Ubiquitin E3 Ligase Itch Negatively Regulates Osteoclast Formation by Promoting Deubiquitination of Tumor Necrosis Factor (TNF) Receptor-associated Factor 6*

Background: TRAF6 activity is crucial for osteoclastogenesis, which is decreased by deubiquitination. The role of Itch in this process is studied. Results: Itch−/− osteoclast precursors formed more osteoclasts, had prolonged RANKL-induced NF-κB activation, and had delayed TRAF6 deubiquitination. Itch bound to the deubiquitinating enzyme cylindromatosis. Conclusion: Itch inhibits osteoclastogenesis by binding to cylindromatosis and prompting TRAF6 deubiquitination. Significance: Itch is a new inhibitor of osteoclastogenesis. Itch is a ubiquitin E3 ligase that regulates protein stability. Itch−/− mice develop an autoimmune disease phenotype characterized by itchy skin and multiorgan inflammation. The role of Itch in the regulation of osteoclast function has not been examined. We report that Itch−/− bone marrow and spleen cells formed more osteoclasts than cells from WT littermates in response to receptor activator of NF-κB ligand (RANKL) and was associated with increased expression of the osteoclastogenic transcription factors c-fos and Nfatc1. Overexpression of Itch in Itch−/− cells rescued increased osteoclastogenesis. RANKL increased Itch expression, which can be blocked by a NF-κB inhibitor. The murine Itch promoter contains NF-κB binding sites. Overexpression of NF-κB p65 increased Itch expression, and RANKL promoted the binding of p65 onto the NF-κB binding sites in the Itch promoter. Itch−/− osteoclast precursors had prolonged RANKL-induced NF-κB activation and delayed TNF receptor-associated factor 6 (TRAF6) deubiquitination. In WT osteoclast precursors, Itch bound to TRAF6 and the deubiquitinating enzyme cylindromatosis. Adult Itch−/− mice had normal bone volume, but they had significantly increased LPS-induced osteoclastogenesis and bone resorption. Thus, Itch is a new RANKL target gene that is induced during osteoclastogenesis. Itch interacts with the deubiquitinating enzyme and is required for deubiquitination of TRAF6, thus limiting RANKL-induced osteoclast formation.

activation of the target protein via conformational changes depending on the ubiquitin linkage. Generally, the ubiquitin molecules are linked through the lysine residue at position 48 or 63 of ubiquitin (known as Lys 48 and Lys 63 polyubiquitination, respectively) (9,10). The different types of polyubiquitin chains have different effects on the substrate: a protein with Lys 48 polyubiquitination is mainly recognized by the 26 S proteasome and undergoes degradation, whereas a protein with Lys 63 polyubiquitination usually becomes activated to mediate downstream signaling events such as kinase activation, alteration of intracellular locations, and DNA repair (11). Protein ubiquitination is carried out in part by ubiquitin E3 ligases, which play an important role in the pathogenesis of autoimmune diseases and inflammation (1,2).
Biochemical experiments demonstrate that TRAF6 functions as a ubiquitin ligase. After cells are stimulated by cytokines such as RANKL, IL-1, and LPS, TRAF6 promotes Lys 63 polyubiquitination on target proteins in NF-B signaling and on itself (9). This Lys 63 -linked polyubiquitination on TRAF6 can be reversed by deubiquitinating enzymes (DUB) such as cylindromatosis (CYLD) and A20 (12,13). Osteoclast precursors from Cyld Ϫ/Ϫ mice are hyper-responsive to RANKL-induced osteoclasts differentiation. CYLD targets TRAF6 for deubiquitination and inhibits the downstream signaling events as part of its mechanism to regulate osteoclast formation. A20 Ϫ/Ϫ mice develop spontaneous inflammation because A20 is required for the termination of IL-1R/Toll-like receptor 4 signaling in immune cells by promoting TRAF6 deubiquitination (13). These studies highlight the importance of deubiquitination as one of the molecular mechanisms to regulate TRAF6 function, which has not been well studied in osteoclast biology.
