The pivotal role of the alternative NF-κB pathway in maintenance of basal bone homeostasis and osteoclastogenesis

Abstract

The alternative NF-κB pathway consists predominantly of NF-κB-inducing kinase (NIK), IκB kinase α (IKKα), p100/p52, and RelB. The hallmark of the alternative NF-κB signaling is the processing of p100 into p52 through NIK, thus allowing the binding of p52 and RelB. The physiologic relevance of alternative NF-κB activation in bone biology, however, is not well understood. To elucidate the role of the alternative pathway in bone homeostasis, we first analyzed alymphoplasic (aly/aly) mice, which have a defective NIK and are unable to process p100, resulting in the absence of p52. We observed increased bone mineral density (BMD) and bone volume, indicating an osteopetrotic phenotype. These mice also have a significant defect in RANKL-induced osteoclastogenesis in vitro and in vivo. NF-κB DNA-binding assays revealed reduced activity of RelA, RelB, and p50 and no binding activity of p52 in aly/aly osteoclast nuclear extracts after RANKL stimulation. To determine the role of p100 itself without the influence of a concomitant lack of p52, we used p100^−/− mice, which specifically lack the p100 inhibitor but still express p52. p100^−/− mice have an osteopenic phenotype owing to the increased osteoclast and decreased osteoblast numbers that was rescued by the deletion of one allele of the relB gene. Deletion of both allele of relB resulted in a significantly increased bone mass owing to decreased osteoclast activity and increased osteoblast numbers compared with wild-type (WT) controls, revealing a hitherto unknown role for RelB in bone formation. Our data suggest a pivotal role of the alternative NF-κB pathway, especially of the inhibitory role of p100, in both basal and stimulated osteoclastogenesis and the importance of RelB in both bone formation and resorption. © 2010 American Society for Bone and Mineral Research

Introduction

NF-κB contributes to a broad range of biologic activities such as regulation of immunity, inflammation, cell growth, and survival. The pivotal role of the NF-κB family of transcription factors in osteoclasts (OCs) has come to light when the deletion of both NF-κB1 and NF-κB2 (Nfkb1^−/−/Nfkb2^−/−) caused osteopetrosis in mice owing to an absence of OCs.1, 2 NF-κB signals through two main pathways called classical (canonical) and alternative (noncanonical).3 The NF-κB family includes RelA, RelB, c-Rel, NF-κB1, and NF-κB2. NF-κB1 and NF-κB2 are synthesized as large precursors, p105 and p100, respectively. They have long C-terminal domains that contain multiple ankyrin repeats, rendering these precursors functionally similar to the inhibitor IκB.3 The inhibitory effect of p105 and p100 is relieved on processing into p50 and p52, respectively.3 Whereas the cleavage of p105 occurs constitutively in the classical pathway, p100-to-p52 processing is tightly regulated via the alternative pathway.

The alternative pathway is limited to a subset of tumor necrosis factor (TNF) cytokine family members such as receptor activator of NF-κB ligand (RANKL), B-cell activating factor (BAFF), CD40L, TWEAK, and lymphotoxin-β. Activation of NF-κB-inducing kinase (NIK) results in activation of IκB kinase α (IKKα) homodimers and phosphorylation and processing of the p100 precursor,3 thus releasing the active subunit p52. p100 is the only IκB that binds to RelB. Removal of the ankyrin repeats from p100, thereby reducing its IκB-like activity, appears to be the major characteristic of NIK in the alternative pathway in OC differentiation.4 Therefore, the processing of p100 allows nuclear translocation of RelB, which may heterodimerize with either p52 or p50 for subsequent gene transcription.

Recent work by Tucker and colleagues5 gives some insight into the role of p100 in the alternative pathway by using mice harboring a novel mutation in nfkb2 (Nfkb2^Lym1/Lym1), which encodes a nonprocessible form of p100. These mice have a complex phenotype with inflammatory lesions in liver and lung.5 The abnormalities in several tissues including lymph nodes (LN), thymus, and spleen of Nfkb2^Lym1/Lym1 mice resemble both NIK-deficient (Nik^−/−) and alymphoplasia (aly/aly) mice, where p100 processing is disturbed,6, 7 resulting in the accumulation of p100. RelB^−/− mice also have altered lymph nodes and defective splenic and thymic stromal cells, mimicking the Nik^−/− and aly/aly mice. The alymphoplasia phenotype is due to a mutation in the C-terminal region of NIK that interacts with IKKα and TNF receptor-associated factors (TRAFs).6 Given the fact that the immunologic phenotype is similar in RelB^−/−, Nfkb2^Lym1/Lym1, Nik^−/−, and aly/aly mice, the question arises as to whether they also have a similar bone phenotype. Whereas RelB^−/− mice and Nfkb2^Lym1/Lym1 mice show increased trabecular bone volume with reduced RANKL-induced in vitro OC formation,4, 5 Nik^−/− mice are reported to show increased bone volume4 with a reduction in RANKL-stimulated osteoclastogenesis compared with wild-type (WT) mice.8 Notwithstanding the preceding observations, the bone phenotype of aly/aly mice is not yet clarified.

