Targeting ANT1 to regulate PINK1/Parkin-mediated mitophagy is an effective treatment of trauma-induced tendon heterotopic ossification

Abstract

Background

Heterotopic ossification (HO) is a common degenerative disease following trauma. Tendon HO is primarily attributed to osteogenic differentiation of stem/progenitor cells within the tendon. However, the precise mechanism underlying this process remains unclear. Recent studies suggest that PTEN induced kinase 1 (PINK1)/Parkin-mediated mitophagy plays a crucial role in biomineralization. Adenine nucleotide translocase 1 (ANT1), an upstream regulator of the PINK1/Parkin pathway, may influence tendon ossification development by modulating mitophagy.

Methods

This study investigated the role of mitophagy in tendon osteogenesis in clinical specimens, mouse tissues, and cells. The impact of ANT1 on tendon osteogenesis through mitophagy regulation was assessed by knocking down solute carrier family 25 member 4 (Slc25a4) both in vitro and in vivo. Furthermore, elamipretide was identified as a potential targeted drug for ANT1 through computer virtual screening and experimental verification. Its therapeutic efficacy on tendon ossification was validated using mouse cells, tissues, and human cells.

Results

This study found that PINK1/Parkin-mediated mitophagy was activated during tendon ossification, and the regulation of mitophagy could impact the osteogenesis of injured tendon-derived progenitor cells (inTPCs). Loss of Slc25a4 inhibited tendon ossification by downregulating the excessive mitophagy. Elamipretide, a targeted drug for ANT1, showed significant efficacy in treating HO.

Conclusion

Modulating PINK1/Parkin-mediated mitophagy by targeting ANT1 mitigated the progression of trauma-induced tendon HO, indicating ANT1 can be a potential therapeutic target for HO, with elamipretide emerging as a promising drug for its treatment.

The translational potential of this article

This study identifies ANT1 as a therapeutic target and supports elamipretide as a promising treatment strategy for HO.

Graphical abstract

1. Introduction

Heterotopic ossification (HO) is a pathological phenomenon characterized by ectopic bone formation in extraskeletal tissues [1]. As a common complication after trauma, tendon HO frequently develops at the site of tendon injury or surgical repair, thereby compromising tendon healing quality [2,3]. The presence of HO disrupts the original tendon structure, impairs biomechanical function, exacerbates clinical symptoms, and increases the risk of tendon rupture [2,4,5]. Nevertheless, current treatments are largely limited to surgical resection and symptomatic management [6]. Although HO has been recognized as a pathological process involving inflammation, angiogenesis, signal transduction, stem/progenitor cell recruitment, chondrogenic and osteogenic differentiation, and eventual bone formation, the precise molecular mechanisms remain poorly defined [7,8]. Therefore, it is of clinical importance to further explore the key regulatory pathways underlying HO and develop targeted and effective therapeutic strategies.

Tendon stem/progenitor cells (TSPCs) play a crucial role in the metabolism, repair, and regeneration of tendons due to their capacity for self-renewal and multi-differentiation [[9], [10], [11]]. As reported, osteogenesis and chondrogenesis of TSPCs are pivotal in HO [9,12]. Furthermore, TSPCs obtained from injured tendons exhibit enhanced osteogenic and chondrogenic potential compared with those from healthy tendons, potentially leading to pathological ossification [13]. Previous studies have demonstrated that injured tendon-derived progenitor cells (inTPCs) found in injured tendons have characteristics of TSPCs and demonstrate a greater ability for chondrogenesis compared with the progenitor cells from healthy tendons [14,15]. The involvement of inTPCs in the regeneration and degeneration of tendons suggests that inTPCs can be considered target cells of tendon repair [14,15]. Therefore, improper differentiation of inTPCs may contribute to inadequate healing of injured tendons, such as heterotopic osteogenesis and chondrogenesis [15]. However, until now, most studies on osteogenic differentiation of tendon-derived cells in HO have focused on cells derived from normal tendons. Exploration of the intrinsic mechanism of osteogenesis in inTPCs during tendon HO progression has been neglected.

Since it is widely recognized, mitochondria are vital organelles responsible for energy metabolism and adenosine triphosphate (ATP) production, which is essential for the physiological functions of the cells [16]. A previous study demonstrated that mitochondrial dysfunction of TSPCs promotes HO formation [17]. Mitophagy, a process that involves the transportation of damaged or failing mitochondria to lysosomes for degradation through autophagy [18], plays a significant role in preserving mitochondrial integrity and function [19]. The best-known mitophagy system is currently regulated by the serine/threonine kinase PTEN induced kinase 1 (PINK1) and the E3 ubiquitin ligase Parkin [20]. PINK1/Parkin-mediated mitophagy has been proven to participate in cell-mediated biomineralization by delivering amorphous calcium phosphate (ACP) to the extracellular matrix [21]. Relevant studies have confirmed that mitophagy contributes to the osteogenesis of bone marrow mesenchymal stem cells (BMSCs) [22,23]. However, the role of mitophagy in tendon repair and ectopic ossification formation has not been reported. Therefore, we investigated whether mitophagy affects the osteogenesis of inTPCs in HO.

Adenine nucleotide translocase 1 (ANT1), encoded by the gene of solute carrier family 25 member 4 (SLC25A4), is located in the inner membrane of mitochondria and is responsible for the exchange of adenosine diphosphate (ADP)/ATP within mitochondria and for the adjustment of mitochondrial function [24]. Atsushi Hoshino et al. confirmed that ANT1 is required for PINK1/Parkin-mediated mitophagy, independent of its ADP/ATP exchange function [25]. Michael J. Bround et al. also presented evidence suggesting the involvement of ANT1 in PINK1/Parkin-mediated mitophagy [26]. These findings highlight the potential role of ANT1 in regulating PINK1/Parkin-mediated mitophagy during tendon ossification development.

In this study, an Achilles tenotomy model was used to induce tendon HO and harvest inTPCs to investigate the key effect of regulating PINK1/Parkin-mediated mitophagy by targeting ANT1 on the treatment of trauma-induced tendon HO and the interrelated mechanisms. Specifically, the results revealed that PINK1/Parkin-mediated mitophagy was activated in tendon ossification of HO patients and mice models, and the regulation of mitophagy could affect the osteogenesis of inTPCs. ANT1, an upstream regulator of PINK1 [25], participates in tendon HO formation by regulating mitophagy. Moreover, as a targeted drug for ANT1, elamipretide demonstrated efficacy in treating HO, further emphasizing the vital role of ANT1 as a therapeutic target for HO.

2. Materials and methods

2.1. Human specimen collection

Human tendon HO specimens were obtained from patients (5 males, 3 females; aged 20–60 years) undergoing tendon HO excision. Human normal tendon specimens were obtained from the remaining tendon graft tissues from autologous tendon transplantation (i.e., cruciate ligament reconstruction surgery) in individuals (7 males, 1 female; aged 16–50 years). These specimens were washed with phosphate-buffered saline (PBS) in a sterile environment and subsequently divided into three parts: one part was rapidly frozen using liquid nitrogen and then stored at −80 °C for extracting tissue proteins and RNA; the second part was fixed in 4 % paraformaldehyde (PFA) for tissue sectioning; the third part was used for isolating human TSPCs (hTSPCs) (tendon HO tissues did not require retention of this part).

2.2. Cell isolation and culture

Normal Achilles tendons were obtained from female C57BL/6 J mice aged 6–8 weeks for TSPCs isolation. The injured Achilles tendons were obtained from the same type of mice that had undergone Achilles tenotomy one week before for inTPCs isolation. Human normal tendons were obtained from excess tendon graft tissues from autologous human tendon transplantation for hTSPCs isolation. TSPCs, inTPCs, and hTSPCs were isolated according to well-established procedures [9,14,15,17,27]. Briefly, the tendon tissue samples were digested sufficiently using type I collagenase (Sigma, C0130). Single-cell suspensions were cultured in low-glucose Dulbecco's modified Eagle medium (DMEM) supplemented with 15 % fetal bovine serum (FBS) (both Gibco, CA, USA) and 1 % penicillin/streptomycin. The cells at passages 3–4 were used in the experiments, with a medium change every three days. The osteogenic differentiation medium (ODM) consisted of high-glucose DMEM (Gibco) with 10⁻⁸ M dexamethasone, 50 μg/mL vitamin C, 10 mM β-glycerol phosphate (all Sigma–Aldrich, St. Louis, MO, USA), 10 % FBS, and 1 % penicillin/streptomycin. It was renewed every two days for osteogenic induction.

