The MC3T3-E1 osteoblast cell line was obtained from Procell Life Science & Technology Co., Ltd. and the primary osteoblasts were obtained from C57BL/6 mice (22). A total of 12 C57BL/6 mice were used in this study and all animals were housed in the SPF-grade Animal Experiment Center of the Institute of Sports Medicine, Shandong First Medical University, with a room temperature of 22°C, relative humidity of 55% and 12-h light/dark cycle, standard mice food feeding, free water intake. Chloral hydrate (3%) was used to anesthetize the mice by intraperitoneal injection. Neonatal C57BL/6 mice were sacrificed by cervical dislocation method and disinfected with 70% ethanol and calvaria was predigested in 0.25% trypsin (MilliporeSigma) at 37°C for 20 min. After the bone fragments were cut into small pieces of ~1 mm² and digested in 0.1% collagenase type II (MilliporeSigma) at 37°C for 50 min to collect the digested product. Digestion of residual bone fragments was repeated once with fresh collagenase. The digested product collected twice was centrifuged at 90 × g at room temperature for 5 min, the supernatant was discarded and the pellet was resuspended and inoculated with Dulbecco's Modified Eagle's Medium (DMEM; Gibco; cat. no. C11995500BT; Thermo Fisher Scientific, Inc.) containing 10% fetal bovine serum (FBS; cat. no. 04-001-1ACS; Biological Industries) and 1% penicillin-streptomycin (MilliporeSigma). The Biomedical Research Ethics Committee of Shandong First Medical University approved this study (Shandong Academy of Medical Sciences; ethics approval no. W202210070243). All cells were cultured in DMEM at 37°C in a 5% carbon dioxide incubator. The medium was changed two to three times per week.

Primary osteoblasts and MC3T3-E1 cells were plated at a density of 1×10⁵ cells/ml in 6-well plates at 37°C in a 5% CO₂ constant temperature incubator. When the cells were 80% confluent, the plasmids were transfected according to the instructions of the Lipofectamine® 2000 reagent (Invitrogen; Thermo Fisher Scientific, Inc.). The short interfering (si)RNA-FGF23 plasmid (5′-GCTATCACCTACAGATCCATA-3′), empty vector plasmid (5′-UUCUCCGAACGUGUCACGU-3′), FGF23 overexpression plasmid (5′-UCCAUGCAGAGGUUAUCUUC-3′) and miR-17-5p inhibitor (5′-CAAACACCUUACACACCAGGUAG-3′) were synthesized by Hanbio Biotechnology Co., Ltd. To prepare the transfection mixture, 4 µg of plasmid was added to 500 µl of serum-free medium and 5 µl of Lipofectamine 2000 (cat. no. 11668019; Thermo Fisher) was added to 500 µl of serum-free medium and mixed well and then allowed to stand at room temperature for 20 min. The cells were washed with phosphate-buffered saline (PBS) three times and then 1 ml of Lipofectamine® 2000 transfection mixture and 1.5 ml serum-free medium were added and the cells were incubated for 6 h at room temperature. The cells were then switched to the regular medium and continued to be incubated for 48 h at 37°C in a 5% carbon dioxide incubator. The transfection efficiency of FGF23 was verified by western blotting. Transfection with an empty vector served as the control for plasmid transfection (23).

The incubators were gassed with nitrogen to maintain an anoxic atmosphere of 5% CO₂ and 95% N₂ (24). The experimental subjects were divided into five groups as follows: Normal control group (NC), hypoxia control group (con), empty vector group (EV), FGF23 overexpression group (FGF23) and FGF23 silencing group (si-FGF23).

Primary osteoblasts and MC3T3-E1 cells were collected from each group, digested with trypsin, plated at 1×10⁴ cells/well in 96-well plates and cultured for 48 h. Three replicate wells were used for each group. Then, 10 µl of 5 mg/ml MTT solution (cat. no. Top0191T; Beijing Biotuoda) were added. After incubation at 37°C for 4 h, 150 µl dimethyl sulfoxide was added and incubated for 10 min. The absorbance value (OD_{490 nm}) at a wavelength of 490 nm was measured using a microplate reader. Three replicate blank samples were used for testing, three times for each group (25).

