CHMP5 controls bone turnover rates by dampening NF-κB activity in osteoclasts

Abstract

Physiological bone remodeling requires that bone formation by osteoblasts be tightly coupled to bone resorption by osteoclasts. However, relatively little is understood about how this coupling is regulated. Here, we demonstrate that modulation of NF-κB signaling in osteoclasts via a novel activity of charged multivesicular body protein 5 (CHMP5) is a key determinant of systemic rates of bone turnover. A conditional deletion of CHMP5 in osteoclasts leads to increased bone resorption by osteoclasts coupled with exuberant bone formation by osteoblasts, resembling an early onset, polyostotic form of human Paget's disease of bone (PDB). These phenotypes are reversed by haploinsufficiency for Rank, as well as by antiresorptive treatments, including alendronate, zolendronate, and OPG-Fc. Accordingly, CHMP5-deficient osteoclasts display increased RANKL-induced NF-κB activation and osteoclast differentiation. Biochemical analysis demonstrated that CHMP5 cooperates with the PDB genetic risk factor valosin-containing protein (VCP/p97) to stabilize the inhibitor of NF-κBα (IκBα), down-regulating ubiquitination of IκBα via the deubiquitinating enzyme USP15. Thus, CHMP5 tunes NF-κB signaling downstream of RANK in osteoclasts to dampen osteoclast differentiation, osteoblast coupling and bone turnover rates, and disruption of CHMP5 activity results in a PDB-like skeletal disorder.

Figure 1.

Chmp5 deletion in osteoclasts results in enhanced osteoclast differentiation and activity. (A) Expression of CHMP5 in long bones. Immunohistochemistry for CHMP5 in the femurs of 4-wk-old male WT mice. TB, trabecular bone. Arrows indicate osteoclasts. (B) Analysis of CHMP5 expression during in vitro osteoclast differentiation of primary BMMs by immunoblotting. (C) Radiography of the hindlimbs (left) and skulls (right) of 6-wk-old male Chmp5^fl/fl and Chmp5^Ctsk mice. The images represent three mice per group. (D, F, and G) μCT analysis of the femurs, the tibias, and the vertebrae of 12-wk-old male Chmp5^fl/fl and Chmp5^Ctsk mice. Shown are representative sagittal sections of femurs (D, top), tibias (D, middle), L4 vertebrae (D, bottom) with quantification of maximal diameter of distal femur and total volume of middle femur (M.TV; F), 3D reconstructions of femurs (G, left), and cross section of L4 vertebrae (G, right). (E) Hematoxylin and eosin staining (top, low power; middle, high power) and TRAP staining (bottom, high power) of 12-wk-old male Chmp5^fl/fl and Chmp5^Ctsk femurs. (H and I) Von Kossa staining (H), TRAP staining (I, top), and hematoxylin and eosin staining (I, bottom) of 12-wk-old male Chmp5^fl/fl and Chmp5^Ctsk tibias (H) and Pagetoid lesions (I). (J) Histomorphometric analysis of 12-wk-old male Chmp5^fl/fl tibias and Chmp5^Ctsk Pagetoid lesions. n > 5 mice per group in D, F, G, and J. All error bars indicate SEM by the Mann-Whitney test. **, P < 0.01; ***, P < 0.001. Bars: (A) 25 μm; (E) 100 μm; (H) 300 μm; (I) 50 μm.

Figure 2.

Chmp5 deletion leads to an increase in osteoclast differentiation and RANKL responsiveness. (A–C) Chmp5^fl/fl and Chmp5^Ctsk BMMs were cultured with 40 ng/ml of M-CSF and 20 ng/ml of RANKL for the indicated time periods, and cell lysates were immunoblotted with the indicated antibodies (A). Alternatively, BMMs were cultured with different concentrations of RANKL and multinucleated cells were stained with TRAP (B), and then TRAP activity was measured by colorimetric analysis (C) after 6 d of culture. ***, P < 0.005 by a Bonferroni-corrected two-tailed Student’s t test. (D) Chmp5^fl/fl and Chmp5^Ctsk BMMs were cultured with 40 ng/ml of M-CSF and 10 ng/ml of RANKL for the indicated time periods and TRAP activity was measured by colorimetric analysis. ***, P < 0.005 by a Bonferroni-corrected two-tailed Student’s t test. (E) Bone resorptive activity of Chmp5^fl/fl and Chmp5^Ctsk osteoclasts. Shown are representative images of in vitro bone resorption assays (right) with quantification of pit number, resorption area, and medium CTX (left, nM). *, P < 0.05; **, P < 0.01 by a two-tailed Student’s t test. Red arrows indicate bone resorption pits. n = 5 fields/slide. (F) Photomicrographs of TRAP-positive osteoclasts in Chmp5^fl/fl trabecular bone and Pagetoid lesions in Chmp5^Ctsk mice (CKO; left). Additionally, the number of nuclei in TRAP-positive osteoclasts was counted (n = 30 cells/slide). **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. (G) Elevated levels of Src transcripts in Chmp5^Ctsk osteoclasts. RT-PCR analysis was performed with total RNAs isolated from cultured Chmp5^fl/fl and Chmp5^Ctsk osteoclasts after 4 d of culture. ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. (H and I) After 4 d of culture, Chmp5^fl/fl and Chmp5^Ctsk osteoclasts were lysed and immunoblotted with the indicated antibodies (H). Alternatively, GTP-bound Rac1 and total Rac 1 were determined by immunoblotting with anti-Rac 1 antibody (I). (J) Enlarged actin rings in Chmp5^Ctsk osteoclasts. Chmp5^fl/fl and Chmp5^Ctsk osteoclasts were immunostained with FITC-phalloidin. n = 3 independent experiments in A–D and F–J. All error bars indicate SEM. Bars: (B) 60 µm; (E) 100 µm; (F) 25 µm; (J) 100 µm.

Figure 3.

Elevated levels of genes involved in osteoclast differentiation and clastokines in Chmp5^Ctsk osteoclasts. (A and D) Enrichment plots (left) and signature genes (right) in Chmp5^fl/fl and Chmp5^Ctsk osteoclast culture (n = 3 mice/group). GSEA analysis displays enrichment of genes involved with osteoclast differentiation and fusion (A) and Ephrin B reverse signaling (D). (B) Expression of the genes involved with osteoclast differentiation and fusion is highly increased in Chmp5^Ctsk osteoclast culture. RT-PCR analysis was performed with total RNAs isolated from cultured Chmp5^fl/fl and Chmp5^Ctsk osteoclasts (n = 3 mice/group). (C and E) Expression of IL-6 (C) and known clastokines (E) in cultured Chmp5^fl/fl and Chmp5^Ctsk osteoclasts was measured by RT-PCR analysis (n = 3 mice/group). Data in B, C, and E represent n = 3 independent experiments. All error bars indicate SEM. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by Bonferroni-corrected two-tailed Student’s t tests.

Figure 4.

Chmp5^Ctsk osteoclasts promote osteoblastogenesis and bone remodeling. (A and B) Primary calvarial osteoblasts were co-cultured with Chmp5^fl/fl (WT-OC) and Chmp5^Ctsk (CKO-OC) BMMs in basal (BM) or osteoblast differentiation medium (OBM) in the presence of M-CSF and RANKL, and then stained with alizarin red (A); alizarin red was then measured by colorimetric analysis for mineralization activity (B). **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. (C) Primary calvarial osteoblasts were cultured under OBM in the presence of conditioned medium (CM) collected from Chmp5^fl/fl (WT) and Chmp5^Ctsk (CKO) osteoclasts, and alkaline phosphatase activity (ALP) was measured to assess osteoblast differentiation. **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. (D and E) Various concentrations of mouse IgG-Fc or EphB4-Fc fusion proteins were added to the co-culture of osteoblasts and Chmp5^fl/fl or Chmp5^Ctsk BMMs in osteoblast differentiation medium containing M-CSF and RANKL. Mineralization activity was assessed by alizarin red staining. *, P < 0.05 by a Bonferroni-corrected two-tailed Student’s t test. (F–H) Total RNA was isolated from 6-wk-old male Chmp5^fl/fl and Chmp5^Ctsk tibias (n = 6 mice/group) for RT-PCR analysis. Data in A–C and E–H represent n = 3 independent experiments. All error bars indicate SEM. **, P < 0.01; ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. Bar: (A) 5 mm.

Figure 5.

Chmp5^Ctsk mice display enhanced osteoblast activity and angiogenesis in Pagetoid lesions. (A) μCT analysis of 12-wk-old male Chmp5^fl/fl and Chmp5^Ctsk tibias. Shown are representative 3D reconstructions of tibias (left) and images showing the cross sections corresponding to the red lines (right). Red box indicates the area analyzed for histomorphometry. (B) Histomorphometric analysis of 12-wk-old male Chmp5^fl/fl tibias and Chmp5^Ctsk periosteal Pagetoid lesions. n > 5 mice/group. **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. (C, D, and F) In situ hybridization for Col1 and Ocn (C) and immunohistochemistry for VEGF and PCAM (D) of Pagetoid lesions in Chmp5^Ctsk femurs. Number of capillaries in cross section of cortical bones (n = 10 fields/slide from 5 randomized mice/group). (F) Cortical bones of 12-wk-old male Chmp5^fl/fl mice were used as WT control. **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. (E) Total RNA was isolated from 6-wk-old male Chmp5^fl/fl and Chmp5^Ctsk tibias, and Vegf-A expression was measured by RT-PCR (n = 6 mice/group). *, P < 0.05 by a Bonferroni-corrected two-tailed Student’s t test. (G) Serum levels of CTX, P1NP, and ALP in 12-wk-old male Chmp5^fl/fl and Chmp5^Ctsk mice (n = 7 mice/group). **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. Bars: (C and D) 50 µm.

Figure 6.

Rank haploinsufficiency reverses PDB-like bone phenotypes in Chmp5^Ctsk mice. (A–C) μCT analysis of the femur and tibia of 12-wk-old male Rank^+/+;Chmp5^fl/fl (WT), Rank^+/Δ;Chmp5^fl/fl (Rank het), Rank^+/+;Chmp5^Ctsk (CKO), and Rank^+/Δ;Chmp5^Ctsk (CKO;Rank het) mice. Shown are representative 3D reconstructions of femurs and tibias (A) with quantification of bone volume/total volume (BV/TV), trabecular separation (Tb.Sp), trabecular number (Tb.N), midshaft volume (M. TV), and maximal femoral diameter. (B and C; n = 7 mice/group). Red arrows indicate maximal femoral diameter. *, P < 0.05; **, P < 0.01; ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. (D and E) Femurs of the indicated mice were stained for histological analysis, including hematoxylin and eosin and TRAP staining, and the images are compilation/mosaics (E). TRAP-positive osteoclasts were counted in trabecular bones of WT and CKO;Rank het mice and Pagetoid lesions in CKO mice (n = 7 mice/group; D). ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. (F) Serum levels of CTX, P1NP, and ALP in 12-wk-old male WT, Rank het, CKO, and CKO;Rank het mice (n = 7 mice/group). *, P < 0.05; **, P < 0.01; ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. All error bars indicate SEM. Bars: (E, top and middle) 100 µm; (E, bottom) 25 µm.

Figure 7.

Antiresorptive treatments reverse PDB-like bone phenotypes in Chmp5^Ctsk mice. (A–C) μCT analysis of the femurs and tibias in 6-wk-old male Chmp5^fl/fl and Chmp5^Ctsk mice. Mice were randomized to treatment with PBS, alendronate (Alen), zolendronate (Zolen), or OPG-Fc (OPG) weekly from 2–5-wk-old via IP injection. Shown are representative 3D reconstructions of femurs and tibias (A) with quantification of bone volume/total volume (BV/TV), trabecular separation (Tb.Sp), midshaft volume (M. TV), and maximal femoral diameter (B and C; n = 5 mice/group). Red arrows indicate maximal diameter. **, P < 0.01; ***, P < 0.001 by a Bonferroni-corrected two-tailed Student’s t test. (D and E) Femurs of the indicated mice were stained for histological analysis, including hematoxylin and eosin and TRAP staining. RAP-positive osteoclasts were counted in trabecular bones of Chmp5^fl/fl and Chmp5^Ctsk mice treated with PBS, alendronate, or OPG-Fc (n = 5 mice/group; E). *, P < 0.05 by a Bonferroni-corrected two-tailed Student’s t test. (F) Serum levels of CTX, P1NP, and ALP in 6-wk-old male Chmp5^fl/fl and Chmp5^Ctsk mice treated with PBS, Alen, Zolen, or OPG (n = 5 mice/group). All error bars indicate SEM. **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. Bars: (D, top and middle) 100 µm; (D, bottom) 25 µm.

Figure 8.

CHMP5 down-regulates RANK-mediated NF-κB activation in osteoclasts. (A, B, and H) Fractionation of the CHMP5 protein complex in osteoclasts. RAW264.7 cells were lysed with 1% CHAPS lysis buffer and fractionated by size exclusion chromatography. The fractions were immunoblotted with the indicated antibodies (A). The fraction in the red box was immunoprecipitated with anti-Chmp5 antibody or IgG control antibody along with protein G–conjugated dynabeads, and immunoblotted with the indicated antibodies (B and H). The images in A, B, and H represent two independent experiments. (C) HEK293 cells were transfected with HA-Chmp5 along with vector or Flag tagged-IκBα or -RelA. After 48-h transfection, the cells were lysed, immunoprecipitated with anti-Flag antibody–conjugated agarose, and immunoblotted with the indicated antibodies. (D) RAW264.7 cells were transfected with Chmp5 in a dose-dependent manner along with PBII-luc and Renilla and stimulated with 50 ng/ml of RANKL (top) or RAW264.7 cells expressing control, or Chmp5 shRNA were treated with different concentrations of RANKL (bottom). Relative luciferase activity was normalized to Renilla. *, P < 0.05; ***, P < 0.0001 by a Bonferroni-corrected two-tailed Student’s t test. (E-G) Chmp5^fl/fl and Chmp5^Ctsk BMMs were cultured in the presence of M-CSF and RANKL, and after 3 d total cell lysates were used for an NF-κB DNA-binding assay (E) or immunoblotted with the indicated antibodies (F). Alternatively, BMMs were infected by lentiviruses expressing vector or the super repressor, HA-IκBα-RR, and osteoclast differentiation was measured by TRAP activity (G). **, P < 0.01; ***, P < 0.0001 by a Bonferroni-corrected two-tailed Student’s t test.

Figure 9.

CHMP5 and USP15 suppress RANKL-induced NF-κB activation and osteoclast differentiation via stabilization of IκBα. (A) HEK293 cells were transfected with Chmp5 or Usp15 along with Flag-IκBα, His-ubiquitin, and Myc-β-Trcp, treated with 10 µM MG132, and subjected to IκBα ubiquitination (top) and protein expression assays (bottom). (B and C) HEK293 cells were transfected with control vector, Chmp5, or Usp15 along with Flag-IκBα, and IκBα stability was determined by pulse–chase labeling with [³⁵S]-methionine followed by autoradiography and quantified by ImageJ (C). (D and E) RAW264.7 cells expressing control vector, Chmp5, or Usp15 shRNA were stimulated with 25 ng/ml of RANKL at different time points, and immunoblotted with the indicated antibodies (D). IκBα degradation was quantified using ImageJ and normalized to Hsp90 expression (E). (F and G) RAW264.7 cells expressing control vector or Usp15 shRNA were transfected with Chmp5 along with PBII-luc and Renilla (F). Alternatively, RAW264.7 cells were transfected with Usp15 in a dose-dependent manner along with PBII-luc and Renilla (G). Luciferase activity was measured and normalized to Renilla. *, P < 0.05; **, P < 0.01; ***, P < 0.0001 by a Bonferroni-corrected two-tailed Student’s t test. (H) RAW264.7 cells expressing control vector or Usp15 shRNA were cultured in the presence of M-CSF and different concentrations of RANKL. TRAP staining was performed after 6 d. (I) WT BMMs were treated with vehicle or NF-κB inhibitor for 1 h before RANKL stimulation. After 48 h of stimulation with RANKL, Chmp5 and Usp15 expression was analyzed by RT-PCR. ***, P < 0.0001 by a Bonferroni-corrected two-tailed Student’s t test. Data in A–I represent n = 3 independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Bar: (H) 60 µm.

Figure 10.

CHMP5 mediates deubiquitination of the VCP/p97 complex via USP15. (A) GST or GST-Chmp5 was incubated with the purified proteins, immunoprecipitated with glutathione-agarose, and immunoblotted with the indicated antibodies (top). Alternatively, GST or GST-VCP was incubated with purified CHMP5 protein (bottom). Input indicates loading controls for USP15, VCP, and CHMP5. (B) RAW264.7 cells expressing a control vector or Chmp5 shRNA were cultured with 5 µg of RANKL for 2 d before treatment with 10 µM MG132. Cell lysates were immunoprecipitated with anti-Vcp antibody and protein G–conjugated dynabeads and immunoblotted with the indicated antibodies. (C) HEK293 cells were transfected with Myc-VCP and HA-ubiquitin in the absence or presence of CHMP5 or USP15. 24 h later, cells were treated with 10 µM MG132 and subjected to ubiquitination of Myc-VCP. (D and E) HEK293 cells were transfected with HA-CHMP5 (D) or USP15 (E) and His-ubiquitin along with vector, Myc-VCP (WT), or Myc-VCP mutants (R155H, A232E). 24 h later, cells were treated with 10 µM MG132, immunoprecipitated with anti-Myc antibody–conjugated agarose, and immunoblotted with the indicated antibodies. (F) HEK293 cells were transfected with Myc-VCP (WT) or Myc-VCP mutants (R155H, A232E) and HA-ubiquitin in the absence or presence of CHMP5 or USP15. 24 h later, cells were treated with 10 µM MG132 and subjected to ubiquitination of Myc-VCP. (G) HEK293 cells were transfected with vector, Myc-VCP (WT), or Myc-VCP mutants (R155H, A232E) along with Flag-RANK, PBII-luc and Renilla in the absence or presence of USP15, and 24 h later, luciferase activity was measured and normalized to Renilla. **, P < 0.01 by a Bonferroni-corrected two-tailed Student’s t test. Data in A–G represent n = 2 independent experiments. **, P < 0.01.