Since the physiologic relevance of the alternative NF-κB pathway in bone homeostasis is also not well characterized, we started out by first investigating the bone phenotype of aly/aly mice and the physiologic role of p100 processing in basal as well as stimulated osteoclastogenesis. To further clarify the role of p100 in a physiologic context without the ambiguity of p52, we used p100^−/− mice that carry a homozygous deletion of the COOH-terminal ankyrin repeats of p100 but still containing a functional p52 protein.9, 10 We show here that p100^−/− mice are osteopenic owing to increased OC activity and impaired osteoblast (OB) parameters, whereas deletion of RelB could rescue the osteopenia in p100^−/− mice. This study shows for the first time a physiologic relevance of p100 in osteoclastogenesis and a role for RelB in OBs.

Materials and Methods

Animals

Eight-week-old male C57BL/6J wild-type (WT) mice and heterozygous for aly (aly/+) and homozygous aly (aly/aly) mice were obtained from Nippon Clea (Tokyo, Japan). Mice were maintained in our animal care facilities. Long bones (tibiae and femora) from 3-week-old WT, p100^−/+, p100^−/−, p100^−/−relB^−/+, and p100^−/−relB^−/− mice were received from the Leibniz Institute for Age Research (Fritz Lipmann Institute, Jena, Germany). The experimental procedures were reviewed and approved by the Animal Care and Use Committee of the Tokyo Medical and Dental University (Tokyo, Japan).

Reagents

GST-RANKL was from Oriental Yeast Co., Ltd. (Shiga, Japan). The vehicle used for the in vivo experiments was sterile PBS. For in vitro experiments cells were incubated in α-minimum essential medium (MEM, Sigma, St. Louis, MO, USA), supplemented with 10% fetal bovine serum (FBS, Moregate Biotech, Bulimba, Australia), 100 IU/mL benzyl penicillin, and 100 µg/mL streptomycin (Sigma). Recombinant human macrophage colony-stimulating factor (M-CSF) was from R&D Systems (Minneapolis, MN, USA). Soluble recombinant human RANKL was from Wako (Osaka, Japan). Anti-p65 (sc-372), RelB (sc-226), and p52 (sc-298) antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA).

In vivo RANKL injections

WT, aly/+ and aly/aly mice were divided into RANKL-injected (n = 6) and vehicle (veh)–injected groups (n = 6). Then 2 mg/kg of GST-RANKL or PBS-veh were injected subcutaneously under anesthesia with medetomidine hydrochloride (0.5 mg/kg; Domitor, Meijiseika, Tokyo, Japan) and ketamine hydrochloride (50 mg/kg; Ketalar, Sankyo, Tokyo, Japan). Heads of anesthetized mice were shaved to receive injections over the center of the calvariae. In order to recover from anesthesia quickly, mice were injected with atipamazole hydrochloride (0.25 mg/kg; Antisedan, Meiji Seika, Tokyo, Japan). Mice were sacrificed by cervical dislocation under anesthesia 5 days after the injections of RANKL.

Radiologic assessment

After sacrifice, calvariae, tibiae, and femora were dissected, and soft tissues were removed. All bones then were fixed in PBS-buffered glutaraldehyde (0.25%)–formalin (4%) fixative (pH 7.4) for 2 days at 4 °C and washed with PBS for further studies. The bone mineral denity (BMD) of tibiae and femora were measured by peripheral quantitative computed tomography (pQCT) (XCT Research SA+, Stratec Medizintechnik GmbH, Germany), as described elsewhere.11 Three-dimensional (3D) reconstruction images of proximal tibiae were obtained by focal micro-computed tomography (µCT) (ScanXmate-E090, Comscan, Kanagawa, Japan), as described previously.11

Histologic preparation and bone histomorphometry

Calvariae and long bones from 8-week-old WT, aly/+, and aly/aly mice and 3-week-old WT, p100^−/+, p100^−/−, p100^−/−relB^−/+, and p100^−/−relB^−/− mice were embedded in mixtures of methyl methacrylate (MMA) and 2 hydroxyethyl methacrylate (GMA) resins as described elsewhere.11, 12 Coronal and sagittal sections (3 µm thick) from calvariae and long bones were prepared, respectively, as described elsewhere with minor modifications.13 Sections were stained with tartrate-resistant acid phosphatase (TRAP) and counterstained with toluidine blue or methyl green. OCs were designated as TRAP+ multinucleated cells (three or more nuclei) located on the bone surface. The number of OCs in calvariae was counted in an area (0.29 × 0.68 mm) at the center of the calvariae including the sagittal suture, as described elsewhere.13 Standard bone histomorphometric analysis was performed in the secondary spongiosa of tibiae starting at 0.3 mm distal from the proximal growth plate to exclude the primary spongiosa by using an image analyzing system (KS400, Carl Zeiss, Jena, Germany). Some sections were stained according to von Kossa with a modified van Giesson method to clarify mineralized tissue.