2.3. Animal model and grouping

Female C57BL/6 J mice, aged 6–8 weeks, were used to develop the Achilles tenotomy model. First, a full transverse cut was performed at the midpoint of the right Achilles tendon in each mouse. The skin was then sutured to allow self-healing of the ruptured tendon. For sham surgery, a skin incision was made without any injury to the Achilles tendon.

A total of 200 mice were randomly assigned to 18 groups. Tenotomy group, sham group (Fig. 1L), vehicle group and elamipretide group (Fig. 8A) each contained four groups of mice. The mice were euthanized at 1, 4, 7, and 10 weeks. The AAV-shSlc25a4 group and AAV-shNC group (Fig. 5A) each contained only one group of mice, which were euthanized at 10 weeks. The treatment of each group was shown in Supplementary Method 1. The right Achilles tendons were collected at each time point for further analyses. Blinding was implemented for group assignments during the different phases of the study.

Fig. 1.

PINK1/Parkin signaling is activated in tendon HO tissues from humans and mice. (A) H&E and Alcian blue staining of normal tendons and tendon HO tissues from humans. Scale bar = 200 μm (original magnification) and 100 μm (insert magnification of the boxed area). (B) Quantification of the mineralized bone area in H&E-stained sections. (C) Quantification of Alcian blue staining. (D, E) qRT–PCR analysis of PINK1 (D) and PRKN (E) expression in normal tendons and tendon HO tissues from humans. (F) Western blotting analysis of PINK1 and Parkin proteins in normal tendons and tendon HO tissues from humans. (G, H) Quantification of western blots of PINK1 (G) and Parkin (H). (I) IHC staining of PINK1 and Parkin in normal tendons and tendon HO tissues from humans. Scale bar = 200 μm (original magnification) and 100 μm (insert magnification of the boxed area). (J, K) IHC scores of PINK1 (J) and Parkin (K). (L) Schematic diagram of animal sample collection after Achilles tenotomy. (M, N) qRT–PCR analysis of Pink1 (M) and Prkn (N) in sham and tenotomy groups of mice at 1, 4, 7, and 10 weeks. (O) IHC staining of PINK1 and Parkin in mice from the sham and tenotomy groups at 10 weeks. Scale bar = 100 μm (original magnification) and 20 μm (insert magnification of the boxed area). (P, Q) IHC scores of PINK1 (P) and Parkin (Q). (R) Representative TEM images showing healthy mitochondria (white arrow) and mitophagy (red arrow) in the sham and tenotomy groups at 4 and 7 weeks. Scale bar = 2 μm (original magnification) and 500 nm (insert magnification of the boxed area). ∗p < 0.05, ns, not significant.

Fig. 8.

Elamipretide downregulates the ANT1/PINK1/Parkin pathway in trauma-induced tendon HO. (A) Schematic diagram of elamipretide administration and animal sample collection after Achilles tenotomy. (B) qRT–PCR analysis of Slc25a4 in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. (C) IHC staining of ANT1 in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. Scale bar = 100 μm (original magnification) and 20 μm (insert magnification of the boxed area). (D) IHC score of ANT1. (E) Representative TEM images showing healthy mitochondria (white arrow) and mitophagy (red arrow) in the vehicle and elamipretide groups at 7 weeks. Scale bar = 2 μm (original magnification) and 500 nm (insert magnification of the boxed area). (F, G) qRT–PCR analysis of Pink1 (F) and Prkn (G) in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. (H) IHC staining of PINK1/Parkin in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. Scale bar = 100 μm (original magnification) and 20 μm (insert magnification of the boxed area). (I, J) IHC scores of PINK1 (I) and Parkin (J). ∗p < 0.05, ns, not significant.

Fig. 5.

Slc25a4deficiency in tendons reduces post-traumatic HO formation. (A) Schematic diagram of AAV injection and animal sample collection after Achilles tenotomy. (B) Micro-CT images of AAV-shNC and AAV-shSlc25a4 groups 10 weeks after injury. (C) Quantification of the volume of the ectopic mineralized bone. (D, E) qRT–PCR analysis of the osteogenic-related genes Ocn, Runx2, and Alpl (D), and the chondrogenic-related genes Acan, Sox9, and Col2a1 (E) in the AAV-shNC and AAV-shSlc25a4 groups at 10 weeks. (F, G) H&E staining (F) and Alcian blue staining (G) of the proximal and distal ends of the tendons in AAV-shNC and AAV-shSlc25a4 groups 10 weeks after injury. Scale bar = 100 μm. (H–J) qRT–PCR analysis of Slc25a4 (H), Pink1 (I), and Prkn (J) in the AAV-shNC and AAV-shSlc25a4 groups at 10 weeks. (K) IHC staining of ANT1, PINK1, and Parkin in the AAV-shNC and AAV-shSlc25a4 groups at 10 weeks. Scale bar = 100 μm (original magnification) and 20 μm (insert magnification of the boxed area). (L–N) IHC scores of ANT1 (L), PINK1 (M), and Parkin (N). ∗p < 0.05.

2.4. Histological, histochemical, and immunohistochemical (IHC) analyses

The human tendon specimens and the distal hind limbs of mice were conventionally fixed, decalcified, dehydrated, paraffin-embedded, and sectioned. Subsequently, the specimens were subjected to histological staining with hematoxylin and eosin (H&E) or Alcian blue in accordance with standard protocols. The calcified HO area in H&E staining and the positive area of Alcian blue staining were quantified using ImageJ software. For IHC detection, the samples were incubated with specific primary antibodies according to standard protocols. The details of the antibodies used in IHC are documented in Supplementary Table 1. The IHC score system was based on the intensity and area of positive staining. It was used for semi-quantification of IHC results, which were categorized as High Positive (3), Positive (2), Low Positive (1), and Negative (0) [28].

2.5. Transmission electron microscopy (TEM) examination

Tendon tissues and cells were collected and fixed with 2.5 % glutaraldehyde, postfixed with 1 % osmium tetroxide, dehydrated in gradient ethanol, and embedded in acetone. Thin sections (100 nm thick) were stained with uranyl acetate and lead citrate. The ultrastructure was examined using a Tecnai G2 Spirit TEM (FEI, Czech). Five fields were randomly selected in each group to count the number of mitochondria, autophagosomes (APs), and autolysosomes (ALs).

2.6. Mitochondrial membrane potential (MMP) (ΔΨm) assay

Mitochondria in cell samples were stained using the MitoMP Detection Kit (JC-1) (Dojindo, MT09, Japan) at 37 °C for 30 min. Subsequently, a fluorescence microscope was used to detect MMP (ΔΨm). The level of ΔΨm was determined by the ratio of the red to green fluorescence intensity with ImageJ software.

2.7. Mitophagy detection assay

Mitophagy Detection Kit (Dojindo, MD01, Japan) was used to examine the colocalization of mitochondria and lysosomes in cells seeded in confocal dishes (NEST, China). Cells were exposed to the Mtphagy and Lyso Dyes working solutions for 30 min at 37 °C separately and were detected using confocal fluorescence microscopy (Carl Zeiss, Germany). Mtphagy Dye accumulates in intact mitochondria, with intensified red fluorescence indicating increased mitophagy. Lyso Dye was applied to confirm the fusion of mitochondria labeled with Mtphagy Dye and lysosomes. Two different channels in one image were split and colored by ImageJ software for pixel-matching colocalization analysis. Additionally, ImageJ software was used to calculate the colocalization ratio based on Pearson's correlation (Rr) and Overlap coefficient (R).

2.8. Gene expression analysis

Total RNA was extracted from animal and cell samples using TRIzol and then reverse-transcribed into complementary DNA (cDNA) using PrimeScriptTM RT Master Mix (both Takara). Quantitative real-time polymerase chain reaction (qRT–PCR) was performed using SYBR Premix Ex TaqTM II (Takara) on an ABI QuantStudio5 (Applied Biosystems, Foster City, CA, USA). The data were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH/Gapdh), and relative gene expression was calculated using the 2^−△△Ct method. The primer sequences of the relevant genes are listed in Supplementary Table 2.

2.9. Western blotting

Proteins were extracted from tissues and cells using RIPA Buffer (Beyotime, China) supplemented with protease and phosphatase inhibitors (Beyotime, China). The lysates were mixed with 5 × protein loading buffer (Dakewe, China) and heated to 100 °C for 10 min. Proteins were then separated by SDS-PAGE at a constant voltage. The polyvinylidene difluoride (PVDF) membranes were then blocked with 5 % defatted milk powder, incubated with primary antibodies, and treated with secondary antibodies. Lastly, the target proteins were detected using an ‌enhanced chemiluminescence (ECL) kit (Dalian Meilun Biotechnology, China). Information on the antibodies used in western blotting is summarized in Supplementary Table 1. The relative protein expression of the bands was analyzed using ImageJ software.

