Icariin Protects against Glucocorticoid-Induced Osteonecrosis of the Femoral Head in Rats

Glucocorticoid (GC)-related osteonecrosis of the femoral head (ONFH) is a common complication following administration of steroids in many diseases, including systemic lupus erythematosus, rheumatoid arthritis, multiple myeloma, acute lymphoblastic leukemia, organ transplantation and severe acute respiratory syndrome [1-6]. There are many mechanisms of GC-related ONFH such as disruption of the balance between osteogenic and adipogenic differentiation [7], fat embolization [8], circulatory impairment [9], cell apoptosis and dysfunction [10], intramedullary pressure changes [11], coagulation disorders [12], and modified artery constriction [13]. Among them, disruption of the balance between osteogenic and adipogenic differentiation of bone marrow mesenchymal stem cells (BMSCs) is a principal mechanism involved in the onset and progression of this disease. When adipocytes increase in number and enlarge, the intraosseous pressure may rise and blood flow rate may decrease, resulting in reduced bone remodeling, causing ONFH. Hence, it is essential for the treatment of GC-related ONFH to restore the balance between osteogenic and adipogenic differentiation of BMSCs, by inhibiting adipogenesis and promoting osteogenesis and thus increasing bone remodeling [14].

Previous epidemiological data showed a remarkable difference in the prevalence of osteonecrosis (mainly knee and hip) in patients recovering from SARS between southern China (5–6%) [15, 16] and northern China (32.7%) [17]. The lower prevalence of osteonecrosis in southern China was suggested to be associated with the potential use of conventional herbal intervention to prevent GC-associated osteonecrosis. The antiviral Herba Epimedii with immunomodulation is one of the medicinal herbs which has been widely used as a “Bone Strengthener and Kidney Tonic” in China for thousands of years [18, 19]. Icariin (ICA) is extracted from Herba Epimedii and is considered to be the major active ingredient. It has been reported that ICA can improve bone mineral density (BMD) and bone strength in ovariectomized rats [20]. Moreover, Li et al. reported that ICA prevented ovariectomy-induced bone loss and lowered marrow adipogenesis [21]. Meanwhile our previous study demonstrated that ICA might have a beneficial effect on the BMSCs of patients with steroid-associated osteonecrosis by demethylation of the adenosine triphosphate-binding cassette (ABC) B1-promoter [14]. However, to date, few investigations have focused on the relevance of ICA to the balance between osteogenic and adipogenic differentiation in the treatment of GC-related ONFH.

The aim of this study was to investigate the effect of ICA on potential treatment to prevent progression of ONFH. In this study, we hypothesized that ICA could prevent the development of GC-induced ONFH by regulating the balance between osteogenic and adipogenic differentiation of BMSCs. Furthermore, we also evaluated the underlying mechanisms of ICA associated with the potential enhancement of osteogenesis and anti-adipogenesis involving BMSCs using methylprednisolone (MPS) to suppress osteogenic activity both in vitro and in vivo.

Bone marrow mesenchymal stem cells (BMSCs) were obtained from the bilateral femurs and tibias of 3-week-old female Sprague-Dawley (SD) rats (Laboratory Animal Center of Huazhong University of Science and Technology, Wuhan, China) according to the method described by Kodama et al. [22]. The cells were cultured in α minimum essential medium (α-MEM; Gibco BRL, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA) and 1% penicillin G–streptomycin at 37°C in a humidified atmosphere of 5% CO₂. BMSCs were passaged when adherent cells reached a density of 80–90%. BMSCs at passages three to six were used in all experiments. All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Tongji Medical College, Huazhong University of Science and Technology.

ICA (purity > 98%) was purchased from Abcam, Cambridge, UK. It was dissolved in dimethyl sulfoxide (DMSO) and stored in the dark at -20°C. The final concentration of DMSO in the medium was 0.1% or less throughout the experiments. All experiments were repeated three times.

