Macrophage-derived exosomes modulate wear particle-induced osteolysis via miR-3470b targeting TAB3/NF-κB signaling

Abstract

Aseptic prosthesis loosening (APL) is one of the most prevalent complications associated with arthroplasty. The main cause is the periprosthetic osteolysis induced by wear particles. However, the specific mechanisms of crosstalk between immune cells and osteoclasts/osteoblasts during osteolysis are unclear. In this study, we report the role and mechanism of macrophage-derived exosomes in wear particle-induced osteolysis. The results of exosomes up-taken experiments revealed that osteoblast and mature osteoclasts capture macrophage-derived exosomes (M-Exo). Next-generation sequencing and RT-qPCR on M-Exo revealed that exosomal microRNA miR-3470b was downregulated in wear particle-induced osteolysis. The results of analysis on Luciferase reporter assays/fluorescence in situ hybridization (FISH)/immunofluorescence (IF)/immunohistochemistry (IHC) and co-culture experiments demonstrated that wear particles induced osteoclast differentiation by increasing the expression of NFatc1 via M-Exo miR-3470b targeting TAB3/ NF-κB signaling. We also illustrate that engineered exosomes enriching miR-3470b facilitated to suppressed the osteolysis; the microenvironment enriching with miR-3470b could suppress wear particle-induced osteolysis via inhibition of TAB3/ NF-κB in vivo. In summary, our findings indicate that macrophage-derived exosomes transfer to osteoclasts to induce osteolysis in wear particle-induced APL. Engineering exosomes enriching with miR-3470b might be a novel strategy for the targeting treatment of bone resorption-related diseases.

Graphical abstract

Highlights

• •Crosstalk between macrophages and osteoclasts/osteoblasts is crucial to induced osteolysis.
• •Uptake of macrophage-derived exosomes show greater osteoclast differentiation.
• •Wear-particles cause macrophages to deliver exosomes with relatively low miR-3470b
• •Exosomal miR-3470b can suppress osteolysis by targeting TAB3/NF-κB signaling.

1. Introduction

Joint arthroplasty, one of the most economical and effective methods, has been widely used to treat severe joint damage and end-stage bone and joint diseases attributed to various causal factors [1]. Aseptic prosthesis loosening (APL) is the most common reason for prosthesis failure and revision surgery, with about 10% incidence rate 10 years postoperatively [2,3]. Wear particles from artificial joint components are released at the bone-implant interface and induce a succession of biological responses, causing osteolysis and progression of APL. Macrophages are the main cells that exert the effects of innate immunity (non-specific immunity) in the APL microenvironment [2,4,5]. Macrophages infiltrate the relevant tissues of APL [2,6] with inflammatory mediator secretion [7,8], thereby increasing bone resorption or decreasing bone formation [5,[7], [8], [9]]. However, the regulatory mechanism networks of intracellular molecules causing an imbalance of osteoclast/osteogenesis in the bone microenvironment around the APL are not fully elucidated, and there is no specific prevention and treatment drug for it.

Exosomes and non-coding RNA (ncRNA) are proved to effect critically on the physiological and pathological processes of human cell development and regulation [[10], [11], [12], [13]]. Exosomes are double-layer lipid cystic vesicles with a diameter ranging from 50–150 nm derived from intracellular compartments or from the shedding of plasma membranes, which can be absorbed by biological fluids near the source cell and far away in the human body and cause various phenotypic reactions [14,15]. In addition to their intracellular functions, emerging research has found that ncRNA can be used as intercellular signals to mediate exosomal transport in intercellular communication. The miR-214-3p from osteoclasts-derived cells has been transferred to osteoblasts to suppress osteoblastic bone formation [12]; miR-155 delivered by exosomes potentially treat osteoporosis [13]. In addition, mesenchymal stem cell (MSC)-derived exosomes fixed on the titanium (Ti)-metal surface could promote osseointegration during the adhesion and proliferation of MSCs [16]. However, the relevant effects and mechanisms of macrophage-derived exosomes (M-Exo) and miRNA networks in regulating inflammatory osteolysis remain unclear.

To determine the function of macrophage-derived exosomes in the regulation of osteolysis, we studied the exosomes secreted by macrophages and related miRNAs in the exosomes. In particularly, we evaluated the role of macrophage-derived exosomes containing miRNA-3470b in inflammatory osteolysis and determined the underlying mechanism of the related effects. This study further clarified the mechanism of action that leads to inflammatory osteolysis and APL. This in-depth understanding of the bone microenvironment around the prosthesis may be critical to develop novel targeted therapies for APL.

2. Materials and methods

2.1. Patients and specimens

Samples were obtained from patients (from February 2008 to October 2022) in the First Affiliated Hospital of Sun Yat-Sen University. All experiments were attained the approvals from the Ethics Committee of the First Affiliated Hospital of Sun Yat-Sen University. Periprosthetic membranes were collected from five APL patients (adults, three males and two females) during their revision surgeries of total hip arthroplasty (THA). After the exclusion on peri-prothesis joint infection by ICM-2018 diagnose criteria [17,18], the diagnosis of APL accorded from clinical symptoms and radiographic evidence, also negative microbiological cultures. The diagnosis of APL may be established by common symptoms (local pain, joint dysfunction), no signs of infection (fever, sinus tract, etc.); no infection signs of lab test results; the culture of synovial fluid and periprosthetic membranes are negative; component migration and radiolucent lines at the implant-bone interface [[19], [20], [21]]. The average time between the primary THAs and revision surgeries was 6.1 years. Human synovial tissues of the control group were selected from the femoral head fracture/necrosis patients who received the primary THAs (adults, two males and two females). Patients with immune system involvement, connective tissue diseases, and infections were excluded. We promptly fixed the tissues in 4% paraformaldehyde (PFA) for 24h. Then tissues were embedded in paraffin after the dehydration with ethanol for the next experiments.

2.2. In situ hybridization (ISH) with fluorescent probe on paraffin section

The tissue in paraffins were sliced and digested with proteinase K (20 μg/ml) at 37 °C for 20–30min. After washing with pure water, slices were incubated in pre-hybridization solution for 1h and then the hybridization solution for 8–12h. The hybridization solution was washed off using 1–2 × SSC. The slices were incubated with DAPI solution for 8min in a continuous dark condition; anti-fluorescence quenching mounting tablets were added dropwise to mount the slides after washing. The slices were observed under a microscope and photographed under an upright fluorescence microscope (Nikon).

