The IkappaB function of NF-kappaB2 p100 controls stimulated osteoclastogenesis

Abstract

The prototranscription factor p100 represents an intersection of the NF-kappaB and IkappaB families, potentially serving as both the precursor for the active NF-kappaB subunit p52 and as an IkappaB capable of retaining NF-kappaB in the cytoplasm. NF-kappaB-inducing kinase (NIK) controls processing of p100 to generate p52, and thus NIK-deficient mice can be used to examine the biological effects of a failure in such processing. We demonstrate that treatment of wild-type osteoclast precursors with the osteoclastogenic cytokine receptor activator of NF-kappaB ligand (RANKL) increases both expression of p100 and its conversion to p52, resulting in unchanged net levels of p100. In the absence of NIK, p100 expression is increased by RANKL, but its conversion to p52 is blocked, leading to cytosolic accumulation of p100, which, acting as an IkappaB protein, binds NF-kappaB complexes and prevents their nuclear translocation. High levels of unprocessed p100 in osteoclast precursors from NIK-/- mice or a nonprocessable form of the protein in wild-type cells impair RANKL-mediated osteoclastogenesis. Conversely, p100-deficient osteoclast precursors show enhanced sensitivity to RANKL. These data demonstrate a novel, biologically relevant means of regulating NF-kappaB signaling, with upstream control and kinetics distinct from the classical IkappaBalpha pathway.

Figure 1.

NIK^−/− mice have defective RANKL-mediated osteoclastogenesis in vitro. (A) Equal numbers of WT and NIK^−/− cells from unfractionated bone marrow were cultured for 8 d in the presence of vitamin D3 and then fixed and stained for TRAP, a marker of OC differentiation. Under these conditions, both stromal cells and OC precursors are present. NIK^−/− cultures fail to form large, spread, TRAP-positive OCs. Bar, 300 μm. (B) Equal numbers of Mφs (expanded from bone marrow in high doses of M-CSF) from WT and NIK^−/− mice were cultured in the presence of M-CSF and the indicated dose of RANKL for 6 d and then fixed and stained for TRAP. Very few, poorly spread TRAP-positive cells are seen in the NIK^−/− cultures. Bar, 100 μm. (C) Osteoclastogenic cultures were generated as in B on an artificial hydroxyapatite matrix (Osteologic slides). On day 6, cultures were stained with methylene blue at basic pH. Resorbed matrix is seen as a clear white area around dark cells. NIK^−/− cultures have negligible matrix resorption. Bar, 300 μm. (D) WT and NIK^−/− Mφs were treated with RANKL for 0–4 d and then harvested for RNA extraction. Semiquantitative RT-PCR was performed for CTR, MMP9, and β3 integrin, all markers of OC differentiation, with GAPDH as control. Although β3 integrin levels are comparable at day 1, all three markers show a failure in induction from days 2–4 in NIK^−/− cultures, confirming the morphological lack of differentiation.

Figure 2.

NIK^−/− mice have normal basal osteoclastogenesis and a blunted response to RANKL in vivo. (A) TRAP-stained sections of femurs from unmanipulated 3–4-mo-old WT and NIK^−/− mice show no difference in trabecular architecture or OC number red-stained cells. Bar, 600 μm. (B) WT and NIK^−/− mice were injected with 500 μg GST-RANKL or PBS daily for 7 d and killed on the 8th day. TRAP-stained coronal sections of calvaria adjacent to the sagittal suture show a dramatic increase in OCs along sutures and sinusoids only in WT RANKL-treated animals. Bar, 150 μm. (C) Quantitation of osteoclasts along inner (sinusoidal) calvarial surfaces for animals described in B confirms the significant difference in the response of WT and NIK^−/− mice to RANKL injection. Error bars indicate SEM. *P < 0.0001 compared with WT sham. **P < 0.001 compared with NIK^−/− sham and WT RANKL.

Figure 2.

