CHIP regulates bone mass by targeting multiple TRAF family members in bone marrow stromal cells

Carboxyl terminus of Hsp70-interacting protein (CHIP or STUB1) is an E3 ligase and regulates the stability of several proteins which are involved in different cellular functions. Our previous studies demonstrated that Chip deficient mice display bone loss phenotype due to increased osteoclast formation through enhancing TRAF6 activity in osteoclasts. In this study we provide novel evidence about the function of CHIP. We found that osteoblast differentiation and bone formation were also decreased in Chip KO mice. In bone marrow stromal (BMS) cells derived from Chip−/− mice, expression of a panel of osteoblast marker genes was significantly decreased. ALP activity and mineralized bone matrix formation were also reduced in Chip-deficient BMS cells. We also found that in addition to the regulation of TRAF6, CHIP also inhibits TNFα-induced NF-κB signaling through promoting TRAF2 and TRAF5 degradation. Specific deletion of Chip in BMS cells downregulated expression of osteoblast marker genes which could be reversed by the addition of NF-κB inhibitor. These results demonstrate that the osteopenic phenotype observed in Chip−/− mice was due to the combination of increased osteoclast formation and decreased osteoblast differentiation. Taken together, our findings indicate a significant role of CHIP in bone remodeling.


INTRODUCTION
Bone remodeling is a dynamic process, involving a balance between the activity of bone-forming osteoblasts, derived from mesenchymal progenitor cells, and the activity of bone-resorbing osteoclasts, derived from hematopoietic stem cells. 1 A disruption of this balance leads to many bone and joint diseases, especially metabolic bone diseases. Although the coupling of bone resorption and bone formation has long been recognized, the mechanism mediating this fundamental process in skeletal homeostasis is not fully understood. The understanding of the regulatory mechanisms of osteoclast formation and osteoblast function is critical for the rational drug design for the treatment of metabolic bone disorders.
Osteoblast is a cell type derived from mesenchymal progenitor cells and is responsible for bone formation. Osteoclast and osteoblast are two major cell types involved in bone remodeling in trabecular bone. Several previous studies have demonstrated that NF-κB signaling plays critical roles in regulation of osteoclast and osteoblast activities in bone. Activation of NF-κB signaling promotes osteoclast formation [13][14][15][16] and inhibits osteoblast function. [17][18][19][20][21][22] CHIP is an E3 ligase, which has been reported to promote ubiquitination and degradation of several substrates involving in different signaling pathways. [23][24][25] In vivo findings demonstrated that Chip deficiency causes increased sensitivity to stress associated hyperthermia, 26 leading to a markedly reduced life span in mice. 27 In addition, Chip −/− mice developed severe defects in heart and neural tissues. 28,29 Moreover, Chip −/− mice exhibit defects in motor sensory, cognitive, and reproductive function. 30 Although the mechanisms of NF-κB regulation during bone remodeling have been extensively investigated, the roles of CHIP in regulation of NF-κB signaling and osteoblast function has not been demonstrated. Our previous study demonstrated that Chip −/− mice develop osteopenic phenotype due to increased osteoclast formation. 12 In the present studies, we investigated the roles of CHIP in regulation of protein stability of multiple TRAF family members, NF-κB signaling, and osteoblast function. Our findings provide new insights into the role of CHIP in regulation of osteoclast and osteoblast functions.

RESULTS
Bone mass is decreased in Chip −/− mice To determine the contribution of CHIP in skeletal development, we performed whole skeletal Alizarin red/Alcian blue staining using postnatal 3-day-old wild-type (WT) and ChipKO mice. The sizes of the skeleton of Chip −/− mice were much smaller than those of WT littermates and Chipheterozygous KO mice (Chip +/-) (Fig. 1a). We also analyzed the appearance of 1-month-old Chip −/− mice. The sizes of Chip −/− mice were also smaller than WT and Chip +/-KO mice (Fig. 1b). The X-ray radiographic analysis showed that the length of the spine was shorter and bone density of vertebral bone was reduced in 1-month-old Chip −/− mice compared to WT and Chip +/-KO mice (Fig. 1c). To further confirm changes in the spine of Chip −/− mice, Safranin O/Fast green staining was performed on vertebral sections. The results showed that vertebral bone volume in spine was significantly reduced in 1month-old Chip −/− mice compared to WT littermates (Fig. 1d).
These results indicate that Chip deficiency in mice results in a bone loss phenotype.