The effect of ICA on cell proliferation was evaluated using the Cell Counting Kit-8 (CCK-8; Nanjing Jiancheng Bioengineering Institute, Nanjing, China) assay according to the manufacturer’s instructions. BMSCs were plated into 96-well plates (5, 000 cells/well) and treated with ICA (10^-9 M to 10^-3 M) for 48 hours. The results were presented as absorbance read at 450 nm using a multifunctional microplate reader (ELX800, BioTeK Instruments Inc., Winooski, VT, USA).

To assay cell viability, BMSCs were plated into 96-well plates (five wells per group) at the same cell density and divided into five groups: (1) control group; (2) MPS group, which was treated with 5 × 10^-5 M MPS (methylprednisolone, Pfizer, New York, USA); (3) MPS + ICA ^-5 group; treated with 5 × 10^-5 M MPS and 10^-5 M ICA; (4) MPS + ICA ^-6 group; treated with 5 × 10^-5 M MPS and 10^-6 M ICA; (5) MPS + ICA ^-7 group; treated with 5 × 10^-5 M MPS and 10^-7 M ICA. The relative cell activity was measured by CCK-8 assay at 48 and 72 hours. The MPS concentration was chosen based on previous studies [23].

BMSC differentiation was induced after the adherent cells reached a density of 80–90%. For osteogenic induction, basic medium was supplemented with 10^-7 M dexamethasone, 10^-2 M sodium β-glycerophosphate and 50 µg/mL L-ascorbic acids. For adipogenic induction, BMSCs were cultured in adipogenic medium (culture medium supplemented with dexamethasone (10^-7 M), indomethacin (6 × 10^-6 M), insulin (5 µg/ mL), and 3-isobutyl-1- methylxanthine (5 × 10^-4 M)). The medium was changed every 3 days for 2 weeks to induce BMSC differentiation.

For CCK-8, alkaline phosphatase (ALP) activity and triglyceride (TG) content assays, BMSCs from all five groups were assayed. For alizarin red S staining, oil red O staining, immunofluorescence staining, western blotting and RT-qPCR, BMSCs from three groups were assayed: control group, MPS group and MPS + ICA (10^-6 M) group. All experiments were repeated three times, with three replicate wells in each group.

After osteogenic induction for 14 days, cells in 12-well plates (three replicate wells per group, three plates) were washed with PBS three times and fixed with 4% paraformaldehyde for 30 min. After fixation, paraformaldehyde was removed from the wells and cells were washed with ddH₂O three times. Then, 0.04 M alizarin red S was used to stain cells for 30 min at room temperature (RT). After rinsing twice with ddH₂O, cells were visualized under a light microscope (three plates).

After adipogenic induction for 14 days (three replicate wells per group, three plates), cells were gently washed with PBS three times, and fixed with 4% paraformaldehyde for 30 min. After fixation, paraformaldehyde was removed from the wells and cells were washed with ddH₂O three times. Then, the cells were stained with oil red O staining solution for 30 min at RT. Thereafter, the samples were photographed under an inverted microscope.

BMSCs were seeded into 24-well plates (three replicate wells per group) at a density of 10, 000 cells/ well and cultured with osteogenic or adipogenic medium. ALP activity was assayed on days 3, 7 and 14 at a wavelength of 405 nm using a multifunctional microplate reader. ALP activity was quantified in cell lysates using an ALP assay kit (Nanjing Jiancheng Biological Engineering Institute) according to the manufacturer’s instructions (three plates). TG content was assayed on day 14 using a triglyceride assay kit (Nanjing Jiancheng Biological Engineering Institute) according to the manufacturer’s instructions at a wavelength of 510 nm using a multifunctional microplate reader (three plates).

BMSCs were seeded onto round cover slips and placed into 12-well plates (three replicate wells per group, three plates). After 2 weeks of osteogenic or adipogenic induction, they were fixed with 4% paraformaldehyde for 20 min, treated with 0.1% Triton X-100 for 20 min and blocked with goat serum for 30 min at 37°C. Then, cells were washed three times with PBS and incubated with primary antibodies (mouse monoclonal anti-Runx2, anti-PPARγ; Abcam) overnight at 4°C. After rinsing with PBS three times, these slips were incubated with an appropriate Alexa Fluor^TM 488 horseradish-peroxidase-conjugated antibody (1: 1, 000, Invitrogen) for 1 h at 37°C. Finally, slips were stained with 4’, 6-diamidino-2-phenylindole (DAPI) for another 10 min, rinsed with PBS and analyzed under a fluorescence microscope (DP72, Olympus, Shinjuku, Tokyo, Japan).

