A novel application perspective of the clinical-used drug verapamil on osteoporosis via targeting Txnip

Abstract

Background

RANKL and SCLEROSTIN antibodies have provided a strong effective choice for treating osteoporosis in the past years, which suggested novel molecular target identification and therapeutic strategies development are important for the treatment of osteoporosis. The therapeutic effect of verapamil, a drug previously used for cardiovascular diseases, on diabetes was due to the inhibition of TXNIP expression, which has also been reported as a target in mice osteoporosis. Whether verapamil-inhibited TXNIP expression is related to osteoporosis and how it works on the molecular level is worthy to be explored.

Methods

The polymorphism genotyping analysis was performed on patients with different degrees of osteoporosis. The responsiveness of bone marrow-derived macrophage cells (bone marrow-derived mesenchymal stem cells) to verapamil was evaluated by CCK-8, TRAP staining assay (ALP and AR staining assay), Bone Resorption Assay, and RNA-Sequencing. The expression and cytoplasmic efflux of ChREBP were determined by western blotting and immunofluorescence. Bilateral ovariectomy models were created, rescued by verapamil injection and the effectiveness was evaluated by Micro-CT and Histological analysis.

Results

Here we discovered that rs7211 single nucleotide polymorphism (SNP) of TXNIP is closely associated with increased femur neck bone mineral density (BMD) and decreased osteoporosis rate, suggesting the importance of TXNIP in the development of osteoporosis. Verapamil suppresses Txnip expression, reduces bone turnover rate and thus rescues ovariectomy-induced mice bone loss. Mechanistically, verapamil promoted ChREBP cytoplasmic efflux, regulated Pparγ expression both mediating Txnip-MAPK, NF-$\kappa$ B axis in osteoclasts, and suppressed the ChREBP-Txnip-Bmp2 axis in osteoblasts.

Conclusions

The results of our study show the correlation of rs7211 TXNIP-T allele with Chinese increased femur neck BMD and decreased osteoporosis rate. In addition, verapamil can rescue mice from osteoporosis by regulateing ChREBP, Pparγ-Txnip-MAPK, NF-$\kappa$ B axis in osteoclasts and ChREBP-Txnip-Bmp2 axis in osteoblasts.

The translational potential of this article

The inhibition of Txnip by verapamil in osteoclasts and osteoblasts leads to low bone turnover and reduced bilateral ovariectomy-induced mice bone loss, which points out its great clinical translation potential on postmenopausal osteoporosis treatment.

Graphical abstract

1. Introduction

Bone undergoes a continuous remodeling process of old bone destruction by osteoclasts and new bone formation by osteoblasts [1]. The disruption of this balance will lead to osteoporosis. A great many achievements have been made in the treatment of osteoporosis in the past thirty years [2]. The identification of Receptor Activator of NF-κB (Nuclear Factor-Kappa B) Ligand (RANKL) in regulating osteoclast formation and later the development of RANKL antibody greatly benefited patients with osteoporosis. In addition, the identification of sclerostin in regulating osteoblast formation and later the development of the SCLEROSTIN antibody provided a new choice for the treatment of osteoporosis [3,4]. Therefore, the identification of novel molecular targets and develop new therapeutic strategies are important for the treatment of osteoporosis.

With the aim to identify new molecular targets for the treatment of osteoporosis, we used the human SNPs study and identified TXNIP as closely related to bone mineral density in this study. TXNIP is a molecule important for the regulation of intracellular ROS levels by interacting with TXN [5,6]. In addition, TXNIP also interacts with NLRP3 to regulate IL-1beta expression [7]. Importantly, TXNIP interacts with GLUT4 to regulate glucose metabolism [8]. Therefore, it has been proved that TXNIP is an important molecular target for the treatment of diabetes and TXNIP knockout prevented diabetes development [9]. However, the function of TXNIP in osteoporosis is yet to be further determined.

Identification of new therapeutic indications of ‘old’ drugs becomes a feasible and attractive strategy for drug discovery [10]. Successful cases such as sildenafil (famous as Viagra for erectile dysfunction treatment but originally developed as an antihypertensive drug), thalidomide (prohibited for use after 4 years of authorization owing to its severe side effects while reused on multiple myeloma in 1999 and improved into lenalidomide later) [10,11] suggest the possibility of drug repurposing on osteoporosis. Verapamil is such a unique drug first used in cardiovascular therapeutics via mediating calcium influx [12] and substantially expanded its indications over time from antiarrhythmic to antihypertensive and angina pectoris [13,14]. More attractively, Anath Shalev and her colleagues’ research achievements over twenty years strongly demonstrated the novel effect of verapamil on type 1 diabetes (T1D) by targeting TXNIP across preclinical studies to clinical trials recently [[15], [16], [17], [18]].

Here in this study, we identified the SNPs of TXNIP as closely associated with human bone mineral density. We further aimed to explore the effect of the clinically used drug verapamil for the treatment of osteoporosis and unveil its underlying mechanisms.

2. Material and methods

2.1. Mouse model

The C57BL/6 background of mice was ensured in this research. All animals were specific-pathogen-free (SPF) grade and raised in the Department of Laboratory Animal Science at Shanghai Ninth People’s Hospital. All animals were healthy without any diseases and all animal experimental procedures were conducted in strict accordance and approved by the Institutional Animal Care and Experimental Committee of Shanghai Jiao Tong University School of Medicine (SH9H-2020-A312-1).

2.2. Collection and determination of human hemocytes sample

Venous blood samples collection and divide, DNA samples extraction from hemocytes were described as previous study [19]. Then, the polymorphism genotyping analysis was performed to determine the TXNIP polymorphisms (rs7211, rs7212) by LightCycler 480 real-time PCR system (Roche Life Science, China), and SNaPshot assay (Supported by GENESKY, Shanghai, China).

2.3. Human study approval

The human study was approved by the Medical Ethics Committees of Shanghai Fifth People’s Hospital (Ethical Document No. 133 approved in 2022), and written informed consents were obtained from the participants before venous blood samples collection.