2.3. Titanium particles (Ti) and reagents

According to previously described methods, the storage solution of 10 mg/ml Titanium particles (Ti, 3.2 ± 2.7 μm) (Alfa Aesar, Ward Hill, MA, USA) was prepared by sterilization and endotoxin removal [2,9]. The endotoxin concentration at a wavelength of 545 nm was determined using a regular Limulus Assistant kit. An endotoxin level of less than 0.1 EU/ml is endotoxin-free. The specific kit (387-ASigma-Aldrich, CA, USA) was used for the tartrate-resistant acid phosphatase (TRAP) staining. The Dulbecco's modified Eagle's medium (DMEM) with Ti particle (100 μg/ml) were prepared for the treatment of wear particle-induced osteolysis model in vitro [22,23].

2.4. Cell culture and osteoclast differentiation

RAW264.7 mouse macrophages were provided by the American Type Culture Collection (ATCC, USA). The Dulbecco's modified Eagle's medium (DMEM) containing 10% FBS were used to culture RAW264.7 cells, until the cells grown to confluence. After the scraping from the bottom of the culture flask, the cells were resuspended in α-MEM medium (containing 10% FBS) and were seeded in 12-well plates at a density of 2 × 10^4 cells/ml. TRAP staining was performed on the 5th day after induction with 30 ng/ml receptor activator of nuclear factor kappa B ligand (RANKL) [23]. TRAP-positive cells are the osteoclasts with more than three nuclei [23,24].

2.5. Exosome isolation, qualification and characterization

The FBS were performed ultracentrifugation at 120,000 g RCF for 18 h (overnight) (SW32 Ti rotor, Beckman Coulter). Then the supernatants were collected as exosome-free FBS [25]. The DMEM containing 10% exosome-free FBS were used to culture RAW264.7 cells, until the cells grown to confluence. According to the previous articles [26,27], the cell supernatant medium was collected in a centrifuge tube, and exosomes were extracted using the commercial exosome extraction kit (ExoQuick-TC) of SBI (System Biosciences). The cell supernatant medium was mixed thoroughly with the appropriate proportion (1:5) of ExoQuick-TC reagent overnight at 4 °C. The mixture was centrifuged at 1500×g for 30min. Aspirate supernatant and spin down residual solution by centrifugation at 1500×g for 5min. Then the exosome pellets were resuspended with PBS for subsequent experiments. Using a transmission electron microscope (TEM), the exosomes' morphology was identified, and their size was measured. Following the manufacturer's instructions, the NanoSight NS300 was used to measure the exosome concentration and particle size.

2.6. Western blotting

According to previously published methods, western blotting was performed with certain protocol [7,28]. The proteins of cells/exosomes were collected, lysed and boiled. After separated using 10–12% SDS-PAGE and then transferred onto a membrane, the protein samples were blocked with 5% defatted milk. The protein bands were observed and analyzed via chemiluminescence imaging system (GE Healthcare) and Image J software, after incubated with a primary antibody and following with secondary antibody. The primary antibodies are as follow: CD9 (SBI, 1:1000), CD81 (SBI, 1:1000), CD63 (SBI, 1:1000), HSP70 (SBI, 1:1000), β-actin (CST, 1:1000), CTSK (Santa Cruz, China, 1:250), NFATc1 (Santa Cruz, China, 1:250), TAB3 (Affinity Biosciences, China, 1:250), P65 (CST, 1:1000), and P–P65 (CST, 1:1000).

2.7. Uptake of exosomes

The macrophages-derived exosomes (M-Exo) were extracted and stained with the specific dye DiI. The fluorescent dye DiI (10 μM, Beyotime) was added to the M-Exo suspensions, then incubated for 20min at room temperature. The stained exosomes then were centrifuged at RCF 100,000 g for 70min, washed twice and resuspended with PBS. Osteoblasts/osteoclasts were added to the stained exosomes for 4h co-cultivation. The nucleus were stained by Hochest/DAPI for 30min. The co-culture systems of osteoclasts were also incubated with the FITC-actin ring for 1h. The samples (including the samples fixed by 4% PFA) were observed under a fluorescence microscope (Leica, Germany). The presence of fluorescence was observed and images were collected for analysis.

2.8. Co-culture assay using Transwell

Approximately 1.2 × 10^5 RAW264.7 were divided equally in a 24-well plate and cultured overnight. Osteoclast formation was induced with RANKL (100 ng/ml) in culture medium (α-MEM) at the panel of the Transwell system for 2 days. Then, according to the previous article described [12], approximately 1 × 10^4 exosomes/1.2 × 10^5 cells were seeded at the other panel of the Transwell system (polycarbonate membrane insert, 0.4 μm pore; Corning Inc.) and cultured in α-MEM in different time according to the experiment requirement. All co-culture experiments were done in complete culture medium with exosome-free FBS.

2.9. TRAP staining

After osteoclast generation, the cells were fixed in 4% PFA for 25min at 4 °C. To determine TRAP + osteoclasts were calculated after TRAP staining (387-A, Sigma-Aldrich) and evaluated under microscope (Leica DM4000B).

2.10. Actin ring formation analysis

Target cells were prepared, washed with PBS, stained with phalloidin (Alexa Fluor 555) for 1h, then stained with DAPI (Sigma-Aldrich) for 20min. The stained samples were visualized under a microscope (BX-63; Olympus) after incubation.

2.11. Osteoblast preparation

After the MC3T3-E1 cells seeded in 24-well plate reached 80–90% confluence in every well, we replaced the α-MEM medium by osteogenic differentiation medium (Cyagen Biosciences, China) to induce osteoblastogenesis or different time according to the experimental requirements. RNA was then extracted to perform RT-qPCR. Co-culture and transfection experiments.

2.12. The RNA Library preparation and sequencing

RNA sequencing was performed by a biotechnology company (Huayin Health, China). After extraction from macrophage-derived exosomes, total RNAs were analyzed for quality and concentration using an Agilent 2200 instrument. Small RNA libraries were established after reverse-transcription PCR amplification. Raw reads from the libraries were collected using the Illumina HiSeq 2500 platform.

2.13. Differential expression analysis of miRNAs and mRNAs

The Exosome RNA Purification Kit (SBI, USA) was used to extract total exosome RNA. The sets of RNA qPCR primer were synthesized by Sangon (Shanghai, China) and RiboBio. The qPCR primers were listed in Table 1, Table 2. The U6-Forward Primer and U6-Reverse Primer for miRNA (ssD0904071006/7, Ribobio Co., Ltd) were used as the standard control primers. Reverse transcription kit and SYBR Green qPCR Kit (AG Biotechnology Co., Ltd, China) were used to measure miR-3470b expression levels in exosomes. We operated the CFX96 System (Bio-Rad, CA, USA) or LightCycle480 II (Roche) with to perform RT-qPCR on the cDNAs and then analyzed the data via the installed software, following the manufacturers’ instructions.