For RT-qPCR, total RNA was extracted with TRIzol (Invitrogen) according to the manufacturer’s instructions, and 1 µg of total RNA was used for RT reaction with EasyScript one-step gDNA Removal and cDNA Synthesis Supermix (TransGen Biotech, Beijing, China). The mRNA was quantified by real-time RT-qPCR using the SYBR Green/ROX qRT-PCR mix (Takara, Tokyo, Japan) and the Applied Biosystems StepOnePlus^TM Real-Time PCR System. The sequences of the PCR primers for Runx2, Osteocalcin (OC), ALP, bone morphogenetic protein (BMP) 2, permeability glycoprotein (P-gp), peroxisome proliferator-activated receptor gamma (PPARγ) and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) are listed in Table 1. After denaturing DNA templates at 95°C for 10 min, the reactions were followed by 40 cycles at 95°C for 15 s, 60°C for 20 s and 75°C for 10 s. Relative expression levels were determined by the 2^－ΔΔCt method.

For western blot analysis, cells were lysed in RIPA buffer (Sigma-Aldrich, St. Louis, MO, USA). Equal quantities of protein were loaded and separated on an SDS-PAGE gel, and transferred onto PVDF membranes. After blocking in TBST containing 5% fat-free milk for 1 h at RT, the membranes were probed overnight at 4°C with the following primary antibodies: mouse monoclonal anti-GAPDH, anti-Runx2, anti-ALP, anti-BMP2, anti-OC, anti-P-gp and anti-PPARγ; all from Abcam. After being washed, the membranes were incubated with relevant horseradish peroxide-conjugated secondary antibodies (1: 10, 000, Dako Cytomation, Glostrup, Denmark) for 1 h at RT. GAPDH was used as loading control. The protein bands were visualized and detected by the ECL system (Pierce, Rockford, IL, USA).

A total of 30 female SD rats (Laboratory Animal Center of Huazhong University of Science and Technology) were randomly separated into three groups: the MPS group (rats were intravenously injected with 10 µg/kg/d lipopolysaccharide (LPS, Sigma) for 2 days; after 24 h, the rats were injected intramuscularly with 20 mg/kg/d MPS for three consecutive days, n = 10), the MPS + ICA group (rats were injected with LPS and MPS as in the MPS group accompanied by consecutive ICA 30 mg/kg/d feed, n = 10), the control group (rats were injected with 0.9% normal saline (equal volume to the MPS group) n = 10). The intervention time was 12 weeks.

At the end of the treatment, blood samples from all the groups were obtained from the tail vein to assay the biochemical parameters. Samples of femoral head were obtained after the rats were sacrificed.

Blood samples were centrifuged at 5, 000 rpm for 10 min at 4°C. The serum was separated, and the serum ALP activity was measured using a specific assay (Nanjing Jiancheng Biological Engineering Institute) according to the manufacturer’s instructions. The absorbance at 405 nm was recorded as relative ALP levels in rats. Serum concentrations of triglyceride (TG) were measured with a full automatic biochemical analyzer (Roche Modular-T; Roche, Basel, Switzerland).

After removing soft tissue, the entire femoral head of each rat was scanned at 15 µm isotropic voxel size by micro CT (Scanco Medical, Brüttisellen, Switzerland) to evaluate bone morphologic changes. Images were reconstructed in 3D using 3D Creator software (Volume Graphics GmbH, Heidelberg, Germany). The region of interest (ROI) for the analysis and comparison of trabecular parameters was determined according to Dong et al [24]. The trabecular bone parameters including bone mineral density (BMD), bone volume (BV), bone volume/total volume (BV/TV), bone surface/total volume (BS/TV), trabecular separation (Tb. Sp), trabecular thickness (Tb.Th) and trabecular number (Tb.N) were quantified to determine the relative amount of bone within the femoral head.