2.4. Cell isolation and culture

The whole bone marrow of 4-week-old and 6-week-old male mice were extracted to obtain primary BMMs and BMSCs. Briefly, cells extracted from the bone marrow of tibial and femoral were suspended in α-MEM (for BMMs: supplemented with 30 ng/ml M-CSF, 10 % FBS, and 1 % penicillin/streptomycin) (R&D Systems, Cat # 416-ML, Cat # 462-TR) (for BMSCs: supplemented with 15 % FBS, and 1 % penicillin/streptomycin) (CST, Cat # 9998; GIBCO, Cat # 12561-056, Cat # 15070-063) and then placed in a humid environment at 37 °C with 5 % CO₂ until they reached 90–95 % confluence.

2.5. Bilateral ovariectomy model and Txnip inhibitor verapamil treatment

Twenty-four 9-week-old female C57BL/6J mice were prepared for prevention experiment. Eighteen of them were casually selected and performed bilateral ovariectomy, which was then divided into three groups randomly for an experiment started 3 days later. PBS group: constant daily intraperitoneal injections of PBS (Hyclone, SH30256.01) for 6 weeks. Low dose verapamil group: 6 weeks daily intraperitoneal injections of verapamil (2.11 mg/kg; APExBIO, B1867). High dose verapamil group: 6 weeks daily intraperitoneal injections of verapamil (4.22 mg/kg). The remaining six mice that underwent the same procedure but without their ovaries removed were considered the sham group.

Thirty-two 9-week-old female C57BL/6J mice were prepared for treatment experiment. Twenty-four of them were casually selected and performed bilateral ovariectomy, which was then divided into three groups randomly for an experiment started six-weeks later. PBS group: constant daily intraperitoneal injections of PBS (Hyclone, SH30256.01) for 6 weeks. Positive control group: tail vein injection of zoledronic acid every 3 weeks (0.5 mg/kg; MCE, 165800-06-6). High verapamil group: 6 weeks daily intraperitoneal injections of verapamil (4.22 mg/kg; APExBIO, B1867). The remaining six mice that underwent the same procedure but without their ovaries removed were considered the sham group.

The post-operative infections and pain of operated mice were prevented by gentamicin (i.m., 10 mg/kg) and tramadol (s.c., 25 mg/kg) for the initial first three days. The Calcein solution (i.p., 8 mg/kg) (Sigma Aldrich, St. Louis, Mo) was intraperitoneal injected 14 days before sacrifice, and Alizarin red (i.p., 20 mg/kg) (Sigma Aldrich) was intraperitoneal injected 7 days before sacrifice. All mice were sacrificed at designed point, and the femurs and tibias were collected. The left tibias were scanned by micro-CT. The right tibias were performed histological experiments. The left femurs were used for bone histomorphometry. All analyses were described later herein.

2.6. Cytotoxic assay

The cytotoxic effects of verapamil on BMMs (BMSCs) were determined by the CCK-8 kit (Dojindo Molecular Technology, Japan). Specifically, the BMMs (BMSCs) were seeded into 96-well plates in triplicate at the density of 8 × 10³ (1.6 × 10⁴) cells/well in presence of 100 μL α-MEM with 30 ng/ml M-CSF, 10 % FBS, and 1 % penicillin/streptomycin (15 % FBS, 1 % penicillin/streptomycin) for 24 h. After that, the cells were treated with 0, 6.25, 12.5, 25, 30, 35, 40, 45, 50, 100 μM (0, 6.25, 12.5, 25, 50, and 100 μM) verapamil for 24, 48, 72 and 96 (12, 48, and 96) hours. After treatment, 100 μL medium containing 10 % CCK-8 buffer was added to the wells and incubated in the dark at 37 °C for 2 h. The absorbance was then detected at 450 nm wavelength (650 nm reference) on a microplate reader.

2.7. Osteoclast differentiation and TRAP staining assay

8 × 10³ cells/well BMMs were seeded in triplicate into the 96-well plate for one day. Afterward, the culture medium was replaced by the α-MEM containing different concentrations of verapamil (0, 6.25, 12.5, 25, and 50 μM) and M-CSF (30 ng/ml) and RANKL (50 ng/ml) to stimulate osteoclast differentiation. The culture medium was replaced every 2 days until the matured osteoclasts could be observed (5–7 days) in the plate. After PBS wash and 15 min 4 % paraformaldehyde fix, TRAP staining was performed immediately at 37 °C until the matured osteoclasts were clearly colored (30–60 min without light). The optical microscope (Olympus, Tokyo, Japan) was used to obtain images and the TRAP-positive cells were quantified the area and number using the ImageJ software (NIH, Bethesda, MD, USA).

2.8. Bone resorption assay

BMMs were seeded at the density of 1 × 10⁴ cells/well, in triplicate, on the Osteo Assay Surface plates (Corning, NY, USA) or bovine bone slices (JoyTech, 2-0002) for 36 h. The culture medium was replaced by α-MEM containing M-CSF (30 ng/ml) and RANKL (50 ng/ml), and various doses of verapamil (0, 6.25, 12.5, 25, and 50 μM). The medium needs to be constantly replaced until the osteoclast has formatted and functioned for 1 day or 7 days. The 5 % sodium hypochlorite were used for 3 min to remove the cells. The total area of resorption was photographed and then calculated by Image J software (NIH, Bethesda, MD, USA).

2.9. Osteoblast differentiation and mineralization

BMSCs were firstly seeded into the 48-well plates for 24 h at a density of 5 × 10⁴ cells/well. The medium was then changed twice a week by Low-glucose DMEM with 15 % FBS, osteogenic stimulation factors (5 mM β-glycerophosphate, 50 μg/mL ascorbic acid, 10⁻⁷mM dexamethasone), and various doses of verapamil (0, 6.25, 12.5, and 25 μM). On day 7 and day 28, the BCIP/NBT kit (Beyotime, Shanghai, China) was used to observe the percent of Alkaline phosphatase (ALP) positive cells and 1 % Alizarin red S solution (Solarbio, Beijing, China) was used to visualize the extracellular matrix calcium deposition.