Table 1.

Reference primer sequences for RT-qPCR.

Gene	Forward primer (5′–3′)	Reverse primer (5′–3′)
GAPDH	AGGTCGGTGTGAACGGATTTG	TGTAGACCATGTAGTTGAGGTCA
Ctsk	CTTCCAATACGTGCAGCAGA	TCTTCAGGGCTTTCTCGTTC
NFatc-1	GACCCGGAGTTCGACTTCG	TGACACTAGGGGACACATAACTG
Bglap	TGCTTGTGACGAGCTATCAG	GAGGACAGGGAGGATCAAGT
Runx2	ATGGGACTGTGGTTACCGTCAT	AAGGTGAAACTCTTGCCTCGTC

Table 2.

Reference primer sequences for RT-qPCR.

Gene name	Forward primer(5′–3′)	Reverse primer(5′–3′)	RT primer(5′–3′)
miR-3470b	GCGCGTCACTCTGTAGACCA	CGAATGCACCTGGGCAAG	GTCGTATCCAGTGCAGGGTCCGAGGTATTCGCACTGGATACGACCC

2.14. Target-gene prediction and KEGG/GO pathway enrichment

TargetScan, miRanda, and miRDB were used to predict the miRNA target genes. The DAVID tool was used to identify significant enrichment pathways of miRNA targets. We carried out the functional annotations through David 6.8 Bioinformatics Resources.

2.15. Luciferase reporter assay

Co-transfecting cells with luciferase vectors carrying the wild-type or mutant 3ʹ-UTR of Tab3 and miR-3470b mimics/inhibitor or miR-Control was accomplished using Lipofectamine 2000 (Invitrogen). 48h after transfection, luciferase activity was assessed using the Dual-Luciferase Reporter Assay System E1910 (Promega).

2.16. Animal experiments

Total of 24 C57BL/6 male mice (7-week-old) with a standard diet and average weight of 21 ± 4 g were used for the vivo experiments in this study. Then we constructed the calvarial model of wear particle-induced osteolysis, based on previous studies [[28], [29], [30]]. Three groups of mice (n = 8 per group) were divided at random as follows: Sham, AgomiR-NC + Ti, and AgomiR-3470b + Ti. 1% pentobarbital sodium were injected into the mice intraperitoneal (i.p.) to anesthetize mice before surgery. When the mice got fully anesthetized, 1 cm-long sagittal incision was made at the top of the skull using a scalpel, with periosteum separated from calvarial. Wear particle (Ti, 20 mg) was placed under the periosteum of the exposed cranial seams. The sham group was not treated intraoperatively. The incisions were sutured layer by layer. Mice in the AgomiR-NC + Ti and AgomiR-3470b + Ti groups were injected locally using AgomiR-NC (5nmol/2days)/AgomiR-3470b (5nmol/2days), respectively, for 2 weeks. The injection dosage was based on previously published literature [31,32]. The mice were euthanized on the 14th day postoperatively. Calvarial samples were collected and fixed in 4% paraformaldehyde. The samples were then subjected to micro-CT scanning, immunohistochemistry, and immunofluorescence. The AgomiR-NC and AgomiR-3470b were purchased from Ribobio Co., Ltd.

2.17. Immunofluorescence staining analysis

After fixed in 4% PFA and embedded in paraffin, the calvaria samples or specimens were sectioned using a tissue microtome (Leica, Germany) into 5 μm-thick. First, sections were deparaffinized and rehydrated. Next, the antigens were retrieved using sodium citrate and blocked with 5% bovine albumin. The primary antibodies used in this study are as followed: Ctsk (Santa Cruz, China, 1:250), NFATc1 (Santa Cruz, China, 1:250), TAB3(Abcam, 1:500), P–P65 (CST, 1:1000).

2.18. Micro-computed tomography analysis

Bone mass was assessed using a micro-computed tomography (micro-CT) system (GE Healthcare, USA) as previously reported with certain modifications [23,28]. The fixed calvirial specimens were scanned by micro-CT (8 μm resolution, a current of 80 μA, 80 kV voltage). Microview V2.1.2 and Dragonfly software were used for data analysis on the quantified parameters, including bone volume per tissue volume (BV/TV) and the number/rate of porosity.

2.19. Immunohistochemistry (IHC) analysis

5 mm-thick histological sections of paraffin-embedded specimens were made in the coronal plane for TRAP and hematoxylin and eosin (H&E) staining. All specimens were examined for CD68, NFATc1, SOX2, and TAB3 expression levels. The number of TRAP-positive osteoclasts with multi-nucleus in the calvarium and specimens was assessed. After that, we observed and photographed under an optical microscope (Leica, Germany) and analyzed the datasets via Image-Pro Plus software (version 6.0).

2.20. Biophotonic imaging observation on bone tissue distribution of exosomes

The fluorescence intensity was quantified using the IVIS Lumina III imaging system (PerkinElmer) to assess the tissue distribution of DiD-labeled exosomes. First, 8-week-old female C57BL/6 mice were fed an alfalfa-free diet for 3–5 days. Exosomes were isolated and purified from the macrophages culture medium using cell and labeled with DiD stain. DiD-labeled exosomes (10^9 per mouse) were injected in nail-vein of the mice. The mice were euthanized 16h after the injection and subjected to biophotonic imaging, with identical illumination settings for all image acquisitions.

2.21. Quantification and statistical analysis

GraphPad Prism was used to carry out the statistical analysis. The significant difference was accessed by the t-test and one-way ANOVA where appropriate (*p < 0.05; **p < 0.01; ***p < 0.001).

3. Results

3.1. Characteristics of exosomes derived from macrophages (M-Exo)

Macrophage-derived exosomes (M-Exo) were isolated and characterized using NanoSight NS300 and TEM. The isolated exosomes showed an oval shape and characteristically ranged between 50 and 150 nm [14,15] (Fig. 1A and B). We also used western blotting to confirm the presence of exosomal markers, transmembrane proteins (HSP70, CD9, CD63, CD81) (Fig. 1C). Generally, these results confirm that vesicles are exosomes based on marker expression, morphology, and size.

Fig. 1.

Exosomes were extracted from the supernatant of macrophage with or without wear particle (Ti) treatment for 24 h for analysis.

A Analysis on particle size of macrophage-derived exosomes. B TEM was used to determine the structure of exosomes. C Exosome markers Hsp70, CD63, CD9 and CD81 were determined using western blotting. Ti: titanium particles. D The design of exosome tracking experiments in vitro was depicted in a schematic diagram. E-F Representative images of the uptake of DiI-macrophage-derived exosomes by osteoclasts (E) and osteoblasts (F) 4 h after coculture. G The design of co-culture experiments with the osteoclasts/osteoblasts (upper panels) and macrophages with/without pre-treatment (lower panels). H ∼ I RT-qPCR analysis of the mRNA expression of Runx2, Bglap in osteoblasts, and NFatc-1 and Ctsk in osteoclasts at 24 h after co-culture with the macrophage exposed to GW4869 (CO-GW) or Ti particles (CO–Ti). The expression levels were normalized by the mean value of control group (CO-NC). J Representative biophotonic images showing the fluorescence signal of DiD-labeled exosomes derived from macrophages (M-Exo), which represents the major organs and bone tissue distribution in mice at 8h after intravenous injection. Compared to the negative control group, *p < 0.05; compared to the positive control group. ^#p < 0.05; indicating a statistically significant difference. GW, exosome inhibitor GW4869; Co, co-culture; NC, control group.

3.2. Macrophage-derived exosomes bone targeting and regulation of osteoclasts/osteoblasts functions in vitro

We further investigated whether macrophage-derived exosomes are involved in osteoblast (OB)/osteoclast (OC) differentiation and regulate their functions. First, to test the endocytosis of exosomes, we used macrophage exosomes labeled with a fluorescent membrane stain (DiI) and incubated them with osteoblasts (OBs) and osteoclasts (OCs) for 4h (Fig. 1D). The labeled-exosomes up-taken in OBs/OCs with nuclei (DAPI) or F-actin were observed under a fluorescence microscope after staining. Our results indicate that exosomes derived from macrophages were internalized by OBs (Fig. 1E), and OCs (Fig. 1F). Next, the macrophages were exposed to the GW4869 blocking exosome synthesis and secretion [33,34] (GW group), and to the wear-particle (100 μg/ml Ti in culture medium) (Ti group) as the osteolysis model in vitro [22,23], respectively. OBs were respectively co-cultured with the different groups of exosomes (Co-NC as control, Co-GW, Co–Ti) for 24h (Fig. 1G). RT-qPCR was performed to estimate the mRNA expressions of osteoblast markers (Osteocalcin (Bglap) and Runt-relatedd transcription factor 2 (Runx2)) and osteoclast markers (nuclear factor of activated T-cells, cytoplasmic 1 (NFatc-1) and cathepsin K (Ctsk)) in OCs and OBs (Fig. 1H and I). In Ti group, the level of Bglap and Runx2 were downregulated; the level of NFatc-1 and Ctsk in OCs were upregulated.

To determine whether M-Exo have bone-targeting specificity, C57BL/6 mice were treated with DiD-labeled exosomes isolated and purified from the supernatant of macrophages. Biophoton imaging revealed that an intraosseous fluorescent signal was detected 8h after administration in mice that received macrophage-derived DiD-exosomes, but not in mice that received PBS (Fig. 2J). In summary, Ti-induced M-Exo suppress the osteogenic differentiation of osteoblast and promoted osteoclast differentiation.

Fig. 2.

Prediction and verification of the differential expression of miR-3470b target genes.

A Heat map analysis of differentially expressed miRNAs in exosomes between the groups (NC vs. Ti). B Volcano diagram showing differentially expressed miRNAs in macrophage-derived exosomes from the wear particle-induced (Ti) loosening group and the control group (NC). C KEGG pathway analysis. D Statistical results of GO classification of network genes between groups. E RT-qPCR analysis showed that the expression of miR-3470b in macrophage exosomes (Exo-Ti) in the prosthesis loosening group was higher than that in the control group (Exo-Con). F miR-3470b binding target gene TAB3 was predicted by analyzing multiple databases. G Schematic of the binding sites between miRNAs and their target genes. H Demonstration of the results of the double-luciferase experiment. *p < 0.05, **p < 0.01, indicates a statistical difference. GO: Gene ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes.

3.3. miRNAs transferred by macrophages-derived exosomes influence osteoclasts/osteoblasts

Secreted exosomes and miRNAs can change the gene expression and function of receptor cells. They are not only closely related to immune inflammation and tumor, but also playing a crucial role in regulating bone formation/resorption^20-23. There is increasing evidence that exosomes carry non-coding RNA to mediate the role of osteoclasts, and they are being studied extensively. In the bone microenvironment around the prosthesis, wear particles induce macrophages to produce a series of intracellular molecules, which play a critical role in the occurrence and development of prosthesis loosening through a complex regulatory network. However, the regulatory effect and mechanism of macrophages on osteoblasts/osteoclasts through exosomes and miRNAs of exosomes in prosthesis loosening are still unclear.

Therefore, clarifying the interaction and key mechanisms may be key to inhibiting the occurrence and development of inflammatory osteolysis and APL. Because macrophages are the main immune cells in the microenvironment of APL, next-generation sequencing was performed to analyze the differentially expressed miRNAs of M-Exo between the in vitro APL group (Ti) and the control group (NC) under the criteria|log2 (FC) | > 2 (FC: fold change) and p < 0.005. There were three upregulated and five downregulated miRNAs in M-Exo in the wear particle-induced APL group (Ti) relative to the control group (NC) (Fig. 2A and B).

Next, we performed KEGG and GO analyses on the target gene prediction of miRNAs using Miranda and Targetscan, R language tools. Considering the different roles of miRNAs in bone homeostasis, 10 potential functional pathways were selected from 20 identified pathways (p < 0.05): TGF-β/NF-κB/Ras, sphingolipid/Toll-like receiver/osteoclast differentiation, and AMPK/WNT/mTOR signaling/cytokine-cytokine receptor interaction pathway. In addition, the dataset indicated that the specific target genes were enriched in TGF-β/osteoclast differentiation/Toll-like receptor signaling pathway (p < 0.05) (Fig. 2C and D). Notably, TGF-β expressed in various bone-related cells has been widely proven to modulate osteoblast/osteoclast activities through a variety of targets and pathways, and is related to the inflammatory pathway NF-κB (nuclear factor kappa-light-chain-enhancer of activated B cells). These results suggest that macrophage-derived exosomal miRNAs strongly and widely modulate bone homeostasis through their target genes or potential pathways. Next, we focused on a specific miRNA: miR-3470b, which is newly discovered and rarely reported. The RT-qPCR analysis results confirmed that the miR-3470b expression level in the M-Exo of the APL group (Exo-Ti) was lower than that of the NC control group (Exo-NC), with significant difference (Fig. 2E).