Alterations of osteoclast formation and osteoblast function in
Chip −/− mice The osteopenic phenotype observed in Chip −/− mice could be due to either increased bone resorption or reduced bone formation. To determine the effect of Chip deficiency on bone formation, bone histology and the calcein double labeling assays were performed. The results showed that the bone volume and bone formation rates were significantly reduced in Chip −/− mice ( Fig. 2a-d). In addition, we also found that osteoblast numbers were significantly reduced in Chip −/− mice (Fig. 2e). To determine the role of CHIP in bone resorption, we performed TRAP staining using histological sections of tibiae from 1-month-old WT and Chip −/− mice. The data revealed that TRAP positive osteoclast numbers were significantly increased in Chip −/− mice (Fig. 2f, g). These results suggest that the bone loss phenotype observed in Chip −/− mice could be due to the combination of increased bone resorption and decreased bone formation. To further determine changes in osteoblast differentiation, we isolated BMS cells from 1-month-old WT and Chip −/− mice, and examined Ki67 expression by real-time PCR and performed alkaline phosphatase (ALP) and Alizarin red staining assays. The results showed that Ki67 expression was significantly reduced and ALP activity and mineralized bone matrix formation were dramatically decreased in Chip-deficient BMS cells compared to BMS cells isolated from WT mice (Fig. 2h, i). These results further demonstrated that osteoblast function was reduced in Chip −/− mice. Fig. 1 Bone mass is decreased in ChipKO mice. a The skeleton of 3-day-old newborn mice was stained with Alizarin red and Alcian blue. The size of Chip −/− mice was significantly smaller than those of WT and Chip +/mice. b The picture showed the appearance of 1-month-old Chip KO mice (Chip −/− and Chip +/-) and WT littermate. The size of Chip −/− mice was smaller than that of WT and Chip +/mice. c X-ray radiographic analysis showed that bone density of the spine of 1-month-old Chip −/− mice was lower compared to that of WT and Chip +/mice (indicated by red arrowhead). CHIP regulates protein stability of multiple TRAF family members Our previous study showed that CHIP promotes TRAF6 degradation and inhibits RANKL-induced NF-κB signaling. 12 To further determine the mechanism by which CHIP regulates TRAF6 signaling, we performed a series of in vitro experiments to determine if CHIP affects TRAF6 interaction with IKK using 293 T cells. We found that CHIP directly interacts with TRAF6 (Fig. 3a). IL-1β enhanced CHIP/TRAF6 interaction (Fig. 3b). Expression of CHIP significantly inhibited TRAF6/IKK-β interaction in 293 T cells (Fig. 3c). These results suggest that CHIP may inhibit TRAF6/IKK-β interaction by inducing TRAF6 degradation.
Given that TRAF2, TRAF3, and TRAF5 belong to the same TRAF family and play critical roles in TNF-α-induced NF-κB signaling, we next examined if CHIP affects stability of TRAF2, TRAF3, and TRAF5. The results of Western blot analysis showed that the protein levels of TRAF2 were decreased when CHIP expression was increased in 293 T cells. In contrast, the mutant CHIP (H260Q), which lacks the E3 ligase ability, failed to affect steady state protein levels of TRAF2 (Fig. 4a). We further examined the half-life of TRAF2 protein by assessing changes in TRAF2 protein levels after treating cells with protein synthesis inhibitor cycloheximide (CHX). The results showed that the halflife of TRAF2 protein was greatly reduced by WT CHIP but not by the mutant CHIP (H260Q) (Fig. 4b). These results indicate that CHIP is a negative regulator of TRAF2 protein. We also examined the proteasome-dependency of CHIP-induced TRAF2 degradation and found that CHIP-mediated