2.10. RNA-sequencing and analysis

The total RNA of osteoclast treated with or without 25 μM verapamil for 5 days or osteoblast treated with or without 25 μM verapamil for 7 days were extracted by TRIzol (Invitrogen, Cat # 15596018). The RNA-seq was then conducted according to the BGISEQ platform’s standard protocol, which including complementary DNA library preparation and sequencing, raw reads, differential gene expression identification, DEGs analysis, GO analysis, KEGG analysis, and GSEA analysis. Global gene expression volcano plot was visualized using Dr. tom online system (http://biosys.bgi.com). Differentially expressed genes were clustered by k-means clustering using the Euclidean distance as the distance and the volcano plot was reformatted by bioinformatics online website (https://www.bioinformatics.com.cn). The Venn plots were generated by the online website Jvenn [20].

2.11. Immunofluorescent staining

After twice PBS washes, cells cultured in the confocal petri dish were fixed with 4 % paraformaldehyde for 10 min. The Triton X-PBS (0.5 %) penetration was added for 5 min after another PBS wash. Then, the cells were blocked with 10 % goat serum dissolved in 1 × TBST at RT for 1 h. The primary antibodies (ChREBP, Abcam, ab253202, 1:100) were next added overnight at 4 °C. The fluorescent secondary antibodies were incubated with the cells on the next day at RT for 1 h. The nuclei were stained in the dark by DAPI at RT for 5 min. After the last PBS wash, the results were observed and photographed by the fluorescence microscope (Leica, Germany). The nuclei localization of proteins were calculated by Image J software (NIH, Bethesda, MD, USA).

2.12. Quantitative real-time PCR analysis

To explore the relative gene expression, total RNA was extracted at indicated time points according to the experiment arrangement by the Axygen RNA Miniprep Kit (Axygen, Cat # AP-MN-RNA, Union City, CA, USA). The bone tissues were soaked in TRIzol at −80 °C overnight, and then mixed with nucleic acid extract = 24:1 (Acmec, AP1014-100 ml) at the ratio of 4:1 for 5mins at RT. After adequate grinding and 10 min centrifugation at 12000×g, the supernatant was 1:1 mixed with isopropanol for 5 min and then follow the experiment steps by the Axygen RNA Miniprep Kit (Axygen, Cat # AP-MN-RNA, Union City, CA, USA). The cDNA was then obtained from the RNA template by using a PrimeScript RT Master Mix kit to reverse transcription (Takara, Cat # RR036A). The primes and samples were prepared according to the instruction of the TBGreen ® Premix Ex Taq ™ kit (Takara, Cat # RR420A) and real-time PCR process (40 cycles: 95 °C for 5 s plus 60 °C for 30 s) was run on ABI 7500 sequencing detection system (Applied Biosystems, Foster City, CA). Specifically, 5 μl TBGreen, 0.2 ul Rox, 0.2 μl forward, 0.2 μl reverse primer, and 3.4 μl ddH2O with 1 μl diluted cDNA were mixed to be the 10 μl total volume liquid for each hole and added into the 384 PCR plate. The specificity of amplification products was judged by melting curve and reverse transcription PCR (RT-PCR). The quantification of each target was normalized to the Actin beta (Actb). The primer sequences are listed in Table S2.

2.13. Western blotting analysis

At designated time points, proteins were extracted using radioimmunoassay (RIPA) lysis buffer (Beyotime, Cat #P0013C, Shanghai, China) plus a mixture of protease and phosphorylase inhibitor cocktail (A + B) (Abmole, Cat #M5293, Cat #M7528). After 13000×g centrifuging for 15 min, the concentration of proteins in the supernatant was detected by Pierce ™ Bicinchoninic Acid (BCA) protein quantitative Kit (Thermo Fisher, Cat # 23225). Then, 5 × SDS-sample loading buffer was used to dilute the supernatant protein and heated at 95° for 10 min. After 4–20 % SDS-PAGE separation and nitrocellulose filter membrane transferration (GE Healthcare Life Sciences, Pittsburgh, PA, USA), the membrane was soaked in 5 % skim milk dissolved in 1 × TBST (Tris-buffered saline with Tween 20) at room temperature for 1 h blocking. The primary antibodies (NFATc1, Santa, sc-7294, 1:1000; c-Fos, CST, 31254, 1:1000; Ctsk, Abcam, ab37259, 1:1000; Txnip, CST, 14715S, 1:1000; ChREBP, Abcam, ab253202, 1:1000; Ppar $\mathrm{\gamma }$, CST, 2435S, 1:1000; phospho-P65, CST, 3033, 1:1000; P65, CST, 8242, 1:1000; phospho-IKK $\alpha$/ $\beta$, CST, 2697, 1:500; IKK $\beta$, CST, 8943, 1:1000; phospho-I $\kappa$ B $\alpha$, CST, 2859, 1:500; I $\kappa$ B $\alpha$, CST, 4814, 1:1000; phospho-JNK, CST, 4688, 1:1000; JNK, CST, 9252, 1:1000; phospho-ERK, CST, 4370, 1:1000; ERK, CST, 4695, 1:1000; phospho-P38, CST, 4511, 1:1000; P38, CST, 8690, 1:1000; BMP2, Abcam, ab214821, 1:1000; Lamin B1, ABWAY, AB0054, 1:1000; $\alpha$-Tubulin, CST, 5335, 1:1000; ACTB, CST, 8457, 1:1000) were next incubated with membranes at 4 °C overnight. Next day, the secondary fluorescence antibody was incubated in the dark at room temperature for 1 h after 3 times TBST wash. After another 3 times TBST wash, the Odyssey v3.0 image scanning was used to develop the result of reactivity (Li Cor. Inc., Lincoln, NE, USA).

