IKKβ Activation Is Sufficient for RANK-Independent Osteoclast Differentiation and Osteolysis

Monocytes differentiate into osteoclasts through stimulation of receptor activator of NF-κB (RANK). Many downstream effectors of RANK play a positive role in osteoclastogenesis, but their relative importance in osteoclast differentiation is unclear. We report the discovery that activation of a single pathway downstream of RANK is sufficient for osteoclast differentiation. In this regard, introduction of constitutively activated IKKβ (IKKβSSEE) but not wild-type IKKβ into monocytes stimulates differentiation of bona fide osteoclasts in the absence of RANK ligand (RANKL). This phenomenon is independent of upstream signals because IKKβSSEE induced the development of bone-resorbing osteoclasts from RANK and IKKα knockout monocytes and in conditions in which NEMO-IKKβ association was inhibited. NF-κB p100 and p105, but not RelB, were critical mediators of this effect. Inflammatory autocrine signaling by tumor necrosis factor α (TNF-α) and interleukin 1 (IL-1) were dispensable for the spontaneous osteoclastogenesis driven by IKKβSSEE. More important, adenoviral gene transfer of IKKβSSEE induced osteoclasts and osteolysis in calvariae and knees of mice. Our data establish the sufficiency of IKKβ activation for osteolysis and suggest that IKKβ hyperactivation may play a role in conditions of pathologic bone destruction refractory to RANK/RANKL proximal therapeutic interventions. © 2010 American Society for Bone and Mineral Research.


Introduction
B one balance depends on the concerted activity of osteoblasts, bone-forming cells, and osteoclasts, bone-resorbing cells. In pathologic conditions such as rheumatoid arthritis and osteoporosis, bone balance favors increased osteoclast activation, (1) resulting in bone pain and increased fracture risk. Therefore, therapies that target the osteoclast are useful in these conditions. (2) On the other hand, gene mutations that disrupt osteoclast differentiation lead to development of osteopetrotic bones that compromise bone homeostasis. (3) Undoubtedly, increasing understanding of the factors that regulate the osteoclast in health and disease will offer important insight into new therapies for bone loss associated with pathologic conditions and for osteopetrosis.
The osteoclast differentiates from monocyte precursors through the action of ligand for the receptor activator for NF-kB (RANKL) and macrophage colony-stimulating factor (M-CSF). (1) On stimulation of their cognate receptors, RANK and c-Fms, a series of signaling events induces activation of transcription factors such as NF-kB, AP-1, and NFATc1, which results in fusion of precursors and expression of genes required for osteoclast function, including through phosphorylation of two IKK activation-loop serines. IKK then phosphorylates IkB, targeting it for proteasomal degradation and allowing NF-kB to enter the nucleus and regulate gene transcription. (14)(15)(16) Pharmacologic inhibition of the IKK association with NEMO abrogates osteoclastogenesis and inflammatory osteolysis. (17,18) Furthermore, mice devoid of IKKa (19) or IKKb (20,21) demonstrate an impaired ability for osteoclast development in vitro. Moreover, mice lacking IKKb displayed osteopetrosis and resistance to inflammatory bone erosion, whereas mice lacking active IKKa showed no skeletal phenotype. (20) This finding implicates IKKb as an important target for therapy in osteoclastmediated disease.
We now report that IKKb is not only necessary for RANKLmediated osteoclastogenesis, but its activation also is sufficient for osteoclast formation. Using retroviral delivery of constitutively active IKKb (IKKb SSEE ), we reveal a signal for differentiation of functional osteoclasts that occurs downstream of, but independent from, RANK. IKKb SSEE , but not wild-type IKKb or IKKa SSEE , induces osteoclast differentiation from monocytes. This phenomenon depends on NF-kB but does not require NEMO, IKKa, or RelB. Finally, adenoviral gene transfer of IKKb SSEE in knees and calvariae of mice is sufficient for development of massive osteolysis. Our findings demonstrate for the first time that a single activated kinase is sufficient for RANK-independent osteoclast differentiation and that active IKKb induces osteolytic disease. These data highlight the centrality of IKKb in osteoclast differentiation and implicate hyperactivation of IKKb in pathologic bone destruction.

