Sprague Dawley (SD) rats (4 weeks old) were purchased from the Experimental Animal Center of Anhui Medical University (SPF grade, weighed 100–130 g) (Hefei, China). A total of ∼50 rats were used for various experimental approaches in the current study. The rats were housed in groups (four to five rats per cage). All rats received a standard rodent diet and tap water ad lib. A constant temperature of 21–23°C and humidity of 50 ± 5% were maintained. Following feeding for 1 week, the rats were killed by cervical dislocation and were confirmed to have no spontaneous breathing and heartbeat. All efforts were made to minimize suffering. The rats were immersed in 75% ethanol and sterilized for 20 min. After removing the femur under aseptic conditions, scissors were used to cut both ends of the femur. A 5-ml syringe equipped with a 25-gauge needle was used to rinse the bone marrow in the medullary cavity with α-modified minimum essential medium (α-MEM, HyClone, Invitrogen, United States). The washed cells were collected, cultured in α-MEM supplemented with 10% fetal bovine serum (Gibco, United States) and 1% penicillin/streptomycin (Gibco, United States), and incubated at 37°C with 5% CO₂. All experiments used cells from passages 3 to 5. The study has been approved by the Ethics Committee of the Second Affiliated Hospital of Anhui Medical University.

A pure titanium foil with a thickness of 2 mm was cut into a square with a length of 35 mm and used as a substrate. The top and bottom surfaces of the foil were sanded with silicon carbide sandpaper. A pure titanium foil was soaked in absolute ethanol and cleaned with an ultrasonic cleaner. After washing again with deionized water, the foil was air-dried at room temperature. The nanotubes were formed in electrolyte solution containing 0.15 M NH₄F at 20 V for 30 min by anodization using a potentiostat. The cathode was a pure platinum foil. After anodization was completed, the sample was rinsed with deionized water.

BMSCs were seeded on the TiO₂ nanotubes–modified titanium specimens at a density of 1 × 104 cells/cm². The cells were cultured for 48 h to allow them to attach and reach approximately 80% confluence, after which the growth medium was replaced, and cyclic mechanical stress was applied. The cyclic mechanical stress parameters used for our manufactured mechanical loading device (Figures 1A,B) are as follows: 0.3 and 0.9% elastic strain magnitude, 5-Hz sinusoidal curve, for 30 min/day (Chang et al., 2019). The specimen (a) with BMSCs cultured on the surface was placed into the fixed slot (b) and fixed. Then, the fixing slot containing the specimen was put into the fixed cylinder (c). The medium was added to the fixed cylinder until the test piece became submerged, so as to ensure that the specimen and the cells on the surface were always immersed in the medium during the process of applying cyclic mechanical stress. The top hammer (d) applied cyclic mechanical stress by pressing the top cover (e) on the top of the fixed slot. The top hammer was connected with the winch (g) through the lever (f). The device was connected through the computer, the winch rotation parameters were set, and the device was started. When the specimen is compressed, the elastic strain magnitude is ((the original height-the height after compression)/the original height) (0.3% and 0.9%). The entire process took 7 days. Cells were harvested immediately after the mechanical stress stimulation was applied.

To study osteogenic differentiation with DKK1, the density of 1 × 104 cells/cm² was seeded to a specimen. The conditioned medium was changed with DKK1 (100 ng/ml, 120–30, Pepro-Tech, United States). The medium was changed every 3 days. The cells were harvested after 7 days of culture for other experiments.

We used PCR to amplify Rat GCN5. XbaI and NotI were used to digest the amplicons before inserted into the pCDH lentiviral vector. The sequences of the primers were forward: AGC​ACT​CCC​ATC​TTC​AGT​CC and reverse: GCT​TCC​TCT​TCT​CTC​CTG​GCA​T. The primers for GCN5 shRNA were forward: CCG​GGC​AGG​GTG​TTC​TGA​ACT​TTC​TCA​AGA​GAA​AAG​TTC​AGA​ACA​CCC​TGC​TTT​TTT​G and reverse: ATT​CAA​AAA​AGC​AGG​GTG​TTC​TGA​ACT​TTT​CTC​TTG​AGA​AAG​TTC​AGA​ACA​CCC​TGC.