In the present study, we examined the effects of Itch deficiency on osteoclastogenesis, NF-B activation, and TRAF6 deubiquitination. We found that Itch Ϫ/Ϫ mice have increased osteoclast numbers and are more responsive to LPS-stimulated osteoclast formation and bone resorption. Osteoclast precursors from Itch Ϫ/Ϫ mice have prolonged RANKL-induced NF-B activation and significantly delayed TRAF6 deubiquitination upon RANKL withdrawal. Itch is a negative regulator of osteoclastogenesis by promoting TRAF6 deubiquitination. Itch-regulated TRAF6 deubiquitination may represent a new mechanism for persistent NF-B activation in immune cells to account for autoimmune phenotypes of Itch Ϫ/Ϫ mice.

EXPERIMENTAL PROCEDURES
Animal Studies-Itch Ϫ/Ϫ mice were generated previously on a C57BL/6J background (4) and were genotyped by PCR analysis. All animal procedures were conducted using procedures approved by the University of Rochester Committee for Animal Resources. For in vivo bone resorption studies, mice (n ϭ 4/group) were given LPS (Sigma) injections (100 g in 25 l of PBS) or 25 l of PBS into the loose subcutaneous tissue overlying the calvaria using a Hamilton syringe (Hamilton Co., Reno, NV) as described previously (14). Injections were performed daily for 3 days. Animals were sacrificed 1 day after the last injection. Calvarial bones were harvested for histology.
Plasmids, Antibodies, and Retrovirus-HA-CYLD and FLAG-Itch expression plasmids were provided by Dr. Shao-Cong Sun (The University of Texas M. D. Anderson Cancer Center) and Dr. Derek Abbott (Case Western Reserve University), respectively. Antibodies to FLAG, HA, and ␤-actin were purchased from Sigma; antibodies to ubiquitin, CYLD, TRAF6, and NF-B p65 were from Santa Cruz Biotechnology Inc.; antibody to Itch was from BD Biosciences; and antibodies to IB␣, phospho-IB␣, JNK, phospho-JNK, ERK, and phospho-ERK were from Cell Signaling Technology. FITC-anti-CD45, PE-Cy7-anti-c-Kit, PE-anti-CD11b, and allophycocyanin-antic-Fms used in fluorescence-activated cell sorting (FACS) were from eBioscience. To generate an Itch retroviral expression vector, the Itch coding region was amplified by PCR from the FLAG-Itch plasmid and cloned into the pMX-GFP retroviral vector at the BamHI and NotI sites to generate the pMX-Itch-GFP expression vector. The pMX-GFP vector was used as a control for infection efficiency. These retroviral vectors were transiently transfected into the Plat-E retroviral packaging cell line, and viral supernatant was collected 48 h later as we described previously (15).
Osteoclastogenesis and Viral Infection-Bone marrow cells or splenocytes were cultured with conditioned medium (1:50 dilution) from an M-CSF-producing cell line (16) for 3 days in ␣-modified essential medium with 10% fetal calf serum (Hyclone Laboratories, Logan, UT) to enrich for OCPs. For osteoclastogenesis, OCPs were cultured with M-CSF conditioned medium and RANKL (10 ng/ml) or LPS (100 ng/ml) for 2-3 or 6 -7 days. For viral infection, OCPs were infected with retroviral supernatants from Itch-, NF-B p65-, or GFP virusinfected Plat-E packaging cells in the presence of M-CSF and Polybrene (8 g/ml) for 2 days. Cells were then cultured with M-CSF and RANKL to form osteoclasts. After multinucleated cells were observed under a microscope, the cells were fixed and stained for TRAP activity. TRAP ϩ cells containing Ն3 nuclei were counted as described (17,18).