Animals and cells
All mice were housed in a controlled barrier facility at Washington University (St Louis, MO, USA). TRACP-Cre mice (22) were from Dr Roodman (University of Pittsburgh, PA, USA). Floxed IKKb (23) mice were from Dr Pasparakis (University of Cologne, Germany). TRACP-Cre floxed/floxed IKKb mice were generated by crossing TRACP-Cre transgenic mice with floxed IKKb mice. IKKa heterozygous mice (24) were obtained from Dr Akira (Osaka University, Japan). RelB knockout (25) and control bone marrow was from Dr Novack (Washington University, St Louis, MO, USA). RANK knockout (26) and control spleens, as well as NF-kB doubleknockout (27) and control spleens were provided by Dr Xing (University of Rochester Medical Center, Rochester, NY, USA) For in vivo experiments, wild-type C57BL/6 mice at 5 to 6 weeks of age were used. Plasmids pMxs retroviral expression plasmid was from Dr T Kitamura (University of Tokyo, Japan). Mouse cDNA for IKKa was kindly provided by Dr Kenneth Marcu (Stony Brook, NY, USA). IKKb and RelB cDNA were purchased from ATCC (Manassas, VA). RelA cDNA was provided by Dr C Sasaki (NIA, Baltimore, MD, USA). All expression constructs were subcloned into pMxs using standard techniques. The following mutations were generated using the QuickChange II Site Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) with primer pairs in parentheses: IKKb SSEE (IKKb_S177_181E_f, GAGCTGGATCAGGGCGAACTGTGCACGGA-ATTTGTGGGGACTCTGC, and IKKb_S177_181E_r, GCAGAGTCCC-CACAAATTCCGTGCACAGTTCGCCCTGATCCAGCTC); IKKb WWAA (IKKb_W739_741A_f, GACTCTAGACGCGAGCGCGTTACAGATG-GAGGATG, and IKKb_W739_741A_r, CATCCTCCATCTGTAAC-GCGCTCGCGTCTAGAGTC); IKKb KM (IKKb_K44M_f, GTGAACA-GATCGCCATCATGCAATGCCGACAGGAGC, and IKKb_K44M_r,  GCTCCTGTCGGCATTGCATGATGGCGATCTGTTCAC); and IKKa SSEE (IKKa_S176_180E_f, GATGTTGATCAAGGAGAGCTCTGTACAGAA-TTTGTGGGAACATTGC, and IKKa-S176_180E_r, GCAATGTTCCCA-CAAATTCTGTACAGAGCTCTCCTTGATCAACATC). Note that the constitutive activating effect of the mutation of IKKb was established previously. (28,29) Generation of monocytes/macrophages Marrow was flushed from long bones into a minimum essential medium (a-MEM). Spleens and day 18.5 fetal livers were crushed into cell suspensions in a-MEM and were centrifuged at 453 rcf. Cell pellets were resuspended in whole medium [a-MEM with 1Â penicillin/streptomycin, 10% heat-inactivated fetal bovine serum (FBS)]. Monocytes/macrophages were produced by growing cell suspensions in the presence of 10 ng/mL of M-CSF. Monocytes/macrophages were allowed to proliferate for 3 days at 378C in 5% CO 2 , at which point they were infected with retrovirus (50% virus supernatant, 50% a-MEM containing 10% FBS, 10 ng/mL of M-CSF, 100 U penicillin/100 mg strep per Liter, and 4 mg/mL hexadimethrine bromide). Twenty-four hours after infection, cells were selected in a-MEM containing 10% FBS, 10 ng/mL of M-CSF, 100 U penicillin/100 mg strep per Liter, and 2 mg/mL puromycin for 72 hours, at which point selection medium was removed, and cells were washed and grown for 24 additional hours without puromycin. At this point, cells were lifted, counted, and plated for downstream experiments.

Generation of retrovirus
The use of Plat-E retrovirus packaging cells stably expressing retroviral structural proteins gag-pol and env for transient production of high-titer retrovirus was described previously. (30) Briefly, 8 mg of pMx vectors expressing our gene of interest was transfected into 5 million Plat-E cells (grown in DMEM supplemented with 10% FBS, 10 ng/mL of M-CSF, and 100 U penicillin/100 mg strep per Liter) using Fugene 6 (Roche, Palo Alto, CA, USA) according to manufacturer's instructions. Twentyfour hours after transfection, the medium was exchanged to remove the transfection reagent. Then 24 and 48 hours after medium exchange, supernatant was collected and pooled for infection of monocytes (see above). Inhibitor studies For inhibition of osteoclastogenesis, cells were treated with 100 ng/mL of OPG/Fc chimera (R&D Systems, Minneapolis, MN, USA), 25 mM TAT-NBD (YGRKKRRQRRR-G-TTLDWSWLQME) or 25 mM of TAT-mutant NBD (YGRKKRRQRRR-G-TTLDASALQME) during the entire course of retroviral transduction and in vitro osteoclast differentiation.

In vitro osteoclastogenesis
RNA isolation and cDNA production RNA was isolated from macrophage or osteoclast cultures using the Total RNA Isolation Mini Kit (Agilent Technologies, Santa Clara, CA, USA) according to manufacturer's instructions. Reverse transcription was described previously. (21) Real-time quantitative PCR The real-time quantitative PCR (qPCR) procedure was described in detail previously. (21) Western blotting The Western blot procedure was described previously. (21) One million cells were used for protein extraction and demonstration of protein expression.

Coimmunoprecipitation
One million cells expressing GFP, flag IKKb WT , flag IKKb WA , flag IKKb SSEE , or flag IKKb SSEE/WA were lysed in immunoprecipitation (IP) buffer [10 mM Tris, pH 7.4, 150 mM NaCl, 0.5% NP-40 (IGEPAL), 1 mM EDTA, 1 mM NaF, 1 mM PMSF, 1 mM Na 3 VO 4 , and 1Â protease inhibitor cocktail] at 48C. Protein was measured by BCA Assay (Pierce, Rockford, IL, USA) and normalized. Nonspecific binding was removed by rocking total cell lysate at 48C with GammaBind G Sepharose beads (GE Lifesciences, Piscataway, NJ) and 100 ng of normal mouse IgG for 2 hours at 48C. Beads and normal antibody were centrifuged, and supernatant was incubated with GammaBind G Sepharose beads and 1 mg/mL of mouse anti-Flag M2 antibody (Sigma) in a total of 700 mL of IP buffer and 1Â protease inhibitor cocktail at 48C for 16 hours. Immune complexes were centrifuged with beads. Supernatant was removed by vacuum suction, and 2Â sample buffer [0.5 M Tris-HCl, pH 6.8, 10% (w/v) SDS, 10% glycerol, 0.05% (w/v) bromphenol blue, 3% b-mercaptoethanol, and distilled water] was added to the beads, which were boiled for 5 minutes to elute the complex components, which were analyzed by Western blot.

Kinase assay
Plat-E cells expressing indicated flag-tagged IKKb constructs were lysed in IP buffer. IKKb was immunoprecipitated with M2 antibody, washed twice with IP buffer and once with kinase assay buffer (Cell Signaling Technologies Danvers, MA, USA), and incubated for 30 minutes at 308C in 30 mL of kinase assay buffer with 1 mg GST-IkBa, 2.5 mM MgCl 2 , and 16 mM ATP. The reaction was terminated with 30 mL of reducing sample buffer. Samples were analyzed by Western blot.

Bone-resorption assays
Osteoclasts were cultured on 5 mm 2 100-mm-thick dentin slices for 5 days in a 48-well tissue culture plate. To visualize resorption pits and tracks, slices were exposed to 0.5 N NaOH, and cells were removed by mechanical agitation. Slices were washed in PBS and stained with 0.1% toluidine blue (w/v) in PBS. Stained slices were rinsed with PBS and blotted dry, and pits were visualized by light microscopy. Resorption of artificial matrix was described previously. (21) Actin ring staining Cells were fixed in 4% paraformaldehyde in PBS for 5 minutes at room temperature. Fixed macrophages or osteoclasts on dentin slices were washed with PBS and permeabilized in 0.2% Triton X-100 in PBS for 10 minutes at room temperature. Dentin slices were washed with PBS and then incubated in a 1:40 dilution of Alexa Fluor-488 phalloidin (Invitrogen Molecular Probes, Eugene, OR, USA) for 10 minutes in a dark, humidified chamber at room temperature. Slices were washed with PBS and mounted onto microscope slides for visualization of actin rings with fluorescent microscopy.

Generation and use of adenovirus
Adenovirus expressing IKKb SSEE was generated by subcloning from the pMx parental vector into Ad5 CMV K-NpA Shuttle using EcoR1 and Not1 restriction endonucleases (New England Biolabs, Ipswich, MA, USA). Recombination, (31) production, and characterization [plaque-forming units (pfus)/particle] of virus were provided by Viraquest, Inc. (North Liberty, IA, USA). For local in vivo gene transfer in mice, 1 Â 10 7 pfus of virus diluted in 10 mL of sterile PBS were injected intraarticularly into the knee joint capsule. Contralateral knees on the same mouse served as experimental (Ad IKKb SSEE ) and control (Ad ntLacZ). Five mice were used for these experiments with comparable results. For calvarial osteolysis, 1 Â 10 7 pfus of virus diluted in 50 mL of sterile PBS were injected supracalvarially. Lipopolysaccharide (LPS) (10 mg) or RANKL (4 mg) diluted in 50 mL total volume of PBS were injected as positive controls. For calvarial experiments, five mice were used for each condition (virus, RANKL, and LPS) with comparable results. In vivo injections were executed with 100-mL insulin syringes with 29G needles. Seven days after injection, knees and calvariae were fixed, decalcified, and analyzed histologically for osteoclasts and osteolysis.

Histology
Bones were collected from mice and fixed in 10% buffered formalin for 24 hours. Bones then were decalcified for 7 days in buffer consisting of [14% (w/v) EDTA and H 4 NOH, pH 7.2], dehydrated in ethanol (30% to 70%), cleared through xylene, and embedded in paraffin. Sections were stained histochemically for TRACP to visualize osteoclasts or hematoxylin and eosine (H&E) to assess tissue architecture. Immunohistochemistry was performed according to the antibody manufacturer's instructions for immunoperoxidase staining.

Microscopy
Cells and histologic sections were imaged under white or ultraviolet (UV) light on an inverted microscope (Olympus IX-51). For f-actin visualization, UV light was passed through an FITC filter cube to localize green phalloidin. Digital images were captured using a CCD camera (Olympus DP70, 12-megapixel resolution).

Statistics
Student's two-tailed t test for comparison between means was used for all analyses.

Constitutively active IKKb induces RANKL-independent osteoclast differentiation from monocytes
We and others have demonstrated the necessity for IKKb in osteoclast differentiation. (20,21) In an effort to identify mutations in IKKb that could prevent or enhance its ability to rescue osteoclast differentiation in cells lacking IKKb, we made the observation that constitutively activated IKKb (IKKb SSEE ), but not the wild-type (IKKb WT ) form, in wild-type or IKKb knockout bone marrow-derived macrophages induced the formation of osteoclasts in the absence of RANKL (Fig. 1A). Levels of IKKb WT and IKKb SSEE protein were comparable ( Figure 1B), whereas IKKb SSEE , but not IKKb WT , was recognized by an antibody specific for IKKb phosphorylated at activation-loop serines (Fig. 1C), suggesting that the kinase domain of IKKb SSEE exists in an active conformation and that mutation of IKKb activation-loop serines 177 and 181 to glutamic acid, and not overexpression of IKKb, is responsible for the formation of osteoclasts in the absence of RANKL.
Further characterization showed that IKKb SSEE induced expression of RelB and c-fos, which are important for normal osteoclast differentiation. (13,32) IKKb SSEE , but not IKKb WT , also induced the expression of b 3 -integrin and cathepsin K, two markers for mature osteoclasts whose products are required for bone resorption (6,33) (Fig. 1C). Real-time qPCR analysis revealed that IKKb SSEE also induced expression of calcitonin receptor, cathepsin K, TRACP, and b 3 -integrin (Fig. 1D). Furthermore, IKKb SSEE -induced osteoclasts form actin rings and resorb artificial (not shown) and authentic bone matrix (Fig. 1A). Expression of IKKb SSEE by RANKL-independent osteoclasts was demonstrated using IKKb SSEE -GFP fusion construct (not shown). These data provide evidence that the TRACP þ multinucleated cells induced through expression of constitutively active IKKb in macrophages are authentic osteoclasts.
Next, we sought to examine whether stimulation of osteoclast differentiation through introduction of IKKb SSEE was a phenomenon restricted to precursors obtained from adult tissue. To this end, IKKb SSEE -infected, but not GFP-or IKKb WT -infected, fetal liver cells formed authentic osteoclasts with visible actin rings that resorbed dentin (not shown). Actin rings and resorption pits were observed in IKKb WT -infected cells only after RANKL administration (not shown). These observations reveal that IKKb SSEE is sufficient to induce an authentic program for functional osteoclasts from adult and fetal precursor cells independent of RANKL. To verify the specificity of the osteoclastogenic effect of the phosphomimmetic mutation, we mutated IKKb activationloop serines to alanine (IKKb SSAA ) and lysine to methionine (IKKb KM ), respectively. These mutations resulted in an activationdeficient molecule that failed to rescue RANKL-induced osteoclastogenesis in IKKb knockout monocytes (Fig. 1E, F). Therefore, phosphomimmetic mutation of IKKb activation-loop serines is a specific inducer of the osteoclast program, and inactivating the kinase domain of this molecule hinders its osteoclastogenic activity.

IKKb SSEE Rescues RANK Knockout Osteoclast Phenotype
Having established that RANKL is dispensable for IKKb SSEEmediated osteoclastogenesis, we tested whether intrinsic RANK signaling played a role in this phenomenon by using the RANKL decoy molecule osteoprotegerin (OPG-Fc) (10,34) and RANK-null cells. OPG-Fc completely inhibited RANKL-induced osteoclastogenesis in IKKb WT -infected macrophages but had no effect on IKKb SSEE -induced osteoclast differentiation, indicating that IKKb SSEE induces osteoclastogenesis without RANKL ( Fig. 2A). Furthermore, IKKb SSEE , but not IKKb WT nor GFP, induced the formation of osteoclasts from RANK knockout cells (Fig. 2B). Importantly, RANK knockout macrophages expressing GFP or IKKb WT failed to form osteoclasts in response to RANKL (Fig. 2B). IKKb SSEE -induced osteoclasts also formed actin rings and resorbed dentin (Fig. 2C). Consistent with this result, Western blot revealed that IKKb SSEE introduction into, but not RANKL treatment of, RANK knockout cells resulted in expression of c-fos and RelB, as well as c-src, b 3integrin, and cathepsin K (Fig. 2D), indicating that IKKb SSEE -induced RANK knockout osteoclasts are indeed bona fide osteoclasts. Real-time qPCR supported this conclusion (Fig. 2E). Therefore, IKKb SSEE functions independent of RANK to induce differentiation of functional osteoclasts. IKKb SSEE acts independently of the classical IKK complex to drive osteoclastogenesis Activation of IKKb by upstream signals requires its association, via two carboxyl-terminal tryptophans (W739 and W741), with NEMO. (35,36) Since IKKb SSEE induces osteoclastogenesis independent of RANK, we tested whether IKKbSSEE also could induce osteoclastogenesis in the absence of NEMO binding. First, we determined that while administration of cell-permeable NBD peptides, which inhibit the association of IKKb with NEMO, blocks RANKL-induced osteoclast differentiation, NBD did not inhibit osteoclastogenesis in response to transduction of IKKb SSEE (Fig. 3A). Second, while mutations of W739 and W741 to alanine in the presence of the S177 and S181 to glutamic acid (IKKb SSEE/WA ) prevent IKKb SSEE from binding to NEMO (Fig. 3B, C), IKKb SSEE/WA is still capable of inducing RANKL-independent osteoclastogenesis (Fig. 3D). This quadruple IKK mutant is expressed properly and retains its kinase activity (Fig. 3E, F). These results solidify the conclusion that IKKb SSEE induces RANKLindependent osteoclastogenesis without binding to NEMO, uncoupling the mechanism of IKKb SSEE -induced osteoclastogenesis from all known upstream stimuli important for osteoclast differentiation. This suggests that in the setting of osteoclast differentiation, IKKb binding to NEMO is important only for IKKb activation-loop phosphorylation, after which point the association is not required.
Based on these results, we hypothesized that IKKb SSEE could induce osteoclastogenesis without the classical IKK complex, which includes IKKa, a kinase that is required for osteoclastogenesis in vitro. (19) We confirmed that IKKa null fetal liverderived macrophages (FLCs) do not differentiate into osteoclasts (Fig. 4A) and fail to express mRNA for osteoclast markers in response to RANKL stimulation (Fig. 4C). However, transduction of IKKa null FLCs with IKKb SSEE restores osteoclastogenesis in the absence of RANKL, restores actin rings and bone resorption (Fig. 4A, phalloidin and tol. blue, respectively), and induces expression of typical signaling proteins (Fig. 4B) and expression of mRNA for osteoclast markers (Fig. 4C). These data indicate that formation of the classical IKK complex and the IKKa-mediated noncanonical NF-kB signaling pathway are not a requirement for IKKb SSEE to stimulate RANK-independent osteoclastogenesis.

Requirement for coordinated NF-kB Activation in IKKb SSEE -Induced Osteoclastogenesis
To identify the mechanism underlying IKKb SSEE -induced osteoclastogenesis, we examined the status of essential NF-kB subunits compared with RANKL-treated conditions. We observed elevated levels of RelB in the cytosol of IKKb SSEE -expressing cells at all time points assessed, including nonstimulated, compared with GFP-and IKKb WT -expressing cells. We also observed reduced levels of IkBa that coincided with an increased level of RelA protein in the nucleus in the absence of RANKL stimulation and at all time points tested in IKKb SSEE -compared with GFP-and IKKb WT -expressing cells, indicating that the constitutively activated form of IKKb induces continuous IkBa processing (Fig. 5A). These data suggest that IKKb SSEE acts through an NF-kBdependent mechanism to induce osteoclast differentiation. To test this, we challenged RelB knockout cells with IKKb SSEE because we observed induction of RelB protein expression in response to IKKb SSEE in macrophages and because RelB expression is required for RANKL-induced osteoclast differentiation in vitro and for stimulated but not basal osteoclast formation in vivo. (13) RelB knockout bone macrophages were capable of differentiating into TRACP þ osteoclasts that express cathepsin K in the absence of RANKL when expressing IKKb SSEE (Fig. 5B, C). Realtime PCR revealed that while induction of expression of mRNA for calcitonin receptor and TRACP in response to RANKL was impaired in RelB null cells, IKKb SSEE rescued the induction to levels equivalent to that in wild-type cells expressing IKKb SSEE (Fig. 5D). Therefore, IKKb SSEE does not require RelB to induce osteoclast differentiation.
Given the observation that IKKb SSEE induces nuclear translocation of RelA and induces increased expression of RelB, we further explored the possibility that these NF-kB subunits mediate the IKKb SSEE effect. However, overexpression of RelA, RelB, or a combination of RelA and RelB (Fig. 5F) did not induce osteoclast differentiation. To verify activity of the RelA and RelB, we observed that RelA induced expression of IkBa and that RelA and RelB alone or in combination induced expression of p100 (Fig. 5E). These results indicate that IKKb SSEE is a specific activator of NF-kB capable of inducing osteoclast differentiation and that ectopic overexpression of RelA and RelB is insufficient to coordinate this effect.
Phosphorylation of T-loop residues is a hallmark of activation for many kinases. (37) Given the specificity of IKKb activation as a mediator of osteoclast differentiation, we asked whether constitutive activation of other kinases through phosphomimmetic mutations also could induce osteoclast differentiation. IKKa and IKKb share significant primary and secondary structural homology, (15) so we reasoned that in contrast to other kinases, constitutive activation of IKKa through phosphomimmetic mutation would be most likely to induce an osteoclast program like IKKb SSEE . We found that when expressed at comparable levels ( Fig. 5E), IKKb SSEE induces osteoclast differentiation from macrophages, whereas IKKa SSEE had no such effect (Fig. 5F), demonstrating that IKKb is the specific kinase activator of the osteoclast program.
It has been established that a combination of both NF-kB1/ p50 and NF-kB2/p52 subunits is required for osteoclast differentiation. (11) We tested whether IKKb SSEE -induced RANKindependent osteoclastogenesis also requires NF-kB1 and -2 by transducing control and NF-kB1 À/À /NF-kB2 À/À (NF-kB doubleknockout) spleen macrophages with GFP, IKKb WT , and IKKb SSEE Fig. 2. IKKb SSEE -induced osteoclastogenesis does not require RANKL/RANK upstream signals. (A) Equal number of macrophages (30,000 cells/well) were cultured in the presence of M-CSF with or without RANKL, each in the absence or presence of OPG/Fc chimera. IKKb SSEE -expressing cells were cultured with M-CSF in the absence or presence of OPG/Fc chimera. Cultures were carried out as described under ''Materials and Methods.'' Cells were stained with TRACP to visualize osteoclasts. (B) Wild-type, RANK þ/? , or RANK -/spleen-derived macrophages were infected with a retrovirus expressing GFP, IKKb WT , or IKKb SSEE . These cells were cultured in the presence of M-CSF alone or in combination with RANKL for 4 days and stained with TRACP to visualize osteoclasts. (C) Wild-type and RANK À/À spleen-derived macrophages were infected with a retrovirus expressing IKKb WT or IKKb SSEE . These cells were cultured in the presence of M-CSF alone or in combination with RANKL on dentin and were stained with phalloidin or toluidine (Tol.) blue to visualize actin rings and resorption pits, respectively. Scale bars indicate relative magnification. Resorbed areas were quantified using Bioquant and expressed as percent area. (D) An equal number of wild-type (þ/ þ ) or RANK knockout (À/À) spleen cells infected with the indicated viruses were cultured in the presence of M-CSF or M-CSF þ RANKL (RL), and an equal number of total cell lysates were analyzed by Western blot for expression of the indicated proteins. The lower panel represents real-time qPCR for RANK mRNA in wild-type and RANK knockout cells. (E) Relative expression of mRNA for osteoclast markers assessed by realtime qPCR. GAPDH served as internal standard normalization. Values are expressed as relative quantity plus SEM. (Fig. 5G) and performing TRACP staining for osteoclasts in the absence of RANKL administration. While control cells expressing IKKb SSEE produced a significant number of osteoclasts capable of resorbing bone coinciding with expression of mRNA for cathepsin K, no osteoclasts were observed in NF-kB doubleknockout cells ( Figure 5H,I) despite constitutive IkBa processing (not shown). We conclude that IKKb SSEE -mediated induction of osteoclastogenesis requires NF-kB-mediated gene regulation.