Follow the instructions for using the caspase-3 protein activation assay kit (Nanjing KeyGen Biotech Co., Ltd.), 50 µl of each group of sample proteins were taken, 50 µl 2X reaction buffer and 5 µl caspase-3 substrate were added and then incubated at 37°C for 4 h in the dark. The absorbance at 405 nm was measured using a microplate reader (2).

The influence of hypoxia on apoptosis in the osteoblasts was tested by flow cytometry (FCM; Beckman CytoFLEX FCM; Beckman Coulter) using a phospholipid-binding protein V (Annexin V-FITC)/propidium iodide (PI) kit (US Everbright Inc.). After the experimental hypoxia treatment ended, the cells were washed three times with phosphate buffer, collected by centrifugation at 90 × g at 4°C for 5 min and resuspended at 5×10⁴ cells/ml in 500 µl binding buffer. The cells were then transferred to Eppendorf tubes and incubated with 5 µl Annexin V-FITC and 5 µl PI for 30 min at room temperature in the dark. Results were analyzed by FlowJo v10.6.2 software. The apoptosis rate is the sum of early apoptosis rate and late apoptosis rate (26).

Transfected cells were seeded in 6-well plates at 1×10⁵ cells/well and washed twice with PBS after 48 h incubation. Then, 100 µl of FDA and EB (MilliporeSigma) were added to each well and incubated at 37°C in the dark for 10 min before imaging with a fluorescence microscope (27).

Cellular autophagy was detected with the Cellular Autophagy Staining Assay Kit (cat. no. C3018S; Beyotime Institute of Biotechnology). Monodansylcadaverine (MDC) is an eosinophilic fluorescent probe that can specifically label autophagosomes by ion capture and specific binding to membrane lipids. After the cells were treated accordingly, 1 ml of MDC staining solution was added to each well and incubated for 30 min at 37°C in a cell incubator protected from light. After washing, 1 ml of Assay Buffer was added and green fluorescence was observed under a fluorescence microscope.

RT-qPCR was used to assess FGF23 expression. First, total RNA was isolated from primary osteoblasts (1×10⁶) and MC3T3-E1 cells (1×10⁶) using an RNA extraction kit (cat. no. BSC52M1; BioFlux) and complementary DNA was synthesized using RT kit (cat. no. BSB07M2B, BioFlux) according to the manufacturer's protocols. The Hieff® qPCR SYBR Green Master Mix (cat. no. 11184ES03; Shanghai Yeasen Biotechnology Co., Ltd.) was used for qPCR. PCR primer sequences are shown in Table I. The prepared cDNA, GAPDH and U6 were used as a template and reference for RT-qPCR reactions. Amplification conditions were as follows: 95°C for 5 min, 95°C for 10 sec, 60°C for 30 sec, 72°C for 30 sec, for a total of 35 cycles. Finally, the relative FGF23 expression was analyzed using the 2^−ΔΔCq method [ΔCq=Cq (target gene)-Cq (reference gene)] (28). Two replicate blank samples were used for each group (29).