In vitro osteoclastogenesis and bone-resorption assays

In vitro osteoclastogenesis and bone-resorption assays were performed as described elsewhere.14 In brief, bone marrow (BM) cells were cultured for 2 days in the presence of M-CSF (20 ng/mL). Bone marrow macrophages (BMMs) were cultured in the presence of M-CSF (20 ng/mL) and RANKL (10, 30, and 50 ng/mL) unless otherwise indicated. TRAP+ multinucleated cells were counted on day 6. Quantitative 3D analysis of bone resorption pits was carried out with a reflection confocal laser microscope (VK-8510K, Keyence Corp, Osaka, Japan) using Win Roof image analyzing software (Mitani Corp., Fukui, Japan), as described elsewhere.15, 16

Quantitative real-time polymerase chain reaction (QRT-PCR)

BMM were cultured for 6 days in the presence of M-CSF (20 ng/mL) and RANKL (50 ng/mL). QRT-PCR was carried out as described elsewhere.14 Each experiment was done in triplicate, and results were standardized against the mRNA level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Primers are listed in Table 1.

Table 1. Oligonucleotides Used for qRT-PCR

Target mouse gene	Sequence	GenBank reference
Note: Forward (F) and reverse (R) primers are listed. gapdh = glyceraldehyde-3-phosphate dehydrogenase; trap = tartrate resistant acid phosphatase; mmp-9 = matrix metalloproteinase-9.
gapdh	(F) 5′-CTCCCACTCTTCCACCTTCG-3′	NM008084
	(R) 5′-TTGCTGTAGCCGTATTCATT-3′
cathepsin K	(F) 5′-TGTATAACGCCACGGCAAA-3′	X94444
	(R) 5′GGTTCACATTATCACGGTCACA-3′
trap	(F) 5′-TACCTGTGTGGACATGACC-3′	BC029644
	(R) 5′-CAGATCCATAGTGAAACCGC-3′
mmp-9	(F) 5′-TCCAGTACCAAGACAAAGCCTA-3′	X72795
	(R) 5′-TTGCACTGCACGGTTGAA-3′
Integrin β3	(F) 5′-TGACTCGGACTGGACTGGCTA-3′	NM016780
	(R) 5′-ACTTCTCACAGGTGTCTCCAT-3′

Western blot analysis

BMMs treated with RANKL for the indicated times were lysed. The lysates were resolved in 4% to 12% gradient gels from Invitrogen (Carlsbad, CA, USA), transferred to Pall Fluoro Trans W membrane (Ann Arbor, MI, USA), and immunoblotted with individual antibodies. Then membranes were washed and incubated with horseradish peroxidase–conjugated secondary antibodies (Santa Cruz Biotechnology). The immunoreactive proteins were visualized using western lightning chemiluminescence reagent (Perkin Elmer, Boston, MA, USA).

NF-κB DNA-binding assay

TransAM NF-κB Family Transcription Factor Assay Kit (Active Motif, Carlsbad, CA, USA) was used to estimate the DNA-binding capacity of NF-κB, which was performed following manufacture's instructions. Briefly, cell extracts were prepared after 15 minutes of RANKL stimulation, as described previously.14 Nuclear extracts were prepared from WT, aly/+, and aly/aly cells using a Nuclear Extract Kit (Active Motif) and incubated in 96 well plates coated with immobilized oligonucleotide containing the NF-κB consensus site (5′-GGGACTTTCC-3′). NF-κB binding to the target oligonucleotide was detected by incubation with primary antibodies specific for the activated form of p65 (RelA), p50, p52, and RelB (Active Motif), followed by anti-IgG horseradish peroxidase conjugate and developing solution and quantified at 450 nm with a reference wavelength of 655 nm. Raji nuclear extracts provided in the kit were used as a positive control. This method has been shown to be more sensitive than electrophoretic mobility shift assays (EMSAs).17

Statistical analysis

All data are presented as means ± SD. The statistical significance of differences among groups was assessed using one-way ANOVA. When significant F values were detected, Fisher's PLSD post hoc test was performed to compare assay groups. Difference was considered significant when p < .05.