Restriction enzymes AgeI and EcoRI were used to digest the PCR product before incorporated into the pLKD vector. Sanger sequencing was performed to verify the inserted fragments. To produce the lentivirus, 293T cell line was co-transfected with two packaging vectors (psPAX2 and pMD2.G) and two plasmid vectors. The supernatant was centrifuged at 1,000 rpm for 10 min to get rid of the cell debris. A 0.45-μm polyethersulfone low-protein-binding filter was used to filter. The titer of the lentivirus was 115 IFU/ml, assessed by a quantification kit (Invitrogen, United States). A vector inserted with scrambled GCN5 was used as a negative control.

BMSCs were seeded onto TiO₂ nanotubes–modified titanium specimens, and cyclic mechanical stress was applied. On day 7, ALP activity was measured using an ALP kit (Hongqiao, China) according to the manufacturer’s instructions. The results were observed under a stereo microscope.

Total RNA was extracted from the cells with TRIzol (Invitrogen, United States) and reverse-transcribed into cDNA using a PrimeScriptTM cDNA Synthesis kit (6210A, Takara, Japan) according to the manufacturer’s instructions. Real-time PCR was performed with SYBR^® Premix Ex Taq™ (Takara, Japan). GAPDH served as an internal control, and the expression level relative to GAPDH was calculated using 2−ΔΔCq method. The primer sequences are listed in Table 1.

The cells were washed three times with PBS and lysed with RIPA buffer supplemented with protease and phosphatase inhibitors for 30 min at 4 C. For the Western blot analysis, 15 μL of the sample was separated by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis and electro-transferred onto nitrocellulose membranes. The primary antibodies used were rabbit polyclonal anti-GCN5 (1:1000, ab231075, Abcam, China); rabbit monoclonal anti-Axin-2 (1:1000, ab109307, Abcam, China); rabbit monoclonal anti-β-catenin (1:1000, ab32572, Abcam, China) and mouse monoclonal phosphorylated β-catenin (1:500, 05-665-AF647, Millipore, United States); rabbit monoclonal anti-H3K9ac (1:1000, ab10812, Abcam, China); rabbit monoclonal anti-RUNX2 (1:1000, ab192256, Abcam, China); rabbit monoclonal anti-OCN (1:1000, ab102936, Abcam, China); and rabbit monoclonal anti-OPN (1:1000, ab63856, Abcam, China). For normalization of protein loading, GAPDH antibody (1:1000, ab245357, Abcam, China) was used. The protein bands were measured with ImageJ 1.48v software and normalized to the corresponding GAPDH bands. The relative density of each target protein normalized to the control was used to represent the changes in expression of target proteins.

ChIP assay was performed following the manufacturer’s instructions of a ChIP assay kit (Millipore, United States). The rabbit monoclonal anti-GCN5 (1:1000, ab231075, Abcam, China) and anti-H3K9ac (1:1000, ab10812, Abcam, China) were used. The rat IgG (I8015, Millipore, United States) and anti-RNA polymerase II (PLA0127, Millipore, United States) were used as the negative and positive control, respectively. The precipitated DNA samples were assessed by real-time PCR. Primers were designed from the region of 500 bp distance to the transcription start sites. The primer sequences are listed in Table 2. The values of the results were normalized to the input value.

GraphPad Prism 7.0 (GraphPad Software) was used for statistical analysis. Data are expressed as mean ± SEM. Student's t test was used for comparing two groups, one-way analysis of variance was used for comparing multiple groups, and the least-significant difference method was used as a post-hoc test for multiple comparisons between groups. p < 0.05 was considered as significant.

Our previous studies have measured the topography of the TiO₂ nanotube–modified titanium specimen surface and related physical parameters (Chang et al., 2019). The topography of the specimen surface is shown in Figure 1C. The schematic diagram and model diagram of the cyclic mechanical strain load device are showed in Figures 1A,B. The diameter of TiO₂ nanotubes is about 80 nm (Figure 1C). We applied three levels of cyclic mechanical elastic strain to titanium sheets planted with BMSCs. After 7 days of continuous treatment, we found that significantly higher levels of ALP activities in the 0.3 and 0.9% groups, especially in the 0.9% group (Figure 2A). Then the results of real-time PCR indicated that the mechanical strain promoted the expression of osteogenesis genes, such as Runx2, OPN, and OCN (Figures 2B–D). The results of the Western blot showed that mechanical strain increased the expression of these osteogenic markers from the protein level (Figure 2E). These results further validate our previous research findings (Chang et al., 2019). These indicated that rat BMSCs are sensitive to external mechanical signal stimulation, and the elastic deformation caused by cyclic mechanical strain on titanium specimen can promote the osteogenic differentiation of BMSCs.

Previous literature reports that high levels of Wnt signaling, an important signaling pathway that regulates osteogenic differentiation, can be activated in osteogenic differentiation of BMSCs and promote bone formation (Wang et al., 2017; Zhu et al., 2019). In osteoporotic mice, the level of Wnt ligand protein was reduced, and the activity of the entire pathway was decreased (Karadeniz et al., 2020). Therefore, the Wnt/β-catenin signaling pathway is a critical signaling pathway in osteogenic differentiation of BMSCs. Wnt acted as an activator of the entire signaling pathway and bound to the membrane surface ligand complex to activate the pathway.

Wnt signaling pathway activity was observed under mechanical strain and nonmechanical strain. Western blot and real-time PCR were used to detect key molecules in the Wnt/β-catenin signaling pathway: Wnt1, Wnt6, Wnt10a, β-catenin, phosphorylated β-catenin [phosphorylated β-catenin could be regarded as inactive because it was about to be degraded (Zhu et al., 2019)], and Axin-2 [a defined target in the Wnt/β-catenin signaling pathway (Zhu et al., 2019)]. Western blot revealed that the level of phosphorylated β-catenin was lower and Axin-2 was higher in the 0.9% group after 7 days of exposure to mechanical strain. The level of β-catenin was stable simultaneously (Figures 3E,F). Real-time PCR indicated that the transcription of Wnt1, Wnt6, Wnt10a, and Axin-2 in the 0.9% group was significantly increased compared with that of the control group (Figures 3A–D). These results indicate that the Wnt/β-catenin signaling pathway was activated by mechanical strain and played a key role in the osteogenic differentiation of BMSCs.

After applying mechanical strain to the titanium specimen, we detected the expression level of GCN5 in BMSCs through real-time PCR and Western blot. GCN5, one of the important members of the HATs family, played a significant regulatory role in a variety of signaling pathways, especially the Wnt signaling pathway (Li et al., 2016; Karadeniz et al., 2020). As the elastic deformation aggravated, the expression level of GCN5 gradually increased. Correspondingly, H3K9 was the major target site of GCN5 (Karadeniz et al., 2020). H3K9 became H3K9ac after being acetylated. The results showed that H3K9ac had the same trend as GCN5 and elastic strain promoted H3K9 acetylation (Figure 4G). In addition, we performed ChIP-qPCR to demonstrate the direct effects of GCN5 and H3K9ac with the Wnt gene expression. The ChIP analysis showed that the enrichment of GCN5 and H3K9ac in the Wnt1, Wnt6, and Wnt10a promoter regions was increased in the 0.9% group (Figures 4A–F, Supplementary Figure S1 and Supplementary Table S1). These results indicate that mechanical strain can promote the expression of GCN5 and increase the level of acetylation of H3K9 by promoting GCN5 binding to the promoter region of Wnt gene.