NIK^−/− mice have normal basal osteoclastogenesis and a blunted response to RANKL in vivo. (A) TRAP-stained sections of femurs from unmanipulated 3–4-mo-old WT and NIK^−/− mice show no difference in trabecular architecture or OC number red-stained cells. Bar, 600 μm. (B) WT and NIK^−/− mice were injected with 500 μg GST-RANKL or PBS daily for 7 d and killed on the 8th day. TRAP-stained coronal sections of calvaria adjacent to the sagittal suture show a dramatic increase in OCs along sutures and sinusoids only in WT RANKL-treated animals. Bar, 150 μm. (C) Quantitation of osteoclasts along inner (sinusoidal) calvarial surfaces for animals described in B confirms the significant difference in the response of WT and NIK^−/− mice to RANKL injection. Error bars indicate SEM. *P < 0.0001 compared with WT sham. **P < 0.001 compared with NIK^−/− sham and WT RANKL.

Figure 3.

RANKL-induced p100 processing does not occur in NIK^−/− OC precursors despite normal induction of nfkb2 transcription. (A) WT and NIK^−/− Mφs were treated with RANKL for increasing times, and total lysates were analyzed by immunoblot to detect changes in p100 and p52 levels, using a monoclonal antibody recognizing the NH₂ terminus common to p100 and p52. The blot was stripped and reprobed with anti-β actin as a loading control (bottom). In WT cultures, p100 levels remain steady, whereas p52 levels are increased by RANKL treatment. Parallel NIK^−/− cultures fail to generate p52 at any time point and accumulate p100. (B) WT Mφs were treated with RANKL or TNFα for the indicated times, and p100/p52 levels were assessed by immunoblot as in A. p100 processing does not occur in response to TNFα. (C) WT and NIK^−/− Mφs were treated with RANKL for increasing times, and RNA was harvested. Levels of nfkb2 mRNA were analyzed by semiquantitative RT-PCR using GAPDH as a control. Numbers beneath lanes indicate fold increase over WT 0 h baseline. Although WT and NIK^−/− cultures show equivalent induction of nfkb2 at 6–12 h, levels are significantly lower in NIK^−/− cultures thereafter.

Figure 4.

NF-κB signaling in NIK^−/− Mφs is intact. (A) Mφs were acutely stimulated with RANKL for the indicated times, and cytoplasmic (top) and nuclear (bottom) fractions were analyzed by immunoblot for IκBα and RelA, respectively. (B) NF-κB DNA binding activity was assessed at 0 and 15 min by EMSA using a κB3 oligonucleotide probe and nuclear extracts from A. NF-κB signaling is comparable in WT and NIK^−/− Mφ cultures in response to RANKL.

Figure 5.

NF-κB does not accumulate in the nucleus of NIK^−/− preOCs. (A) PreOCs, generated by treating Mφs with RANKL for 48 h followed by a 3 h starvation, were stimulated with RANKL for the indicated times, and cytoplasmic (top) and nuclear (bottom) fractions were isolated and analyzed by immunoblot for IκBα and RelA, respectively. Although IκBα gets degraded in both WT and NIK^−/− cultures, RelA does not accumulate in the nucleus in the absence of NIK. (B) Nuclear extracts prepared as in A were analyzed for NF-κB DNA binding activity by EMSA. Whereas WT preOC cultures have detectable NF-κB activity at baseline and further induction by RANKL, NIK^−/− preOC cultures show little binding activity with or without RANKL. (C) RelA-specific NF-κB DNA binding activity was assessed by ELISA (EZ Detect kit; Pierce Chemical Co.) using the same preOC nuclear extracts as in A. In this assay, NF-κB–specific oligos are bound to a plate. Nuclear extract is then applied followed by anti-p65 (RelA) antibody and an HRP-conjugated secondary. The assay is developed with chemiluminescent reagents and read on a luminometer/plate reader. Each sample was run in triplicate. Unmutated (UnMu) and mutant (Mu) soluble oligos were added to the WT 30-min sample to confirm specificity. The results of this assay mirror the EMSA in B and confirm lower baseline NF-κB activity and lack of RANKL-mediated induction in NIK^−/− preOCs. Binding activity is abolished by addition of the unmutated oligo and unaffected by the mutant oligo (last two bars).