Results

Alymphoplasia (aly/aly) mice show osteopetrotic phenotype

Aly/aly mice carrying a mutated nik gene are characterized by the systemic absence of lymph nodes (LNs) and Peyer's patches, as well as disorganized splenic and thymic structures combined with immunodeficiency.6 Since NIK is necessary for the alternative NF-κB pathway in in vitro osteoclastogenesis,8 we examined the possibility of whether the mutation in NIK also may result in bone abnormalities in vivo. Radiologic observations of soft X-rays showed that the bones of aly/aly mice were slightly wider than those of WT and aly/+ controls, whereas the lengths of both tibiae and femora were almost comparable between WT and aly/aly mice (data not shown). Increased radiopacity in long bones was observed in soft X-ray images of aly/aly compared with WT and aly/+ mice (data not shown). µCT of tibiae confirmed the increased bone mass phenotype in aly/aly mice (Fig. 1A). Quantitative measurement of BMD by pQCT showed significantly increased trabecular and total BMD in aly/aly mice (see Fig. 1B, C). Abundant bone tissue was apparent in the cancellous region of tibiae by staining with von Kossa (see Fig. 1D). There was, however, no obvious defect in tooth eruption in aly/aly mice (data not shown).

Figure 1. Alymphoplasia (aly/aly) mice show increased radiodensity, BMD, and bone volume parameters. (A) µCT reconstruction images at the trabecular-rich region of proximal tibiae of 8-week-old wild-type (WT), aly/+, and aly/aly mice. Notice the increased bone density in the tibial metaphysis of aly/aly compared with WT and aly/+ mice. Bar represents 1 mm. (B, C) The trabecular and total BMD measured by pQCT. (D) Sagittal section of representative tibiae from 8-week-old WT, aly/+ and aly/aly mice stained according to the method of von Kossa. The mineralized matrix is shown in black. Bar represents 500 µm. Histomorphometric analyses of proximal tibiae were performed. White, gray, and black bars show WT, aly/+, and aly/aly mice, respectively. (E) BV/TV = trabecular bone volume (% bone volume). (F) Tb.Th = trabecular thickness. (G) Tb.N = trabecular number. (H) Tb.Sp = trabecular separation. Data are expressed as means ± SD. n = 6 for each group. *p < .05; **p < .001 versus WT control.

To further clarify the bone phenotype, standard bone histomorphometric analysis of tibiae was performed. Higher trabecular bone volume was observed in aly/aly mice (see Fig. 1E). The increase in trabecular bone volume was largely due to increased trabecular thickness (see Fig. 1F) and increased trabecular number (see Fig. 1G), whereas trabecular separation was reduced (see Fig. 1H). Both OC number (Fig. 2A, B) and OC-surface-to-bone-surface ratio (Fig. 2C) were significantly reduced in aly/aly mice. Gain of bone volume also may result from high bone formation caused by changes in OB differentiation and function. To evaluate the function of OBs, fluorescent labeling was performed (see Fig. 2D) to measure mineral apposition rate (MAR) and bone-formation rate (BFR). The results demonstrated mildly increased MAR (see Fig. 2E) and significantly increased BFR (see Fig. 2F) in aly/aly mice. Both OB number (see Fig. 2G) and OB-surface-to-bone-surface ratio (see Fig. 2H) were significantly increased in aly/aly mice. Collectively, these data indicate an osteopetrotic phenotype in aly/aly mice. Although OCs and OBs contributed to the osteopetrotic phenotype in aly/aly mice, the presence of abundant cartilage remnants (data not shown), which is a typical feature of osteopetrosis, indicated an impairment in OC function as the primary cause for the bone phenotype in aly/aly mice.

Figure 2. Bone resorption and bone formation indices of aly/aly mice. (A) Undecalcified sections (3 µm) of proximal tibiae from WT, aly/+, and aly/aly mice stained with TRAP and counterstained with methyl green. The arrowheads show the TRAP+ multinucleated cells. Bar represents 100 µm. (B) N.Oc /BS = osteoclast number/bone surface. (C) Oc.S/BS = osteoclast surface (% of bone surface). (D) Fluorescent images of undecalcified sections of tibiae. Bar represents 200 µm. (E) Mineral apposition rate (MAR). (F) Bone-formation rate (BFR). (G) N.Ob /BS = osteoblast number/bone surface. (H) Ob.S/BS = osteoblast surface (% of bone surface). Data are expressed as means ± SD. n = 6 for each group. *p < .05; **p < .001 versus WT control.

In vivo RANKL-induced osteoclastogenesis is blunted in aly/aly mice

To determine whether p100 processing through NIK has a role in RANKL-induced OC formation, we injected GST-RANKL over the calvariae of WT, aly/+, and aly/aly mice for 3 days. On day 5 after the first injection, mice were sacrificed, and calvariae were subjected to histomorphometry. Decalcified sections of calvariae revealed a blunted response to RANKL-induced osteoclastogenic stimuli in aly/aly mice (Fig. 3A). WT and aly/+ mice showed an increase in TRAP+ multinucleated cells along calvarial sinuses on stimulation with RANKL, whereas RANKL-injected aly/aly mice showed only a few TRAP+ multinucleated cells compared with vehicle-injected controls (see Fig. 3A, B).