Total protein of each group was lysed with RIPA lysis solution (cat. no. R0010; Beijing Solarbio Science & Technology Co., Ltd.). The protein concentration was determined by bicinchoninic acid (BCA; cat. no. PC0020; Beijing Solarbio Science & Technology Co., Ltd.) method. RIPA lysis buffer (Beijing Solarbio Science & Technology Co., Ltd.) was added to extract the total protein fully and ~30 µg per lane of protein samples were resolved by 10% sodium dodecyl sulfate-polyacrylamide gel, after which the proteins were transferred to polyvinylidene (MilliporeSigma) membranes for 1 h at 100 V and blocked for 1 h in 5% skimmed milk powder at room temperature. Primary antibodies against, anti-hypoxia-inducible factor-1α (HIF-1α; 1:1,000; cat. no. 14179; Cell Signaling Technology, Inc.), anti-Bax (1:1,000; cat. no. GB11007; Wuhan Servicebio Technology Co., Ltd.), anti-Bcl-2 (1:1,000; cat. no. sc-492; Santa Cruz Biotechnology, Inc.), anti-caspases-3 (1:1,000; cat. no. sc-7148; Santa Cruz Biotechnology, Inc.) and −9 (1:1,000; cat. no. 9504; Cell Signaling Technology, Inc.), anti-LC3B (1:1,000; cat. no. 3868; Cell Signaling Technology, Inc.), anti-LC3A (1:1,000; cat. no. 4599; Cell Signaling Technology, Inc.) anti-Beclin-1 (1:1,000; cat. no. 3738; Cell Signaling Technology, Inc.) and anti-β-actin (1:1,000; cat. no. GB15001; Wuhan Servicebio Technology Co., Ltd.) were added followed by overnight incubation at 4°C. After washing with TBST (0.1% Tween 20; cat. no. ST673, Beyotime Institute of Biotechnology), the secondary antibodies, goat anti-mouse IgG (1:2,000; cat. no. GB23301; Wuhan Servicebio Technology Co., Ltd.) and goat anti-rabbit IgG (1:2,000; cat. no. GB23303; Wuhan Servicebio Technology Co., Ltd.), were incubated for 1 h at room temperature. Protein bands were visualized using an ECL kit (cat. no. P10200; New Cell & Molecular Biotech Co., Ltd.). The relative expression levels of the proteins were analyzed by ImageJ v1.8.0 software using β-actin as a reference (30).

Complementary binding sites between miR-17-5p and FGF23 were predicted using the TargetScan database (http://www.tragetscan.org/) and StarBase database (http://www.starbase.sysu.edu.cn/). Nucleotide sequences containing miR-17-5p binding sites in wild-type (WT) or mutant (MUT) FGF23 3′-UTR were cloned into the pGL3 luciferase reporter plasmid (Invitrogen; Thermo Fisher Scientific, Inc.). MC3T3-E1 cells in the logarithmic growth phase were seeded in 24-well plates (2×10⁵ cells/well) and the miR-17-5p mimic and miR-NC were co-transfected with WT or MUT reporter plasmids into MC3T3-E1 cells using Lipofectamine® 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) and cultured for 48 h. The fluorescence activity was detected 48 h post-transfection using a dual-luciferase kit (cat.no.E1910; Promega). The firefly and Renilla Luciferase activity were detected by the dual luciferase reporter gene detection system and the luciferase activity of the reporter genes was calculated and the luciferase value=Firefly Luciferase activity/Renilla Luciferase activity (31).

Each experiment was repeated three times. GraphPad Prism v6.02 statistical software (Dotmatics) was used to analyze the data. Normally distributed measurement data are expressed as mean ± standard deviation. The unpaired Student's t test was used for two samples and on way analysis of variance (ANOVA) was used for multiple samples. Tukey's post hoc tests were performed using the Bonferroni method for two samples comparisons between groups. P<0.05 was considered to indicate a statistically significant difference.

The levels of FGF23 in the cells after apoptosis induction under different hypoxic conditions were evaluated by western blotting and RT-qPCR. Due to alterations in hypoxia time, the expression levels of FGF23 protein and mRNA in osteoblasts showed significant fluctuations. Western blotting showed that FGF23 protein expression was the highest at 4 h and then gradually decreased with the extension of hypoxia time in in MC3T3-E1 cells (P<0.05; Fig. 1A). The RT-qPCR results were consistent with the western blotting results (Fig. 1B). Additionally, HIF-1α expression increased under prolonged hypoxia for up to 12 h (Fig. 1A). FGF23 protein and mRNA expression in primary osteoblasts was highest at 4 h and then gradually decreased (Fig. 1C and D). These results indicated that the expression of FGF23 is influenced by hypoxic environments and associated with time. Therefore, hypoxia for 4 h was selected for the following experiments. MC3T3-E1 cells, a mouse osteogenic precursor cell line, are commonly used to study ONFH (32,33).