After micro-CT scanning, the femoral heads were treated by decalcification and paraffin embedding. Then they were sectioned at a thickness of 5 µm in the coronal plane. Sections of the left femoral heads in each group were stained with hematoxylin and eosin (H&E). Oil red O was carried out to detect the lipid droplet formation in adipogenesis, the sections of femoral heads were stained with oil red O working solution for 15 min after fixing and pre-wetting with 60% isopropanol. The positive area was quantitative measured with the Image J 1.47 software. The diagnosis of osteonecrosis was established based on the presence of empty lacunae or pyknotic nuclei of osteocytes in the bone trabeculae, accompanied by surrounding bone marrow necrosis [25]. Sections of the right femoral heads were deparaffinized, antigen retrieved, incubated with primary antibodies (mouse monoclonal anti-Runx2, anti-PPARγ; Santa Cruz Biotechnology) and then incubated with the appropriate horseradish peroxide-conjugated secondary antibodies. Finally, sections were colored with DAB and counterstained with hematoxylin. Photomicrographs were acquired using a Leica DM 4000. Then, Image-Pro Plus 6.0 was used for quantitative analysis, while target protein and total area of trabecular bones was measured based on integrated optical density (IOD). The results were defined as mean density (IOD/area).

Data were presented as mean ± SD. Differences between groups were analyzed by one-way ANOVA, using SPSS 17.0 software (SPSS, Inc., Chicago, IL, USA). P < 0.05 was considered as statistically significant.

We first examined the individual effects of ICA (10^-9 M–10^-3 M) on proliferative activity in BMSCs by CCK8 assay (Fig. 1A). ICA at 10^-3 M significantly decreased cell proliferation compared to the control group (P < 0.05). In contrast, lower concentrations of ICA (10^-9 M–10^-5 M) significantly increased cell prolifration (P < 0.05). ICA at 10^-4 M increased cell proliferation, but the difference was not statistically significant (P > 0.05). Our results indicated that ICA promoted proliferation of BMSCs, especially at 10^-6 M.

The CCK8 assay showed that BMSC proliferation was significantly suppressed by 5 × 10^-5 M MPS, while this inhibition was partly antagonized by ICA, especially at a concentration of 10^-6 M (Fig. 1B).

ALP activity is a marker of osteoblastic differentiation. As shown in Fig. 2A, after treatment of BMSCs with MPS or MPS + ICA (10^-5 M–10^-7 M) for 3, 7 or 14 days, ALP activity was significantly down-regulated by MPS (P < 0.05), while the suppression was antagonized when combined with different concentrations of ICA, especially at 10^-6 M (P < 0.05).

Alizarin red S staining showed fewer calcium nodules in the MPS group compared with the control group, while the mineralization of BMSCs improved significantly in those treated with a combination of ICA and MPS compared to those treated with MPS alone (Fig. 2B).

We analyzed the expression of osteogenic-associated genes by RT-qPCR to verify whether ICA improved osteogenic differentiation of BMSCs treated with MPS. We found that mRNA levels of Runx2, ALP, BMP2 and OC were all significantly down-regulated by MPS (P < 0.05), while ICA at 10^-6 M partly antagonized this down-regulation of Runx2, ALP, BMP2 and OC expression (Fig. 2C). Western blotting showed that MPS down-regulated the expression of Runx2, ALP, BMP2 and OC, while MPS and 10^-6 M ICA increased their expression (Fig. 2D). Immunofluorescence staining revealed that the MPS group exhibited a decrease in Runx2 expression compared to the control group, while the MPS＋ICA ^-6 Mgroup showed an increase in the protein compared to the MPS group (Fig. 4A).

As shown in Fig. 3A, after treatment of BMSCs with MPS or MPS + ICA (10^-5 M–10^-7 M) for 14 days, the TG content was significantly higher in the MPS-treated group than the control group (P < 0.05), and a lower content of TG was observed in the MPS + ICA group compared to the MPS group (P < 0.05), especially in the MPS＋ICA^-6 M group.