3.4. Macrophage-derived exosomal miR-3470b promoted osteoclast differentiation and inhibited osteoblast differentiation by targeting Tab3

To reveal the molecular mechanism of miR-3470b regulated bone homeostasis, we predicted target genes through regular miRNA prediction software (Diana, Targetscan, miRDB, and miRwalk) (Fig. 2 F) and identified TGF-β-activated kinase 1 (MAP3K7)-binding protein 3 (TAB3), which plays an important role in inflammation, tumors, and osteogenesis/osteoclasts. In addition, according to the bioinformatics analysis results, there is a conserved 8-nt binding site for miR-3470b in Tab3 is in its 3′ UTR (Fig. 2G). Therefore, a dual-luciferase reporter gene system was utilized to ascertain whether miR-3470b targets Tab3 directly. In the experiment on the transfection and cloning of Tab3 3′ UTR gene, both the blank group and NC group were statistically different from the miR group, respectively (*p < 0.05), indicating that mmu-miR-3470b can affect its fluorescence activity by binding to Tab3 3′ UTR gene. When comparing the fluorescence activity between the blank group and NC inhibitor group, the mmu-miR-3470b inhibitor group appeared to be significantly different in statistical analysis (*p < 0.05 and **p < 0.01, respectively), showing that the miR-3470b inhibitor can block miR-3470b and bind the 3 ‘UTR of the Tab3 gene and affect its fluorescence activity. There was no statistically significant difference between the blank group and NC group (p > 0.05), following the transfection and cloning mut Tab3’ UTR plasmid. No statistically significant differences were observed among the miR-3470b inhibitor group, blank group, and NC inhibitor group (p > 0.05). These results indicated that miR-3470b could not affect its fluorescence activity by binding mut-Tab3 ‘UTR gene, and the mutation of the binding site was complete. (Fig. 2H).

3.5. Expression of Tab3 and differentiation of osteoclasts increased as miR-3470b in periprosthetic tissues of APL patients decreased

To illustrate the effect of miR-3470b on regulating bone homeostasis in APL, we performed FISH assay of miR-3470b on interface membrane tissue specimens from the APL and control groups. We observed the expression of miR-3470b in the interface membrane tissue specimens of the control group (con), whereas the expression of miR-3470b increased and TAB3 was up-regulated in the APL group (*p < 0.05) (Fig. 3A and B). TRAP + osteoclast increased in the periprosthetic interface membrane of APL group, with a significant difference (*p < 0.05) (Fig. 3C).

Fig. 3.

Correlation between the differential expression of miR-3470b/TAB3 and osteoclast differentiation in periprosthetic tissues. A Representative images of miR-3470b expression in tissue samples of the control group and prosthesis loosening group. B Immunohistochemical representative images of TAB3 differential expression between Con and APL groups and the comparative analysis of the positive expression regions of TAB3. C TRAP staining representative images of osteoclasts and the comparative analysis of TRAP positive counts between the Con and APL groups. D ∼ F Immunohistochemical representative images of CD68/SOX2/NFATc-1 differential expression between Con and APL groups and the comparative analysis of the positive expression regions of CD68/SOX2/NFATc-1, respectively. G Representative images of immunofluorescence staining combined with TAB3/NFATc1/P65, showing DAPI (blue)/TAB3 (green)/NFATC1 (red)/P65 (yellow) and their overlay in human tissues. Compared with the control group, *p < 0.05 indicates a statistical difference. Control group: Con; Prosthesis loosening group: APL. Image magnification, × 50–200 × ; scale bar, 100–500 μm.

To evaluate osteoclastic bone resorption in the microenvironment around the prosthesis, specimens of the periprosthetic interface membrane tissue were prepared in paraffin sections for further experiments including IHC and IF. The expression of the macrophage marker CD68 and osteoclastogenesis-related proteins was detected using IHC. A few macrophages were observed sporadically in the layers of the synovium and around the vessels in the specimens of the control group (Con), whereas the expression of CD68, osteoclast-related SOX2, and NFATc-1 increased in the APL group (APL) (*p < 0.05, **p < 0.01) (Fig. 3D–F). Furthermore, we performed immunofluorescence staining in human sample tissues from the APL group and control groups, which combined with DAPI (blue)/TAB3 (green)/NFATc1 (red)/P65 (yellow) (Fig. 3G). In summary, these results revealed that macrophages may importantly modulate the progression of osteolysis via exosomal miR-3470b, TAB3/NF-κ-B in APL.

3.6. Macrophage regulates the osteoclastogenesis via exosomal miR-3470b in vitro

Since miR-3470b was the most prevalent miRNA we found in macrophage-derived exosome and has seldom ever been studied in previous studies, we sought to investigate the role it played in wear particle-induced osteolysis. We constructed the engineered exosomes from macrophages via miRNA mimics or inhibitors transfecting. The expression of target miRNA (miR-3470b) in engineered exosomes transfected with miR-3470b inhibitors was significantly lower than that in exosomes transfected with miRNA-NC inhibitors (con-inhibitor) (Fig. 4A). Also, the miR-3470b expression of exosomes in miR-3470b mimics group was significantly higher than that in con-mimics (transfected with miRNA-NC mimics) (Fig. 4B). Also, wear particle (Ti) treatment decreased the expression of miR-3470b in engineered exosomes (Fig. 4A and B).

Fig. 4.

Functional verification of miR-3470b-loaded macrophage exosomes in vitro. A, B RT-qPCR analysis results of miR-3470b expression loaded in engineered macrophage-derived exosome. C, D RT-qPCR analysis results of osteoclast marker NFatc-1 in osteoclast after the co-culture with different engineered macrophage-derived exosomes. E, F TRAP staining and FITC-actin-ring fluorescence staining representative images of osteoclasts of the groups co-cultured with different engineered macrophage-derived exosomes. G, H The TRAP positive osteoclast area ratio (%) statistical analysis chart, compared with negative control group (con-inhibitor-(exo)/con-mimic-(exo)), *p < 0.05, **p < 0.01; Compared with positive control group(con-inhibitor-Ti(-exo)/con-mimic-Ti(-exo)), ^#p < 0.05, ^##p < 0.01. Image magnification, × 50–200 × ; scale bar, 100–500 μm.