Increased autophagosomes were found in primary osteoblasts and MC3T3-E1 cells following hypoxia, and FGF23 knockdown partially reversed this result (Fig. 2A). Western blotting demonstrated that the protein levels of FGF23 were altered by transfection (Fig. 2B and D). Significant increases were observed in the levels of microtubule-associated protein light chain-phosphatidylethanolamine conjugate (LC3-II), Beclin-1 and LC3-II/LC3-I ratio, under hypoxic conditions (P<0.05). Furthermore, compared with the EV group, Beclin-1, LC3-II and the LC3-II/LC3-I ratio were increased in the FGF23 group; however, they were significantly decreased in the si-FGF23 group (Fig. 2B and D). In addition, autophagy-related gene expression also showed corresponding alterations (Fig. 2C and E). Taken together, these results indicate that FGF23 can promote hypoxia-induced autophagy levels in osteoblasts.

MTT assay was used to measure the influence of different factors on cell viability. The viability was reduced in the con, FGF23 overexpression and si-FGF23 groups compared with the NC group. The FGF23 overexpression group had the lowest cell survival rate (P<0.05; Fig. 3A). FGF23 could significantly increase activity of caspase-3 compared with that in the EV group (P<0.05; Fig. 3B). FCM and FDA/EB staining showed increased apoptosis of osteoblasts in the EV group. However, the FGF23 group showed significantly increased apoptosis compared with the EV group, whereas apoptosis decreased slightly in the si-FGF23 group (Fig. 3C-D). These results indicated that FGF23 might activate the apoptosis pathway and promote cell death during hypoxic cellular stress.

Western blotting results showed that hypoxia led to a significant increase in levels of pro-apoptotic proteins Bax, caspases-3 and −9 and a decrease in levels of anti-apoptotic protein Bcl-2 in primary osteoblasts and MC3T3-E1 cells. Western blotting results showed significantly increased levels of the pro-apoptotic proteins Bax and caspases-3 and −9 and decreased levels of anti-apoptotic Bcl-2 (Fig. 4A and C). The RT-qPCR data showed enhanced expression of the pro-apoptotic (Bax and caspases-3 and −9) mRNA. By contrast, the apoptosis suppressor Bcl-2 mRNA was diminished in osteoblasts compared with the NC group in response to FGF23. The differences were statistically significant (Fig. 4B and D). Apoptosis levels were further elevated following FGF23 overexpression, whereas, FGF23 knockdown attenuated the apoptosis level in osteoblasts compared with the con group. The above results indicated that knockdown of FGF23 can ameliorate hypoxia-induced apoptosis in osteoblasts.

To further investigate the molecular mechanisms by which FGF23 regulates osteoblast apoptosis, a potential binding site between FGF23 and miR-17-5p was identified by the online database TargetScan and StarBase (Fig. 5A). Subsequent overexpression or inhibition of miR-17-5p was performed for dual luciferase reporter gene assay. As shown in Fig. 5B, luciferase analysis revealed that the FGF23 wild-type luciferase activity in the miR-17-5p group was lower than that in the miR-NC group (P<0.05). By contrast, there was no statistically significant difference in the FGF23 MUT luciferase activity between the miR-17-5p and miR-NC groups (P>0.05). In addition, RT-qPCR results showed that the expression level of miR-17-5p was downregulated after overexpression of FGF23; it was significantly upregulated after the silencing of FGF23 (Fig. 5C). Detection of cell transfection efficiency revealed that transfection with miR-17-5p inhibitor downregulated miR-17-5p expression (Fig. 5D).

Compared with the con group, the apoptosis rate (Fig. 6A and B), Bax, caspase-3 and caspase-9 protein (Fig. 6C) expression were significantly decreased, and Bcl-2 protein expression was increased, in MC3T3-E1 cells in the si-FGF23 group. Conversely, compared with the si-FGF23 group, Bax, caspase-3 and caspase-9 protein expression and apoptosis rate were significantly increased. Bcl-2 expression was downregulated in MC3T3-E1 cells in the si-FGF23 + miR-17-5p inhibitor group (Fig. 6).

FGF23 knockdown reduced the number of autophagosomes, which was partly reversed by the addition of miR-17-5p inhibitor (Fig. 7A). The expression of the autophagy-associated proteins Beclin-1 and LC3 was detected by western blotting and RT-qPCR. The results showed that Beclin-1 and LC3-II/LC3-I expression decreased after FGF23 silencing, whereas this result was partially reversed after miR-17-5p inhibition (Fig. 7B and C).