Figure 3. Alymphoplasia mice have a blunted response to RANKL in vivo and in vitro. (A) WT, aly/+, and aly/aly mice were injected with 2 mg/kg GST-RANKL or PBS daily for 3 days and sacrificed on day 5 after the first injection. TRAP-stained coronal sections of calvariae show a dramatic increase in OCs along sutures and the periphery of the bone marrow cavity only in WT and aly/+ RANKL-treated animals. Arrowheads show TRAP+ multinucleated cells. Bar represents 100 µm. (B) Number of TRAP+ multinucleated (three or more) cells in the calvariae. The counting was performed in the region of interest (0.29 × 0.68 mm) including the sagittal suture, confirming the difference in the response of WT, aly/+, and aly/aly mice to RANKL injection. n = 6 for each group. **p < .001 versus vehicle (veh)–injected group in each genetic background. (C) Bone marrow macrophages (BMMs) were cultured for 6 days in the presence of M-CSF (20 ng/mL) and RANKL (10, 30, and 50 ng/mL) and then fixed and stained for TRAP, a marker for OC differentiation. WT and aly/+ cells formed large multinucleated cells, whereas aly/aly cell cultures failed to form large, spreading TRAP+ OCs. (D) Number of TRAP+ multinucleated cells (MNCs) (more than three nuclei). Data are expressed as the mean ± SD, and the representative of two independent experiments is shown. ***p < .0001 versus WT control.

In vitro osteoclastogenesis and activity are blunted in aly/aly mice by the reduction of NF-κB binding activity

Since we observed impaired RANKL-induced osteoclastogenesis in aly/aly mice in vivo, we carried out in vitro osteoclastogenesis assays to determine whether p100 processing also has a role in OC differentiation in vitro. BMMs of WT, aly/+, and aly/aly mice were cultured in the presence of M-CSF and various concentrations of RANKL. TRAP+ multinucleated cells were increased in a dose-dependent manner in WT and aly/+ cultures (see Fig. 3C). Cultures of BMMs of aly/aly mice, however, generated only a few TRAP+ multinucleated cells (see Fig. 3C). Quantitative analysis of TRAP+ multinucleated cells in cultures confirmed the preceding observations (see Fig. 3D). BMMs were cultured on dentine slices for 6 days in the presence of M-CSF and RANKL to examine the ability of OCs to form pits. Volume, area, and depth were measured using a reflection confocal laser microscope. Large coalesced lacunae on the dentine slices appeared in WT and aly/+ groups, whereas the resorption lacunae became smaller and sparse in aly/aly cultures (Fig. 4A). The volume of dentine excavated by aly/aly OCs was significantly reduced compared with that of WT and aly/+ controls (see Fig. 4B). The decrease in pit volume resulted from a decrease in both area (see Fig. 4C) and depth (see Fig. 4D). To further analyze the extent to which RANKL-mediated OC differentiation and activity were inhibited by the absence of functioning NIK, we assessed the expression of OC-specific genes using QRT-PCR in BMMs treated with RANKL and M-CSF for 6 days. Expression levels of trap, mmp-9, β₃, and cathepsin K were significantly decreased in aly/aly cultures compared with WT and aly/+ cultures (see Fig. 4E). The NIK mutation in aly/aly mice impairs p100 processing, resulting in the absence of p52.18, 19 Therefore, we next assessed the effect of RANKL on protein levels of p100 and p52 in WT and aly/aly cultures. In WT cultures, exposure to RANKL resulted in generation of p52 beginning at 3 hours (see Fig. 4F). In contrast, RANKL exposure for as long as 24 hours failed to induce the appearance of p52 in aly/aly cultures. Moreover, expression levels of RelB were slightly reduced in aly/aly compared with WT controls (see Fig. 4F).

Figure 4. Alymphoplasia mice have defective RANKL-stimulated activity in vitro and RANKL-stimulated p100 processing. (A) BMMs were cultured for 6 days on dentine in the presence of M-CSF (20 ng/mL) and RANKL (50 ng/mL). Representative microscopic views of dentine slices by reflection confocal laser scanning microscope are shown. Bar represents 50 µm. (B) Resorbed volume. (C) Resorbed area. (D) Average pit depth. (E) WT, aly/+, and aly/aly cultures were treated with RANKL for 6 days and then harvested for RNA extraction. Quantitative RT-PCR was performed with GAPDH as control. (F) WT and aly/aly samples were treated with RANKL for increasing times, and lysates were analyzed by immunoblot to detect changes in p100, p52 (arrowhead), RelA, and RelB. (G) DNA-binding activity of NF-κB subunits. NF-κB DNA-binding activity was assessed by TransAM NF-κB Family Transcription Factor Assay Kit using nuclear extracts of OCs after 15 minutes of RANKL stimulation. Data are expressed as the mean ± SD, and the representative of two independent experiments is shown. *p < .01; **p < .001; ***p < .0001 versus WT control.

To elucidate the effect of impaired p100 processing for NF-κB activation, NF-κB DNA-binding assays were performed. After 3 days of RANKL treatment, OCs were further stimulated for 15 minutes. Nuclear extracts from OCs were prepared and assayed using the TransAM NF-κB Kit as described in “Materials and Methods.” We observed decreased RelB-specific DNA binding in aly/aly compared with WT and aly/+ OCs. Interestingly, p50- and RelA-specific NF-κB activation also was reduced, whereas p52 DNA binding was undetectable in aly/aly nuclear extracts (see Fig. 4G).

Reduced BMD and trabecular bone in p100^−/− mice that can be rescued by deletion of one allele of the relB gene

The Nfkb2 knockout mouse, which lacks both p100 and p52, has normal basal osteoclastogenesis in vivo,1 suggesting that p52 itself may be dispensable for OC differentiation.8 Although the bone phenotype in aly/aly mice is due to impaired p100 processing, still we could not exclude the possibility that the reduction of p52 may have a role in the osteopetrotic bone phenotype in aly/aly mice. To address the role of p100 without the influence of p52, we used p100^−/− knockin mice, which lack p100 but still express functional p52.9, 10 First, to show that the p100^−/− mice lack the inhibitor p100 but still express p52, we checked the expression levels of p100 and p52 in WT and p100^−/− mice. p100 was readily detected in WT OCs (Fig. 5), and exposure to RANKL resulted in the generation of WT p52 (single arrowhead). In p100^−/− cultures, p100 was absent, whereas the knockin p52 (double arrowhead) was readily detected (see Fig. 5). It has been shown previously that the nuclear binding of p52 and RelB was strongly increased in tissues from p100^−/− mice compared with WT controls, whereas the nuclear activity of RelA remained largely unchanged.9, 10 We also analyzed p52, RelB, and RelA protein levels in extracts from p100^−/− and WT mouse embryonic fibroblasts (MEFs). Nuclear levels of p52 and RelB were clearly increased in the absence of the p100 inhibitor, whereas RelA protein levels were comparable with those of WT controls (AL, manuscript in preparation).

Figure 5.  p100^−/− mice lack p100 but still express p52. WT and p100^−/− mice were treated with RANKL for the indicated time periods, and lysates were analyzed to detect changes in p100 and p52 levels. In WT cultures, p100 levels remained steady, whereas p52 levels were increased by RANKL treatment (single arrowhead). p100^−/− cultures show a strong signal for the knockin p52 protein at all time points (double arrowheads).

To further clarify the physiologic importance of p100, we analyzed the bone phenotype of p100^−/− mice. Though newborn p100^−/− mice were grossly indistinguishable from their littermates, they could be recognized by their smaller size and hunched posture by postnatal days 10 to 14.10 First, we subjected the long bones of WT, p100^−/+, and p100^−/− mice to radiologic analysis. Increased radiolucency was observed in soft X-ray images of p100^−/− compared with WT and p100^−/+ mice (data not shown). Proximal tibiae were subjected to µCT analysis. Reconstructed 3D images revealed a marked reduction of trabecular bone in p100^−/− compared with WT and p100^−/+ mice (Fig. 6A). Thus aly/aly mice accumulate the p100 inhibitor and develop an osteopetrotic phenotype, whereas p100^−/− mice show increased activation of p52 and RelB correlating with osteopenia.

Figure 6. p100^−/− mice develop osteopenia that is rescued by deletion of one allele of the relB gene. (A) µCT reconstruction images at the trabecular-rich region of proximal tibiae. Bar represents 400 µm. (B) BMD measured by pQCT. (C) Sagittal section of representative tibiae from 3-week-old WT, p100^−/+, p100^−/−, p100^−/−relB^−/+, and p100^−/−relB^−/− mice stained according to the method of von Kossa. Bar represents 400 µm. (D) BV/TV = trabecular bone volume (% bone volume). (E) Tb.Th = trabecular thickness. (F) Tb.N = trabecular number. (G) Tb.Sp = trabecular separation. Data are expressed as means ± SD. n = 3 to 4 for each group. *p < .05; **p <.001; ***p < .0001; ^##p < .001; ^###p < .0001 versus WT control.