2.10. Alkaline Phosphatase (ALP) and Alizarin Red S (ARS) staining

For ALP staining, after osteogenic induction, the cells were fixed in 4 % PFA for 30 min and stained using the BCIP/NBT ALP color development kit (Beyotime, C3206, China) according to the manufacturer's instructions. For ARS staining, the cells were fixed in a special fixative of the kit for 20 min and stained with the osteoblast mineralization nodules ARS staining kit (Beyotime, C0148S, China) following the provided guidelines. Both ALP and ARS stainings were quantified by ImageJ software for relative positive ratios.

2.11. Cell viability assay

Cell viability was assessed using a Cell Counting Kit-8 (CCK-8) (Dalian Meilun Biotechnology, China). The cells were seeded into 96-well plates and exposed to carbonyl cyanide 3-chlorophenylhydrazone (CCCP), cyclosporin A (CsA), and elamipretide at varying concentrations for 48 h. Subsequently, the cells were incubated in 10 % CCK-8 solutions at 37 °C for 1 h. The absorbance of the solutions was measured at 450 nm using a microplate reader. The formula (1) applied to compute cell viability was as follows:

$$\begin{array}{c}Cellviability(\%)=\frac{{OD}_{experimental}‐{OD}_{blank}}{{OD}_{control}‐{OD}_{blank}}\times 100\%\end{array}$$ (1)

2.12. Apoptosis assay

Apoptosis trials were conducted using the Annexin V-FITC/PI kit (KeyGen Biotech, China) in accordance with the standard procedure. The cell samples were incubated with PI and Annexin V-FITC at room temperature in the dark and analyzed using FACS Verse™ (BD Bioscience, USA).

2.13. Small interfering RNA (siRNA)-mediated knockdown and cell transfection

Slc25a4-specific siRNAs (si-Slc25a4) (GCACAUUAUCGUGAGCUGGAU for mouse) and negative control siRNAs (si-NC) (Tsingke, China) were used to knock down Slc25a4 in cells. The transfection of siRNA oligonucleotides was performed using Lipofectamine® RNAiMAX Transfection Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. Slc25a4/ANT1 expression levels were assessed using qRT–PCR and western blotting assays.

2.14. Adeno-associated virus (AAV) for Slc25a4 knockdown

AAV-shSlc25a4 (pAAV-U6-shSlc25a4-CMV-EGFP) and AAV-shNC (pAAV-CMV-EGFP) were constructed by Tsingke (China). To knock down Slc25a4, each mouse was administrated with 1E + 12 vgs/mL AAV in a 6 μL volume via local tendon injection on the first day after Achilles tenotomy (n = 20). The specimens were harvested at 10 weeks for detection (Fig. 5A).

2.15. Molecular docking, molecular dynamic simulation and surface plasmon resonance (SPR)

The molecular docking approach for the complexes was detailed in Supplementary Method 2. The molecular dynamics simulation of ANT1 complexes with BKA, CATR, and elamipretide was performed using Gromacs2020 software. The SPR assay was conducted using a Biacore T200 instrument (Cytiva, USA) with the CM5 sensor chip. Additional details were available in Supplementary Method 3 and Supplementary Method 4.

2.16. Elamipretide administration

After Achilles tenotomy, the mice were injected intraperitoneally (i.p.) with elamipretide (GLPBIO, GC30014, USA) dissolved in PBS every day (2.5 mg/kg, 50 μL/mouse) from day 1 to week 6. The vehicle group received the same volume of PBS as the elamipretide solvent (Fig. 8A).

2.17. Micro-computed tomography (Micro-CT)

At 10 weeks after Achilles tenotomy, distal hind limbs from mice treated with elamipretide (n = 8) and vehicle (n = 8) were fixed in 4 % PFA at 4 °C and scanned using a micro-CT scanner (Skyscan 1176; Bruker) at 50 kV and 200 μA. The detection data of ectopic mineralized bones were analyzed with a threshold of 255.

2.18. Statistical analyses

Data from at least three independent experiments are presented as mean ± standard deviation (SD). For two-group comparisons, a two-tailed Student's t-test or Welch's t-test was used for normally distributed data, depending on variance homogeneity, while the Mann–Whitney U test was applied for non-normally distributed data. For multiple-group comparisons, normally distributed data were analyzed using one-way or two-way analysis of variance (ANOVA) when the assumption of homogeneity of variances was met, and Welch's ANOVA when it was violated. Post hoc analyses were conducted using Tukey's test under homoscedasticity and the Games–Howell test under heteroscedasticity. For non-normally distributed data, the Kruskal–Wallis H test followed by Dunn's post hoc test was employed. Study results were considered statistically significant when ∗ p < 0.05.

3. Results

3.1. PINK1/parkin is upregulated in tendon HO tissues

From clinical specimens, it was observed that tendon HO tissues presented more areas of ossification (Fig. 1A and B) and significant positive areas of Alcian blue staining compared with normal tendons (Fig. 1A and C). To identify the PINK1/Parkin expression level in human tissues, qRT–PCR, western blotting, and IHC staining were performed. At the gene level, qRT–PCR showed that the gene expression of PINK1/PRKN significantly increased in tendon HO (Fig. 1D and E). At the protein level, western blotting (Fig. 1F–H) and IHC staining (Fig. 1I–K) implied that PINK1/Parkin expression was significantly upregulated in tendon HO.

In the mouse model of tendon HO induced by Achilles tenotomy, the Achilles tendon specimens were harvested at 1, 4, 7, and 10 weeks after surgery for further analyses (Fig. 1L). The qRT–PCR analysis showed that the gene expression of Pink1 increased at all time points (Fig. 1M), and Prkn increased at 4 and 10 weeks but showed no significant change at 7 weeks after the reduction at 1 week (Fig. 1N). At 10 weeks after tenotomy, IHC showed that PINK1/Parkin was significantly positively stained in ossification tissues (Fig. 1O–Q). Additionally, the TEM images directly demonstrated the appearance of mitophagy in injured tendons at 4 and 7 weeks when ectopic ossification gradually progressed (Fig. 1R). These results suggested that PINK1/Parkin-mediated mitophagy is remarkably activated in trauma-induced tendon HO.

3.2. InTPCs show higher levels of PINK1/parkin-mediated mitophagy and osteogenic differentiation than TSPCs

TSPCs and inTPCs were compared to simulate the difference between normal and injured tendons in vitro. According to the characteristics of mitophagy observed by TEM summarized by previous researchers [29], we used TEM in this study to examine TSPCs and inTPCs and demonstrated that undifferentiated inTPCs exhibited fewer healthy mitochondria and more APs and ALs containing phagocytosed mitochondria than undifferentiated TSPCs. InTPCs exposed to ODM for two days still exhibited fewer healthy mitochondria and more APs and ALs than TSPCs. Exposed to ODM for two weeks, both inTPCs and TSPCs showed few mitochondria, and inTPCs still showed more APs and ALs than TSPCs (Fig. 2A–C). These results suggested that based on the difference in the level of mitophagy in the normal state, inTPCs likely produced more mitophagy than TSPCs under osteogenic induction. To evaluate the damage to mitochondria, JC-1 staining was performed to detect ΔΨm. The red/green fluorescence intensity ratio of inTPCs was decreased compared with that of TSPCs after two days of osteogenic induction (Fig. 2D–G), demonstrating that ΔΨm of inTPCs was lower than that of TSPCs. Fluorescence colocalization staining of mitochondria and lysosomes showed that the degree of colocalization correlation of inTPCs was higher than TSPCs (Fig. 2H–K). Expression of mitophagy-related molecules was also detected in this study. qRT–PCR showed that the expression levels of mitophagy-related genes, Pink1, Prkn, microtubule-associated protein 1 light chain 3 beta (LC3b), and Sequestosome 1 (Sqstm1)/p62, were upregulated in inTPCs after two weeks of ODM exposure compared with TSPCs (Fig. 2L). Analysis of western blotting revealed similar expression trends (Fig. 2N and O). Simultaneously, the gene (Fig. 2M) and protein (Fig. 2P and Q) expression levels of osteogenic biomarkers (such as OCN, RUNX2, and ALP) of inTPCs were significantly higher than those of TSPCs. ALP staining after one week of osteogenic induction (Fig. 2R and S) did not, but ARS staining after two weeks of osteogenic induction (Fig. 2T and U) implied that inTPCs possessed a stronger osteogenic differentiating ability than TSPCs. Briefly, these results suggested that inTPCs exhibited more mitophagy and higher osteogenic differentiation potential, and the synchronous increase in mitophagy and osteogenesis likely implied the effect of mitophagy on osteogenic differentiation of inTPCs in injured tendons.