Deubiquitination Assay-OCPs were treated with RANKL plus N-ethylmaleimide for 48 h to block deubiquitination. Whole cell lysates were incubated with UbiQapture-Q matrix (Biomol) to pull down all ubiquitinated proteins according to the manufacturer's instructions. Briefly, cell lysates containing 200 g of protein were incubated with washed UbiQapture-Q matrix by gentle agitation at 4°C overnight. After washing three times, captured proteins were eluted with 2ϫ SDS-PAGE loading buffer and analyzed by Western blotting using anti-TRAF6 antibody as described previously (19,20).
Immunoprecipitation and Western Blotting-Proteins from cell lysates were quantitated using a kit from Bio-Rad. Proteins were subjected to immunoprecipitation as described in the technical bulletin from Sigma. Briefly, 300 g of proteins in 1 ml of cell lysis buffer was mixed with 1 g of antibody, incubated for 1 h at 4°C, and then incubated with prewashed EZview Red Protein G Affinity Gel beads from Sigma for 1 h at 4°C. The bound antigens were eluted from the beads by boiling samples for 5 min. Eluted samples were fractionated by SDS-PAGE and transferred to PVDF membranes. For NF-B and ERK signaling analysis, OCPs were starved without serum for 8 h and then treated with PBS or RANKL for various time periods. Cells were harvested and lysed with lysis buffer containing 1 mM Na 3 VO 4 , 1 mM PMSF, and 1 mM NaF. After protein quantification, 50 g of protein was fractionated by SDS-PAGE and transferred to PVDF membranes. Immunoblotting was carried out as described previously (20).
Chromatin Immunoprecipitation (ChIP)-A ChIP assay was performed with the MAGnify Chromatin Immunoprecipitation System (Invitrogen) according to the manufacturer's instructions. OCPs were treated with RANKL or PBS for 48 h. Cells were fixed with 1% formaldehyde for 15 min and sonicated on ice 8 times with a 20-s on and 20-s off cycle at high power using the Bioruptor UCD-200 sonicator. Antibody against p65 (Santa Cruz Biotechnology Inc.) or control rabbit IgG (Invitrogen) were used in the immunoprecipitation step. The amounts of each specific DNA fragment in immunoprecipitates were determined by quantitative PCR. The following primers were used: NF-B binding site 1 (Ϫ3465 to Ϫ3473): forward, 5Ј-GCAGAAATGTCCCAAAGA-3Ј; reverse, 5Ј-TGG-AAAGCCAGCAAAGC-3Ј; NF-B binding site 2 (Ϫ3183 to Ϫ3175): forward, 5Ј-TATGGAGATTATTAGGCTGG TG-3Ј; reverse, 5Ј-GAGCGGGGTCTACAAAGT-3Ј.
NF-B Activation-The DNA binding capacity of NF-B (p65 subunit) was measured in the nuclear extracts of OCPs using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo, Rockford, IL) and TransAMe NF-B kits (Active Motif, Carlsbad, CA) according to the manufacturers' instructions. Briefly, the assay is based on a 96-well plate to which an oligonucleotide containing the NF-B consensus binding site (5Ј-GGGACTTTCC-3Ј) has been immobilized. The activated NF-B contained in nuclear extracts specifically binds to this nucleotide. By using an antibody that is directed against an epitope on p65 that is accessible only when NF-B is bound to its target DNA, the NF-B bound to the oligonucleotide is detected. Addition of a secondary antibody conjugated to horseradish peroxidase provides a sensitive colorimetric readout that is quantified by densitometry.
FACS-Cells were harvested, and red blood cells were lysed. Cells were stained with FITC-anti-CD45, PE-Cy7anti-c-Kit, PE-anti-CD11b, and allophycocyanin-anti-c-Fms for 30 min and subjected to FACS analysis according to the manufacturer's instructions (21). Results were analyzed by FlowJo7 software.