2.14. Micro-Computed Tomography scanning

The tibias were dissected from mice and fixed under room temperature (RT) with 4 % PFA for 48 h. Then stored in 75 % ethanol after washing with 1 × PBS. When scanning in the Skyscan 1176 μCT machine (Bruker®, Belgium), the conditions were set as 50 kV, 500 μA, 9 μm resolution, 0.5 mm aluminum filter to acquire 10 mm bone in length for each sample. Use the Skyscan Nrecon program to reconstruct the scanned 3D image. For another high-resolution micro-CT machine (μCT-100, SCANCO Medical AG, Switzerland). Set the conditions to 75 % ethanol environment, 200uA X-ray intensity, 70 kVp X-ray tube potential, and 300 ms integration time. The bone trabecular threshold is 211 parts per thousand. The bone microstructure indicators (BV/TV; Tb. N; Tb. Th; Tb. SP; Conn. D) were calculated by evaluating and analyzing three-dimensional regions of interest (ROI) (starting from the first horizontal segment 0.5 mm away from the proximal growth plate of the tibia, measure the contour of the trabecular bone area downwards by 1.5 mm) by software (version: 6.5–3, SCANCO Medical AG, Switzerland).

2.15. Bone histomorphometry and histological analysis

According to the experimental requirements, the mice were euthanized, the femurs and tibias were separated, and fixed under RT with 4 % PFA for 48 h. After washing overnight with flowing water, femurs were embedded in methyl methacrylate (MMA) resin. Cut the slices using a Leica RM2255 slicing machine (Leica, Germany), 5 μM thick pieces were then used for Von Kossa staining and double fluorescence staining. The bone tissue morphology analysis conducted by Bioquant Osteo Software (Bioquant, USA) includes regions of interest (ROI) (all trabecular bones 0.5 mm away from the proximal growth plate of the tibia without cortical bone). As for tibias, the 10 % EDTA was used to decalcify the fixed tibias for 14–21 days until the bone tissue could easily punctured by a syringe needle. Then the tissues were processed by dehydrating in different concentrations of ethanol, infiltrating with xylene, and embedding with paraffin. Hematoxylin and eosin (H&E) and TRAP staining were then performed on these prepared histological sections. The specimens were observed and photographed under a high-quality microscope (Leica DM4000B). Each sample’s number of osteoclasts and osteoclasts surface per bone surface were calculated.

2.16. Statistical analysis

The n used in whether mice per group or human specimens indicates the number of biologically independent samples. All the cell experiments and animal sample sizes were predetermined to reach at least 3 biological repeats without prior power calcs and any exclusion criteria. Histological analyses and cell experiment result analyses were done in a blinded fashion and the statistical results were calculated by Prism 8 (GraphPad Software Inc, San Diego, CA, USA). The data were presented mainly in two forms: mean $\pm$ SD, and violin plots with all points. The inter-group variances of most evaluation parameters were similar and the Kolmogorov–Smirnov test was used to determine the normality of the data. For the two groups’ comparisons, significance was analyzed by the 2-tailed, unpaired Student’s t-test. When analyzing data among more than two groups, One-way analysis of variance (ANOVA) with turkey’s post-hoc-test was used. Two-way analysis of variance (ANOVA) with sidak’s post hoc test was used. The standard of statistically significant (ns, not significant; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001).

3. Results

3.1. TXNIP polymorphisms are associated with human osteoporosis in Chinese cohort

TXNIP has been reported as an important therapeutic target for the treatment of diabetes. So it is interesting to investigate whether TXNIP plays a role in other degenerative diseases such as osteoporosis. The expression-related polymorphisms (rs7211, rs7212) of TXNIP had been reported to be associated with metabolic syndrome [21], such as arterial stiffness, blood glucose, and blood pressure in a Brazilian general population [22,23] and coronary artery disease in a Chinese Han cohort [24]. We questioned whether TXNIP polymorphisms was associated with human osteoporosis or bone mineral density (BMD). The association of TXNIP polymorphisms (rs7211, rs7212) with osteoporosis of 1305 Chinese patients were then performed (Table S1). Both copies of the rs7211 TXNIP-T allele were associated with increased femur neck BMD and decreased osteoporosis rate in our Chinese cohort (Table 1).

Table 1.

Association between the TXNIP polymorphisms (rs7211, rs7212) and BMD or osteoporosis in the Chinese cohort, analyzed by logistic regression.

Sample (N = 1305)	rs7211			rs7212
	CC (N = 946)	CT (N = 324)	TT (N = 35)	CC (N = 951)	GC (N = 354)	GG (N = 0)
Age, year	63.58 ± 9.10	64.07 ± 8.87	61.54 ± 7.96	63.56 ± 9.10	63.88 ± 8.82	/
BMI, kg/m2	23.85 ± 3.18	23.79 ± 3.07	24.17 ± 3.04	23.85 ± 3.18	23.85 ± 3.05	/
Lumbar BMD, g/cm3	1.054 ± 0.200	1.048 ± 0.206	1.058 ± 0.180	1.054 ± 0.200	1.049 ± 0.203	/
Total Hip BMD, g/cm3	0.890 ± 0.145	0.877 ± 0.146	0.917 ± 0.126	0.889 ± 0.144	0.881 ± 0.144	/
Femoral Neck BMD, g/cm3	0.818 ± 0.133	0.803 ± 0.137	0.849 ± 0.133a	0.818 ± 0.133	0.808 ± 0.137	/
Wards Triangle BMD, g/cm3	0.631 ± 0.144	0.617 ± 0.144	0.653 ± 0.132	0.631 ± 0.144	0.620 ± 0.143	/
Osteoporosis Rate	18.90 %	20.70 %	11.4 %a (p = 0.027)	19.00 %	19.50 %	/

^ap < 0.05.

3.2. Verapamil intervention inhibits osteoclast Txnip expression, differentiation and bone resorption

After identifying the safe concentration without cytotoxicity effect on primary bone marrow-derived macrophages (BMMs) (Fig. S1A), verapamil ranging from 0 to 50 $\mu$ M was used for further exploration. The mRNA level and protein expression of Txnip were inhibited by verapamil in a time- and concentration-dependent manner during osteoclastogenesis (Fig. 1A–C). Compared with the control group, BMMs cultured with verapamil showed a remarkable suppression of multinucleated trap-positive osteoclast formation even at a dose of 6.25 μM. Verapamil treatment also decreased bone resorption. More than 80 % of bone resorption was suppressed starting from 12.5 μM compared with the control group (Fig. 1D–G, Figs. S1B and S1C). To identify the change in mRNA level, the total RNA of osteoclast treated with or without 25 μM verapamil for 5 days was extracted and the RNA-seq was conducted. The results showed 1370 genes down-regulated >2-fold and 421 genes up-regulated >2-fold in the verapamil intervention group. The expression of osteoclast-specific genes, such as c-Fos, Ctsk, Acp5, Calcr were downregulated after verapamil treatment (Fig. 1H). The qPCR experiments validated that the expression of osteoclast-specific genes, such as Nfatc1, c-Fos, Ctsk, Acp5, and Dcstamp were inhibited in a time- and concentration-dependent manner (Figs. S1D and S1E). Western blotting results for Nfatc1, c-Fos, and Ctsk expression further supported the conclusion (Fig. 1I). Together, these data confirmed the inhibition effect on Txnip expression and osteoclastogenesis after verapamil intervention.