To further demonstrate that the increased level of GCN5 caused by mechanical strain can promote osteogenic differentiation, we transfected the empty vector, the GCN5 lentiviral overexpression vector, and the knockdown GCN5 vector into the BMSCs (Supplementary Figure S2 and Supplementary Table S2). Elastic mechanical strain was applied to the titanium specimen of BMSCs which were transfected with the empty vector and the knockdown GCN5 vector. BMSCs in the non-strain group were transfected with the GCN5 lentiviral overexpression vector. We used the ALP staining experiment to detect ALP activity. The expression levels of GCN5 and osteogenic differentiation markers OCN, OPN, and RUNX2 were determined by Western blot and real-time PCR. The results of the ALP staining experiments showed that knockdown of GCN5 significantly inhibited the ALP activity of BMSCs exposed to mechanical strain. However, in the control group without mechanical strain, the overexpression of GCN5 saved partial ALP activity (Figure 5A). The results of Western blot and real-time PCR showed that in the 0.9% group, osteogenesis-related markers were significantly repressed with knockdown of GCN5. At the same time, the overexpression of GCN5 could increase the levels of osteogenesis-related markers in the control group. And the overexpression of GCN5 possessed almost the same bone-promoting ability as mechanical strain (Figures 5B–G). These results indicate that GCN5 is a key node for mechanical strain to promote osteogenic differentiation.

To further verify that high levels of GCN5 caused by mechanical strain can enhance the Wnt pathway, the levels of markers associated with the Wnt pathway were measured at the same time. In the 0.9% group, the levels of phosphorylated β-catenin were significantly increased and Axin-2 was repressed by the knockdown of GCN5 (Figures 6E,F). The real-time PCR assay showed knockdown of GCN5 inhibited the transcription of related Wnt gene and Axin-2 (Figures 6A–D). In the control group without mechanical strain, the overexpression of GCN5 increased the expression level of Axin-2, repressed the expression of phosphorylated β-catenin (Figures 6E,F), and also promoted the transcription of the Wnt gene and Axin-2 (Figures 6A–D). The ChIP-qPCR directly demonstrated that knockdown of GCN5 could inhibit the enrichment of GCN5 and H3K9ac in the promoter regions of Wnt1, Wnt6, and Wnt10a under the circle mechanical strain (Figures 6G–I).

We further verified whether GCN5 which was stimulated by mechanical strain regulated osteogenic differentiation of BMSCs via the Wnt pathway. As an inhibitor of the Wnt pathway, DKK1 binds to membrane receptors and prevents Wnt ligands from binding to membrane surface receptor complexes, thereby inhibiting Wnt signaling pathway activity and osteogenic differentiation of BMSCs (Scott et al., 2013). We assessed ALP activity on BMSCs with DKK1 treatment and non–DKK1 treatment. As a result, DKK1 significantly suppressed ALP activity in the control group as well as in the 0.9% group (Figure 7A). The Western blot and real-time PCR results showed that DKK1 reduced the activity of the Wnt signaling pathway during mechanical strain application. The expression of Axin-2 in the pathway decreased and phosphorylated β-catenin increased due to DKK1 (Figures 7C,E). Not only the activity of the Wnt signaling pathway but also the expression of downstream osteogenic differentiation markers was repressed, including Opn, Ocn, and Runx2 (Figure 7D). However, DKK1 did not significantly affect the levels of GCN5 and H3K9ac in terms of protein expression (Figure 7B). The results of the ChIP analysis demonstrated that DKK1 had no significant effect on the enrichment of GCN5 and H3K9ac in the promoter regions of Wnt1, Wnt6, and Wnt10a. After mechanical strain stimulation, the enrichment of GCN5 and H3K9ac in the Wnt1, Wnt6, and Wnt10a promoter regions was still increased compared to the control group (Figure 8). These results indicate that DKK1 can only affect the downstream Wnt/β-catenin pathway activity but not upstream GCN5. Combined with the abovementioned results of overexpression and knockdown of GCN5, it is indicated that the high level of GCN5 caused by mechanical strain regulates the osteogenic differentiation of BMSCs through the Wnt/β-catenin signaling pathway.