Oil red O staining showed the presence of more fat droplets in the MPS group compared with the control group, while the number and size of fat droplets were both significantly decreased after treatment with MPS and 10^-6 M ICA compared to the MPS alone group (Fig. 3B).

Our results indicated that the mRNA level of PPARγ was significantly up-regulated by MPS (P < 0.05), as shown in Fig. 3C, while expression in the MPS＋ICA^-6 M group was decreased compared to that in the MPS group (P < 0.05). Neither MPS nor ICA affected the mRNA expression of P-gp in adipogenic differentiation of BMSCs (P > 0.05). As shown in the western blot analysis (Fig. 3D), the expression of PPARγ was up-regulated by MPS, while expression in the MPS + ICA^-6 M group was lower than in the MPS group. In contrast P-gp expression showed no difference among the three groups. Immunofluorescence staining revealed that the expression of PPARγ in the MPS group was higher than in the control group, while compared with the MPS group, the MPS + ICA^-6 M group showed a decrease in the expression of PPARγ (Fig. 4B).

Serum levels of ALP and TG were measured after 12 weeks’ intervention. As shown in Fig. 5A, the ALP level in the MPS group was lower than in the control group, but the difference was not significant (P > 0.05). However, we detected a significantly higher ALP level in the rats treated with MPS and ICA (P < 0.05). As shown in Fig. 5B, the MPS group had higher serum levels of TG compared to the control group (P < 0.05), while the addition of ICA clearly decreased its production.

We used micro-CT to analyze trabecular changes in the subchondral area of the femoral heads. After 12 weeks, the incidence of osteonecrosis was 0 (0/10), 70 % (7/20) and 30% (3/10) in the control group, MPS group and MPS + ICA group respectively (P < 0.05). Fig. 6A shows representative micro CT images of osteonecrosis. The BMD of the rats in the MPS group was 657.5 ± 39.4 mg/cm3, which was significantly lower than the control group (861.0 ± 27.0 mg/cm3), while supplementation with ICA significantly restored the BMD (763.7 ± 20.8 mg/cm3) (Fig. 6B). Meanwhile, quantitative analysis of the trabeculae showed that other bone parameters such as bone volume/total volume (BV/TV), bone surface/total volume (BS/TV), trabecular number (Tb. N), trabecular thickness (Tb. Th) and trabecular spacing (Tb. Sp) were similar among the three groups (Fig. 6C-G). In addition, H&E and oil red O staining showed fat tissue invasion and obvious subchondral necrosis in the subchondral bone trabeculae in the MPS group. In contrast, no such changes were observed in the MPS + ICA group in which there were few smaller fat cells and empty lacunae (Fig. 7A, 7B, 7E). Consistent with H&E and oil red O staining observations, immunohistochemical staining showed more positive staining of Runx2 expression and less expression of PPARγ in the MPS + ICA group compared with the MPS group (Fig. 7C, 7D, 7F, 7G).

Recently, there has been strong interest in studying the efficacy of ICA as a potential alternative therapy for improving bone formation, preventing bone loss and treating osteoporosis [26, 27]. However, there are few studies focusing on ONFH. To our knowledge, this is the first study to demonstrate that ICA can be an effective agent for prevention of ONFH. In this study, we showed that ICA improved cell proliferation, enhanced osteogenic differentiation and suppressed adipogenic differentiation of MPS-treated BMSCs. Moreover, our in vitro study demonstrated that ICA increased the expression of osteogenic-associated genes of BMSCs which were inhibited by MPS and decreased the expression of adipogenic-associated genes of BMSCs stimulated by MPS. In addition, we observed that treatment with ICA clearly prevented bone loss and reduced the incidence of ONFH at 12 weeks in a rat model of GC-associated ONFH.