Fig. 1.

Verapamil suppresses osteoclast Txnip expression, differentiation and bone resorption.

(A) Relative Txnip mRNA level of osteoclast in a time manner (with or without 25 μM verapamil for 0, 1, 3, 5 days) and dose manner (with 0, 12.5, 25 μM verapamil for 5 days). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(B) Txnip protein expression of osteoclast treated with or without 25 μM verapamil for 0, 1, 3, 5 days.

(C) Txnip protein expression of osteoclast treated with 0, 12.5, 25 μM verapamil for 5 days.

(D) Representative tartrate-resistant acid phosphatase (TRAP) staining images show the differentiation of negative control or different concentration verapamil-treated BMMs with 5–7 days stimulated by M-csf (30 ng/ml) and Rankl (50 ng/ml). Scale bar = 50 μm.

(E) Representative bovine bone slices images show the bone resorption function of negative control or different concentration verapamil-treated BMMs with 6–8 days stimulated by M-csf (30 ng/ml) and Rankl (50 ng/ml). Scale bar = 25 μm.

(F) Quantification of TRAP ⁺ cells per cm² and TRAP ⁺ area ratio of (D). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(G) Quantification of bone resorption area of total area of (E). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(H) volcano plot of differential genes of osteoclast treated with or without 25 μM verapamil for 5 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(I) Nfatc1, c-Fos, and Ctsk protein expression of osteoclast t treated with 0, 12.5, 25 μM verapamil for 5 days and the relative grey level of them. (n = 3 per group, 1 technical replicate of 3 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 by One-way ANOVA (A, F, G, I) or by Two-way ANOVA (A)).

3.3. Verapamil intervention inhibits osteoblast Txnip expression, differentiation and mineralization

The cell viability on primary bone marrow-derived mesenchymal stromal cells (BMSCs) (Fig. S2A) after verapamil intervention was also detected, and the safe doses of verapamil ranged from 0 to 25 $\mu$ M was used. Verapamil suppressed the expression of Txnip in a time- and concentration-dependent manner (Fig. 2A–C). Compared with the control group, BMSCs cultured with osteogenic cultured medium and verapamil showed impaired osteoblast differentiation starting from a dose of 12.5 μM. The influence of verapamil on osteoblast mineralization was further performed and a similar inhibition started from 12.5 μM was detected. More than 70 % of bone mineralization was suppressed at 25 μM compared with the control group (Fig. 2D–G).To identify the change in mRNA level, the total RNA of osteoblast treated with or without 25 μM verapamil for 7 days was extracted and the RNA-seq was performed. The Gene Ontology (GO) enrichment analysis results showed significant enrichment of biological processes on “osteoblast differentiation”, “bone mineralization”, and “extracellular matrix organization” (Fig. 2H). Consistent downregulation of osteogenic genes such as Sp7, Bglap, Col1a1, Bmp2, and Dmp1 was observed on differential expression genes (DEGs) analysis (Fig. 2I). Osteogenic genes, such as Alpl, Bglap, Col1a1, and Spp1 were inhibited in a time- and concentration-dependent manner after verapamil treatment (Figs. S2B and S2C). We identified the suppressive effect of verapamil on Txnip expression and osteogenesis.

Fig. 2.

Verapamil suppresses osteoblast Txnip expression, differentiation and mineralization.

(A) Relative Txnip mRNA level of osteoblast in a time manner (with or without 25 μM verapamil for 0, 3, 7 days) and dose manner (with 0, 12.5, 25 μM verapamil for 7 days). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(B) Txnip protein expression of osteoblast treated with or without 25 μM verapamil for 0, 3, 7 days.

(C) Txnip protein expression of osteoblast treated with 0, 12.5, 25 μM verapamil for 7 days.

(D) Representative Alkaline phosphatase (ALP) staining shows the differentiation of negative control or different concentration verapamil treated BMSCs after 7 days of osteogenic medium induction. Scale bar = 10 μm.

(E) Representative Alizarin red (AR) staining shows the mineralization of negative control or different concentration verapamil treated BMSCs after 21 days of osteogenic medium induction. Scale bar = 10 μm.

(F) Quantification of ALP⁺ area ratio and ALP⁺ area of Osteoblasts of (D). (n = 4 per group, 1 technical replicate of 4 biological replicates).

(G) Quantification of AR⁺ area ratio and AR⁺ area of Osteoblasts of (E). (n = 4 per group, 1 technical replicate of 4 biological replicates).

(H) GO enrichment analysis of osteoblast treated with or without 25 μM verapamil for 7 days. (n = 3 per group, 1 technical replicate of 3 biological replicates). (I) volcano plot of differential genes of osteoblast treated with or without 25 μM verapamil for 7 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 by One-way ANOVA (A, F, G) or by Two-way ANOVA (A)).

