 was not bound to this mutant (Fig.  7B, fourth lane). Similarly, a small fraction of Vav3 was recruited to the GST-RANK cyto -LAAF mutant lacking the IVVY motif by TRAF6 (Fig. 7B, sixth lane). However, neither TRAF6 nor Vav3 binding was detected for GST-RANK cyto -AAA/LAAF, which lacks both T6BSs and the IVVY motif ( To further confirm the formation of the receptor complex by the TRAF6-Vav3 interaction, we generated FLAG-tagged fulllength RANK (FLAG-RANK) mutants that lacked either T6BSs or the IVVY motif (Fig. 8A). Consistent with the data shown in These results indicate that TRAF6 is recruited to RANK cyto mutants that lack their own T6BSs through the Vav3 interaction and that Vav3 is also recruited to a RANK cyto mutant lacking the IVVY motif through the TRAF6 interaction. Taken together, we conclude that TRAF6 recruits Vav3 to the RANK signaling complex and vice versa via a direct proteinprotein interaction that is independent of the binding motifs in RANK cyto .  SEPTEMBER 23, 2016 • VOLUME 291 • NUMBER 39

JOURNAL OF BIOLOGICAL CHEMISTRY 20647
The TRAF6-Vav3 Interaction Enhances Osteoclastogenesis-To address the role of the TRAF6-Vav3 direct interaction in osteoclastogenesis, we determined whether the TRAF6-Vav3 interaction in the RANK signaling complex induces OC differentiation and function. First, we generated retroviral hCD40/mRK chimeric receptor mutants that lacked either T6BSs or the IVVY motif to examine the effect of the TRAF6-Vav3 interaction under conditions of binding motif deficiency (Fig. 9A). BMMs transduced with a retroviral hCD40/mRK mutant and TRAF6 were differentiated into OCs by treatment with an anti-human CD40 monoclonal antibody to exclude the possibility that endogenous RANKL signaling affects osteoclastogenesis. TRAF6-expressing hCD40/mRK-WT BMMs treated with anti-human CD40 stimulation for 3 days exhibited increased (1.8 Ϯ 0.08 and 209.3 Ϯ 23.5, respectively) osteoclastogenesis compared with the mock control (1.5 Ϯ 0.05 and 114.3 Ϯ 14.6, respectively), as indicated by both TRAP activity and TRAP staining (Fig. 9B, left panel). Similar to previous reports that both T6BSs and the IVVY motif are crucial for osteoclastogenesis (12,13,16,17), OC formation was not detected in hCD40/ mRK mutants that lacked either T6BSs (hCD40/mRK-AAA) or the IVVY motif (hCD40/mRK-LAAF) after 3-day culture (Fig. 9B, left panel). However, when the OC culture was
To further confirm the effect of the TRAF6-Vav3 interaction on osteoclastogenesis, we conducted an OC differentiation  Puromycin-selected Vav3-expressing BMMs were cultured for 2 h under serum-free conditions. The BMMs were stimulated with RANKL (100 ng/ml) for the indicated times and subsequently analyzed via IB with antibodies that recognized phosphorylated and total NF-B p65 (Ser 536 ), p38 (Thr 180 /Tyr 182 ), ERK (Thr 202 /Thr 204 ), and JNK (Thr 183 /Tyr 185 ). The cell lysates were normalized for total protein content. The level of Vav3 expression was detected via anti-FLAG IB. The relative level of phosphorylated forms was calculated after normalization to total protein input (right). *, p Ͻ 0.05. All experiments were performed at least three times with similar results.

TRAF6-Vav3 Interaction Enhances Osteoclastogenesis
assay with BMMs transduced with retroviral hCD40/mRK mutants and Vav3. Consistent with the data obtained after RANKL stimulation (shown in Fig. 5A (Fig. 10A,  left panel). In a 3-day culture, OC formation was also not detected in hCD40/mRK mutants that lacked either T6BSs (hCD40/mRK-AAA) or the IVVY motif (hCD40/mRK-LAAF) (Fig. 10A, left panel). However, when the OC culture was extended to day 6, OC formation by either the hCD40/mRK-AAA or hCD40/mRK-LAAF mutant was enhanced (0.6 Ϯ 0.2  , original magnification, ϫ50). The summarized data from the resorption pit assays are shown in the bottom panel. C, the effect of Vav3 expression on the activation of osteoclastogenic factors. OCs were differentiated from BMMs with RANKL (100 ng/ml) and M-CSF (50 ng/ml) for the indicated times and subsequently analyzed via immunoblotting (IB) with antibodies that recognized phosphorylated or total NFATc1 (7A6), c-Fos (D-1), and CREB (Ser 133 ). The cell lysates were normalized for total protein content. The level of Vav3 expression was detected via anti-FLAG IB. The relative protein level was calculated after normalization to ␤-actin protein input (right). The relative level of phosphorylated CREB was calculated after normalization to total CREB protein input (right). ␤-Actin was used as the loading control. D, the effect of Vav3 on the expression of osteoclastogenic markers. Total RNA was prepared from OCs and subjected to real time PCR analysis. The data were normalized to ␤-actin. All experiments were performed at least three times with similar results. SEPTEMBER Fig. 10B, right panel). However, OC differentiation and bone resorption were not detected in the hCD40/mRK-AAA/LAAF mutant that lacked both T6BSs and the IVVY motif (Fig. 10, A and B, right panels). Finally, we analyzed whether the TRAF6-Vav3 interaction enhances RANK signaling. Upon RANKL stimulation, phosphorylation of NF-B p65, p38, ERK, and JNK was significantly reduced after expression of the Vav3 ⌬[196 -370] mutant compared with Vav3 (Fig. 10C). Taken together, these data indicate that the direct TRAF6-Vav3 interaction in the RANK signaling complex enhances OC differentiation and bone-resorbing function.

Discussion
The crucial role of T6BSs in RANK-mediated osteoclastogenesis through the recruitment of TRAF6 is well documented (2,12,13). An additional HCR motif in RANK cyto was also recently demonstrated to play a pivotal role in RANK-mediated osteoclastogenesis (16,17). However, the functional cross-talk between T6BSs and HCR through the recruitment of specific adaptors in the RANK signaling complex remains to be elucidated. Thus, in this study, we screened Vav3 as a novel TRAF6 binding partner and identified the functional importance of the TRAF6-Vav3 interaction in the RANK signaling complex. We demonstrated that the interaction between the CC domain in TRAF6 and the DH domain in Vav3 synergistically activates RANK-mediated NF-B and MAPK signaling, as well as NFATc1 induction during osteoclastogenesis. The CC domain of TRAF6 is important for TRAF6 autoubiquitination and downstream NF-B activation via binding to Ubc13/Uev1A or NEMO (24,25). Moreover, Lys 63 -linked TRAF6 autoubiquitination induced by RANKL stimulation is associated with the activation of downstream adaptors/kinases, including TAK1, TAK1-binding protein 1/2, NEMO, IKKs, and MAPKs, in RANK-mediated signaling (10,11). Thus, it was important to examine whether the TRAF6-Vav3 interaction is involved in the TRAF6 ubiquitination pathway. However, the TRAF6-Vav3 interaction was not dependent on TRAF6 ubiquitination (Fig.  3E), and the E3 Ub ligase activity of TRAF6 was not affected by Vav3 expression (Fig. 3F). Vav3-deficient BMMs have been consistently reported to behave normally in terms of both RANKLstimulated NF-B activation and M-CSF-stimulated ERK activation, but mice lacking Vav3 display an osteopetrotic phenotype caused by defective OC function (26). Thus, it is possible that Vav3 per se is not directly involved in RANKmediated signaling but may be required for the regulation process in RANK-mediated osteoclastogenesis. Interestingly, it has been proposed that Vav3 is involved in both OC fusion and actin ring formation via recruitment to the HCR motif during osteoclastogenesis (20). Hence, considering our results and those of previous studies, it is reasonable to presume that the direct TRAF6-Vav3 interaction strengthens the TRAF6 signaling complex via cross-talk between T6BSs and the HCR motif in RANK cyto , thereby further enhancing TRAF6-induced downstream NF-B and MAPK signaling in osteoclastogenesis.
Previous studies of RANK cyto deletion mutants demonstrated that the IVVY motif in the HCR plays a crucial role in RANK-mediated osteoclastogenesis by inducing NFATc1 (16,17,21). Blocking peptides that harbor the IVVY motif attenuate RANK-mediated osteoclastogenesis, possibly by inhibiting the recruitment of Vav3 to the RANK signaling complex (18,20). Moreover, Vav3 can bind to the IVVY motif in vitro (18,20). Thus, Vav3 is a strong candidate for an adaptor molecule that binds directly to the IVVY motif. However, the IVVY motif per se is not directly involved in NF-B and MAPK activation (16,17,20), and treatment with the IVVY peptide alone is not sufficient for inhibiting OC differentiation (20). Moreover, a requirement for an additional adaptor molecule in the Vav3-IVVY motif interaction has been proposed by Kim et al. (20) because Vav3 does not directly interact with RANK cyto in a yeast two-hybrid system. Thus, the direct binding of Vav3 to the IVVY motif remains unclear. Further studies are required to elucidate the Vav3-IVVY motif interaction.
Interestingly, Vav3 and Gab2 share an IVVY binding motif to form the RANK signaling complex, indicating that Gab2 may be involved in the binding of Vav3 to the IVVY motif, although Gab2 binding requires additional binding residues in the HCR of the RANK cyto (16,18). Moreover, Gab2 can interact with TRAF6 and recruits PLC␥2 to the RANK signaling complex (16,27). RANK-mediated NF-B, MAPK, and NFATc1 activation are reduced in Gab2-deficient OCs (19). Mice lacking Gab2 also exhibit an osteopetrotic phenotype, with decreased bone resorption caused by partially impaired osteoclastogenesis (19). Thus, Gab2 functions as a crucial molecular scaffold in the RANK signaling complex (19,27). However, whether Gab2 associates directly with RANK cyto and how Gab2 function is linked to TRAF6 or Vav3 in the RANK signaling complex remain unclear. Choi et al. (28) recently identified early estrogen-induced gene 1 (EEIG1) as an another HCR-binding adaptor molecule. EEIG1 also shares an IVVY binding motif to form the RANK signaling complex with Gab2 and PLC␥2, although Vav3 recruitment to this signaling complex has not been defined. Interestingly, EEIG1 is involved in RANK-mediated PLC␥2 activation and NFATc1 induction but not NF-B or MAPK activation (28). Thus, the activation of RANK signaling by EEIG1 can be distinguished from TRAF6, Gab2, or Vav3 signaling (19,28). These previous studies and our current observations emphasize the need for further investigations of  , original magnification, ϫ50). The summarized data from the resorption pit assays are shown in the bottom panel. *, p Ͻ 0.05. D, the effect of Vav3 KD expression on the activation of osteoclastogenic factors. OCs were differentiated from Vav3 KD -BMMs with RANKL (100 ng/ml) and M-CSF (50 ng/ml) for the indicated times and subsequently analyzed via IB with antibodies that recognized phosphorylated or total NFATc1 (7A6), c-Fos (D-1), and CREB (Ser 133 ). The cell lysates were normalized for total protein content. The level of Vav3 expression was detected via anti-Vav3 IB. The relative protein level was calculated after normalization to ␤-actin protein input (right panel). The relative level of phosphorylated CREB was calculated after normalization to total CREB protein input (right panel). ␤-Actin was used as the loading control. E, the effect of Vav3 KD on the expression of osteoclastogenic markers. Total RNA was prepared from Vav3 KD -OCs and subjected to real time PCR analysis. The data were normalized to ␤-actin. All experiments were performed at least three times with similar results. SEPTEMBER 23, 2016 • VOLUME 291 • NUMBER 39 the functional relevance of the adaptor molecules TRAF6, Gab2, PLC␥2, EEIG1, and Vav3 for the formation of the RANK signaling complex via cross-talk between T6BSs and the HCR.
In conclusion, in this study, we demonstrated that T6BSs cooperate with the IVVY motif to enhance RANK-mediated osteoclastogenesis via the TRAF6-Vav3 interaction in the RANK signaling complex. We demonstrated that the CC domain of TRAF6 interacts directly with the DH domain of Vav3 to form the RANK signaling complex independent of the TRAF6 ubiquitination pathway. Via cross-talk between T6BSs and the IVVY motif in RANK cyto , the TRAF6-Vav3 interaction enhances downstream TRAF6 signaling and NFATc1 induction by further strengthening the TRAF6 signaling complex, thereby inducing osteoclastogenesis.

TRAF6-Vav3 Interaction Enhances Osteoclastogenesis
RANK cyto ) expression plasmid was described previously (20). To generate T6BS-deficient mutants, all consensus Glu residues in the T6BSs were mutated to Ala by PCR amplification as previously described (39), and the mutated fragments were subcloned into pFLAG-CMV1 (FLAG-RANK-AAA) or pEBG (GST-RANK cyto -AAA). The retroviral expression plasmids for the chimeric receptor of human CD40 and murine RANK cyto (hCD40/mRK) and its T6BS-deficient mutant (hCD40/mRK-AAA) were described previously (12). To generate the IVVY motif mutants, the LAAF mutation was introduced into the IVVY motif in the RANK cyto by PCR amplification, and the fragments of the LAAF mutant were subcloned into pFLAG- Puromycin-selected BMMs expressing Vav3 or DH domain-deficient Vav3 were cultured for 2 h under serum-free conditions. The BMMs were stimulated with RANKL (100 ng/ml) for the indicated times and subsequently analyzed via immunoblotting (IB) with antibodies that recognized phosphorylated and total NF-B p65 (Ser 536 ), p38 (Thr 180 /Tyr 182 ), ERK (Thr 202 /Thr 204 ), and JNK (Thr 183 /Tyr 185 ). The cell lysates were normalized for total protein content. The level of Vav3 expression was detected via anti-FLAG (M2) IB. The relative level of phosphorylated forms was calculated after normalization to total protein input (right). ␤-Actin was used as a loading control. The relative level of phosphorylated forms was calculated after normalization to total protein input (right). *, p Ͻ 0.05. All experiments were performed at least three times with similar results. SEPTEMBER 23, 2016 • VOLUME 291 • NUMBER 39 JOURNAL OF BIOLOGICAL CHEMISTRY 20657 CMV1 (FLAG-RANK-LAAF), pEBG (GST-RANK cyto -LAAF), or pMXs-puro (hCD40/mRK-LAAF). The expression plasmids carrying a double mutation of both T6BS and the IVVY motif were constructed by PCR amplification and insertion in pFLAG-CMV1 (FLAG-RANK-AAA/LAAF), pEBG (GST-RANK cyto -AAA/LAAF), and pMXs-puro (hCD40/mRK-AAA/ LAAF), respectively. The plasmids p(B) 3 -NF-B-Luc (NF-B-Luc), pAP-1-luciferase (AP-1-Luc), and pcDNA3.0HisLacZ (Invitrogen) were described previously (38,39). The expression plasmid of HA-tagged ubiquitin (HA-Ub) was described previously (38,39). For the Vav3 knockdown (Vav3 KD ) plasmid DNA, a shRNA targeting the murine Vav3 gene was designed as previously described (39)  Osteoclastogenesis and OC Analysis-Murine BMMs were prepared as previously described (37,41). In brief, bone marrow cells isolated from the femurs of 8-week-old C57BL/6 male mice were differentiated into BMMs by culturing them in ␣-minimum essential medium containing 10% FBS and M-CSF (150 ng/ml) for 2 days. To infect the BMMs with a retrovirus carrying a foreign gene (Vav3 or TRAF6) and puromycin-resistant expression cassettes, Vav3-or TRAF6-retroviral expression vectors were transfected into the retroviral packaging cell line PlatE using TurboFect transfection reagent (Fermentas). The BMMs were infected by retroviral supernatants harboring Vav3 or TRAF6 expression and puromycin-resistant cassettes with Polybrene (8 g/ml) for 6 h, and the infected BMMs were then selected with puromycin (2 g/ml) and M-CSF (150 ng/ml) for 2 days. Then BMMs were analyzed via flow cytometry with a FITC-conjugated anti-CD11b antibody (eBioscience, San Diego, CA). The puromycin-resistant BMMs (CD11b ϩ cells with a purity of Ͼ99.0%; 1.5 ϫ 10 4 cells/well) in 96-well plates were differentiated into OCs in the presence of M-CSF (50 ng/ml) and RANKL (100 ng/ml) or the anti-human CD40 monoclonal antibody G28 -5 (100 ng/ml) for 3-6 days. To analyze OCs, the cells were fixed in 10% formalin and an ethanol/ acetone (50:50 v/v) mixture. To measure TRAP activity, the cells were incubated with p-nitrophenyl phosphate substrate (Sigma-Aldrich) solution for 30 min, and the solution was then mixed with 1 N NaOH. The TRAP solution activity of the collected supernatant was measured at 405 nm using a microplate reader (Bio-Rad). For TRAP staining, the TRAP solution supernatant was discarded from the cells, and the cells were incubated in TRAP staining solution containing naphthol AS phosphate (Sigma-Aldrich) and fast red violet (Sigma-Aldrich) for 20 -30 min at room temperature. Finally, the morphology of the TRAP-stained OCs was photographed under a microscope. TRAP ϩ MNCs with more than three nuclei were considered multinucleated OCs. Analysis of the RANKL-induced signaling pathway was performed as previously described (37,41). For the bone resorbing activity of OCs, retroviral-transduced BMMs (1.5 ϫ 10 4 cells/well) were cultured with M-CSF (50 ng/ml) and RANKL (100 ng/ml) or the anti-human CD40 monoclonal antibody G28 -5 (100 ng/ml) on dentin slices for 3-6 days, as previously described (37,41). The cells were removed with 10% sodium hypochlorite solution to permit observation of the resorption pit. The resorption pit was stained with 1% toluidine blue and photographed under a microscope, and the resorbing activity was evaluated by measuring the area of the resorption pit. To analyze osteoclastogenic gene expression, real time PCR was performed as previously described (37,41). The primers used for real time PCR amplification were as follows: Protein-Protein Interaction and IP Assay-For mass spectrometry analysis, 293T cells (1 ϫ 10 6 cells/ml) in 100-mm culture plates were transfected with FLAG-TRAF6 (20 g) using TurboFect transfection reagent (Fermentas, Hanover, MD). At 24 h post-transfection, the transfected cells were lysed in lysis buffer (250 mM Tris-HCl, pH 7.5, 1 mM EDTA, 150 mM NaCl, 0.5% Triton X-100, and 5% glycerol). After centrifugation, the total protein concentrations of the supernatants were determined using BCA protein assay kits (Pierce). An equal amount of each supernatant (normalized for total protein content) was precipitated with FLAG beads (50% slurry) for 16 h at 4°C. The beads were washed three times with lysis buffer and then suspended in loading buffer for 10% SDS-PAGE. The gel was stained with Coomassie Blue, and the selected bands were analyzed by nanoscale liquid chromatography coupled with tandem mass spectrometry by the proteomics core facility at the University of Pennsylvania School of Medicine. Protein-protein interaction and IP assays were performed as described previously (39). In brief, 293T cells (3.5 ϫ 10 5 cells/ml) in 6-well plates were co-transfected with epitope-tagged TRAF6 (1.0 g) and Vav3 (1.0 g) using TurboFect (Fermentas). At 24 h posttransfection, the transfected cells were lysed in lysis buffer, and the lysate (normalized for total protein content) was precipitated with GST beads, Myc beads, or FLAG beads (50% slurry) for 16 h at 4°C. The beads were washed three times with lysis buffer. The samples were separated by 10% SDS-PAGE, transferred to PVDF membranes, and immunoblotted with antibodies. For the endogenous protein-protein interaction assay, endogenous TRAF6 in murine BMM lysates (normalized for total protein content) was immunoprecipitated with an anti-TRAF6 antibody and protein G beads (50% slurry) overnight at 4°C, followed by immunoblotting analysis with antibodies.
Subcellular Localization-293T cells (1.0 ϫ 10 4 cells/ml) were plated on sterile glass coverslips in 6-well plates. After 12 h of culture, the cells were co-transfected with RFP-TRAF6 (1.5 g) and Vav3-Myc (1.5 g). At 24 h post-transfection, the cells were stained with an anti-Myc primary antibody and a FITCconjugated anti-rabbit IgG secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA). Finally, the fluorescent signals were photographed under an Axiovert 135M laser scanning confocal microscope (Carl Zeiss).
Ubiquitination Assay-The TRAF6 ubiquitination assay was performed as previously described (39). In brief, the expression plasmids for FLAG-TRAF6 (0.5 g), HA-Ub (0.5 g), and Vav3-Myc (0.5 g) were transfected into 293T cells seeded (3.5 ϫ 10 5 cells/ml) in 6-well plates. At 24 h post-transfection, the cells were lysed. The cell lysates (normalized for total protein content) were subsequently subjected to IP and immunoblotting analysis with antibodies.
Statistical Analysis-All experiments were performed at least three times. The data represent the means Ϯ standard deviations (n ϭ 3 or n ϭ 4/group). Student's t test was used to determine the significance of the differences between the experimental samples.