Fig. 2.

InTPCs have higher levels of PINK1/Parkin-mediated mitophagy and osteogenic differentiation than TSPCs. (A) Representative TEM images showing healthy mitochondria (white arrow) and mitophagy (red arrow) in the TSPC and inTPC groups when undifferentiated and when exposed to ODM for two days and two weeks, respectively. Scale bar = 1 μm (original magnification) and 200 nm (insert magnification of the boxed area). (B) Quantification of the number of mitochondria per TEM field. (C) Quantification of the number of APs and ALs per TEM field. (D) Representative JC-1 fluorescence images of TSPCs and inTPCs after two days of osteogenic induction (red: 561 nm/green: 488 nm). Scale bar = 100 μm. (E–G) Quantification of the JC-1 fluorescence intensity ratio. (H) Representative colocalization images of mitochondrial and lysosomes in TSPCs and inTPCs after two days of osteogenic induction (red: 594 nm/green: 488 nm). Scale bar = 10 μm (original magnification) and 2 μm (insert magnification of the boxed area). (I, J) Pixel matching colocalization analysis in TSPCs (I) and inTPCs (J). (K) Pearson's correlation and Overlap coefficient of colocalization. (L, M) qRT–PCR analysis of the mitophagy-related genes Pink1, Prkn, LC3b, and p62 (L) and osteogenic-related genes Ocn, Runx2, and Alpl (M) in TSPCs and inTPCs after two weeks of osteogenic induction. (N, O) Western blotting analysis (N) and quantification (O) of mitophagy-related proteins, PINK1, Parkin, LC3BⅡ, and p62 in TSPCs and inTPCs after two weeks of osteogenic induction. (P, Q) Western blotting analysis (P) and quantification (Q) of osteogenic-related proteins, OCN, RUNX2, and ALP in TSPCs and inTPCs after two weeks of osteogenic induction. (R) ALP staining of TSPCs and inTPCs under weeklong osteogenic induction. Scale bar = 200 μm. (S) Quantification of ALP staining. (T) ARS staining of TSPCs and inTPCs under two weeks of osteogenic induction. Scale bar = 200 μm. (U) Quantification of ARS staining. ∗p < 0.05, ns, not significant.

3.3. Mitophagy regulates osteogenesis in inTPCs

The mitophagy inducer CCCP and the inhibitor CsA were used to regulate mitophagy. As the osteogenic differentiation experiment is relatively time-consuming, relatively low concentrations of CCCP (1 μM) and CsA (5 μM) were used in this study to stimulate inTPCs, respectively. These concentrations did not cause poisoning (Fig. 3A and B) or apoptosis (Fig. 3C and D) of the inTPCs. According to a previous research [21], the experimental groups were cultured for two days in ODM containing CCCP (1 μM) or CsA (5 μM), and the vehicle group was cultured for two days in ODM containing the same volume of solvent as the experimental groups. All groups were cultured in pure ODM for an additional two days, and this four-day treatment was considered an administration cycle and applied to minimize the toxicity of CCCP or CsA on osteogenic differentiation of cells. During the experimental period, the administration cycle was repeated three times for qRT–PCR, western blotting, and ALP staining; and four times for ARS staining. The qRT–PCR analysis demonstrated that the expression of mitophagy-related genes, Pink1, Prkn, and LC3b, and osteogenic-related genes, Ocn, Runx2, and Alpl, was increased in inTPCs in response to CCCP treatment but was suppressed in response to CsA treatment (Fig. 3E and F). Conversely, CCCP suppressed p62 gene expression, whereas CsA promoted it (Fig. 3E). Similarly, western blotting demonstrated that the expression of mitophagy-related proteins, PINK1, Parkin, and LC3BⅡ, and the expression of osteogenic-related proteins OCN, RUNX2, and ALP were upregulated with p62 suppression in response to CCCP treatment, but were downregulated with p62 accumulation in response to CsA treatment (Fig. 3G–J). Furthermore, CCCP increased ALP and ARS staining, indicating that osteogenic differentiation was promoted by mitophagy upregulation, while CsA reduced ALP and ARS staining, suggesting that osteogenic differentiation was suppressed by mitophagy downregulation (Fig. 3K–N). Briefly, osteogenesis of inTPCs was greatly regulated by mitophagy. Therefore, inTPCs are likely responsible for tendon ossification via PINK1/Parkin-mediated mitophagy in this model.

Fig. 3.

Mitophagy regulates the osteogenic differentiation of inTPCs. (A) Cell viability of inTPCs treated with CCCP at different concentrations (0, 10, 50, 100, 500, and 1000 nM) for 48 h. (B) Cell viability of inTPCs treated with CsA at different concentrations (0, 0.1, 0.5, 1, 5, and 10 μM) for 48 h. (C) Annexin V-FITC/PI assay for inTPCs treated with vehicle, 1 μM CCCP, and 5 μM CsA. (D) Percentage of viable cells for inTPCs. (E, F) qRT–PCR analysis of mitophagy-related genes Pink1, Prkn, LC3b, and p62 (E), and osteogenic-related genes Ocn, Runx2, and Alpl (F) in inTPCs, respectively treated with vehicle, 1 μM CCCP, and 5 μM CsA for three stimulation cycles. (G, H) Western blotting analysis (G) and quantification (H) of mitophagy-related proteins PINK1, Parkin, LC3BⅡ, and p62 in inTPCs treated with vehicle, 1 μM CCCP, and 5 μM CsA for three stimulation cycles. (I, J) Western blotting analysis (I) and quantification (J) of osteogenic-related proteins OCN, RUNX2, and ALP in inTPCs, respectively, treated with vehicle, 1 μM CCCP, and 5 μM CsA for three stimulation cycles. (K) ALP staining of inTPCs treated with vehicle, 1 μM CCCP, and 5 μM CsA for three stimulation cycles. Scale bar = 200 μm. (L) Quantification of ALP staining. (M) ARS staining of inTPCs treated with vehicle, 1 μM CCCP, and 5 μM CsA for four stimulation cycles. Scale bar = 200 μm. (N) Quantification of ARS staining. ∗p < 0.05, ns, not significant.

3.4. ANT1 is involved in the osteogenesis of inTPCs via PINK1/parkin-mediated mitophagy in trauma-induced tendon HO

As the transmembrane protein of mitochondria, ANT1 promotes PINK1/Parkin-mediated mitophagy both in vitro and in vivo [25]. At the in vivo level, human and mouse specimens were used to test the expression level of ANT1. The results displayed that the gene (Fig. 4Aand C) and protein (Fig. 4B and D–F) expression of ANT1 was significantly activated in the tendon ossification tissues. At the in vitro level, the ANT1 expression level was compared between TSPCs and inTPCs. The results indicated that the gene (Fig. 4G) and protein (Fig. 4H and I) expression of ANT1 also distinctly increased in inTPCs of injured tendons, suggesting that ANT1 is implicated in trauma-induced tendon HO. To assess the effects of Slc25a4 loss on the osteogenesis and mitophagy of inTPCs, siRNA was transfected into inTPCs to knock down Slc25a4. After 48 h of transfection under osteogenic induction, the knockdown efficiency of Slc25a4/ANT1 was confirmed (Fig. S1A–C). With the downregulation of ANT1 expression, the expressions of PINK1, Parkin, and LC3BⅡ were inhibited, p62 expression was increased, and the expressions of the osteogenic-related biomarkers, RUNX2 and ALP, were reduced, indicating the inhibition of mitophagy and osteogenesis. As a mitophagy agonist, CCCP did not affect ANT1 expression, promoted the expression of PINK1, Parkin, LC3BⅡ, RUNX2, and ALP, and suppressed p62 expression. Transfection of si-Slc25a4 reversed the effect of CCCP on activating osteogenic protein expression and the PINK1/Parkin pathway of inTPCs (Fig. 4J–S), indicating that ANT1 likely contributes to the osteogenesis of inTPCs by regulating PINK1/Parkin-mediated mitophagy. To further determine whether ANT1 affects tendon HO formation by regulating the PINK1/Parkin axis in vivo, a mouse model of Slc25a4 deficiency was established by AAV local injection in tendon (Fig. 5A). After 10 weeks of Achilles tenotomy, micro-CT imaging (Fig. 5B and C) and H&E staining (Fig. 5F) demonstrated significantly less ectopic bone generation in the AAV-shSlc25a4 group than in the AAV-shNC group. Meanwhile, the expression of osteogenic-related and chondrogenic-related genes in injured tendons was reduced by AAV-shSlc25a4 injection (Fig. 5D and E). Alcian blue staining disclosed that AAV-shSlc25a4 suppressed cartilage tissue formation in injured tendons (Fig. 5G). Briefly, the deficiency of Slc25a4 in tendons remarkably restrained post-traumatic HO formation. qRT–PCR (Fig. 5H–J) and IHC (Fig. 5K–N) analyses revealed that AAV-shSlc25a4 significantly reduced ANT1, PINK1, and Parkin expression in injured tendons compared with AAV-shNC. Therefore, these results suggested that ANT1 may contribute to the formation of trauma-induced tendon HO via the PINK1/Parkin pathway. Moreover, ANT1 may serve as a therapeutic target for tendon HO.