3.4. Verapamil mediates ChREBP, Ppar $\gamma$-Txnip-MAPK, NF-$\kappa$ B axis in osteoclast and ChREBP-Txnip-Bmp2 axis in osteoblast

The ChREBP cytoplasmic efflux promotive effect of verapamil was demonstrated to be the reason for Txnip transcription downregulation in pancreatic islet $\beta$-cell [16]. Our further exploration was started by identifying whether this molecular mechanism also exists in osteoclasts. Our Western Blotting and immunofluorescence staining results showed that ChREBP cytoplasmic efflux inhibited by osteoclast stimulation factors was promoted by verapamil treatment (Fig. 3A–D). Moreover, Txnip transcription factors predicted by the Genecard website were further intersected with the DEGs found by our previous osteoclastic RNA-seq, which screened 6 specific Txnip transcription factors (Ppar $\gamma$, Vdr, E2f8, Ahr, Hes1, Hopx) change significantly after verapamil intervention in osteoclast (Fig. 3E). The heatmap displayed the highest expression change of Ppar $\gamma$, Txnip transcriptional suppression factor, among these 6 genes (Fig. 3F). The upregulated mRNA level and protein expression of Ppar $\mathrm{\gamma }$ after verapamil treatment were next validated in vitro (Fig. 3G–K). On the whole, our data demonstrated the combined suppression effect of ChREBP and Ppar $\gamma$ on Txnip transcription in osteoclasts.

Fig. 3.

Combined suppression effect of ChREBP and Ppar$\mathit{\gamma }$on Txnip transcription by verapamil treatment in osteoclasts.

(A) ChREBP protein expression in cytoplasm and nucleus of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 30 min.

(B) Relative grey level of nuclear ChREBP of (A). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(C) Representative immunofluorescent staining images show that ChREBP protein cytoplasmic efflux of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 30 min. Scale bar = 15 μm.

(D) Qualitative evaluation of colocalization of ChREBP and Nucleus of (C). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(E) Venn plot of differential gen es found by osteoclastic RNA-seq and Txnip transcription factors predicted by the Genecard website. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(F) Heatmap displayed the expression change of 6 specific Txnip transcription factors (Ppar$\gamma$, Vdr, E2f8, Ahr, Hes1, Hopx). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(G) Relative Ppar$\gamma$ mRNA level of osteoclast in a time manner (with or without 25 μM verapamil for 0, 1, 3, 5 days) and dose manner (with 0, 12.5, 25 μM verapamil for 5 days). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(H) Ppar $\mathrm{\gamma }$ protein expression of osteoclast treated with or without 25 μM verapamil for 0, 1, 3, 5 days.

(I) Ppar $\mathrm{\gamma }$ protein expression of osteoclast treated with 0, 12.5, 25 μM verapamil for 5 days.

(J) Relative grey level of (H). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(K) Relative grey level of (I). (n = 3 per group, 1 technical replicate of 3 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001 by One-way ANOVA (B, G, K) or by Two-way ANOVA (G, J)).

To search downstream response signaling pathways after Txnip inhibition in osteoclasts, the GO enrichment analysis was conducted and the results showed significant enrichment of biological processes on “inflammatory response”, “chemokine-mediated signaling pathway”, and “positive regulation of ERK1 and ERK2 cascade” (Fig. 4A). More specifically, the Gene Set Enrichment Analysis (GSEA) analysis showed multiple inflammation-related signaling pathways like MAPK, Toll-like receptor, and TNF signaling pathway were all downregulated in the verapamil-treated group (Fig. 4B–D). We therefore detected the protein expression of nuclear factor kappa-B (NF-$\kappa$ B) and MAPK signaling pathways. The transduction factors of classical NF-$\kappa$ B signaling pathway, such as the inhibitor of NF-κB kinase (IKK), inhibitor of NF-$\kappa$ B (I $\kappa$ B), and NF-$\kappa$ B were mainly detected by Western Blotting. The phosphorylation of IKK, I $\kappa$ B, and P65 was inhibited over time with verapamil treatment (Fig. 4E). The decreased nuclear translocation of phosphorylated P65 was further demonstrated by both Western Blotting and immunofluorescence staining experiments (Fig. 4F–H, Figs. S3A and S3B). Meanwhile, the phosphorylation of the MAPK signaling pathway including phosphorylated ERK, JNK, and P38 was also inhibited by verapamil compared with the control group (Fig. 4I, and Figs. S3C–S3E). Collectively, these results indicated the verapamil mediated ChREBP, Ppar $\gamma$-Txnip-MAPK, NF-$\kappa$ B axis in osteoclast.

Fig. 4.

Txnip downregulation suppresses both NF-$\kappa$ B and MAPK signaling pathway in osteoclasts.

(A) GO enrichment analysis of osteoclast treated with or without 25 μM verapamil for 5 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(B-D) MAPK, Toll-like receptor, and TNF signaling pathway by GSEA analysis of osteoclast treated with or without 25 μM verapamil for 5 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(E) NF-$\kappa$ B signaling pathway protein expressions of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 0, 10, 20, 30, 60 min.

(F) The p-P65 protein expression in cytoplasm and nucleus of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 30 min.

(G) Relative grey level of nuclear p-P65 of (F). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(H) Representative immunofluorescent staining images show that p-P65 protein nuclear transportation of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 30 min. Scale bar = 15 μm.

(I) MAPK signaling pathway protein expressions of BMMs pretreated with or without 25 μM verapamil for 4 h and added with Rankl for 0, 10, 20, 30, 60 min.

Data are expressed as mean $\pm$ SD (∗p < 0.05; ∗∗∗∗p < 0.0001 by One-way ANOVA (G)).

In osteoblasts, the Txnip transcription factors predicted by the Genecard website were also intersected with the DEGs of our previous osteoblastic RNA-seq to find a novel transcriptional regulation mechanism of verapamil on Txnip. However, no specific Txnip transcription factor was screened (Fig. S4A). We next examined whether the change of ChREBP cytoplasmic efflux also exists in osteoblasts. The Western Blotting and immunofluorescence staining results showed that ChREBP cytoplasmic efflux inhibited by osteogenic induction medium was also increased after verapamil treatment (Fig. 5A and B, Fig. S4B). To explore important signaling pathways mediated osteoblast differentiation and mineralization suppression after Txnip inhibition, the expression of DEGs enriched in the “osteoblast differentiation” and “bone mineralization” by GO analysis was displayed as a heatmap. The Bmp2 level was significantly decreased in the verapamil treatment group (Fig. 5C). Consistently, the Tgf-$\beta$ signaling pathway was downregulated in the verapamil-treated group as shown by GSEA analysis (Fig. 5D). The downregulated mRNA level and protein expression of Bmp2 in a time-dependent and dose-dependent manner after verapamil treatment were next determined in vitro (Fig. 5E–I). Moreover, the decreased Bmp2 expression was rescued after Txnip overexpression (Fig. S4C). In a word, our experiments demonstrated the verapamil-mediated ChREBP-Txnip-Bmp2 axis in osteoblast.