Quantitative Real Time RT-PCR-Total RNA was extracted from cell cultures using TRIzol reagent (Invitrogen). cDNAs were synthesized using the iSCRIPT cDNA Synthesis kit (Bio-Rad). Quantitative real time RT-PCR amplifications were performed in the iCycler (Bio-Rad) real time PCR machine using iQ SYBR Green Supermix (Bio-Rad) according to the manufacturer's instruction. Sequence-specific primers of c-fos, Nfatc1, and ␤-actin used for quantitative real time PCR amplification were described previously (15). The following sequence-specific primers of Itch were used: forward, 5Ј-TCACT-TGGGCATAGGTCTCT-3Ј; reverse, 5Ј-TGTGCCCAGACA-CTGAGTTA-3Ј. The primer sequences for c-fos and Nfatc1 were described previously (15). To determine the number of copies of target DNA in the samples, purified PCR fragments of known concentration were serially diluted and served as external standards for each experiment. Data were normalized to Gapdh levels.
Micro-computed Tomography, Histology, and Histomorphometry of Bone Sections-Femurs obtained from 3-monthold mice were dissected free of soft tissue, fixed overnight in 70% ethanol, and scanned at high resolution (10.5 m) on a VivaCT40 micro-computed tomography scanner (Scanco Medical, Bassersdorf, Switzerland) using an integration time of 300 ms, energy of 55 peak kilovoltage, and intensity of 145 A. The three-dimensional images were generated using a constant threshold of 275 for all samples.
The calvariae and limbs were removed from mice after sacrifice, fixed in 10% buffered formalin, decalcified in 10% EDTA, and embedded in paraffin. Sections (4 m thick) were then stained with H&E and for TRAP activity. Histomorphometric analysis of osteoclast numbers (expressed as the number of osteoclasts per millimeter of bone surface) and of eroded surfaces (expressed as a percentage of the eroded surface of total bone surface) was performed in calvarial sections using an Osteometrics image analysis software system (Osteometrics, Atlanta, GA) ( 22).
Statistics Analysis-Data are presented as mean Ϯ S.D., and all experiments were repeated at least twice with similar results. Statistical analyses were performed with Stat View statistical software (SAS, Cary, NC). Differences between the two groups were compared using unpaired Student's t test, and more than two groups were compared using one-way analysis of variance followed by a Bonferroni/Dunnett test. p values less than 0.05 were considered to be statistically significant.

Itch Ϫ/Ϫ Mice Have Increased Osteoclast Formation in Vitro-
To examine whether Itch regulates osteoclastogenesis, we cultured bone marrow cells or spleen cells from Itch Ϫ/Ϫ mice and their WT littermates with M-CSF and RANKL in osteoclastogenic assays. Cells from bone marrow of Itch Ϫ/Ϫ mice formed significantly more osteoclasts than WT cells at all cell densities tested (Fig. 1A). Increased osteoclast formation was also observed when Itch Ϫ/Ϫ splenocytes were used (number of osteoclasts/well, 360 Ϯ 24 versus 20 Ϯ 12 in WT cells). The expression levels of osteoclast transcription factors c-fos and Nfatc1 were increased in Itch Ϫ/Ϫ osteoclasts (Fig. 1B). Osteoclasts are derived from myeloid precursors. Several studies (23,24) including reports from our group (25,26) demonstrate that OCPs can be identified by a combination of cell surface markers such as CD45 ϩ /c-Kit ϩ /c-Fms ϩ cells. Furthermore, bone marrow OCPs include both CD45 ϩ /c-Kit ϩ /c-Fms ϩ /CD11b ϩ and CD45 ϩ /c-Kit ϩ /c-Fms ϩ /CD11b Ϫ cells, whereas peripheral OCPs only contain CD45 ϩ /c-Kit ϩ /c-Fms ϩ /CD11b ϩ cells (18,24,25). To determine whether the increased osteoclast formation in the absence of Itch is due to increased OCPs in Itch Ϫ/Ϫ mice, we examined the frequency of OCPs in bone marrow and spleen cells (peripheral OCPs) by FACS analysis. A similar percentage of OCPs was detected in bone marrow and spleen cells from Itch Ϫ/Ϫ mice and WT littermates (Fig. 1C).