Since Nfkb2^−/− (p100^−/−p52^−/−) mice have no bone phenotype,1 we speculated whether the deletion of RelB, the predominant dimerization partner of p52 in the alternative pathway, could rescue the reduction of bone mass in p100^−/− mice. Supporting our notion, the osteopenia in p100^−/− mice was recovered by deletion of one allele of the relB gene (p100^−/−relB^−/+), whereas p100^−/−relB^−/− mice had a high bone mass compared with WT controls (see Fig. 6A). To confirm these observations of proximal tibiae, BMD measurements of long bones were performed. Analysis by pQCT showed that the BMD in p100^−/− mice was significantly reduced compared with WT and p100^−/+ controls, whereas the BMD in p100^−/−relB^−/+ mice was similar to that in WT controls (see Fig. 6B). Confirming the µCT data, the BMD in p100^−/−relB^−/− mice was significantly increased compared with that in WT mice (see Fig. 6B). Less bone tissue appeared in the cancellous region of p100^−/− tibiae by staining with von Kossa (see Fig. 6C), which was restored in p100^−/−relB^−/+ bones. In p100^−/−relB^−/− mice, however, abundant bone tissue was present (see Fig. 6C).

To further clarify the bone phenotype, standard bone histomorphometric analysis of tibiae was performed. Significant reduction of trabecular bone volume was observed in p100^−/− mice (see Fig. 6D). The reduction in trabecular bone volume was largely due to reduced trabecular thickness (see Fig. 6E) and trabecular number (see Fig. 6F), whereas trabecular separation was increased significantly (see Fig. 6G). Deletion of a one allele of relB in p100^−/−relB^−/+ mice recovered the bone structural indices in p100^−/− mice (see Fig. 6D–G), whereas deletion of both alleles of relB resulted in significantly increased bone volume and trabecular numbers compared with WT controls (see Fig. 6D, F).

Histologic sections showed many TRAP+ multinucleated cells in the cancellous region of proximal tibiae of p100^−/− mice (Fig. 7A). Both OC number (see Fig. 7B) and the OC-surface-to-bone-surface ratio (see Fig. 7C) were increased significantly in p100^−/− mice, and these were restored to control levels by deletion of a one relB allele (see Fig. 7A–C). Bone-formation parameters such as number of OBs and OB-surface-to-bone-surface also were reduced in p100^−/− mice compared with WT controls (see Fig. 7D, E). In the same manner, they were restored by deletion of one allele of the relB gene.

Figure 7. Bone-resorption and bone-formation parameters of WT, p100^−/+, p100^−/−, p100^−/−relB^−/+, and p100^−/−relB^−/− mice. (A) Undecalcified sections (3 µm) of proximal tibiae stained with TRAP and counterstained with methyl green. The arrowheads show the TRAP+ multinucleated cells. Bar represents 200 µm. (B) N.Oc /BS = osteoclast number/bone surface. (C) Oc.S/BS = osteoclast surface (% of bone surface). (D) N.Ob/BS = osteoblast number/bone surface. (E) Ob.S/BS = osteoblast surface (% of bone surface). Data are expressed as means ± SD. n = 3 to 4 for each group. *p < .01; **p < .001; ^##p < .001versus WT control.

Discussion

Alymphoplasia (aly/aly) mice have a defective NIK and are unable to process p100 to p52. To date, no defects of aly/aly mice have been described in other systems besides the immune system. In this study we used aly/aly mice and uncovered a requirement for the processing of p100 in the maintenance of bone homeostasis in addition to osteoclastogenesis under physiologic as well as stimulated conditions. Here, we showed that the defective NIK in aly/aly mice led to an osteopetrotic bone phenotype owing to defects in both OCs and OBs. In addition, we observed reduced BMD and bone volume parameters correlating with increased basal osteoclastogenesis in p100^−/− mice. Taken together, these results suggest that p100 plays an important role in the physiologic regulation of murine osteoclastogenesis.

Alymphoplasia (aly/aly) mice have some similarities with Nfkb1^−/−/Nfkb2^−/− double-knockout (dKO) mice. Whereas Nfkb1^−/− mice have partial defects in LN formation, all LNs are absent in dKO and aly/aly mice. In contrast to single-knockout mice, dKO mice are severely osteopetrotic and suffer from impaired tooth eruption.1, 2 A recent study by Tucker and colleagues5 shows that Nfkb2^Lym1/Lym1 mice, which express a nonprocessable form of p100, are mildly osteopetrotic. These mice have significantly increased bone volume and trabecular number. In addition, tooth eruption is not impaired in Nfkb2^Lym1/Lym1 mice. Similar to Nfkb2^Lym1/Lym1 mice, aly/aly mice also showed significantly increased bone volume and trabecular number in the cancellous region of tibiae (see Fig. 1). OC number and the OC-surface-to-bone-surface ratio was reduced in aly/aly mice (see Fig. 2), indicating the importance of p100 processing in physiologic osteoclastogenesis in vivo. In addition, mimicking Nfkb2^Lym1/Lym1 mice, aly/aly mice also have normal tooth eruption, indicating a mild osteopetrotic phenotype. Increased OB activity combined with reduced basal osteoclastogenesis also contributed to the osteopetrotic phenotype in aly/aly mice.