Fig. 5.

ChREBP- Txnip- Bmp2 axis in osteoblasts.

(A) Representative immunofluorescent staining images show that ChREBP protein cytoplasmic efflux of BMSCs pretreated with or without 25 μM verapamil for 4 h and added with osteogenic induction medium for 2 h. Scale bar = 15 μm.

(B) Qualitative evaluation of colocalization of ChREBP and Nucleus of (C). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(C) Heatmap of DEGs enriched in the “osteoblast differentiation” and “bone mineralization” in osteoblastic RNA-seq. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(D) Tgf-$\beta$ signaling pathway by GSEA analysis of osteoblast treated with or without 25 μM verapamil for 7 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(E) Relative Bmp2 mRNA level of osteoblast in a time manner (with or without 25 μM verapamil for 0, 3, 7 days) and dose manner (with 0, 12.5, 25 μM verapamil for 7 days). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(F) Bmp2 protein expression of osteoblast treated with or without 25 μM verapamil for 0, 3, 7 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(G) Bmp2 protein expression of osteoblast treated with 0, 12.5, 25 μM verapamil for 7 days. (n = 3 per group, 1 technical replicate of 3 biological replicates).

(H) Relative Bmp2 grey level of (F). (n = 3 per group, 1 technical replicate of 3 biological replicates).

(I) Relative Bmp2 grey level of (G). (n = 3 per group, 1 technical replicate of 3 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001 by One-way ANOVA (F, J) or by Two-way ANOVA (E, I)).

3.5. Verapamil prevents and treats mice from bilateral ovariectomy (OVX)-induced bone loss

To investigate the effect of pathological bone loss by verapamil in vivo, the OVX mice model was selected. The effect was first evaluated by treating mice directly after surgery (Fig. 6A). Micro-CT scanning, hematoxylin-eosin (H&E) staining, and bone parameter analyses (BV/TV; Tb.N; Tb.Th; Tb.Sp) demonstrated that the trabecular bone was significantly maintained both in the low dose and high dose-treated verapamil groups compared with the PBS treated group (Fig. 6B–F). The therapeutic effect was next evaluated by treating mice six weeks later after surgery (Fig. 7A). Micro-CT scanning, Von Kossa staining and bone parameter analyses (BMD; BV/TV; Tb.N; Tb.Th; Tb.Sp) showed a significantly increased bone mass in the positive control group (zoledronic acid) and a moderate bone mass enhancement in the verapamil-treated group compared with the PBS treated group (Fig. 7B–E, Fig. S4D). Although the bone formation remains equal to the PBS and positive control groups, the verapamil treatment decreased osteoclast resorption closely to the positive control group, which might be a more important reason for bone loss treatment (Fig. 7F–I). In summary, our research results indicate that the commonly used clinical drug verapamil has preventive and therapeutic effects on pathological osteoporosis mice by targeting Txnip.

Fig. 6.

Verapamil prevents the OVX-induced mice bone mass loss.

(A) Schematic diagram of the prevention experiment of verapamil on OVX-induced mice bone loss.

(B) Representative three-dimensional reconstructed micro-CT images and hematoxylin-eosin (H&E) staining showing the cortical and trabecular from sham group mice, OVX group mice with pbs, low dose verapamil, or high dose verapamil-treated for six weeks.

(C-F) Quantification of the trabecular bone parameters including bone volume per tissue volume (BV/TV), trabecular number, trabecular thickness, and trabecular spacing. (n = 6 per group, 1 technical replicate of 6 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001 by One-way ANOVA (C-F)).

Fig. 7.

Verapamil treats the OVX-induced mice bone mass loss.

(A) Schematic diagram of the therapeutic experiment of verapamil on OVX-induced mice bone loss.

(B) Representative three-dimensional reconstructed micro-CT images showing the cortical and trabecular from sham group mice, OVX group mice with pbs, zoledronic acid, or high dose verapamil-treated for six weeks.

(C) Quantification of the trabecular bone parameters including bone volume per tissue volume (BV/TV), trabecular number, trabecular thickness, and trabecular spacing. (n = 8 per group, 1 technical replicate of 8 biological replicates).

(D) Representative Von Kossa staining images from sham group mice, OVX group mice with pbs, zoledronic acid, or high dose verapamil-treated for six weeks. Scale bar = 20 μm, 25 μm, and 10 μm.

(E) Quantification of bone volume per tissue volume (BV/TV). (n = 8 per group, 1 technical replicate of 8 biological replicates).

(F) Representative images of bone growth rates as determined by Calcein and Alizarin red labeling from sham group mice, OVX group mice with pbs, zoledronic acid, or high dose verapamil-treated for six weeks. Scale bar = 20 μm, 25 μm, and 10 μm.

(G) Quantification of mineral apposition rate (MAR). (n = 7 per group, 1 technical replicate of 7 biological replicates).

(H) Representative tartrate-resistant acid phosphatase (TRAP) staining of tibias from sham group mice, OVX group mice with pbs, zoledronic acid, or high dose verapamil-treated for six weeks. Scale bar = 2.5 μm.

(I) Quantification of the number of osteoclasts per bone surface. (n = 8 per group, 1 technical replicate of 8 biological replicates).

Data are expressed as mean $\pm$ SD (ns, no significance; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001 by One-way ANOVA (C, E, G, I)).