Our study demonstrated that treatment of BMSCs with ICA (10^-9 M–10^-5 M) significantly increased cell proliferation, especially at 10^-6 M. Many studies indicated that ICA at different concentrations enhanced cell proliferation. Mok et al. found that ICA significantly enhanced proliferation at concentrations ranging from 10^-14 M to 10^-6 M in UMR 106 cells [26]. Moreover, a dose-dependent study by Ma revealed that the optimal concentration of ICA for stimulating osteogenesis was 10^-5 M in ROB cells [28]. These differences in the effective concentration of ICA were due to the differences in the cells used. However, few studies have reported the effect of ICA on BMSCs. Moreover, we also discovered that ICA clearly increased proliferation of BMSCs inhibited by MPS. As BMSCs are essential for the healing response [23, 27] and decreasing activity of BMSCs after exposure to high-dose steroids might be a triggering step of GC-associated ONFH [29], our results on BMSC proliferation with ICA treatment might suggest a potential remedy for the insufficient repair mechanism in GC-associated ONFH.

In the MPS groups, the actual GC dose is that of dexamethasone (10^-7 M) + MPS. However, under physiological conditions, serum cortisol concentrations reach 10^-8–10^-7 M in humans [30] and 2.7-5.4x10^-8 M in rats [31]. The requirement for dexamethasone for adequate osteogenesis and adipogenesis in in vitro experiments has been reported previously [32, 33]. Dexamethasone regulates gene expression in differentiating cells and induces the affinity of the GC receptor for its target sequence in the genome [34]. Moreover, the medium in all groups shared the same concentration of dexamethasone, and the influence of MPS on osteogenesis of BMSCs treated with dexamethasone has been previously reported [35]. In addition, the dose of MPS was 500 times greater than that of dexamethasone in the present study. The effects of ICA on osteogenesis and adipogenesis following treatment with MPS would not be affected by dexamethasone.

BMSCs are multipotent cells that can differentiate into multiple cell lineages. Osteogenic and adipogenic differentiation of BMSCs maintains dynamic balance in the physiological state, but this balance may be disrupted by steroids. Long-term use of high doses of GC can directly destroy the balance by downregulating osteogenic-related genes and upregulating the expression of adipogenic-associated genes [7]. This is considered to be a principal mechanism leading to GC-related ONFH. Accordingly, how to inhibit adipogenic differentiation of BMSCs and promote their osteogenic differentiation is a key issue in the management of ONFH progression.

Our study indicated that MPS clearly induced the differentiation of BMSCs into adipocytes and inhibited their differentiation into osteoblasts. Although the direct effects of ICA in modulating osteogenic differentiation and inhibiting adipogenic differentiation have been reported [36, 37], its effects on BMSCs exposed to MPS and treated with ICA have not been studied. Our study further demonstrated that ICA promoted osteogenic differentiation and suppressed adipogenic differentiation of BMSCs treated with MPS.

We found that ICA promoted the osteogenic effect of BMSCs inhibited by MPS and enhanced the biomechanical strength of the femurs in a rat model of GC-associated ONFH. A previous study has reported that osteoblasts differentiated from BMSCs can facilitate new bone formation [38]. However, there is also the possibility that BMSCs might differentiate into the adipocyte lineage when exposed to steroids, and then inhibit osteogenic differentiation, thus resulting in inadequate bone repair in GC-induced ONFH [39]. These findings are consistent with our experimental results, in which we found that lesions in rats of the MPS group showed empty lacunae accompanied by surrounding marrow cell necrosis and occupation of adipocytes during histopathological examination. After ICA treatment, the ratio of empty lacunae and the area of bone marrow occupied by adipocytes were significantly reduced, suggesting that improvement in local adipogenesis was induced by ICA. In addition, ICA increased bone formation and improved the trabecular microarchitecture in the femoral head. These findings suggested that ICA could prevent bone loss induced by MPS.

There are many signaling pathways involved in promoting osteogenic and/or adipogenic lineage differentiation of BMSCs, of which the two most important are the BMP2/Runx2 and the PPARγ signaling pathways [40]. Runx2 is the key transcription factor in osteogenesis and is regulated by BMP-2 [41]. Likewise, PPARγ is the master regulator of adipogenesis and also has been well described to have anti-osteoblastogenic effects. Together, they regulate various cytokines to determine the choice of osteogenic versus adipogenic BMSCs differentiation. Typically, stimulating one pathway is associated with suppression of the other [42, 43]. Although studies have shown that ICA produces higher ALP activity and expression of Runx2, OC and BMP2 levels in vitro [27, 28]; the combined effect of ICA and MPS on BMSCs has rarely been reported. In addition, PPARγ is well known to be upregulated as an adipogenic marker in GC-induced ONFH [44], but few studies have reported the effect of ICA on adipogenic differentiation of BMSCs induced by GC. The present study first demonstrated that ICA regulated both osteogenic and adipogenic differentiation of BMSCs treated with MPS in both in vitro and in vivo experiments. The ICA + MPS group showed higher expression of Runx2 and BMP2, and lower PPARγ expression than the MPS group, suggesting that ICA might mediate promotion of osteogenesis in the BMP2/Runx2 signaling pathway and inhibit adipogenesis through the PPARγ-mediated pathway in GC-induced ONFH.

The mechanism via which ICA regulates the pathways of BMP2/Runx2 and PPARγ in GC-induced ONFH is uncertain. Glucocorticoid has a positive relation to oxidative stress production [45] and the presence of DNA oxidation injury was confirmed in the early period of corticosteroid treatment [46]. The important point of oxidative stress modulates osteogenic and adipogenic differentiation has received much attention [47]. Recently, the activity of ICA supplementation in reducing oxidative stress has been reported [48] and confirmed in our previous study [14]. Thus, we speculate that ICA regulates the pathways of BMP2/Runx2 and PPARγ by reducing oxidative stress in GC-induced ONFH. Moreover, estrogen-like activities of ICA have been widely reported [26] and estrogen-like analogues can also reduce oxidative stress [49]. Such a mechanism might be partly involved in regulating oxidative stress through the ER. However, this remains to be studied further.

GC induced osteoblast apoptosis was regarded as the common mechanism in the progress of osteoporosis and osteonecrosis. However, balance disruption between osteogenic and adipogenic differentiation and increased bone marrow fat were also important mechanisms in GC-induced ONFH [7, 50]. In our present study, we not only showed the inducing osteogenic effects of ICA, but also demonstrated its' inhibiting adipogenic effects. Feng R et al. showed that ICA prevented GC induced apoptosis in a GC-induced osteoporosis animal model [51]. The establishment of animal model is different with our animal model of GC-induced ONFH. Moreover, osteoporosis is a systemic disease and ONFH is the disease limited to the femoral head. Osteonecrosis may occur without GC-induced osteoporosis [52]. There are many effective drugs for the treatment of GC-induced osteoporosis such as, bisphosphonates, teriparatide, denosumab and etc. The effect is limited when prescribed these drugs in ONFH [53, 54]. GC-induced ONFH occurs predominantly in younger patients and it is the most common cause of total hip replacement in young adults [50]. Since these replacements have about a 10-year lifespan, any delay in the need for surgery would be welcome [54]. For these reasons, efforts should be made to preserve the joint and delay hip replacement. Because of the potential effect of ICA in GC-induced osteoporosis treatment, it is possible and necessary to research the effect of ICA in GC-induced ONFH. As far as we know, the present study was first to evaluate the effect of ICA in GC-induced ONFH. We will establish the GC-induced ONFH model of core decompression + ICA sustained-release for further understanding of this compound in GC-induced ONFH. If confirmed, ICA will be a promising drug for the treatment and prevention of early stage of GC-induced ONFH.

In conclusion, the results of our study provided, for the first time, evidence that ICA is an effective compound for promoting bone repair and preventing or delaying the progression of GC-associated ONFH by regulating the BMP2/Runx2 and PPARγ-mediated signaling pathways. Thus, ICA should be considered as a potential candidate for management of GC-associated ONFH.

This work was supported by the Natural Science Foundation of Hubei Province (2013CFB376, 2018CFB095). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The authors declare that they have no competing interests.