We then examined whether overexpression of Itch can rescue the increased osteoclastogenesis phenotype of Itch Ϫ/Ϫ cells using a retroviral delivery approach. Overexpression of Itch in WT OCPs reduced RANKL-induced osteoclast formation (number of osteoclasts/well, 27 Ϯ 12 versus 40 Ϯ 22 in GFP controls). Overexpression of Itch in Itch Ϫ/Ϫ OCPs prevented increased osteoclast formation (number of osteoclasts/well, 138 Ϯ 18 versus 259 Ϯ 32 in GFP-infected cells) ( Fig. 2A). Itch expression after retroviral infection in WT and Itch Ϫ/Ϫ cells was confirmed by Western blotting (Fig. 2B).
RANKL Up-regulates Itch Expression via NF-B-To investigate the regulation of Itch during osteoclast differentiation, we treated WT OCPs with RANKL for 4 -72 h. RANKL significantly increased Itch mRNA expression starting at 4 h and peaking at 24 h (Fig. 2C), and this was confirmed by Western blot (Fig. 2D). RANKL did not affect TRAF6 mRNA expression and slightly increased TRAF6 protein levels at later time points.
NF-B is one of the downstream transcription factors of RANKL in osteoclasts and regulates the transcription of a number of target genes. To examine whether NF-B mediates RANKL-induced Itch expression, we first treated WT OCPs with the NF-B inhibitor pyrrolidine dithiocarbamate in the presence or absence of RANKL. Pyrrolidine dithiocarbamate itself did not affect Itch expression, but it remarkably reduced RANKL-induced Itch expression (Fig. 2E). We then overexpressed NF-B p65 in WT OCPs and found that p65 increased Itch expression (Fig. 2F). Finally, we searched NF-B binding sites within Ϫ4000 bp of the murine Itch promoter (NCBI Ref-Seq accession number NC_000068) using TFSEARCH software (version 1.3) and identified two putative NF-B binding sites at the Ϫ3473/Ϫ3465 and Ϫ3183/Ϫ3175 location. We performed ChIP assays using anti NF-B p65 antibody to pull down protein-chromatin complexes and two primer sets around these NF-B binding sites. Results showed p65 binding to both NF-B binding sites that was enhanced by RANKL treatment (Fig. 2G).
Itch Negatively Regulates RANKL-induced Activation of NF-B Signaling-Itch has been reported to be a negative regulator of the NF-B pathway in T cells by prolonging TNFinduced NF-B activation (6) NF-B is essential for osteoclast formation (27). We examined the effect of Itch deficiency on RANKL-induced activation of NF-B signaling in osteoclasts. OCPs from Itch Ϫ/Ϫ or WT mice were treated with RANKL (10 ng/ml) for different times, and the expression of IB␣ and phosphorylated IB␣, a commonly used readout for activation of canonical NF-B signaling, was examined. RANKL increased IB␣ phosphorylation in WT cells at 15 min, and then it subsequently declined. However, RANKL-induced IB␣ phosphorylation persisted for 240 min in Itch Ϫ/Ϫ cells (Fig. 3A). We observed increased JNK phosphorylation in WT cells at 15 min, and then it subsequently declined, whereas RANKL-induced JNK phosphorylation persisted for 240 min. However, the change of ERK phosphorylation was similar. RANKL increased ERK phosphorylation at 15 min, and it decreased gradually from 30 to 240 min. However, the basal level of ERK phosphorylation in Itch Ϫ/Ϫ cells was higher than that in WT cells (Fig.  3A). The NF-B activation was also examined using an ELISAbased assay in which the DNA binding capacity of NF-B p65 subunit was assessed. In WT cells, RANKL-triggered NF-B activation peaked at 15 min and declined thereafter, whereas in Itch Ϫ/Ϫ cells, RANKL-triggered NF-B activation persisted for 1 h (Fig. 2B). RANKL/RANK interaction recruits several TRAF adaptor proteins including TRAF2, -3, -5, and -6 to mediate downstream signaling. TRAF6 is a unique adapter protein for RANKL signaling in osteoclasts, and RANKL induces TRAF6 ubiquitination, thereby transducing positive signaling from the initial receptor/ligand interaction to downstream signaling proteins (28,29). We examined the ubiquitination of TRAF6. WT cells showed an increased ubiquitination of TRAF6 at 15 min of RANKL treatment, whereas Itch Ϫ/Ϫ cells showed a constitutive ubiquitination from 15 min to 1 h of RANKL treatment (Fig. 3C).
Itch Binds to TRAF6 and Limits TRAF6 Deubiquitination-To examine whether Itch affects RANKL-induced TRAF6 signaling, we first determined whether Itch interacts with TRAF6 in WT osteoclasts in the presence or absence of RANKL. Cells were treated with RANKL for 2 days, and whole cell lysates were immunoprecipitated with anti-TRAF6 antibody or IgG control and blotted with anti-Itch antibody. In the absence of RANKL, we observed small quantities of Itch associating with TRAF6. RANKL greatly increased the interaction between Itch and TRAF6 (Fig. 4A). RANKL has been reported to induce TRAF6 ubiquitination and activate downstream signals (28,29), which are limited via deubiquitination (30). To investigate whether Itch is involved in the regulation of TRAF6 ubiquitination and deubiquitination, we treated WT and Itch Ϫ/Ϫ OCPs with RANKL to promote TRAF6 ubiquitination and subsequently  removed RANKL to facilitate its deubiquitination using a published protocol (12). RANKL induced TRAF6 ubiquitination in WT and Itch Ϫ/Ϫ OCPs to the same degree after 2 days of treatment (Fig. 4B). However, ubiquitinated TRAF6 disappeared after RANKL withdrawal in WT cells, but it persisted in Itch Ϫ/Ϫ cells, suggesting dysregulation of TRAF6 deubiquitination in the absence of Itch (Fig. 4C).
Itch Increases the Interaction of Deubiquitinating Enzyme, Cylindromatosis, and TRAF6-Deubiquitination is carried out by deubiquitinating enzymes including CYLD (31). In osteoclasts, CYLD promotes TRAF6 deubiquitination and negatively regulates osteoclast formation (12). Itch does not have deubiquitinating activity (32). To determine whether Itch promotes binding of CYLD to TRAF6 to explain the reduced TRAF6 deubiquitination in Itch-deficient cells, we overexpressed FLAG-Itch and HA-CYLD in 293T cells and performed Western blotting and immunoprecipitation (IP) assays in whole cell lysates using anti-HA antibody for CYLD and anti-FLAG to detect Itch. A weak Itch band was detected in the IP complex by the IgG control, whereas a much stronger Itch band was observed in the IP complex by anti-HA (CYLD) (Fig. 5A).
To determine whether Itch binds to CYLD in osteoclasts, WT OCPs were treated with PBS or RANKL and subjected to immunoprecipitation using an anti-CYLD antibody. In the absence of RANKL, a weak Itch band was detected that was enhanced by RANKL treatment (Fig. 5B). To test whether Itch could facilitate CYLD binding to TRAF6, we immunoprecipitated TRAF6 with anti-TRAF6 antibody in OCPs from WT and Itch Ϫ/Ϫ mice in the presence and absence of RANKL for 48 h. RANKL increased the binding of CYLD to TRAF6 in WT OCPs, but this binding was attenuated in Itch Ϫ/Ϫ cells. RANKL-increased expression of CYLD was not changed in Itch Ϫ/Ϫ cells (Fig. 5C).