Figure 6.

RelB and p52 fail to accumulate in the nucleus in NIK^−/− preOCs. Immunoblot of nuclear extracts from Mφ and preOC cultures treated acutely with RANKL shows accumulation of p52 only in WT preOCs (top). Levels of nuclear RelB are highest in WT preOCs, with minimal acute induction by RANKL. WT Mφs show low levels of nuclear RelB, whereas NIK^−/− Mφs and preOCs lack this NF-κB subunit. A background band (bkgd) in RelB immunoblots shows equivalent signal in all lanes.

Figure 7.

Association of RelA and p100 is increased in NIK^−/− preOCs. PreOCs (generated as in Fig. 5) were starved for 3 h and restimulated with RANKL for the indicated times before lysis. Total lysates were immunoprecipitated with anti-RelA antibody and analyzed by immunoblot for p100, RelA, and IκBα. In both WT and NIK^−/−cultures, association of RelA and IκBα were reduced upon RANKL treatment. In WT cells, RelA associates with p100 only transiently. In contrast, RelA is bound to p100 in NIK^−/− cultures before addition of RANKL, and this association is increased with RANKL treatment as IκBα is degraded.

Figure 8.

Expression of a noncleavable form of p100 in WT Mφs inhibits NF-κB signaling and osteoclastogenesis. (A) WT Mφs were transduced with retrovirus bearing empty pMX or pMX-p100ΔGRR, a noncleavable form of p100, overnight and selected in puromycin for 3 d before lysis. Immunoblots for p100/p52 demonstrate expression of the retroviral driven p100ΔGRR (lane 2) at levels three to fourfold above endogenous p100 (lane 1), similar to that seen in RANKL-treated NIK^−/− Mφs. (B) NF-κB activity was assessed by EMSA on Mφs transduced with pMX (control) or pMX-p100ΔGRR as above. Lanes 1–2, cells transduced with pMX, unstimulated (lane 1), or treated with RANKL for 15 min (lane2); lanes 3–4, cells transduced with pMX-p100ΔGRR, unstimulated (lane 3), or treated with RANKL for 15 min (lane 4); lane 5, RANKL-stimulated vector control, with unlabeled probe at 50-fold excess; and lane 6, RANKL-stimulated vector control supershifted with polyclonal anti-RelA antibody. The supershifted band is indicated (ss). Expression of p100ΔGRR in Mφs dramatically decreases NF-κB signaling by RelA in response to RANKL. (C) After the 3 d puromycin selection, retrovirally transduced Mφs were cultured in RANKL for 5 d and then stained for TRAP. Expression of p100ΔGRR inhibits osteoclastogenesis in these WT cultures. Bar, 300 μm.

Figure 9.

RANKL-mediated osteoclastogenesis in vitro is enhanced in the absence of p100. Mφs from WT (A) and nfkb2^−/− (B) mice were cultured with increasing doses of GST-RANKL for 3, 4, or 5 d and then fixed and stained for TRAP. The number of mature OCs was counted for one 4× field per well, 4 wells per condition. At high doses of RANKL (50–150 ng/ml), WT OCs begin to form on day 4 and are confluent on day 5. In contrast, nfkb2^−/− OCs begin to form on day 3 with as little as 25 ng/ml RANKL and are confluent on day 4. The peak numbers of OCs are slightly lower for nfkb2^−/− cultures than for WT cultures because the nfkb2^−/− OCs are larger. This is also the case for the nfkb2^−/− cultures at 25 ng/ml RANKL on day 5 in which the OCs are smaller than at the higher doses. Thus, nfkb2^−/− cultures, which lack both p100 and p52, show accelerated osteoclastogenesis at lower doses of RANKL compared with WT controls.

Figure 10.

Model for p100-mediated blockade of NF-κB signaling in NIK-deficient preOCs. See Discussion.