Then, the engineered exosome enriching miR-3470b and low-expressing miR-3470b exosome were respectively co-cultured with osteoclasts for 48h in the co-culture system as described previously. Next, the formation/differentiation of osteoclasts was assessed by quantifying the mRNA expression of osteoclast marker NFatc-1 (Fig. 4C and D) and by analysis on TRAP and F-actin ring staining (Fig. 4E∼H). Following the lack of exosomal miR-3470b, there was a substantial decrease in the ratio of the integrated optical density to the area of TRAP staining and the number of osteoclasts, which was then reversed by miR-3470b-enriching exosomes (*p < 0.05) (Fig. 4F, H). Overall, these results show that the low-expressing miR-3470b exosomes derived from macrophages could induce osteolysis, while the miR-3470b-enriching exosomes derived from macrophages could inhibit osteoclasts formation and wear particle-induce osteolysis, in vitro.

3.7. Macrophage-derived exosomal miR-3470b regulates the osteoclast differentiation via the Tab3/NF-κB pathway

The NF-κB pathway is considered a critical pathway in inflammation and osteoclasts. Based on bioinformatics prediction of the target site of miR-3470b in TAB3, a luciferase reporter vector containing the 3’-UTR of TAB3 was constructed, and the reporter assay results revealed that miR-3470b could significantly inhibit luciferase expression. However, the mutation of nucleotides in the 3′-UTR of TAB3 caused complete abrogation of the repressive effect (Fig. 2H).

Bioinformatic analysis (KEGG, GO analysis) on the dataset GSW2803868 (the mRNA sequencing performed on osteolysis model in vitro) from GEO indicated that NF-kB signaling is one of the key pathways to wear particle-induced osteolysis (Fig. 5ÃC). Studies revealed that TAB3 regulates the NF-κB pathway. The results of western blotting also verified that the miR-3470b-enriching exosomes decreased the level of P–P65, TAB3, and the osteoclastogenesis marker (NFatc-1) (*p < 0.05) (Fig. 5D and E). These results showed that M-Exo could suppress osteoclast differentiation via miR-3470b targeting TAB3/NF-κB.

Fig. 5.

Bioinformatic analysis on the mRNA sequencing (from GEO: GSW2803868) and western blotting assay on osteolysis model in vitro. A Statistical results of GO classification of network genes between groups. B KEGG pathway analysis. C Image of connection between pathways. D Protein expression bands of NFatc-1, Tab3, and NF-κB (P-p65/p65). E The gray values of bands in each group were quantitatively analyzed and compared (control group EXO-mimic-NC/wear particles EXO-mimic-NC + Ti group/EXO-miR-3470b mimic group/EXO-miR-3470b mimic + Ti group). Compared with the control group EXO-mimic-NC, *p < 0.05, *p < 0.01 indicated a statistical difference. Compared with EXO-mimic-NC + Ti, ^#p < 0.05, ^##p < 0.01indicates a statistical difference.

3.8. miR-3470b repress wear particle-induced osteolysis in vivo

Next, we evaluated the effect of miR-3470b in an osteolysis model (surgical treatment with Ti (+Ti)) to further verify the results. The inflammatory infiltrating macrophages, lymphocytes and TRAP + osteoclasts were observed in the calvarial tissue after the H&E/TRAP staining (Fig. 6 A). The number of osteoclasts increased in the calvarial samples of AgomiR-NC + Ti group, compared to the Sham group (*p < 0.05) (Fig. 6 B). By analyzing the Micro-CT scanning dataset, we found that bone resorption increased on the calvaria surface of in the control group with wear particle (AgomiR-NC + Ti) from the increased number/ratio of porosity (Fig. 6 A), decreased BV/TV, in comparison with Sham group (*p < 0.05) (Fig. 6C, D). However, after the administration of AgomiR-3470b, the rate of porosity decreased and BV/TV increased in the AgomiR-3470b + Ti group. Summarily, osteoclastic bone resorption decreased in the AgomiR-3470b + Ti group compared to that in the AgomiR-NC + Ti group, with a statistically significant difference (^#p < 0.05, ^##p < 0.01) (Fig. 6C, D).

Fig. 6.

miR-3470b inhibited osteolysis induced by wear-particles in vivo. A Representative images of HE (top panel), TRAP staining (medium panel), and microCT scan (3D reconstruction) (last panel) of calvarial samples in each group. B Statistical analysis of TRAP-positive osteoclasts in each group. C ∼ D Statistical analysis of BV/TV and rate of porosity on the calvarial samples. The degree of osteolysis on AgomiR-3470b + Ti group was lower than that of the positive control group (AgomiR-NC + Ti group). E Fluorescence intensity comparative analysis results of miR-3470b FISH assay. F Representative FISH assay images of specific miR-3470b in each group. G Representative immunofluorescence images of DAPI (blue)/Tab3 (green)/NFatc1 (red) in each group, with white arrows pointing at the areas co-expressing of Tab3 and NFatc-1. H Representative overlay images of immunofluorescence images of DAPI (blue)/Tab3 (green)/NFatc-1 (red)/P-p65 (yellow) in each group, with white arrows pointing at the area co-expressing Tab3, NFatc-1 and P-p65. And the representative immunofluorescence images of Cstk in each group, with white arrows pointing at the areas expressing of Ctsk. I, J Fluorescence intensity comparative analysis results of the expression of Tab3, NFatc-1 (I) and Ctsk (J). K, L Representative IHC images of IL-β (K) in each group and the comparative analysis on their IOD/positive area (L). Compared with the Sham group, *p < 0.05, **p < 0.01; compared with the positive control group (AgomiR-NC + Ti group), ^#p < 0.05, ^##p < 0.01. Image magnification, 50 × ∼200 × ; scale bar, 100–500 μm.

3.9. Ti particle stimulated osteolysis by decreasing the expression of miR-3470b and increasing Tab3 expression in vivo

To evaluate whether miR-3470b can diminish Tab3 expression in vivo, FISH assay was performed on calvaria samples to assess the expression level of miR-3470b (Fig. 6 E, F), which significantly decreased in the AgomiR-NC + Ti group. We also performed immunofluorescence staining of Tab3/NFATc1, showing DAPI (blue)/Tab3 (green)/NFATC1 (red)/P-p65 (yellow) and their overlay in the calvarias of each group (Fig. 6G and H). The immunofluorescence analysis of osteolysis model revealed that the labelling intensities levels of target marker protein Tab3/NFatc-1, Ctsk were significantly lower in mice treated with AgomiR-3470b (AgomiR-3470b + Ti group) when compared with those in mice treated with AgomiR-NC (AgomiR-NC + Ti group) (Fig. 6I and J). and we further observed the relation between activated NF-κB signaling (P-p65) and the expression of Tab3 and NFatc-1 (Fig. 6H). We also performed the IHC analysis on the expression of inflammatory factors IL-1β (Fig. 6K and L). These results are consistent with those presented in the previous section. These results indicate that miR-3470b disrupts bone homeostasis, promotes osteoclastic bone resorption, and promotes osteolysis in patients with APL. Accordingly, our findings suggest the following potential mechanisms: M-Exo containing miR-3470b target Tab3/NF-κB and inhibit wear particle-induced inflammatory osteolysis and APL.

4. Discussion

As the most efficient procedure for treating end-stage joint diseases [35], the frequency of primary total joint arthroplasties (TJA) continues to rise globally [36,37]. However, the risks associated with postoperative infection, aseptic prosthesis loosening (APL), and bone defects persist. One of the most common post-arthroplasty complications is prosthesis loosening, which is also the primary cause of prosthesis failure [38,39], seriously restricting the service life of the artificial joint. As life expectancy increases, long-term prosthetic survival in total hip arthroplasty is becoming increasingly important. According to the findings of a multicenter clinical trial, APL accounted for 38.7% and 49.2%, respectively, of revision operations following 53,150 and 92,588 total knee and hip arthroplasty procedures [40]. However, in addition to having no effective prevention and treatment drugs, this condition necessitates revision surgery or even multiple operations in most patients. This is highly strenuous to the patient's body and spirit, and taxing on families, society, and the health care system. Therefore, elucidating the mechanism of the occurrence and development of APL and exploring the corresponding molecular targeting, potential treatment strategies, and prevention of APL is of great clinical and academic significance.

Increasing evidence reveals an emerging paradigm with cross-talk between macrophages and bone homeostasis in the microenvironment of APL [5]. After wear particles are produced, macrophages are first stimulated, and are the main cells that exert innate immunity (non-specific immunity) around the prosthesis. Studies found that macrophages activated by wear particles infiltrated and further caused the activation of Nuclear Factor-Kappa B (NF-κB) and secretion of various inflammatory mediators (such as TNF-α, IL-1β, and IL-6), which increased osteoclast differentiation in the relevant tissues of patients with joint prosthesis loosening [2,[6], [7], [8]]. A complex regulatory network exists in the bone microenvironment around prostheses. Wear particles can stimulate macrophages around the prosthesis to produce various intercellular transfer active molecules, leading to an osteoclast/osteogenesis imbalance and osteolysis around the prosthesis. Macrophages are crucial to the emergence and progression of prosthesis loosening. However, the regulation and mechanism of macrophages in osteoblasts/osteoclasts during prosthesis loosening is unclear. Therefore, clarifying the interaction and key mechanism of inflammatory osteolysis during the occurrence and development of prosthesis loosening may be the key to our research study.

Recent studies have greatly elucidated exosomes and non-coding RNA (ncRNAs) for their critical role in physiological and pathological processes, which are currently considered to be one of the most promising research directions on intracellular communication, receiving widespread attention. Exosomes and microRNAs (miRNAs) are closely related to tumors, immune inflammation, and bone formation/resorption regulation [[10], [11], [12], [13],21,41]. The macrophages are proved as the main cells that exert the effects of innate immunity (non-specific immunity) in the immune microenvironment of aseptic prothesis loosening for its adverse effect on the disease progression. According to the recent study, Macrophage exosomes are associated with the progression of bone resorption-related diseases [42]. It's suggest exosomal lncRNAs may potentially play their roles in aggravating the osteoclastogenesis/osteogenesis imbalance [43]. Also, Studies have investigated the therapeutic effects of exosomes from macrophages. Macrophage membrane encapsulated stem cell-derived exosomes could promote the prevention and therapeutic effect on osteolysis [44]. We believe that the study of macrophage exosomes is important for both prevention and treatment of periprosthetic loosening. In this study, we further studied the role of exosomes in loosening joint prostheses and clarified the interaction and regulatory network of inflammatory osteolysis between macrophages and surrounding tissues, by exploring the regulatory role and mechanism of M-Exo in inflammatory osteolysis. We first isolated exosomes from macrophages, examined their size distribution, and confirmed the existence of ‘exosomal markers’ using western blotting. The isolated extracellular vesicles were confirmed to be exosomes by clarifying their morphology, size, and marker expression. (Fig. 1A∼C). We then performed exosomal tracking and series of co-culture experiments. DiD/DiI-labeled M-Exo were injected intravenously into mice and incubated with osteoblasts/osteoclasts, and we found the distributions of major organs and bone tissue in vivo, and endocytosis of exosomes in both osteoclasts and osteoblasts in vitro (Fig. 1D∼F, J). Moreover, the results of PCR revealed that the osteoblast markers were downregulated n the wear particle group, whereas the osteoclast markers were upregulated (Fig. 1G∼I) (*p < 0.05). Ti-induced M-Exo accelerate osteoclast differentiation and inhibit osteoblast differentiation.

Currently, studies verified that exosomes derived from macrophages activate rapid transcription of genes that are crucial for diseases and are highly related to tumor invasion, proliferation, and angiogenesis, as well as the promotion of tumorigenesis and metastasis [45]. Macrophages are delivered through extracellular vesicles (EV) of specific lncRNAs, which stabilize HIF-1α lncRNA (HISLA) to facilitate the tumorigenesis of breast cancer via increasing aerobic glycolysis and apoptosis resistance [26]. Wei et al^. [46] found that exosomes secreted by M2 macrophage-derived facilitate glucose tolerance and insulin sensitivity through miRNA loading, and act as insulin sensitizers in vivo and in vitro. Continuous in-depth research on the transport function of exosomes has increasingly shown that ncRNAs in exosomes derived from macrophages may be potential diagnostic markers and effective therapeutic carriers for various diseases. However, the regulatory effect and mechanism of macrophages on osteoblasts/osteoclasts through exosomes and miRNAs of exosomes in prosthesis loosening are unclear. Therefore, clarifying the interaction and key mechanisms may be key to inhibiting the occurrence and development of inflammatory osteolysis and APL. In this study, we performed next-generation sequencing to analyze miRNAs from macrophage exosomes and filtered for significantly differential miRNAs (Fig. 2A–D). Notably, based on bioinformatics analysis of the sequencing dataset, TGF-β expressed in various bone-related cells has been proven to activate osteoblasts and inhibit osteoclastogenesis, which are closely related to the NF-κB signaling pathway. The results of qPCR analysis confirmed the expression level in the exosomes of the prosthesis loosening group; we focused on a specific miRNA, miR-3470b (Fig. 2E). These results formed the basis for the selection of the macrophage-derived exosomal miRNA, miR3470b, as a target for modulating bone homeostasis in the microenvironment of APL.