TRAF2 degradation could be reversed by the addition of proteasome inhibitor, MG-132 ( Fig. 4c), suggesting that CHIP-induced TRAF2 degradation is proteasome-dependent. In addition, protein levels of TRAF5 were also decreased when CHIP expression is increased (Fig. 4d). The IP result demonstrated that TRAF5 was co-precipitated by CHIP in 293 T cells (Fig. 4e). To determine if CHIP-induced TRAF5 degradation is ubiquitin/proteasome-dependent, we treated cells with proteasome inhibitor MG-132. The results showed that CHIP-induced TRAF5 degradation could be reversed by the treatment of MG-132 (Fig. 4f). In contrast, CHIP has no significant effect on TRAF3 degradation (data not shown). Taken together, CHIP regulates protein stability of multiple TRAF family members. Chip −/− mice showed a decrease in bone formation rates (d). e Results of histomorphometric analysis also showed that osteoblast numbers were also significantly reduced in Chip −/− mice. f, g TRAP staining was performed using tibial sections of 1-month-old WT and Chip −/− mice. TRAP-positive cells were indicated by blue arrowheads. The numbers of TRAP-positive osteoclasts were higher in Chip −/− mice (g). h, i Bone marrow stromal (BMS) cells were isolated from 1-month-old WT and Chip −/− mice, and were cultured with ascorbic acid and βglycerophosphate for 3 days for Ki67 expression (h) and ALP staining (i). The BMS cells were also cultured with osteoblast differentiation medium in the presence of BMP-2 (50 ng·mL -1 ) for 2 weeks and Alizarin red staining was performed (i). Ki67 expression, ALP activity, and mineralized bone matrix formation were significant decreased in Chip −/− mice. *P < 0.05, **P < 0.01, unpaired Student's t-test, n = 6.
CHIP inhibits TRAF-mediated NF-κB signaling Since CHIP regulates TRAF2 and TRAF5 protein levels and TRAF2 and TRAF5 play an important role in TNF-α-induced NF-κB signaling, we then examined whether CHIP affects TNF-αinduced I-κBα phosphorylation and p65 nuclear translocation. The results showed that overexpression of CHIP markedly inhibited the TNF-α-induced I-κBα phosphorylation (Fig. 5a). Notably, the total I-κBα protein levels were increased after CHIP transfection. This is probably because the phosphorylationinduced I-κBα degradation is decreased in CHIP-expressing cells. Next, we performed Western blot analysis using the fractions of the cytoplasm and nuclear proteins and examined changes in p65 localization. The results demonstrated that overexpression of CHIP prevented p65 nuclear translocation when the cells were treated with TNF-α for 40 min (Fig. 5b). These results suggest that CHIP inhibits TNF-α-induced I-κBα phosphorylation. To obtain additional evidence about the role of CHIP in p65 nuclear translocation, we performed immunostaining assay and found that transfection of CHIP inhibited p65 nuclear translocation (Fig. 5c).
To further determine the role of CHIP in NF-κB signaling, we performed NF-κB luciferase assay. The NF-κB reporter construct was transfected with HA-CHIP or empty vectors into BMS cells isolated from 1-month-old WT mice treated with or without TNF-α. The results showed that NF-κB reporter activity was significantly increased by the treatment of TNF-α. However, NF-κB activity was significantly inhibited by the transfection of CHIP ( Fig. 5d). To determine the role of CHIP in TRAF2, TRAF5, and TRAF6-regulated NF-κB activity, we transfected TRAF2, TRAF5 or TRAF6 expression plasmids with or without CHIP into BMS cells. The result showed that the TRAF2, TRAF5, and TRAF6-induced NF-κB reporter activity was significantly inhibited by CHIP co-expression ( Fig. 5e-g). Taken together, these results suggest that CHIP negatively regulates TRAF2, TRAF5, and TRAF6-induced NF-κB activity.

CHIP regulates osteoblast gene expression
In an effort to investigate the role of CHIP in osteoblast differentiation, we first examined changes in levels of several proteins critical for osteoblast differentiation, such as β-catenin, Runx2, and Smad3. Previous in vitro studies have shown that CHIP regulates protein stability of these proteins in different cell types. 23,25,31 Surprisingly, the protein levels of these molecules were not significantly changed in Chip-deficient BMS cells (data not shown). We next examined expression of a panel of osteoblast marker genes by the real-time PCR assay. The mRNA levels of Osterix(Osx) and alkaline phosphatase (Alp)were significantly decreased in BMS cells derived from 1-month-old Chip −/− mice (Fig. 6a, b). As a consequence, the expression of early osteoblast differentiation markers, such as type I collagen(Col1) and osteopontin(OPN), were also significantly decreased in Chipdeficient BMS cells (Fig. 6c, d). The expression levels of late stage osteoblast marker genes, such as bone sialoprotein(BSP) and osteocalcin(OC), were also significantly reduced in Chip-deficient The immunoprecipitates were prepared with the anti-Myc antibody followed by the Western blot analysis. The data showed that CHIP interacts with TRAF6, but not p65. b IL-1β enhances CHIP/TRAF6 interaction. 293 T cells were transfected with the Flag-TRAF6 and Myc-CHIP as indicated and treated with IL-1β. The cells lysates were immunoprecipitated with the anti-Myc antibody followed by the Western blot analysis using the anti-Flag antibody. c HA-IKK-β and Flag-TRAF6 were co-transfected with Myc-CHIP into 293 T cells. IP assay was performed using the anti-Flag antibody followed by the Western blot analysis using the HA antibody. The TRAF6/IKK-β interaction was attenuated by CHIP. To prevent TRAF6 degradation, proteasome inhibitor MG-132 was added to the cell culture in the TRAF6/IKK-β co-IP assay.
In this study we also examined the mRNA expression of osteoprotegerin (OPG) and Rankl in BMS cells. We found that OPG expression was slightly reduced in Chip-deficient BMS cells (Fig. 6g). In contrast, Rankl expression was significantly increased in Chip-deficient BMS cells (Fig. 6h). These results suggest that reduction in osteoblast differentiation contributes to the bone loss phenotype observed in Chip −/− mice.
To further determine the specific role of CHIP in osteoblast function, we generated Chip flox/flox mice. We isolated BMS cells from 1-month-old Chip flox/flox mice and infected these cells with Ad-CMV-Cre virus. The results showed that in vitro deletion of Chip in BMS cells significantly decreased CHIP protein levels (Fig. 7a) and inhibited expression of osteoblast marker genes, including Osx, Col1a1, OPN, BSP,and OC, in Chip-deficient BMS cells (Fig. 7b-f).

CHIP-regulated osteoblast function is NF-κB-dependent
To determine if Chip-deficiency mediated inhibition of osteoblast differentiation is NF-κB-dependent, we isolated BMS cells from 1-month-old Chip flox/flox mice and infected these cells with Ad-CMV-Cre or Ad-CMV-GFP (control virus) and treated cells with or without NF-κB inhibitor, BP-1-102. We examined changes in expression of several osteoblast marker genes in these cells and found that treatment with NF-κB inhibitor significantly reversed the inhibitory effects on the expression of osteoblast marker genes observed in Chip-deficient BMS cells (Fig. 8). BP-1-102 inhibits nuclear NF-κB retention both in vitro and in vivo, thereby attenuating NF-κB activation. 32 These results suggest that CHIP regulates osteoblast differentiation through a NF-κB-dependent mechanism.

DISCUSSION
Bone remodeling is a highly coordinated process involving bone resorption and bone formation. The dynamic process of bone remodeling was impaired in Chip −/− mice, suggesting that CHIP plays a critical role in bone remodeling. Our previous findings showed that the bone loss phenotype observed in Chip −/− mice is due to abnormally increased osteoclast formation and bone resorption. In the present studies we provided novel evidence that in addition to the enhanced osteoclast formation, Chip deficiency in mice also leads to reduced bone formation.
NF-κB and its signaling molecules play critical roles in regulation of osteoclast formation [13][14][15][16] and osteoblast function, [17][18][19][20][21][22] and TRAF family members are critical mediators in NF-κB signaling. [7][8][9][10] Our previous study showed that TRAF6 protein levels are increased in Chip-deficient osteoclasts and we have characterized the effect of CHIP on regulation of TRAF6 protein stability. 12 In the present studies, we further investigated the role of CHIP in TRAF6/ IKKβ interaction and TRAF2 and TRAF5 protein stability and functions. We found that 1) CHIP interferes with TRAF6/IKKβ interaction; 2) CHIP induces TRAF2 and TRAF5 protein degradation; 3) CHIP inhibits TRAF2 and TRAF5-induced NF-κB signaling. These results suggest that CHIP functions as a critical regulator in NF-κB signaling through regulation of multiple TRAF family members.
Osteoblast/osteoclast cross talk regulates bone formation and bone resorption with remarkable precision. For example, the key regulator for osteoclast bone resorption is RANKL, which expressed by activated osteoblasts. 33 Its action on the RANK receptor in osteoclasts is regulated by OPG, a decoy receptor, which is also expressed in osteoblasts. 34 In Chip-deficient BMS cells, Rankl expression was significantly increased (Fig. 6j). These results demonstrate that in addition to the direct regulation of CHIP inhibited TNFα-induced I-κBα phosphorylation. b After treatment with TNF-α in Hela cells stably transfected with HA-CHIP or empty vector, p65 protein levels in the cytoplasm (Cyto) and nucleus were detected. CHIP inhibited p65 nuclear translocation. c 293 T cells were transfected with or without Myc-CHIP. After treatment with IL-1β for 30 min, the cells were stained with the anti-Myc antibody (green) or anti-p65 antibody (red). The nuclear was stained by DAPI (blue). CHIP inhibited p65 nuclear translocation. The white dotted circles in b3 and d3 showed the nuclear region of cells with or without Myc-CHIP transfection. Scale bar: 10 μm. d Luciferase assay was performed in bone marrow stromal (BMS) cells transfected with indicated plasmids and treated with TNF-α for 8 h. CHIP inhibited TNF-α-induced NF-κB signaling. *P < 0.05, **P < 0.01, one-way ANOVA followed by Tukey Post-Hoc test, n = 4, compared between control and TNF-α groups; ## P < 0.01, oneway ANOVA followed by Tukey Post-Hoc test, n = 4, compared between empty vector and CHIP transfection groups. e-g NF-κB reporter construct was transfected into BMS cells with TRAF2, TRAF5 or TRAF6. CHIP inhibited TRAF2, TRAF5 or TRAF6-induced NF-κB signaling. **P < 0.01, one-way ANOVA followed by Tukey Post-Hoc test, n = 4, compared between empty vector and TRAF2, TRAF5 or TRAF6 transfected groups; # P < 0.05, ## P < 0.01, one-way ANOVA followed by Tukey Post-Hoc test, n = 4, compared between TRAF2, TRAF5 or TRAF6 transfection alone or with CHIP co-transfection group.
bone resorption via promoting TRAF degradation, CHIP also indirectly enhances osteoclast formation via controlling osteoblast/osteoclast cross talk.
Although previous studies demonstrated that CHIP regulates protein stability of β-catenin, Runx2, and Smad3 in vitro, 23,25,31 in the present studies, we did not detect significant changes in the steady state protein levels of these molecules, except slight reduction of β-catenin protein levels in Chip-deficient BMS cells (data not shown). It is known that activation of NF-κB signaling inhibits osteoblast function; [17][18][19][20][21][22] however, the detail mechanism remains undefined. Recent studies showed that transcriptional enhancers of cell type specific genes could serve as signal integration hubs binding to signal-dependent transcription factors, such as NF-κB, and lineage-determining transcription factors, such as Runx2, to control the gene expression program and determine cell differentiation fate. 35,36 This may explain why NF-κB regulates osteoblast function without altering Runx2 protein levels.
It is well known that NF-κB signaling plays an important role in inflammation and in rheumatoid arthritis (RA). 37,38 In addition to the inflammation, activation of NF-κB signaling also leads to bone loss by upregulation of osteoclast formation and inhibition of osteoblast function. In the present studies, we demonstrated that CHIP inhibits NF-κB signaling by inducing the degradation of multiple TRAF family members, including TRAFs 2, 5, and 6. In Chip −/− mice osteoclast formation was increased 12 and osteoblast differentiation was inhibited, demonstrated by the present studies. These findings suggest that CHIP may play an important role in the development of RA disease. Expression or function of CHIP may be inhibited during RA development, leading to  activation of NF-κB signaling and bone loss phenotype in joint tissues due to increased osteoclast formation and reduced osteoblast function. To further determine the specific role of CHIP in joint tissues, we have recently generated Chip flox/flox mice. Using tissue-specific knockout approach, we will further dissect the specific effects of CHIP in specific cell populations in bone and cartilage tissues, including mesenchymal stem cells (MSCs), osteoblast and osteoclast lineage cells, and in synovial cells.
In summary, in this study we demonstrate that bone loss phenotype observed in Chip −/− mice is due to the combination of increased bone resorption and reduced bone formation. Since abnormal increase in osteoclast formation and reduced osteoblast function may lead to bone and joint diseases, our findings suggest that CHIP may serve as a novel target for the drug development to treat bone loss associated diseases.

Mice
Chip knockout (KO) mice were obtained from NIH. The first three coding exons of the Chip gene were targeted by homologous recombination. Both wild-type (WT) and Chip −/− mice were maintained in a C57BL/6 and 129SvEv background. Chip flox/flox mice were generated in Nanjing Biomedical Research Institute of Nanjing University, Nanjing, China. In these mice the Chip gene was floxed at the flanking sites of exon 1 and exon 3.
Luciferase assays BMS cells were transiently transfected with pGL3/NF-κB-luc reporter and expression plasmids as indicated. The luciferase assay was performed to examine the effect of CHIP in NF-κB signaling.

Histology
Tibiae and vertebrae were harvested and fixed in 10% NB-formalin for 3 days and decalcified for 14 days in 14% EDTA and then paraffin-embedded. Three micrometers sections were cut and Alcian blue/H&E Orange G and Safranin O/Fast green staining and TRAP staining were performed.

Fig. 8
Inhibition of NF-κB signaling reverses the inhibitory effects of osteoblast differentiation observed in Chip-deficient BMS cells. Bone marrow stromal (BMS) cells were isolated from 1-month-old Chip flox/flox mice and infected with Ad-CMV-Cre or Ad-CMV-GFP (control virus) and cultured with osteoblast differentiation medium for 3 days in the presence or absence of NF-κB inhibitor, BP-1-102. Expression of osteoblast marker genes, including Osx, Col1a1, BSP,and OC was inhibited in Chip-deficient BMS cells. Addition of NF-κB inhibitor (BP-1-102) reversed inhibitory effects of expression of osteoblast marker genes observed in Chip-deficient BMS cells. *P < 0.05, **P < 0.01, one-way ANOVA followed by Tukey Post-Hoc test, n = 4, compared between WT and Chip-deficient BMS cells. # P < 0.05, one-way ANOVA followed by Tukey Post-Hoc test, n = 4, compared between Chip-deficient BMS cells treated with or without BP-1-102.
Cell culture 293 T cells and Hela cells were cultured in DMEM supplemented with 10% fetal calf serum (FCS) at 37°C under 5% CO 2 . In CHIP/TRAF5 co-IP assay, proteasome inhibitor MG-132 was added (10 μM). BMS cells were isolated from 1-month-old Chip flox/flox mice and cultured with αMEM and 10% FCS. BMS cells were then infected with Ad-CMV-Cre or Ad-CMV-GFP (control virus) in the presence of osteoblast differentiation medium for 3 days. 39 Expression of osteoblast marker genes was examined by real-time PCR assay. For mineralized bone matrix formation assay, BMS cells were cultured with osteoblast differentiation medium with BMP-2 (50 ng·mL -1 ) for 2 weeks. Alizerin red staining was then performed. NF-κB inhibitor, BP-1-102, was obtained from Calbiochem (EMD Millipore, Billerica, MA).

Immunoprecipitation and Western blot analysis
The immunoprecipitation (IP) and Western blot assays were performed as previously described. 12,23 Alizarin red staining BMS cells were isolated from femur and tibia of 1-month-old WT and Chip −/− mice (or Chip flox/flox mice infected with Ad-CMV-Cre or Ad-CMV-GFP), and were cultured with αMEM supplemented with 10% FCS and treated with ascorbic acid (50 μg·mL -1 ) and βglycerophosphate (10 mmol·L -1 ) and BMP-2 (50 ng·mL -1 ) for 2 weeks. At the end of the cell culture, BMS cells were fixed in 10% formalin and Alizarin red staining was performed. 40,41 Calcein labeling Calcein (20 mg·kg -1 , i.p.) labeling was performed 1 and 4 days before 1-month-old WT and Chip −/− mice were sacrificed. At the end of the experiments, tibiae were removed, fixed in 75% ethanol, and embedded in plastic. Transverse sections were cut at 5-μm thickness and unstained sections were viewed under fluorescent microscope. 40,42