Fig. 4.

ANT1 participates in the osteogenesis of inTPCs via PINK1/Parkin-mediated mitophagy. (A) qRT–PCR analysis of SLC25A4 expression in normal tendons and tendon HO tissues from humans. (B) IHC staining of ANT1 in normal tendons and tendon HO tissues from humans. Scale bar = 200 μm (original magnification) and 100 μm (insert magnification of the boxed area). (C) qRT–PCR analysis of Slc25a4 in sham and tenotomy mice groups at 1, 4, 7, and 10 weeks. (D) IHC staining of ANT1 in sham and tenotomy mice groups at 10 weeks. Scale bar = 100 μm (original magnification) and 20 μm (insert magnification of the boxed area). (E) IHC score of ANT1 in human tendon tissues. (F) IHC score of ANT1 in mouse tendon tissues. (G) qRT–PCR analysis of the Slc25a4 gene in TSPCs and inTPCs after two weeks of osteogenic induction. (H, I) Western blotting analysis (H) and quantification (I) of the ANT1 protein in TSPCs and inTPCs after two weeks of osteogenic induction. (J–O) Under two days of osteogenic induction, western blotting analysis (J) and quantification of ANT1 (K) protein and mitophagy-related proteins PINK1 (L), Parkin (M), LC3BⅡ (N), and p62 (O) in inTPCs after 100 nM si-Slc25a4 transfection for two days or a combination of 100 nM si-Slc25a4 transfection for two days and 1 μM CCCP added on the second day. (P–S) Under two days of osteogenic induction, western blotting analysis (P) and quantification of osteogenic-related proteins OCN (Q), RUNX2 (R), and ALP (S) in inTPCs after 100 nM si-Slc25a4 transfection for two days or a combination of 100 nM si-Slc25a4 transfection for two days and 1 μM CCCP added on the second day. ∗p < 0.05, ns, not significant.

3.5. Elamipretide stably interacts with the ANT1 protein as a targeted drug

To screen the ideal drug targeting the ANT1 protein based on molecular docking technology, high-throughput virtual screening was performed from a drug library (TargetMol Chemicals Inc, China) containing 5585 compounds with known biological activities. Elamipretide, a mitochondrial-targeted peptide with a binding energy of −9.0 kcal mol⁻¹, was selected for further research because of its strong ANT1 binding ability, safety, and efficacy. As a tetrapeptide, elamipretide was reported to reinvigorate mitochondrial function through direct interaction with ANT1 [30]. BKA and CATR are known inhibitors of ANT1. Molecular docking revealed that the binding energies of BKA, CATR, and elamipretide to the ANT1 protein were −7.0, −8.8, and −9.0 kcal mol⁻¹, respectively, which were all less than or equal to −7 kcal mol⁻¹, indicating that the binding of these three ligands to ANT1 was strong. Among them, BKA is bound to ANT1 at amino acid residues Asn-88 and Arg-188 of the ANT1 protein with three hydrogen bonds and forms hydrophobic interactions with several surrounding amino acids (Fig. 6A). CATR formed six hydrogen bonds with the amino acid residues Arg-80, Asn-88, Ser-127, Arg-188, and Arg-280 of the ANT1 protein, and there were multiple hydrophobic interactions around CATR (Fig. 6B). Elamipretide formed 11 hydrogen bonds with amino acid residues Arg-80, Asn-88, Ala-123, Ser-127, Gly-225, Ser-228, Arg-235, Arg-236, and Asn-277 of the ANT1 protein and exhibited multiple hydrophobic interactions (Fig. 6C). All three docking results formed multiple interaction forces. The elamipretide group exhibited the largest number of hydrogen bonds and the best binding energy, indicating that elamipretide demonstrated better binding to ANT1 than BKA and CATR. Thus, elamipretide is very likely to affect the structure, function, and biological activity of the ANT1 protein.

Fig. 6.

Elamipretide stably interacts with the ANT1 protein as a targeted drug. (A–C) Molecular docking models for the ANT1 protein with BKA (A), CATR (B), and elamipretide (C). (D–G) Molecular dynamic simulation analysis showing the RMSD (D), RMSF (E), Rg (F), and hydrogen bonds number (G) of ANT1 protein complexes with BKA, CATR, and elamipretide. (H–J) Gibbs FEL diagram of ANT1 protein complexes with BKA (H), CATR (I), and elamipretide (J). (K–M) SPR analysis for the binding of ANT1 with BKA (K), CATR (L), and elamipretide (M). (N) Cell viability of inTPCs treated with elamipretide at different concentrations (0, 0.01, 0.05, 0.1, 0.5, and 1 μM) for 48 h. (O) qRT–PCR analysis of Slc25a4 gene in inTPCs treated with vehicle or 1 μM elamipretide after two weeks of osteogenic induction. (P, Q) Western blotting analysis (P) and quantification (Q) of ANT1 protein in inTPCs treated with vehicle or 1 μM elamipretide after two weeks of osteogenic induction. ∗p < 0.05, ns, not significant.

To further prove the degree and stability of binding between compounds and the ANT1 protein, molecular dynamic simulations were performed for each set of molecular docking results. Root mean square deviation (RSMD) is an indicator of stability estimation of protein and ligand complexes. The RMSD curves of the ANT1 protein and CATR complex and the ANT1 protein and elamipretide complex all fluctuated within 1 nm without substantial fluctuations, indicating that these two complexes formed stable structures. However, the RMSD curve of the ANT1 protein and BKA complex produced large fluctuations at 50–60 ns, and the degree of fluctuations was close to 1 nm, which indicated that it lacked stability (Fig. 6D). Root mean square fluctuation (RMSF) represents the extent to which amino acid residues in the protein fluctuate during dynamic simulation. The RMSF curves of the three complexes fluctuated within 1 nm, indicating that the addition of small ligand molecules demonstrated little effect on the overall structural stability of the protein. Only the amino acid residues in sequence numbers 200–210 produced a slight fluctuation which was because these amino acids were located at the edge of the protein structure (Fig. 6E). Radius of gyration (Rg) was used to characterize the compactness and stability of the structure. The Rg curves of the three complexes fluctuated within the range of 2.0–2.2 nm. The curves of CATR and elamipretide groups did not fluctuate significantly, while that of the BKA group fluctuated slightly (Fig. 6F). Therefore, CATR and elamipretide were more tightly bound to the complex with ANT1, and the BKA group was slightly less stable. The number of hydrogen bonds between ANT1 and BKA, and CATR was stable from one to three throughout the process, whereas the number of hydrogen bonds formed between ANT1 and elamipretide was up to 12 (Fig. 6G). In terms of hydrogen bonding interactions, elamipretide demonstrated the strongest binding stability with ANT1. The Gibbs free energy landscape (FEL) diagram was used to describe the conformation with the least energy during the entire process of the complex structure dynamic simulation. Multiple minimum energy clusters were formed in the FEL of the ANT1–BKA complex, and the energy clusters were scattered (Fig. 6H), indicating that the complex was poorly stable. In the FEL of the ANT1 complex with CATR and elamipretide, a nearly single and smooth minimum energy cluster was formed (Fig. 6I and J), suggesting that the complexes formed by ANT1 with CATR and elamipretide were more stable. According to the above analysis, elamipretide is very suitable as a targeted drug for ANT1. SPR experiments were also conducted to test the binding affinities between the recombinant ANT1 protein and drugs. The dissociation constant (K_D) of ANT1 with BKA, CATR, and elamipretide were 19.287, 2.032, and 4.349 μM, respectively (Fig. 6K–M). As ANT1 inhibitors, CATR displayed a stronger binding affinity to ANT1 than BKA, whereas the affinity of elamipretide to ANT1, despite being slightly inferior to CATR, was also quite excellent. Collectively, in addition to the recognized toxic ANT1 inhibitors, elamipretide was identified as a nontoxic drug that commendably targets ANT1. These findings support the robust binding interactions of ANT1 with elamipretide. Subsequent experiments verified the inhibition of elamipretide on ANT1 expression in inTPCs under 2-week osteogenic induction (Fig. 6N–Q).