Our study shows a significant reduction in OC formation in aly/aly mice in response to RANKL in vivo and in vitro (see Fig. 3). This reduction in osteoclastogenesis resulted in reduced bone resorption (see Fig. 4A–D). Similar to our study, in vivo RANKL-induced osteoclastogenesis is blunted in Nik^−/− mice.8 Moreover, Nik^−/− mice show significantly reduced in vitro osteoclastogenesis that cannot be rescued by increasing concentrations of RANKL. The ability of Nfkb2^Lym1/Lym1 bone marrow cells to form OCs in response to RANKL in vitro is also reduced significantly.5 Therefore, we confirmed the importance of p100 processing in RANKL-stimulated osteoclastogenesis.

It has been shown that the NIK mutation in aly/aly mice impairs p100 processing, resulting in the accumulation of p100 in the cytoplasm and the absence of p52.18, 19. Using BMMs of WT and aly/aly mice, we show that processing of p100 is impaired in RANKL-stimulated cells of aly/aly compared with WT mice (see Fig. 4F). Consistent with this finding, Novack and colleagues8 showed that RANKL stimulation enhances p100 accumulation in the cytoplasm of Nik^−/− OCs. Accumulated p100 binds to other NF-κB subunits and prevents their DNA-binding activity.8, 20–22 Our study also showed the NF-κB-specific DNA-binding activity of RelA, p50, and RelB was impaired in aly/aly cells. These data suggest that the lack of p100 processing in aly/aly mice and the impaired DNA-binding activity of other NF-κB subunits resulted in impaired osteoclastogenesis.

To further clarify the physiologic role of p100, we analyzed p100^−/− mice with a homozygous deletion of the COOH-terminal ankyrin repeats of p100 but still containing a functional p52 protein9, 10 (see Fig. 5). NF-κB-binding activity of nuclear extracts from WT and p100^−/− mice showed that elimination of the ankyrin domain of the p100 precursor dramatically increased constitutive κB-binding activity in lymphoid and nonlymphoid tissues.10 p100^−/− mice had a significantly reduced BMD, with bone volume parameters (see Fig. 6) indicating an osteopenic phenotype. The reduced bone volume was due to increased basal osteoclastogenesis followed by reduced OB number and OB-surface-to-bone-surface ratio (see Figs. 6 and 7). Taken together, our data clearly indicate an important role for p100 in physiologic maintenance of bone homeostasis.

We observed clearly increased nuclear levels of p52 and RelB in the absence of the p100 inhibitor (AL, manuscript in preparation). In a previous paper, Guo and colleagues (2008) showed that the nuclear binding of p52 and RelB was increased in p100^−/− cultures compared with WT controls.9 Deletion of one relB allele could restore the RelB activity to WT levels, further showing that the increased RelB activity might be the reason for the immunologic phenotype in these mice.9 Consistent with these observations, we show that deletion of one allele of the relB gene was sufficient to restore the BMD and reduced bone volume parameters as well as bone resorption and bone formation indices to WT control levels (see Figs. 6 and 7). Deletion of both relB alleles significantly increased the bone mass in the absence of the inhibitor p100 (see Figs. 6 and 7). Since RelB^−/− mice have a significantly increased bone volume with no difference in OC number and OC surface in unmanipulated bones,4 the increased bone mass and bone-formation parameters in p100^−/−relB^−/− mice further support a hitherto unknown role for Rel B in bone formation.

In summary, we have shown that p100 processing by NIK is necessary for both basal and stimulated osteoclastogenesis. Lack of p100 processing in aly/aly mice reduces the DNA-binding activity of NF-κB subunits, resulting in impaired osteoclastogenesis and OC activity. Most important, the fact that defective NIK results in the arrest of p100 processing along with the osteopenic phenotype of p100^−/− mice suggests that both NIK and p100 are viable therapeutic targets for treating diseases of accelerated bone loss.

Disclosures

All the authors state that they have no conflicts of interest. Oriental Yeast Co. Ltd. (Shiga, Japan) provided GST-RANKL.

Acknowledgements

This study was supported by the grants from the Japanese Society for the Promotion of Science (JSPS) and the Ministry of Education, Culture, Sports, Science and Technology of Japan, to HS (18390497), EJ (20390473), KO (19390471), and KA (19390472). NSS and NA contributed equally to this paper.