Since Valadi et al. [47] extracted miRNA from exosomes for the first time, exosomes have been found to have the ability to transport miRNA, mRNA, and other substances to target cells, and to express genes in new locations. Squadrito et al. [48] showed that exosomes can enrich specific miRNAs through a “sorting” mechanism; when the miRNA expression of bone marrow macrophages changes, the corresponding miRNAs in exosomes become larger. The same change trend in the amplitude is called the “miRNA sponge effect”. In addition, the membrane structure of exosomes protects miRNAs from RNase degradation [49]. To figure out the molecular mechanism by which miR-3470b regulates bone homeostasis by, online miRNA prediction software (TargetScan, Diana, miRwalk, and miRDB) was used to predict the potential targets and we identified TAB3 (Fig. 2F). TAB3 is widely regarded as a modulator of tumors, inflammation, and osteogenesis/osteoclasts. In addition, the bioinformatics analysis results revealed a conserved 8-nt binding site of miR-3470b in Tab3 in its 3ʹ-UTR. Therefore, miR-3470b directly targets Tab3 via the dual-luciferase reporter gene system (Fig. 2 G and H). Our findings indicated that macrophage-derived exosomal miR-3470b promoted osteoclast differentiation and inhibited osteoblast differentiation by targeting TAB3. Next, by investigating bone resorption and its relationship with miR-3470b around the prosthesis in interface membrane tissue specimens, our results suggested that the expression of Tab3 and differentiation of osteoclasts increased as miR-3470b decreased in the periprosthetic tissues of APL patients (Fig. 3). Furthermore, we performed functional experiments in vitro to elucidate the role of exosomal miR-3470b in the progression of osteoclasts and wear particle-induced osteolysis. We found that macrophage-derived exosomal miR-3470b effectively repressed osteoclast differentiation and osteolysis in vitro (Fig. 4). Overall, these results revealed that wear particle-induced osteolysis was regulated by macrophages compared with low-expressing miR-3470b exosomes, which target TAB3 in osteoclasts.

NF-κB is a ubiquitous transcription factor with rapid post-translational activation of numerous pathogenic signals. Its direct involvement in cytoplasmic/nuclear signaling, and its potency in activating the transcription of multiple genes encoding immune-related proteins are important in immune response [[50], [51], [52]]. The NF-κB pathway is considered a critical pathway in inflammation, osteoclasts, and periprosthetic osteolysis [53]. Osteoclast differentiation is essentially controlled by the RANKL, mainly via NF-κB and NFATc1. The homologous protein TAB3 links TAK1 to the upstream linker, which is involved in the osteoclast differentiation of RANKL through p65 phosphorylation and NFATc1 induction and is critical for the activation of signaling cascades mediated by IL-1, TNF, and RANKL [54,55]. NF-κB and AP-1 transcription factors could be activated by TAB3, which is a component of the NF-κB pathway functioning upstream of TRAF-6/TGF-β-activated kinase [56]. The potential target site of miR-3470b in TAB3 was predicted by bioinformatic analysis as described previously; the results related with luciferase reporter vector containing the 3′-UTR of TAB3 showed that miR-3470b could significantly inhibit luciferase expression, whereas the mutation of nucleotides in the 3′-UTR of Tab3 caused fully abolition of the suppressive effect (Fig. 2). Previous studies have shown that Tab3 is a regulator of the NF-κB pathway. To verify that Tab3 directly regulates NF-κB signaling, Western blot verified that the miR-3470b-loaded exosomes decreased the level of P-p65, Tab3 and osteoclastogenesis marker NFatc-1 (Fig. 5). The results showed that miR-3470b-enriched exosomes could suppress the NF-κB pathway and osteoclast differentiation. Moreover, we estimated this mechanism on the wear particle-induced osteolysis in vivo and we found that the miR-3470b could effectively decreased the bone resorption by reducing the expression of Tab3/NFatc1/P-p65, Ctsk and IL-1β (Fig. 6).

In conclusion, our finding that macrophage-derived exosomes (M-Exo) modulate osteolysis via miR-3470b targeting TAB3/NF-κB to directly regulate inflammatory osteolysis in APL sheds light on a novel mechanism for regulating bone resorption. Exosomal miR-3470b could be crucial mediators that facilitate bone homeostasis and, ultimately, may be a new targeting method for the prevention and treatment of bone resorption-related diseases. Thus, we may further explore the engineering exosomes transferring target miRNA (miR-3470b) to achieve the satisfied effect of prevention and therapy on osteolysis/APL or more bone resorption-related diseases. Considering the feature on specific targeting and the biocompatibility of macrophage-derived exosomes, further investigation into the engineering exosomes may focus on the specific exosomal membrane marker and components to promote the uptake efficiency, pharmacokinetics and biodistribution and capacity of production, to provide scientific basis for significant clinical applications.

Funding

This work was supported by the National Natural Science Foundation of China [grant numbers 82172405, 81972050, 81802179].

Disclaimer

All the authors declare that they have no financial or personal relationships with other people or organizations that could inappropriately influence our work. There is no professional or other personal interest of any nature or kind in any product, service, and/or company that could be construed as influencing the position presented in, or the review of, the manuscript.

Ethics approval and consent to participate

All experiments on specimens were performed in accordance with the Guidelines of the Declaration of Helsinki, and experiments were approved by the Ethics Committee of the First Affiliated Hospital of Sun Yat-Sen University. All specimens were obtained with the informed consent of the participants in this study. All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Sun Yat-Sen University and experiments were approved by the Institutional Animal Care and Use Committee of Sun Yat-Sen University.

CRediT authorship contribution statement

Baiqi Pan: designed and performed most of the experiments and wrote the manuscript. Ziji Zhang: designed part of the experiments and revised the manuscript. Xiaoyu Wu: performed part of the in vitro/vivo experiments. Guoyan Xian: performed part of the in vitro/vivo experiments. Xuantao Hu: performed part of the in vivo experiments. Minghui Gu: provided technical and resource assistance throughout the project. Linli Zheng: provided technical and resource assistance throughout the project. Xiang Li: provided technical and resource assistance throughout the project. Lingli Long: designed part of the experiments and revised the manuscript. Weishen Chen: coordinated the study, designed the experiments and provided funds. Puyi Sheng: coordinated the study, designed the experiments and provided funds.

Declaration of competing interest

No potential conflict of interest was reported by the authors.