3.6. Elamipretide inhibits PINK1/Parkin-mediated mitophagy and osteogenic differentiation of inTPCs

Elamipretide was selected to intervene in the osteogenic differentiation of inTPCs. TEM exhibited it directly that the elamipretide group presented more healthy mitochondria than the vehicle group under two-day osteogenic induction but without significant statistical difference (Fig. 7A and B), probably because the osteogenic induction time was too short to produce therapeutic effects of drugs. However, the number of mitochondrial APs and ALs in the elamipretide group was significantly reduced under two-day osteogenic induction (Fig. 7A and C). Elamipretide visibly improved mitochondrial health (Fig. 7A and B) and decreased mitophagy (Fig. 7A and C) in inTPCs after exposure to ODM for two weeks. JC-1 staining showed that elamipretide decreased ΔΨm of inTPCs under two-day osteogenic induction (Fig. 7D–G), indicating that elamipretide effectively alleviated mitochondrial damage. Fluorescence colocalization staining for mitophagy detection suggested that elamipretide suppressed the mitophagy of inTPCs during osteogenic differentiation (Fig. 7H–K). After two weeks of osteogenic culture, the gene expression of Pink1, Prkn, and LC3b was inhibited, while p62 expression was promoted by elamipretide in inTPCs (Fig. 7L). The protein expression level showed similar decreased trends, except for increased p62 expression (Fig. 7N and O), signifying the suppression of mitophagy. More importantly, elamipretide downregulated the gene (Fig. 7M) and protein (Fig. 7P and Q) expression levels of the osteogenic markers OCN, RUNX2, and ALP, of inTPCs. Additionally, weaker ALP (Fig. 7R and S) and ARS (Fig. 7T and U) staining was observed in the elamipretide group compared with the vehicle group, suggesting that elamipretide demonstrated the ability to attenuate osteogenic differentiation of inTPCs. Briefly, elamipretide showed the potential to inhibit PINK1/Parkin-mediated mitophagy and osteogenic differentiation of inTPCs.

Fig. 7.

Elamipretide inhibits PINK1/Parkin-mediated mitophagy and osteogenic differentiation of inTPCs. (A) Representative TEM images showing healthy mitochondria (white arrow) and mitophagy (red arrow) in inTPCs treated with vehicle or 1 μM elamipretide when exposed to ODM for two days and two weeks, respectively. Scale bar = 1 μm (original magnification) and 200 nm (insert magnification of the boxed area). (B) Quantification of the number of mitochondria per TEM field. (C) Quantification of the number of APs and ALs per TEM field. (D) Representative JC-1 fluorescence images of inTPCs treated with vehicle or 1 μM elamipretide after two days of osteogenic induction (red: 561 nm/green: 488 nm). Scale bar = 100 μm. (E–G) Quantification of the JC-1 fluorescence intensity ratio. (H) Representative colocalization images of mitochondria and lysosomes in inTPCs treated with vehicle or 1 μM elamipretide after two days of osteogenic induction (red: 594 nm/green: 488 nm). Scale bar = 10 μm (original magnification) and 2 μm (insert magnification of the boxed area). (I, J) Pixel matching colocalization analysis in inTPCs treated with vehicle (I) or 1 μM elamipretide (J). (K) Pearson's correlation and Overlap coefficient of colocalization. (L, M) qRT–PCR analysis of the mitophagy-related genes Pink1, Prkn, LC3b, and p62 (L) and the osteogenic-related genes Ocn, Runx2, and Alpl (M) in inTPCs treated with vehicle or 1 μM elamipretide after two weeks of osteogenic induction. (N, O) Western blotting analysis (N) and quantification (O) of mitophagy-related proteins PINK1, Parkin, LC3BⅡ, and p62 in inTPCs treated with vehicle or 1 μM elamipretide after two weeks of osteogenic induction. (P, Q) Western blotting analysis (P) and quantification (Q) of osteogenic-related proteins OCN, RUNX2, and ALP in inTPCs treated with vehicle or 1 μM elamipretide after two weeks of osteogenic induction. (R) ALP staining of inTPCs treated with vehicle or 1 μM elamipretide under weeklong osteogenic induction. Scale bar = 200 μm. (S) Quantification of ALP staining. (T) ARS staining of inTPCs treated with vehicle or 1 μM elamipretide under two weeks of osteogenic induction. Scale bar = 200 μm. (U) Quantification of ARS staining. ∗p < 0.05, ns, not significant.

3.7. Elamipretide suppresses the ANT1/PINK1/Parkin pathway and effectively treats trauma-induced tendon HO

To evaluate the therapeutic effect in vivo, elamipretide was administered to the mice starting from day one after Achilles tenotomy and continued for six weeks. Tendons were collected at 1, 4, 7, and 10 weeks (Fig. 8A). Further analysis revealed that Slc25a4 gene expression was restricted in response to elamipretide treatment at 1, 4, and 7 weeks when ectopic bone gradually developed in the tendons (Fig. 8B). IHC revealed that elamipretide significantly reduced ANT1 protein expression at 4, 7, and 10 weeks (Fig. 8C and D). After the completion of the treatment course, TEM results showed that elamipretide reduced the occurrence of mitophagy and increased the number of mitochondria in the injured tendon at 7 weeks when ossification was initially formed (Fig. 8E). qRT–PCR showed that elamipretide suppressed the gene expression of Pink1 and Prkn at 4, 7, and 10 weeks (Fig. 8F and G). At the protein level, PINK1 expression in injured tendons was restricted by elamipretide at all time points (Fig. 8H and I), and Parkin expression was restricted by elamipretide at 1, 7, and 10 weeks (Fig. 8H and J). These data demonstrated the inhibitory effect of elamipretide on the ANT1/PINK1/Parkin pathway. More importantly, elamipretide showed efficacy in treating trauma-induced tendon HO. As exhibited by micro-CT at 10 weeks, elamipretide strongly attenuated the formation of ectopic mineralized bones in injured tendons (Fig. 9A and B). At 1, 4, 7, and 10 weeks, the expression of osteogenic-related genes, Ocn, Runx2, and Alpl in injured tendons were all downregulated with elamipretide treatment (Fig. 9C–F). Histologically, H&E staining disclosed that elamipretide reduced ectopic ossification formation in the injured tendon (Fig. 9G). Ectopic bone formation may result from endochondral ossification [31]. The expression of cartilage-related genes Acan, Sox9, and Col2a1 in injured tendons was also downregulated by elamipretide (Fig. 9H–K). To detect cartilage nodules formation, the tendon samples were stained with Alcian blue and it was found that tendon HO progressed through endochondral ossification, and elamipretide suppressed cartilage tissue formation during HO formation (Fig. 9L). In summary, elamipretide is likely to treat trauma-induced tendon HO by targeting the ANT1/PINK1/Parkin pathway at the in vivo level.

Fig. 9.

Elamipretide effectively treats trauma-induced tendon HO. (A) Micro-CT images of the vehicle and elamipretide groups 10 weeks after injury. (B) Quantification of the volume of the ectopic mineralized bone. (C–F) qRT–PCR analysis of osteogenic-related genes Ocn, Runx2, and Alpl in the vehicle and elamipretide groups at 1 (C), 4 (D), 7 (E), and 10 (F) weeks. (G) H&E staining of the proximal and distal ends of the tendons in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. Scale bar = 100 μm. (H–K) qRT–PCR analysis of the chondrogenic-related genes Acan, Sox9, and Col2a1 in the vehicle and elamipretide groups at 1 (H), 4 (I), 7 (J), and 10 (K) weeks. (L) Alcian blue staining of the proximal and distal ends of the tendons in the vehicle and elamipretide groups at 1, 4, 7, and 10 weeks. Scale bar = 100 μm. ∗p < 0.05.

3.8. Elamipretide inhibits osteogenic differentiation of hTSPCs

The hTSPCs were isolated from human normal tendon tissue samples and treated with elamipretide under osteogenic induction (Fig. 10A). After two weeks of culture, the gene (Fig. 10B) and protein (Fig. 10C and D) expression of osteogenic markers was suppressed after elamipretide treatment. Moreover, elamipretide reduced ALP and ARS staining after 1 and 2 weeks, respectively (Fig. 10E–H), indicating that the osteogenic differentiation of hTSPCs was restricted by elamipretide. Based on these results, it can be speculated that the therapeutic effects of elamipretide may have potential clinical application value for human tendon HO.

Fig. 10.

Elamipretide inhibits osteogenic differentiation of hTSPCs. (A) Schematic diagram of hTSPCs isolation and treatment. (B) qRT–PCR analysis of the osteogenic-related genes OCN, RUNX2, and ALPL in hTSPCs treated with vehicle or 10 μM elamipretide for two weeks of osteogenic induction. (C, D) Western blotting analysis (C) and quantification (D) of osteogenic-related proteins, OCN, RUNX2, and ALP in hTSPCs treated with vehicle or 10 μM elamipretide for two weeks of osteogenic induction. (E) ALP staining of hTSPCs treated with vehicle or 10 μM elamipretide for weeklong osteogenic induction. Scale bar = 200 μm. (F) Quantification of ALP staining. (G) ARS staining of hTSPCs treated with vehicle or 10 μM elamipretide for two weeks of osteogenic induction. Scale bar = 200 μm. (H) Quantification of ARS staining. ∗p < 0.05.

Accordingly, the therapeutic effect of elamipretide on tendon osteogenesis was confirmed at multiple levels using in vitro (inTPCs), in vivo (tendons), and ex vivo (hTSPCs from human) models, providing a theoretical basis for clinical trials with elamipretide for treating tendon HO.

4. Discussion

The tendon contains few cells and blood vessels; therefore, its endogenous repair is less effective [32]. Tendon HO is generally considered to be the outcome of failed tendon repair; incorrect differentiation of stem/progenitor cells is assumed to be the cause as the pluripotent stem/progenitor cells differentiate into other cell types instead of tenocytes [33]. Recent studies have identified distinct tendon stem/progenitor subpopulations, such as nestin⁺, Ctsk⁺Scx⁺ and CD26⁺ cells, that play critical roles in tendon regeneration and ectopic ossification, emphasizing the importance of lineage fate decisions following injury [[34], [35], [36]]. Therefore, early intervention in abnormal differentiation of stem/progenitor cells is very important in preventing tendon ossification. Whereas, the intrinsic mechanisms regulating osteogenic differentiation during HO remain incompletely understood, making it difficult to identify effective therapeutic targets and corresponding treatment strategies. Our study proved that tendon injury-induced mitophagy contributes to variation in cell osteogenic differentiation, which is reflected in the imbalance in tendon tissue formation. What's more, ANT1 affects the osteogenesis of inTPCs by regulating PINK1/Parkin-mediated mitophagy during HO, and genetically or pharmacologically targeting ANT1 can effectively reduce mitophagy level and, more importantly, inhibit the osteogenic differentiation of inTPCs, consequently playing a therapeutic role in HO.

As reported, injury can drive excess reactive oxygen species (ROS) production in the tendon and cause mitochondrial dysfunction, eventually leading to tendon HO [17,37], implying that mitochondrial damage plays a key role in the formation of tendon HO. If mitochondria damage is too serious to repair, they can be selectively degraded by autophagy, which is known as mitophagy [38]. Mitophagy is macroautophagy in which superfluous or damaged proteins and organelles are recognized and degraded through the autophagosome-lysosome pathway to achieve cellular self-renewal and homeostasis [39]. In addition to mitochondrial dysfunction, we went beyond past research to demonstrate that injury lowered ΔΨm of tendon-derived progenitor cells and activated mitophagy, ultimately resulting in tendon osteogenesis. The removal of damaged mitochondria by mitophagy is regarded as a cellular adaptation to mitochondrial dysfunction [40]. Consequently, it was inferred that tendon injury-induced mitophagy should also be a cellular self-protection mechanism during cell stress. In other words, during the process of tendon ossification, mitochondrial damage exceeds the limit of maintaining homeostasis through repair; accordingly, mitophagy occurs and participates in osteogenesis. To the best of our knowledge, this is the first report on the role of mitophagy in tendon ossification.

PINK1/Parkin axis is the most widely studied pathway in mitophagy to date. Under normal conditions, PINK1 translocates to the inner mitochondrial membrane (IMM), where it is rapidly degraded by several proteases. When mitochondria are damaged with reduced membrane potential, PINK1 degradation is blocked, leading to its accumulation on the outer mitochondrial membrane (OMM). Subsequently, PINK1 phosphorylates ubiquitin and other OMM proteins and promotes Parkin recruitment, which ubiquitinates multiple OMM proteins. These ubiquitinated proteins recruit and bind to the ubiquitin-binding protein p62 and then anchor mitochondria into APs through eventual binding with the autophagic adaptor protein LC3 [41]. The formed mitochondrial APs merge with the lysosomes to form mitochondrial ALs, resulting in the degradation of the damaged mitochondria [19]. Therefore, during mitophagy, the respective number of mitochondria, mitochondrial APs, and mitochondrial ALs can reflect the level of mitophagy. TEM results showed that the mitophagy level in cells increased with the duration of osteogenic induction and was higher in injured tendon-derived cells than in normal tendon-derived cells, which indicated that mitophagy is required for osteogenesis in the tendon. Notably, in the PINK1/Parkin pathway, ubiquitinated OMM proteins can bind to the phagophore by p62 binding to LC3 embedded in the phagophore membrane [42], as well as through direct binding to LC3, and then perform mitophagy [43]. Contradictory reports indicate that Parkin depends [42,44,45] or does not depend [46,47] on p62 to promote mitophagy. This suggests that while p62 can mediate mitophagy, it is not strictly essential for the process. This context may explain why inTPCs expressed more p62 than TSPCs in our study, despite also displaying overall elevated mitophagy levels.

Osteoblasts are differentiated from BMSCs, which participate in bone formation in the skeleton. Osteoblast mitophagy is a vital cellular process for osteogenic differentiation and mineralization [41]. Mitophagy also contributes to BMSC osteogenesis [22,23]. Normally, osteogenesis in the skeleton is a physiological process of “keeping to the beaten track.” Conversely, heterotopic osteogenesis in non-osseous tissues is a pathological process “out of the common template.” The role of mitophagy in heterotopic osteogenesis in the musculoskeletal system has not been previously reported. Vascular calcification (VC), a type of heterotopic mineralization, usually involves an osteogenic reaction [48]. It has been proven that mitophagy alleviates osteogenic differentiation of vascular smooth muscle cells (VSMCs) calcification [49]. There are also reports claiming that excessive mitophagy accelerates osteogenesis of the aorta [50] and aortic valve interstitial cells (AVICs) [51], which is similar to the results of this study. This study revealed that excessive mitophagy promotes the osteogenesis of inTPCs. It follows that the role of mitophagy in osteogenesis cannot be generalized; it may depend on the cell and disease type and may also be related to the degree of mitophagy. Regarding the extent of mitophagy, it is important to note that both overinhibited and overactivated mitophagy are abnormalities that can disrupt mitochondrial homeostasis and drive disease progression [51]. A reasonable number of mitochondria is essential for healthy biological function; therefore, organisms typically modulate mitophagy with a highly precise and complex system [52], and specific molecular regulation may become a new direction for treating some diseases.

Adenine nucleotide translocase (ANT) has been shown to be required for PINK1/Parkin-mediated mitophagy in many cell types and in vivo. ANT regulates presequence translocase TIM23 via interaction with TIM44 to stabilize PINK1 and then promote mitophagy. Notably, contrary to the genetic deletion of ANT1, ANT1 inhibitors impede ADP/ATP transport and accelerate mitophagy, demonstrating that the role of ANT in mitophagy is independent of its ADP/ATP exchange function [25]. ANT1 expression alterations are related to many diseases, but their role remains ambiguous. As previously reported, ANT1 has antiapoptotic and cytoprotective effects on cardiomyocytes [53], whereas it has both proapoptotic and antiapoptotic stimuli in tumor cells [54]. ANT1 overexpression contributes to mitochondrial dysfunction and oxidative stress in muscle cells and impairs muscle and heart development [55]. Conversely, ANT1 upregulation restored mitophagy suppressed by benzo(a)pyrene-7, 8-dihydrodiol-9, 10-epoxide and improved luteal function [56]. In this study, ANT1 gene knockdown was used to regulate mitophagy in inTPCs and tendons. Most importantly, Slc25a4 loss showed the curative ability to alleviate ossification in inTPCs and tendons via the PINK1/Parkin pathway. Mitophagy-associated studies to date have not revealed strategies for treating diseases by specifically targeting mitophagy. Demonstrating the role of ANT1 in mitophagy provides a promising target for modulating mitophagy in the HO remedy. In this study, siRNA and AAV were used to knock down ANT1. Considering the target specificity, using ANT1 knockout (KO) (Slc25a4^−/−) mice can provide stronger evidence for our findings. However, global ANT1 KO mice exhibit phenotypes with exercise intolerance, myopathy, and cardiomyopathy [57], which may affect tendon development and interfere with the experiment results of our models. Future studies can use conditional knockout (CKO) mice to investigate further the role of ANT1-specific expression in tendon HO and the molecular mechanisms behind it. Despite this limitation, the present study is the first to propose that ANT1 contributes to HO via the modulation of mitophagy and can serve as a new cure for abnormal tendon osteogenesis.

In addition to gene therapy, as ANT1 inhibitors, BKA and CATR are toxic agents unsuitable for clinical use. By screening, this study found that elamipretide can bind to and interact with ANT1 in the injured tendon. Elamipretide, also referred to as SS-31, MTP-131, or Bendavia, is a mitochondrial-targeted peptide that selectively binds to cardiolipin on the IMM [58]. By interacting with cardiolipin, elamipretide stabilizes the mitochondrial cristae structure, thereby increasing ATP synthesis and reducing ROS production [59]. Elamipretide has been reported to effectively treat kidney disease, heart failure, and multiple age-associated degenerative diseases [60] and has shown a significant improvement effect on primary mitochondrial myopathy and Barth syndrome in clinical trials [61,62]. Recent reports have shown that elamipretide can also improve mitochondrial function in degenerative tenocytes and promote tendon healing in tendinopathy [63,64]. Tendinopathy also has the potential to progress to tendon ossification [33], both of which are degenerative diseases. This study provides further evidence of the therapeutic efficacy of elamipretide in tendon ossification. Specifically, elamipretide altered the fate of inTPCs and suppressed osteogenic differentiation and ectopic bone formation in the tendon. This process was achieved through the regulation of ANT1-mediated mitophagy by elamipretide. This finding is consistent with reports that elamipretide can directly act on ANT1 [30]. Compared with traditional ANT1 inhibitors, elamipretide has several advantages as a potential ANT1-targeting agent. First, unlike the pharmacological inhibition of ADP/ATP transport by ANT1 inhibitors, elamipretide improved mitochondrial function by direct association with ANT1 and stabilization of the ATP synthasome, of which ANT1 and ATPase are essential components [30]. Although elamipretide inhibits ANT1 expression by interaction, it does not negatively affect ATP production. This allows elamipretide to regulate mitophagy through its interaction with ANT1 specifically. Actually, it may also regulate mitophagy by improving mitochondrial structure. An improvement in the mitochondrial status after the drug administration to inTPCs was also observed in this study. This may be a consequence or cause of the reduction of excess mitophagy, which is attributed to the fact that elamipretide can improve mitochondrial morphology and function and act on ANT1 to affect downstream PINK1/Parkin-mediated mitophagy. This study demonstrated that elamipretide can inhibit mitophagy via ANT1, which was not observed with ANT1 inhibitors. Second, existing evidence indicates that cardiolipin is redistributed from the IMM to the OMM via phospholipid scramblase-3 (PLS3) after mitochondrial damage and then binds to LC3 to engage mitophagy, which is a kind of PINK1/Parkin-independent mitophagy [65]. Accordingly, elamipretide may also regulate mitophagy in the tendon by stabilizing cardiolipin, which needs to be confirmed by further studies. Third, elamipretide has a wide range of efficacy and safety effects. Elamipretide targets all mitochondria and has no detrimental effects on normal mitochondria. Systemic administration of elamipretide is expected to target mitochondria in all tissues and exert beneficial effects [63]. Due to high water solubility, elamipretide does not accumulate in tissues, accounting for its lack of toxicity [58]. Moreover, in previous human studies, elamipretide has shown an excellent safety profile with no untoward “off-target” effects [63,[66], [67], [68]]. Finally, this study demonstrated the efficacy of elamipretide on abnormal osteogenesis in individuals at the cellular level, which can provide instructive information for future human clinical trials of HO treatment and supplement clinical observations. The efficacy and safety of elamipretide have been confirmed in several clinical trials. In the future, including HO as an indication for elamipretide in clinical trials is a possibility worth looking forward to.

However, this study has several limitations. First, we focused exclusively on the PINK1/Parkin axis as the principal mitophagy pathway, without evaluating other mitophagy-related pathways, such as BNIP3/NIX and FUNDC1. A more comprehensive comparison of additional pathways could offer a broader mechanistic understanding of tendon ossification. Second, although we uncovered a functional role of ANT1/PINK1/Parkin-mediated mitophagy in tendon HO, the mechanistic depth remains limited. The precise molecular mechanism by which mitophagy modulates osteogenic signaling requires further investigation. Third, the number of human samples was limited, and larger-scale clinical validation is necessary to confirm the generalizability of our findings. These limitations highlight key directions for future research and should be addressed in subsequent studies to enhance both the mechanistic and translational significance of this work.

5. Conclusion

The present study defined the role of the ANT1/PINK1/Parkin pathway-mediated mitophagy in trauma-induced tendon HO, accounting for a novel mechanism of HO. Tendon injury induced excessive mitophagy in injured tendon progenitor cells, leading to inappropriate activation of cellular osteogenic differentiation and promoting HO formation. Loss of Slc25a4 significantly inhibited tendon osteogenesis by downregulating excessive PINK1/Parkin-mediated mitophagy. Mitochondrial-targeted tetrapeptide elamipretide interacted with ANT1 to modulate mitophagy and reduce the progression of tendon osteogenesis, thus achieving a therapeutic effect on trauma-induced tendon HO. The findings highlight the importance of ANT1 in influencing HO development by regulating mitophagy, providing fresh insights into novel therapies for HO.

CRediT authorship contribution statement

Guanzhi Li: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. Tong Li: Data curation, Formal analysis, Investigation, Methodology, Software, Validation. Ye Deng: Investigation, Methodology, Software, Validation, Visualization. Xiao Deng: Investigation, Methodology, Software. Chao Chen: Investigation, Methodology, Funding acquisition. Bin Yu: Conceptualization, Project administration, Resources, Supervision, Writing – review & editing. Kairui Zhang: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing.

Ethics approval and consent to participate

The human specimen collection adhered to the principles of the Declaration of Helsinki and was approved by the Medical Ethics Committee of Nanfang Hospital of Southern Medical University (No. NFEC-2023-472). Informed consent was obtained from all the patients. The animal study was approved by the Animal Ethic Committee of Nanfang Hospital of Southern Medical University (No. NFYY-2021-0304).

Funding

This study was funded by the National Natural Science Foundation of China (82072430 and 82374605), the Administration of Traditional Chinese Medicine of Guangdong Province, China (20241199), and the Guangdong Basic and Applied Basic Research Foundation (2025A1515012457)

Conflicts of interest

The authors declare no conflicts of interest.

Abbreviations

HO

Heterotopic ossification

PINK1

PTEN induced kinase 1

ANT1

Adenine nucleotide translocase 1

Slc25a4

Solute carrier family 25 member 4

InTPCs

Injured tendon-derived progenitor cells

TSPCs

Tendon stem/progenitor cells

ATP

Aadenosine triphosphate

ACP

Amorphous calcium phosphate

BMSCs

Bone marrow mesenchymal stem cells

ADP

Adenosine diphosphate

PBS

Phosphate-buffered saline

PFA

Paraformaldehyde

HTSPCs

Human tendon stem/progenitor cells

DMEM

Dulbecco's modified Eagle medium

FBS

Fetal bovine serum

ODM

Osteogenic differentiation medium

IHC

Immunohistochemical

H&E

Hematoxylin and eosin

TEM

Transmission electron microscopy

APs

Autophagosomes

ALs

Autolysosomes

MMP/ΔΨm

Mitochondrial membrane potential

qRT–PCR

Quantitative real-time polymerase chain reaction

GAPDH

Glyceraldehyde-3-phosphate dehydrogenase

PVDF

Polyvinylidene difluoride

ECL

Enhanced chemiluminescence

ALP

Alkaline phosphatase

ARS

Alizarin Red S

CCK-8

Cell Counting Kit-8

CCCP

Carbonyl cyanide 3-chlorophenylhydrazone

CsA

Cyclosporin A

siRNA

Small interfering RNA

si-Slc25a4

Slc25a4-specific siRNAs

si-NC

Negative control siRNAs

AAV

Adeno-associated virus

SPR

Surface plasmon resonance

Micro-CT

Micro-computed tomography

SD

Standard deviation

ANOVA

Analysis of variance

Sqstm1

Sequestosome 1

IMM

Inner mitochondrial membrane

OMM

Outer mitochondrial membrane

VC

Vascular calcification

VSMCs

Vascular smooth muscle cells

AVICs

Aortic valve interstitial cells

CKO

Conditional knockout

PLS3

Phospholipid scramblase-3

RMSD

Root mean square deviation

RMSF

Root mean square fluctuation

Rg

Radius of gyration

FEL

Free energy landscape

Appendix A. Supplementary data

The following is/are the supplementary data to this article.