Itch Ϫ/Ϫ Mice Have Increased Osteoclast Formation in Response to LPS Stimulation-Data from the above in vitro studies suggest that Itch Ϫ/Ϫ mice may have an osteoporotic phenotype due to increased osteoclast formation. However, when we examined bones from 3-and 6-month-old Itch Ϫ/Ϫ mice using micro-computed tomography and histology, we found no difference in bone volumes or other bone mass parameters between Itch Ϫ/Ϫ mice and WT littermates (see Fig.  6A for data from 3-month-old mice; data from 6-month-old mice are not shown). Biomechanical testing did not detect any differences in bone strength between Itch Ϫ/Ϫ mice and WT littermates (data not shown). TRAP-stained femoral sections showed that Itch Ϫ/Ϫ mice have 30% higher osteoclast numbers than WT littermates (Fig. 6B). To examine whether Itch mediates increased osteoclast formation under pathological conditions such as periodontal disease, we injected LPS into the loose subcutaneous tissues overlying the calvaria of WT and Itch Ϫ/Ϫ mice. No significant difference was observed between osteoclast numbers in WT and Itch Ϫ/Ϫ mice in response to PBS. However, LPS injection induced significantly more osteoclasts and a greater eroded surface in Itch Ϫ/Ϫ mice than in WT mice (Fig. 6C). The osteoclast formation with LPS (100 ng/ml) in the absence of Itch expression in vitro was also examined. The number of TRAP ϩ osteoclasts was increased by more than 6-fold in cells from Itch Ϫ/Ϫ mice compared with WT cells (Fig.  6D).

DISCUSSION
In the present study, we demonstrated that the ubiquitin E3 ligase Itch inhibits RANKL-induced osteoclast formation. Using osteoclast precursors from Itch Ϫ/Ϫ mice, we identified two molecular mechanisms mediating the inhibitory effect of Itch on osteoclastogenesis. Itch limits TRAF6 deubiquitination by binding the deubiquitinating enzyme CYLD. In the absence of Itch, RANKL-induced NF-B activation is prolonged. Itch transcription is directly regulated by RANKL during osteoclast differentiation via NF-B. Furthermore, LPS-induced osteoclastic bone resorption is significantly increased in Itch Ϫ/Ϫ mice, indicating that Itch negatively regulates osteoclast formation in vivo under pathological conditions such as periodontal disease and bacterial infection-induced bone loss.
Similar to ubiquitination, deubiquitination is another important post-translational mechanism to regulate protein stability and thereby cell function. Protein deubiquitination is carried  out by DUBs, a group of proteases carrying DUB domains. Genomic and functional scans of the human genome found 95 DUBs including A20 and CYLD. Itch does not contain a DUB domain, suggesting that Itch itself does not have DUB activity (32). However, Itch interacts with A20 (6) and CYLD (this study), indicating that it may affect protein deubiquitination through these DUBs. A recent bioinformatics study indicates that a given DUB interacts with multiple adaptor proteins (33), highlighting the complexity of the deubiquitination process in regulating cellular functions. In T cells and mouse embryonic fibroblasts, Itch negatively regulates TNF-induced NF-B activation by facilitating A20 deubiquitination of the Lys 63 -linked ubiquitination of the adaptor protein receptor-interacting protein and switching receptor-interacting protein ubiquitination toward Lys 48 linkage, leading to its proteasomal degradation (6). A20 also deubiquitinates TRAF6 in liver cells (34) and terminates Toll-like receptor-mediated responses (13). A20 Ϫ/Ϫ mice have premature death due to severe inflammation and cachexia (35). Silencing A20 increases TRAF6 protein and NF-B activity in osteoclasts in the presence of LPS (36). Whether this is due to alteration of TRAF6 deubiquitination is unknown. Given the known interaction of Itch and A20 in other cell types, it is likely that Itch may facilitate the inhibitory effect of A20 on osteoclast formation.
Our study indicates that, apart from the interaction with A20, Itch binds to CYLD, the first DUB shown to deubiquitinate TRAF6 in osteoclasts (12). This Itch/CYLD interaction promotes CYLD-mediated TRAF6 deubiquitination in osteoclasts. p62 protein enhances CYLD binding to TRAF6, thereby facilitating its deubiquitination (12). It is not known whether Itch interacts with CYLD-p62-TRAF6. However, the findings that Itch Ϫ/Ϫ mice develop autoimmune disease and have enhanced osteoclast formation in response to LPS suggest that Itch may play an indispensible role in limiting inflammation and osteoclastogenesis in vivo.
We found that adult Itch Ϫ/Ϫ mice have normal bone volumes, similar to p62 Ϫ/Ϫ mice (37). Thus, basal osteoclastogenesis is not affected by the loss of p62 or Itch. Interestingly, both Itch Ϫ/Ϫ and Cyld Ϫ/Ϫ mice develop multiorgan inflammation (6,38), but they have different bone phenotypes. Cyld Ϫ/Ϫ mice have osteoporosis at 8 weeks of age (12), whereas 3-6-monthold Itch Ϫ/Ϫ mice have normal bone volume. This difference in bone volume may be due to different effects of Itch and CYLD on osteoblasts. Osteoblast differentiation is similar in WT and Cyld Ϫ/Ϫ mice (12). We found that Itch Ϫ/Ϫ mice have increased osteoblast numbers and bone formation rate. The bone marrow stromal cells from Itch Ϫ/Ϫ mice have increased osteoblast differentiation but a normal percentage of mesenchymal stem cells. At the molecular level, Itch promotes degradation of JunB protein, a transcription factor in osteoblasts (39). Increased osteoblast differentiation of Itch Ϫ/Ϫ cells may also be related to the known function of Itch-mediated degradation of bone mor- phogenetic protein (40) and TGF␤ signaling proteins (41). Consequently, increased osteoblastic bone formation compensates for increased osteoclastic bone resorption in Itch Ϫ/Ϫ mice. What is the potential biological relevance of Itch in bone? First, modification of Itch levels specifically in osteoclasts or osteoblasts may be useful to treat bone diseases such as osteoporosis.
Second, local Itch inhibition may be helpful for the bone repair process because it creates a high bone turnover microenvironment by increasing the function of both osteoclasts and osteoblasts. Finally, Itch Ϫ/Ϫ mice may provide a new model for investigating the involvement of the deubiquitination process in bone cell regulation. Furthermore, because the Itch Ϫ/Ϫ mice have autoimmune disease and multiorgan inflammation, cellspecific depletion of Itch using the Cre/Loxp system will be required to delineate the functions of Itch in osteoclasts and osteoblasts in vivo. RANKL increases Itch expression in OCPs within 24 h, consistent with a direct effect on transcription. Interestingly, other proteins that play a role in controlling TRAF6 deubiquitination including CYLD (12), A20 (36), and p62 (42) are also up-regulated by RANKL in osteoclasts. The A20 promoter contains functional NF-B binding sites (43). We found NF-B binding sites at the Itch promoter. Similarly, the Cyld and p62 promoters also contain NF-B binding sites (Ϫ3812/Ϫ3804 bp in Cyld and Ϫ2882/Ϫ2874 and Ϫ2416/Ϫ2408 bp in p62). These findings suggest that RANKL may up-regulate the expression of genes that encode DUBs and adaptor proteins through NF-B to terminate NF-B activation. Thus, deubiquitination of TRAF6 is a negative feedback mechanism to limit RANKL-mediated osteoclast formation (Fig. 7). Briefly, Itch Ϫ/Ϫ mice have increased osteoclast formation due to continuous activation of NF-B signaling. Itch promotes TRAF6 deubiquitination by recruiting CYLD to the TRAF6 signal transduction complex. Itch expression is up-regulated by RANKL, which is involved in NF-B signaling. Thus, Itch is a new RANKL target gene that is induced during osteoclastogenesis and functions as a deubiquitinating adaptor molecule to limit RANKL-induced osteoclast formation via the deubiquitinating enzyme CYLD.