Constitutively active IKKb is sufficient for the establishment of in vivo osteolysis
To determine the relevance of IKKb hyperactivation in vivo, we injected mice with adenovirus expressing IKKb SSEE or LacZ supracalvarially or intraarticularly into the knee joint (Supplemental Fig. 1). While LacZ did not induce an osteoclast response in either calvariae or knees, IKKb SSEE stimulated a massive local osteolytic response in both settings characterized by bone destruction and the appearance of osteoclasts at sites of bone erosion (Fig. 6A, B). To support the role of the kinase activity of IKKb in mediating this effect, joints injected with adenoviral IKKb SSEE showed intense immunostaining for phosphorylated IkBa at sites of osteoclastic articular bone erosion, whereas LacZinfected knees stained negatively for articular osteoclasts and phosphorylated IkBa (Fig. 6B).

Discussion
We provide evidence that osteoclast differentiation can be triggered by an autonomous intracellular signal downstream yet independent of RANK. IKK has been implicated in RANKLinduced osteoclastogenesis, (17)(18)(19)(20)(21) but the sufficiency of this single enzyme to independently induce osteoclastogenesis is surprising. The explanation for this phenomenon is likely to involve complex signal regulation that mimics NF-kB activation by RANKL. It is also possible that IKKb SSEE takes on functions not performed by IKKb in normal settings. In support of this, we observe that infection of monocytes with IKKb SSEE results in activation of p100 NF-kB (JO and YA, unpublished observations), which is usually considered to be a function of IKKa. (38) Perhaps atypical functions such as this contribute to its osteoclastogenic activity. Nevertheless, the ability of IKKb SSEE to induce the osteoclast depends on kinase activity because mutation of the ATP-binding lysine to methionine in the kinase domain abrogated the IKKb SSEE osteoclastogenic function ( Fig. 1E and Supplemental Fig. 2).
Differentiation of the osteoclast requires NF-kB. (11) To determine whether the phenotype we observed also requires NF-kB, we tested the ability of IKKb SSEE to drive osteoclastogenesis in NF-kB1/2 double-knockout monocytes, in which it failed. In addition to NF-kB, other transcription factors may play a role in the IKKb SSEE effect. Interestingly, we observed expression of NFATc1 protein, the master regulator of osteoclastogenesis, (39) induced by IKKb SSEE in monocytes (Supplemental Fig. 3). Whether NFATc1 is required for osteoclastogenesis induced by active IKKb or whether IKKb controls NFATc1 activity directly in the differentiating osteoclast is unknown.
NF-kB is also a critical regulator of inflammatory signals, (40) and inflammatory cytokines enhance osteoclast function. (41,42) Since tumor necrosis factor a (TNF-a) and interleukin 1 (IL-1) induce osteoclast differentiation in certain settings, (43) and since we observed both TNF-a and IL-1b expression by monocytes transduced with IKKb SSEE (JO and YA, unpublished observations), we sought to determine whether these inflammatory factors were required for IKKb SSEE to induce osteoclast differentiation. Using IKKb SSEE -transduced TNF-a or IL-1 receptor knockout monocytes, we found that TNF-a and IL-1 are not required for IKKb SSEE to accomplish its effect in osteoclast differentiation (JO and YA, unpublished observations). Therefore, IKKb SSEE -induced osteoclastogenesis in vitro is uncoupled from inflammatory signaling with respect to TNF-a and IL-1. Nevertheless, given that IKKb SSEE does induce secretion of these factors, we must consider the possibility that this kinase modulates osteoclast activation at sites of inflammation through inflammatory signals, the nature of which will be investigated in future studies.
Consistent with our in vitro findings, adenoviral gene-transfer experiments revealed that IKKb SSEE is sufficient for the establishment of osteolysis in vivo. The clinical significance of our findings is highlighted by our observations that IKKb SSEE -induced osteoclastogenesis is refractory to intervention with OPG and deletion of RANK/RANKL. In this regard, a number of conditions in human patients are associated with heightened bone turnover in the setting of inflammation for which a cause has not been identified. (44) Given the potency with which activated IKKb induces osteoclast appearance and bone destruction in this model, it is important to consider IKKb activation as an independent cause and a target in therapy for conditions of inflammatory bone destruction.
Our data highlight the critical role of IKKb in osteoclast differentiation and osteolysis. We have found that constitutively active IKKb unfolds the osteoclast program in the absence of upstream signals. We report the first evidence of RANKindependent osteoclast differentiation that is induced through a single kinase, and we propose that hyperactivation of human IKKb may lead to diseases resulting in bone destruction that would be refractory to treatments targeting receptor-proximal signaling molecules.

Disclosures
All the authors state that they have no conflicts of interest.