4. Discussion

Osteoporosis causes a substantial socio-economic burden and a decrease in patient quality of life [25]. A lot of pharmaceutics have been created for its therapy in the past thirty years [2]. Here we identified TXNIP as a potential molecular target for the treatment of osteoporosis. We identified the correlation between TXNIP and human osteoporosis. Although the expression change following the rs7211 SNPs of TXNIP was still debatable in comparison to the enhanced expression after the rs7212 TXNIP SNP, its ambivalent role between causing hypertriglyceridemia in diabetes mellitus and keeping lower fasting glucose levels in nondiabetic subjects [21] and acting as a protective factor against obesity in nondiabetic subjects and women was of great interest [26]. The influence of TXNIP polymorphisms (rs7211, rs7212) on both diabetes and hypertension in a Brazilian general population [22,23] and on coronary artery disease in a Chinese Han cohort [24] implied its role in metabolic diseases. Here we first demonstrated the correlation of rs7211 TXNIP-T allele with increased femur neck BMD and decreased osteoporosis rate through the SNPs analysis on a Chinese cohort.

The function of verapamil that can inhibit TXNIP expression is further examined for the treatment of osteoporosis. The exploration of verapamil’s effect on bone mass under physiological state began in the 1980s. A study published in 1992 proposed an increase in bone volume and a decrease in bone mass in female rats, but the opposite effect was observed in male rats [27]. When research on mechanical load-induced bone formation flourished in the 2000s, the promoted effect of verapamil on rat bone formation and bone mineral density at non-loaded sites instead of loaded sites [28] and the inhibitory effect of verapamil on the mechanically induced increase in bone formation [29] were reported by researches. As for pathological bone loss, a recent study reported the promoted properties of verapamil on bone mass, microstructure, and mechanical in type 2 diabetes mellitus rats [30]. Although chronic verapamil treatment does not affect reducing oxyradicals in female spontaneously hypertensive rats in both physiological estrous and OVX [31], the effect of verapamil on OVX-induced bone loss is still blank and needs to be further determined. Here in our study, the prevention and therapeutic effects of verapamil on OVX-induced bone loss by downregulated bone turnover were demonstrated by our Micor-CT and bone morphological analysis. Even bone mineral rate decline in the treatment experiment was not significant in the verapamil group compared with the OVX group (extremely low bone formation ability after six weeks of OVX treatment may be the reason for this in vivo phenomenon), the decreased bone resorption ability still contributes to the rescue of verapamil treatment on OVX-induced bone loss.

Inspired by Anath Shalev and colleagues’ results calcium influx blocked by verapamil inhibits the Txnip transcription via increasing ChREBP cytoplasmic efflux in pancreatic islet $\beta$-cells [16]. The ChREBP cytoplasmic efflux increase was detected and demonstrated in our osteoclasts. More interesting, the unique upregulation of Ppar $\gamma$, a transcriptional inhibitor of Txnip [32], was found in verapamil-treated osteoclast by our interactive analysis of RNA-seq results and Txnip transcription factor and further identified in vitro. Three pieces of evidence (verapamil inhibits TNF-$\alpha$ mediated NF-κB signaling activation in inflammatory macrophages [33]; Txnip plays an important role in multiple inflammation signaling pathways [6]; our GO and GSEA analysis showed multiple inflammation-related signaling pathways like MAPK, Toll-like receptor, and TNF signaling pathway were all downregulated in the verapamil-treated group) drives us to further detected the MAPK, NF-κB signaling pathway change after verapamil induced Txnip downregulation. The expression and nuclear transportation decrease of phosphorylation P65 and the downregulation of phosphorylation ERK, JNK, P38 were further demonstrated. In osteoblasts, the upregulated cytoplasmic efflux of ChREBP was also validated. The significant change of specific Txnip transcription factor in osteoblasts was not detected. Considering controversial conclusions, such as verapamil inhibits Bmp2 expression in six adult non-human primates Chacma baboon [34], Txnip upregulates Bmp2 expression in vascular smooth muscle cells (VMSCs) [35], and si-Txnip upregulates bone morphogenetic protein signaling in mice VMSCs [36], exist among different species and cells, the exact effect of verapamil on Bmp2 was further demonstrated to be inhibitory in mice osteoblasts by our experiments. In summary, specific ChREBP, Ppar $\gamma$-Txnip axis in osteoclast and ChREBP-Txnip-Bmp2 axis in osteoblast mediated by verapamil were proven in this paper.

In conclusion, the correlation of rs7211 TXNIP-T allele with Chinese increased femur neck BMD and decreased osteoporosis rate was reported. The ChREBP cytoplasmic efflux was promoted by verapamil treatment in both mice osteoclasts and osteoblasts. The inhibition of Txnip in osteoclasts and osteoblasts leads to low bone turnover and reduced bone loss after verapamil treatment, which points out a great clinical translation potential of verapamil on postmenopausal osteoporosis treatment.

Authors' contributions

J.Z., A.Q., and P.M. conceived and designed the framework; XK.C., K.R., and Y.L. obtained the data; XK.C., K.R., Y.L., P.Z., K.L., L.C., S.F., Q.H., X.Y., and H.Z. analyzed the data; A.Q., J.Z., and XF.C. provided funding support; XK.C., and K.R. drafted the manuscript; A.Q., J.Z., P.M. wrote in part and revised the manuscript, and supervised the studies.

Ethics approval and consent to participate

Animals Ethics approval were received from the Institutional Animal Ethics Review Board of Shanghai Ninth People's Hospital, Shanghai Jiao Tong University School of Medicine (Approval# SH9H-2020-TK194-1). The human study was approved by the Medical Ethics Committees of Shanghai Fifth People’s Hospital (Ethical Document No. 133 approved in 2022), and written informed consents were obtained from the participants before venous blood samples collection.

Consent for publication

Written informed consent for publication was obtained from all participants.

Availability of data and materials

RNA-Seq data in this paper are deposited in the online repository and the GEO accession number is GSE242742 and GSE242744. Any additional information required to reanalyze the data reported in this paper is available from the corresponding author upon request.

Funding

This study was supported by grants from the National Natural Science Foundation of China (Grant No. 92068102; 82372430; 82130073; 82372429; 81972136), Shanghai Leading Talents Program in 2020 (No. 110), and Shanghai Frontiers Science Center of Degeneration and Regeneration in Skeletal System.

Declaration of competing interest

The authors have declared that no competing interests exist.

Appendix A. Supplementary data

The following are the Supplementary data to this article: