Carnosic acid serves as a dual Nrf2 activator and PTEN/AKT suppressor to inhibit traumatic heterotopic ossification

Abstract

Background

Heterotopic ossification (HO) pathogenesis involves ROS-driven stem cell differentiation. Carnosic acid (CA), a natural antioxidant, remains unexplored for HO.

Methods

In vitro, tendon-derived stem cells (TDSCs) were stimulated with IL-1β, and CA was used for intervention to assess its effects on differentiation and ROS production via real-time quantitative PCR (qPCR), western blotting (WB), and immunofluorescence. Additionally, a burn and Achilles tendon transection-induced mouse model of traumatic HO was established to evaluate the therapeutic potential of CA.

Results

In vitro, CA activated nuclear factor erythroid 2-related factor 2 (Nrf2) and inhibited nicotinamide adenine dinucleotide phosphate oxidase 1 (NOX1), leading to increased antioxidant enzyme activity and reduced intracellular ROS levels. CA also regulated the PTEN/AKT signaling pathway, suppressing osteogenic and chondrogenic differentiation of TDSCs. In vivo, micro-computed tomography (Micro-CT) and histological analyses demonstrated that CA activated Nrf2 and enhanced antioxidant enzyme expression, thereby inhibiting osteogenic and chondrogenic factor expression in Achilles tendon tissue and reducing HO formation.

Conclusions

CA is a novel HO therapeutic by dual targeting of oxidative stress and differentiation pathways.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13287-025-04886-2.

Introduction

Heterotopic ossification (HO) refers to the abnormal presence of bone tissue in connective tissues or muscles. The occurrence of HO within joints can lead to symptoms such as swelling, pain, nerve compression, and joint mobility disorders. HO is a common complication observed in patients who have sustained severe trauma, particularly following surgical intervention [1, 2]. Although the precise mechanisms underlying HO remain incompletely understood, research has identified three essential factors for its development: osteogenic precursor cells, specific inducing stimuli, and a conducive microenvironment [3]. A growing body of evidence suggests that the pathological differentiation of stem cells or progenitor cells within skeletal muscle may play a key role in HO development [4].

Currently, pharmacological anti-inflammatory interventions and radiation therapy (RT) are commonly used to reduce the incidence of HO. Nonsteroidal anti-inflammatory drugs (NSAIDs) represent the most frequently utilized pharmacological strategy for the prevention of acquired HO. However, NSAIDs, RT, and other treatments are associated with potential side effects, including gastrointestinal, renal, and cardiovascular complications, delayed bone and wound healing, electrolyte imbalances, jaw necrosis, and an increased risk of tumorigenesis [5–7]. Furthermore, while surgical excision remains a key treatment option for HO, reoperation may induce additional trauma, particularly in patients with nerve injuries, and carries the risk of HO recurrence [8, 9]. Consequently, the current treatment options for HO are limited, underscoring the need for further investigation into alternative therapeutic strategies.

Previous studies have demonstrated the pivotal role of the IL-1 signaling pathway in the pathogenesis of HO [10]. Specifically, IL-1β, predominantly expressed by CD68 macrophages, drives ectopic bone formation by promoting calcium mineralization and upregulating RUNX2 expression in fibro-adipogenic progenitors (FAPs) within injured musculature. Experimental evidence further revealed that recombinant IL-1β directly enhances the osteogenic differentiation capacity of human-derived FAPs in vitro, whereas this process is effectively suppressed by IL-1 receptor antagonist (IL-1RA) or neutralizing antibodies targeting IL-1 [10]. These findings collectively identify IL-1β as a central mediator of post-traumatic HO, underscoring the therapeutic potential of targeting the IL-1 signaling axis for HO prevention.

Carnosic acid (CA), a phenolic diterpene, has been isolated from Rosmarinus officinalis and Salvia species. The United States Food and Drug Administration (FDA) has approved CA as a food additive [11]. This lipophilic compound is widely recognized for its potent antioxidant properties and has applications across various industries, including food, beverages, personal care, nutrition, and health [12]. Notably, CA can readily cross the blood-brain barrier, exerting protective effects through its robust antioxidant activity [13]. Despite these promising findings, the role of CA in the inflammatory pathological state of HO remains inadequately understood. The present study aimed to investigate the effects of CA on the differentiation of TDSCs into osteogenic, chondrogenic, and tenogenic lineages. We also evaluated the preventive effect of CA on HO in an animal model. Our results demonstrate that CA inhibits trauma-induced HO by suppressing the ROS-mediated PTEN/AKT signaling pathway.

Methods

Reagents and materials

CA was obtained from Push Biotech Co., Ltd. (Chengdu, China, Catalog No. 3650-09-7) with a purity greater than 98%. Mouse TDSCs were procured from Procell Life Sciences Co., Ltd. (Wuhan, China, Catalog No. CP-M176). IL-1β was sourced from Neobioscience Biological Technology Co., Ltd. (Shenzhen, China). The Shanghai, China-based business MedChemExpress (MCE) provided the Cell Counting Kit-8 (CCK-8). The fetal bovine serum (FBS), anti-ACAN and anti-COL2α1 antibodies, high glucose Dulbecco’s Modified Eagle Medium (DMEM), and penicillin-streptomycin solution were obtained from Gibco (Thermo Fisher Scientific, Waltham, MA, USA). Dimethyl sulfoxide (DMSO), bovine serum albumin (BSA), radioimmunoprecipitation assay (RIPA) lysis buffer, Triton X-100, Alizarin Red staining (ARS) solution, 4% paraformaldehyde (PFA) tissue fixative, phosphate-buffered saline (PBS), and phenylmethylsulfonyl fluoride (PMSF) were acquired from Solarbio (Beijing, China). Beyotime Biotechnology Co., Ltd. (Beijing, China) provided the BCIP/NBT Alkaline Phosphatase Staining Kit and the Enhanced Mitochondrial Membrane Potential Assay Kit (JC-1). Primary antibodies (anti-P65, anti-p-P65, anti-AKT, anti-GSK-3β, anti-Runx-2, and anti-ALP) were provided by Cell Signaling Technology (Danvers, MA, USA). Zenbio Biotechnology Co., Ltd. in Chengdu, China, supplied p-PTEN, PTEN, and Nrf2 antibodies. The antibodies against β-actin, IκB-α, NQO1, CAT, HO-1, and GSR were obtained from Abcam (Cambridge, UK). Novus (USA) supplied the anti-NOX1 and anti-Keap1 antibodies. Anti-SOX9 and anti-TNMD antibodies were supported by Bioss Biotechnology Co., Ltd. (Beijing, China).

Animal ethics

C57BL/6J mice were obtained from the Animal Experiment Center at Guangxi Medical University. The Guangxi Medical University Animal Ethics Committee approved all animal research (Approval No.: 202307005) according to the univeristy guidelines.

TDSCs viability assay

TDSCs were cultured in 96-well plates with a density of 8 × 10³ cells per well. They were then grown in a DMEM complete medium consisted of 10% FBS and 1% penicillin/streptomycin. Various concentrations of CA (0, 1.25, 2.5, 5, 10, 20 µM) were introduced after a 24-hour period. The cells were cultured under identical conditions for 7 days. Following this, the medium was replaced with complete medium containing 10% CCK-8 reagent and incubated for 2 h. The TriStar2 LB 942 multimode reader (Berthold Technologies GmbH & Co. KG, Baden) was employed to measure the absorbance at 450 nm.

TDSCs osteogenic, chondrogenic, and tendon differentiation induction

To investigate the effects of CA on the osteogenic, chondrogenic, and tenogenic differentiation of TDSCs, we used specific induction media for each differentiation pathway: osteogenic induction medium (1 µM dexamethasone, 50 µM vitamin C, 10 mM β-glycerophosphate, 10% FBS, and 1% penicillin/streptomycin), chondrogenic induction medium (0.1 µM dexamethasone, 1% insulin-transferrin-selenium (ITS), 50 µg/mL vitamin C, 1 mM sodium pyruvate, 10 ng/mL transforming growth factor-β3, 10% FBS, and 1% penicillin/streptomycin), and tenogenic induction medium (1% non-essential amino acids, 50 µM vitamin C, 10% FBS, and 1% penicillin/streptomycin). TDSCs were resuspended and counted in complete medium, then divided into four groups. After 24 h of culture, each group underwent different interventions: the control group received the corresponding induction medium, the IL-1β group received IL-1β (5 ng/mL) in the induction medium, and the CA intervention groups received both CA and IL-1β in the induction medium. The culture media were replaced every three days throughout the experimental period.

ALP and Alizarin red staining

After 7 days of culture in osteogenic induction medium, cells were fixed with 4% PFA and washed with PBS. Cells were stained using the BCIP/NBT/ALP staining kit at 37 °C for 60 min. After 21 days of osteogenic induction, ARS was performed. Cells were fixed and washed as before, then stained with 0.1% Alizarin Red solution at room temperature for 30 min until cells turned dark red. Images were captured using a Cytation 5 imaging system (BioTek, USA), and staining areas were quantified using Image J software.

Immunofluorescence

After CA treatment, cells were fixed with 4% paraformaldehyde and washed with PBS. Then, the cells were permeabilized with 0.1% Triton and blocked with 3% BSA for 60 min. Primary antibodies against Col2α1 (chondrogenic marker), TNMD (tendon differentiation marker), Nrf2 (antioxidant-related marker), and p-AKT (signaling pathway marker) were added, and the cells were incubated overnight at 4 °C in the dark. After removing the primary antibodies and washing with PBS, a 1:200 dilution of fluorescent secondary antibody was added, and the cells were incubated at room temperature in the dark for 30 min. The cells were washed again with PBS and stained with DAPI for nuclear visualization. After drying, images were captured using a Cytation 5 imaging system.

Reverse transcription-polymerase chain reaction (RT-PCR)

Total RNA was isolated from cells using TRIzol reagent and reverse transcribed into cDNA using the RevertAid First Strand cDNA Synthesis Kit. β-actin was used as the reference gene, and relative mRNA expression was analyzed by RT-PCR. cDNA was used as the template for RT-PCR on the LightCycler^® 96 system, with cycling conditions including an initial 10-minute denaturation at 95 °C, followed by 55 cycles of 15 s at 95 °C, 15 s at 60 °C, and 40 s at 72 °C. RT-PCR results were calculated using the 2^-ΔΔCT method. The primer sequences are listed in Table 1.

Table 1.

RT-PCR primers used in this study

Target Gene	Forward (5′−3′)	Reverse (5′−3′)
Egr1	GAAGGCGATGGTGGAGACG	TTGGTCATGCTCACGAGGC
Mkx	GGCAGAATGGAGGGAAGGT	GGGTACGGGTTGTCACGGT
Col14α1	AGAATCGGCTTGGCACAGT	CGGACAGCATCAATCACCT
Nfe2l2	GGTTGCCCACATTCCCAAAC	TATCCAGGGCAAGCGACTCA
Nqo1	GGTAGCGGCTCCATGTACTC	CGCAGGATGCCACTCTGAAT
Hmox1	GGCTTTAAGCTGGTGATGGCT	GGCGTGCAAGGGATGATTTC
Gsr	TGGCACTTGCGTGAATGTTG	CGAATGTTGCATAGCCGTGG
Actb	TCTGCTGGAAGGTGGACAGT	CCTCTATGCCAACACAGTGC

Western blotting

Cells were lysed on ice for 30 min with RIPA buffer containing protease and phosphatase inhibitors. After centrifugation, the mixture was heated and stored at −20 °C. Proteins were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred to a nitrocellulose membrane. The membrane was incubated with 5% milk for 1 h, washed with TBST, and incubated overnight at 4 °C with primary antibodies (anti-Runx-2, anti-ALP, anti-Sox9, anti-ACAN, anti-NOX1, anti-Nrf2, anti-Keap1, anti-β-actin, anti-IκB-α, anti-NQO1, anti-CAT, anti-HO-1, anti-GSR, PTEN/AKT, and NF-κB signaling pathway-related antibodies). The next day, the membrane was incubated with secondary antibodies for 1 h, washed with TBST, and imaged using the Image Quant LAS-4000 system. The images were analyzed with ImageJ software.

Intracellular ROS detection

The cells were distributed into 96-well plates at a density of 8 × 10³ cells per well. After subjecting the cells to CA intervention at various concentrations for 48 h, they were next exposed to serum-free medium containing 10 µM H2DCFDA in a dark environment for 30 min at a temperature of 37 °C. Subsequently, they were rinsed with serum-free medium and remained in darkness for 15 min. The fluorescence of DCF was quantified using a Cytation 5 imaging system (BioTek, USA) at an excitation wavelength of 488 nm. The analysis of the images was performed using the Image J software.

Enhanced mitochondrial membrane potential detection (JC-1)

Cells were seeded in 96-well plates at a density of 8 × 10³ cells/well. After intervention with different concentrations of CA, cells were cultured in a 37 °C, 5% CO₂ incubator for 48 h. JC-1 was diluted to prepare a staining working solution by adding 5 µL of JC-1 to 1 mL of JC-1 staining buffer. The culture medium was removed, and cells were washed with PBS. Then, 50 µL of culture medium and 50 µL of JC-1 working solution were added to each well, mixed, and incubated at 37 °C for 20 min. The supernatant was removed, and cells were washed with JC-1 buffer. JC-1 aggregate and monomer fluorescence signals were detected using the Cytation 5 imaging system, and images were analyzed with Image J software.

Animal model

A total of 90 seven-week-old C57BL/6J mice were randomly divided into 10 groups: 2 sham-operated groups, sacrificed at 4 and 8 weeks post-operation (Sham group, n = 6 for 4 weeks, n = 12 for 8 weeks); 2 heterotopic ossification groups, sacrificed at 4 and 8 weeks post-operation (HO group, n = 6 for 4 weeks, n = 12 for 8 weeks); 2 indomethacin (10 mg/kg) treatment groups, sacrificed at 4 and 8 weeks post-operation (INDO group, n = 6 for 4 weeks, n = 12 for 8 weeks); 2 low-concentration carnosic acid (10 mg/kg) treatment groups, sacrificed at 4 and 8 weeks post-operation (Low-CA group, n = 6 for 4 weeks, n = 12 for 8 weeks); and 2 high-concentration carnosic acid (20 mg/kg) treatment groups, sacrificed at 4 and 8 weeks post-operation (High-CA group, n = 6 for 4 weeks, n = 12 for 8 weeks). Achilles tendon tissue was collected at each time point for further analysis. The mice were kept in a controlled setting free from pathogens, known as a specific pathogen-free (SPF) environment. The environment had a 12-hour cycle of light and darkness, a constant temperature of 25 °C, and a relative humidity of 60%. Food and drink were provided freely and without restriction. The investigations were conducted after a week of acclimatization.

Construction of the heterotopic ossification model

Based on previous studies, this research establishes a model of HO in mice through Achilles tendon transection and burn injury [14]. C57BL/6J mice were anesthetized intraperitoneally with a 2% tribromoethanol solution (Sigma, USA). A surgical incision was made along the midline of the posterior aspect of the lower leg. The skin was carefully incised with a scalpel, and the Achilles tendon was bluntly dissected and exposed. Hemostatic forceps were used to clamp the Achilles tendon approximately 20 times from top to bottom, after which the tendon was transected at the midpoint with scissors without suturing. A heated aluminum block (60 °C, 2 cm × 2 cm × 3 cm) was then applied to the shaved dorsal skin of the mouse for 16 s. The sham group underwent only the skin incision without Achilles tendon transection or burn injury.

Interventions

CA and INDO were dissolved in DMSO and then diluted with saline. The Sham and HO groups received saline + DMSO solution; the INDO group received saline + DMSO solution + indomethacin (10 mg/kg); the low concentration CA group received saline + DMSO solution + CA (10 mg/kg); the high concentration CA group received saline + DMSO solution + CA (20 mg/kg). All drugs were administered to animals via intraperitoneal injection once daily, with body weight measured weekly for two weeks continuously. At the end of the experiment, the execution methods for C57BL/6 were euthanasia and CO₂ asphyxiation. The body was harmlessly disposed of and medical waste was incinerated.

Micro-computed tomography (micro-CT) scanning

Micro-CT (Skyscan1176m, Bruker, Billerica, MA, USA) was used to evaluate the formation of heterotopic ossification in the Achilles tendon 8 weeks after modeling. The lower limbs of each mouse were separated and preserved in 4% PFA for 2 days, then stored in absolute ethanol. The micro-CT scanning parameters were 9 μm isotropic pixel size, current 379 µA, voltage 65 KV. The mouse lower limb bones were reconstructed using software like CTvox to create 3D images. These images were used to compare bone volume (BV) parameters within the Achilles tendon tissue of each group.

Histological analysis

The tissues were fixed in a solution containing 4% PFA for 48 h. They were then subjected to a decalcification process using a solution containing 10% ethylenediaminetetraacetic acid (EDTA) for 3 weeks. Subsequently, the tissues were embedded and sliced into slices with a thickness ranging from 3 to 5 μm. Then, these sections were stained with hematoxylin, eosin (H&E), and safranin fast green staining. For immunohistochemistry (IHC) and immunofluorescence (IF), a 5% BSA solution was used to block the samples at room temperature for 30 min. The primary antibodies (Nrf2, NQO1, Runx2, SOX9, and TNMD) were incubated overnight at a temperature of 4 °C. The secondary antibodies were incubated at 37 °C for 20 min. Subsequently, the cell nuclei were labeled with DAPI. Images were collected and examined under a microscope.

Statistical analysis

GraphPad Prism 9.0 was employed to conduct the statistical analysis. The mean ± standard deviation was used to represent the experimental data. Tukey’s post-hoc analysis was implemented when required, and one-way ANOVA or Student’s t-tests were implemented for statistical comparisons. Each experiment was conducted at least three times. The significance levels were established: *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001. The work has been reported in line with the ARRIVE guidelines 2.0.

Results

CA inhibits inflammation-induced osteogenic differentiation of TDSCs

The experimental data demonstrated that after 7 days of CA treatment, at concentrations below 20 µM, CA exhibited no significant toxicity and did not impact TDSC proliferation (Fig. 1A). Based on prior studies and the CCK-8 results obtained in this investigation, we selected 10 µM and 20 µM concentrations of CA for further analysis [15, 16]. ALP and ARS staining revealed that, following IL-1β treatment, the number of positive cells significantly increased, whereas CA treatment resulted in a marked reduction in the number of positive cells (Fig. 1B–E). These findings suggest that under conditions of osteogenic induction and IL-1β-induced inflammation, 10 µM and 20 µM CA effectively inhibit the osteogenic differentiation of TDSCs. Additionally, the expression of osteogenesis-related proteins, including ALP and Runx2 was significantly suppressed in TDSCs following CA treatment, as confirmed by Western blot analysis (Fig. 1F–H).

Fig. 1.

CA inhibits inflammation-induced osteogenic differentiation of TDSCs. A The effect of CA intervention at 7 days on the proliferation of TDSCs. B–C Quantitative analysis of ALP and ARS staining. D–E ALP and ARS staining images. F Western blot analysis of osteogenic-related protein expression. G–H Quantitative analysis of the ratio of the gray value of ALP, Runx2 to β-actin bands. All data are presented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001

CA inhibits inflammation-induced chondrogenic differentiation and promotes tendon differentiation of TDSCs

Immunofluorescence analysis revealed that, following IL-1β treatment, Col2α1 expression was significantly elevated, whereas its expression markedly decreased after CA intervention (Figs. 2A and C). Western blot analysis further demonstrated that CA treatment influenced chondrogenic markers, with a significant reduction in the expression of ACAN and SOX9 as CA concentration increased (Fig. 2B). Quantitative analysis revealed that, in the high-concentration CA group, the inhibitory effect on these proteins was significantly more pronounced compared to the low-concentration CA group (Figs. 2D and E).

Fig. 2.

CA inhibits inflammation-induced chondrogenic differentiation of TDSCs and promotes tendon differentiation function. A Representative image showing the inhibition of Col2α1 expression by CA. B Western blot analysis of chondrogenic-related protein expression. C Fluorescence quantitative analysis of Col2α1. D–E Quantitative analysis of the ratio of the gray value of ACAN, SOX9 to β-actin bands. F Representative image showing the promotion of TNMD expression by CA. G Fluorescence quantitative analysis of TNMD. H–J qPCR detection of Mkx, Egr, and Col14α1 expression levels in each group of cells. All data are presented as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001

Tenomodulin (TNMD), a specific marker for tendon cells, is crucial for tendon tissue formation, cell proliferation, and collagen maturation. Immunofluorescence results for TNMD showed a significant increase in expression in the CA intervention groups compared to the IL-1β group (Figs. 2F and G). Additionally, qPCR analysis indicated a significant upregulation of tendon-related markers, including Mkx, Egr1, and Col14α1, following CA intervention (Figs. 2H, I, and J).

CA reduces intracellular ROS via Keap1/Nrf2 signalling and increases antioxidant enzyme expression

The intracellular ROS levels in each group of cells were measured using H2DCFDA. The experimental results demonstrated that IL-1β stimulation led to an increase in ROS levels. However, following treatment with CA, the ROS levels were significantly reduced (Figs. 3A and B). qPCR analysis indicated that CA upregulated the expression of Nfe2l2 and promoted the expression of downstream antioxidant-related genes, including Nqo1, Gsr, and Hmox1 (Figs. 3C–F). JC-1 staining revealed that after IL-1β stimulation, the ratio of JC-1 aggregates to monomers decreased, whereas CA intervention significantly increased this ratio (Figs. 3G and H). Immunofluorescence analysis of Nrf2 showed that red fluorescence in the nucleus of TDSCs was weak in both the unstimulated and IL-1β-stimulated groups. In contrast, CA treatment significantly enhanced Nrf2 nuclear translocation (Fig. 3I).

Fig. 3.

CA reduces intracellular ROS levels by promoting the expression of antioxidant enzymes through the Keap1/Nrf2 signaling pathway. A Effect of CA on intracellular ROS levels during osteogenic differentiation of TDSCs. B Quantitative analysis of DCF fluorescence intensity. C–F qPCR detection of the expression of antioxidant-related genes Nfe2l2, Nqo1, Gsr, and Hmox1. G Quantitative analysis of the fluorescence intensity ratio of JC-1 aggregates to monomers. H Representative image of JC-1 detection of mitochondrial membrane potential in cells. I Representative image of CA-induced nuclear translocation of Nrf2. J Representative image of NOX1 expression detected by Western blotting. K Representative image of Keap1, Nrf2, NQO1, HO-1, GSR, and CAT expression detected by Western blotting. L–Q Quantitative analysis of the ratios of NOX1, NQO1, HO-1, GSR, CAT to β-actin, and the ratio of Nrf2 to Keap1. All data are expressed as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001

Western blot analysis demonstrated that IL-1β stimulation markedly increased NOX1 expression, while CA treatment led to a significant reduction in NOX1 protein levels, with a more pronounced effect observed at higher CA concentrations (Figs. 3J and L). Moreover, IL-1β stimulation resulted in a significant reduction in the expression of antioxidant enzymes HO-1, CAT, NQO1, and GSR in TDSCs. However, CA treatment notably enhanced the expression of these proteins (Figs. 3K and M–P). Additionally, CA treatment led to a significant decrease in the expression of Keap1 in TDSCs (Fig. 3K). In contrast to the trend observed for Keap1, Nrf2 expression decreased following IL-1β stimulation but was significantly upregulated after CA treatment (Fig. 3K and Q). These findings suggest that CA activates the Keap1/Nrf2 signaling pathway, thereby enhancing the expression of Nrf2-related antioxidant enzymes, such as HO-1, CAT, NQO1, and GSR, which contributes to a reduction in intracellular ROS levels during the osteogenic differentiation of TDSCs.

CA alleviates trauma-induced HO formation in mouse Achilles tendon tissue

To enable an effective comparison, this study uses the clinically relevant HO preventive drug INDO as a comparator [17]. At 8 weeks post-surgery, no significant toxic effects were observed following CA administration in mice (Supplementary Fig. 1). Micro-CT analysis revealed that, compared to the Sham group, the HO group exhibited distinct heterotopic ossification in the Achilles tendon tissue (Fig. 4A). However, following treatment with INDO and varying concentrations of CA, the volume of heterotopic ossification in the Achilles tendon was significantly reduced (Fig. 4D). Histological analyses via H&E and safranin-fast green staining corroborated the CT imaging results, showing a lower number of chondrocytes in the INDO and CA-treated groups compared to the HO group (Figs. 4B and E). Immunohistochemical staining indicated that CA treatment enhanced the expression of Nrf2 (Figs. 4C and F), and the content of NQO1 in the Achilles tendon tissue was also increased (Figs. 4C and G). To examine the early stages of HO formation in the Achilles tendon tissue of HO mice, H&E and safranin-fast green staining were performed on tendon samples at 4 weeks post-surgery, with the results shown in Supplementary Fig. 2.

Fig. 4.

CA alleviates the formation of HO in mouse Achilles tendon tissue following injury. A Micro-CT imaging of HO formation in mouse Achilles tendon tissue at week 8 post-surgery. B HE staining and Safranin Fast Green staining showing the size of HO areas in mouse Achilles tendon tissue. C Representative images of Nrf2 and NQO1 immunohistochemical staining in Achilles tendon tissue of each group. D Quantitative analysis of the volume of HO formation in Achilles tendon tissue of each group. F Quantitative analysis of the HO area from Safranin Fast Green staining in each group. F–G Quantitative analysis of Nrf2 and NQO1 immunohistochemical staining. All bar graphs are expressed as mean ± standard deviation.*P < 0.05, **P < 0.01, ***P < 0.001

CA inhibits inflammation-induced osteogenic differentiation of TDSCs via the PTEN/AKT signaling pathway

As illustrated in Fig. 5A, IL-1β stimulation resulted in an increase in both the fluorescence intensity and nuclear translocation of phosphorylated AKT (p-AKT). However, following CA intervention, both the fluorescence intensity and nuclear translocation of p-AKT were significantly reduced. Phosphorylation levels of PTEN, AKT, and GSK-3β were further analyzed, revealing that 30 min post-IL-1β stimulation, p-AKT expression peaked. This peak was notably diminished at the same time point following CA intervention (Figs. 5B and D). In contrast, at 20 min after IL-1β stimulation, the expression of p-PTEN decreased, whereas CA intervention significantly upregulated p-PTEN expression (Figs. 5B and F). Although the phosphorylation level of GSK-3β showed a slight increase after CA intervention, this change was not statistically significant (Figs. 5B and E).

Fig. 5.

CA inhibits osteogenic differentiation of TDSCs via the PTEN/AKT signaling pathway. A Representative images showing CA inhibition of p-AKT expression and nuclear translocation. B–C Western blot analysis of PTEN/AKT and NF-κB pathway-related proteins. D–F Quantification of the p-AKT/AKT, p-GSK-3β/GSK-3β, and p-PTEN/PTEN band density ratios. G–H Quantification of the p-P65/P65 and IκB-α/β-actin band density ratios. All data are expressed as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001

As shown in the bands in Fig. 5C, five minutes following IL-1β stimulation, the expression of phosphorylated P65 (p-P65) was elevated, but CA intervention did not alter the phosphorylation level of P65 (Figs. 5C and G). Additionally, the expression of IκB-α was reduced following IL-1β stimulation, with a significant decline observed at the 10-minute time point. However, CA intervention did not affect the expression of IκB-α (Figs. 5C and H).

CA inhibits the expression of Runx2 and SOX9 and promotes the expression of TNMD in Achilles tendon tissue of HO mice

Immunofluorescence staining results demonstrated that, 8 weeks post-modeling, the expression levels of Runx2 and SOX9 in the Achilles tendon tissue of HO mice were significantly elevated compared to those in the Sham group. However, following treatment with INDO and CA, the expression of both Runx2 and SOX9 in the Achilles tendon tissue was markedly reduced compared to the HO group (Figs. 6A–B and D–E). Furthermore, the fluorescence intensity of TNMD in the Achilles tendon tissue of HO mice was lower than that observed in the Sham group. In contrast, after treatment with INDO and CA, TNMD expression was significantly higher than in the HO group (Figs. 6C and F). These findings indicate that CA, similar to INDO, effectively inhibits the expression of Runx2 and SOX9 in Achilles tendon tissue of HO mice while promoting the expression of the tendon-specific marker TNMD in vivo.

Fig. 6.

CA inhibits the expression of Runx2 and SOX9 in the Achilles tendons of HO mice and promotes TNMD expression. A–C Immunofluorescence staining showing the expression of Runx2, SOX9, and TNMD in tendon tissues from each group. D–F Quantification of Runx2, SOX9, and TNMD fluorescence in tendon tissues from each group. All bar graphs are expressed as mean ± standard deviation. *P < 0.05, **P < 0.01, ***P < 0.001

Discussion

In the present study, we used INDO as a reference drug to evaluate the therapeutic potential of CA in preventing traumatic HO. Our results demonstrate that CA, at both low (10 mg/kg) and high (20 mg/kg) doses, significantly reduced HO formation in a mouse model, with efficacy comparable to that of INDO. Micro-CT and histological analyses revealed that CA treatment markedly decreased ectopic bone volume and chondrocyte numbers, enhanced the expression of antioxidant markers such as Nrf2 and NQO1, and suppressed osteogenic and chondrogenic factors including Runx2 and SOX9.

Beyond efficacy, the safety profile of CA deserves particular attention. Although NSAIDs such as INDO are widely used for HO prophylaxis, their clinical utility is limited by well-documented adverse effects, including gastrointestinal ulceration, renal impairment, and cardiovascular risks. Notably, a recent study by Danisman et al. demonstrated that CA exhibits gastroprotective effects in an indomethacin-induced gastric ulcer model, attenuating oxidative stress and pro-inflammatory cytokine production (e.g., IL-1β, IL-6, TNF-α) while enhancing the Nrf2/HO-1 antioxidant pathway [18]. This suggests that CA not only lacks the ulcerogenic potential of INDO but may actively protect against NSAID-induced mucosal injury.

Furthermore, as a natural diterpene derived from rosemary and sage, CA has gained significant regulatory recognition, having been approved by the European Union (EU) and granted Generally Recognized as Safe (GRAS) status by the U.S. FDA [11]. Studies in rats indicate that CA has remarkably high oral bioavailability, estimated between 40% and 65%—significantly higher than many other well-known dietary phytochemicals such as curcumin or green tea polyphenols, which often exhibit bioavailability below 5% [19–22]. Plasma concentrations achieved after oral administration in these animal studies (ranging from 26 to 126 µM) fall within the range that elicits pharmacological responses in cell culture models [23]. Based on dose translation from animals to humans, a human equivalent dose for a 60 kg adult could range from 630 mg to 875 mg of CA to achieve these effective plasma levels. Such a dose “could easily be achieved in two capsules,” indicating the practical feasibility of formulating CA as a dietary supplement [21, 22].

The pro-inflammatory cytokine IL-1β has been increasingly recognized as a key mediator in trauma-induced HO. For instance, Li et al. reported that IL-1β, released by pyroptotic macrophages, promotes cellular senescence and osteogenic differentiation of TDSCs via the NF-κB and HMGB1/TLR9 pathways [24]. Similarly, Geng et al. demonstrated that mechanical stress upregulates IL-1β, which in turn downregulates miR-337-3p, leading to increased expression of NOX4 and IRS1, and subsequent activation of JNK and ERK signaling in TDSCs [25]. These pathways converge to promote chondro-osteogenic differentiation and ectopic ossification. In our study, IL-1β stimulation significantly increased ROS levels and decreased antioxidant enzyme expression, which was reversed by CA treatment. This aligns with previous reports that IL-1β enhances NADPH oxidase activity, particularly NOX1 and NOX4, leading to oxidative stress that facilitates stem cell differentiation toward osteogenic and chondrogenic lineages [25].

As a secondary messenger within cells, ROS have multiple downstream signaling pathways. Previous literature has reported that ROS acts upstream of the PI3K/AKT signaling pathway [26, 27]. In numerous experimental models, CA has been demonstrated to possess antioxidant, anti-inflammatory, and anti-apoptotic properties. For example, studies indicate that CA can stimulate the PI3K/Akt signalling pathway, resulting to the overexpression of Nrf2 and then enhancing the production of phase II detoxification enzymes and antioxidants [28]. Another study found that CA could protect neuroblastoma cells from the neurotoxicity caused by paraquat by turning on and increasing the production of antioxidant enzymes through the PI3K/Akt pathway and the Nrf2 activation mechanism [29]. Consistently, CA was found to inhibit the expression of pro-inflammatory cytokines including TNFɑ, IL-1β ect. in collagen-induced arthritis [30].

Furthermore, the PTEN/PI3K/AKT axis is essential for the regulation of cellular processes, including growth, apoptosis, and metabolism, and is associated with a variety of physiological and pathological conditions [31]. DONG et al. found that overexpression of PTEN led to downregulation of markers associated with osteogenic lineage commitment and early differentiation in cells, and the PTEN/PI3K/AKT signaling pathway was found to play a key role in the development of HO formation [32]. Iwasa et al. discovered that in adult chondrocytes, PTEN knockdown triggered PI3K/AKT and enhanced the synthesis and expression of chondrocyte markers such as proteoglycan, aggrecan, and collagen II [33]. PI3K/AKT is also involved in osteoblast activity and bone formation. Previous research has shown that BMP2-induced osteoblastogenesis can be blocked by the PI3K inhibitor LY294002, demonstrating the critical function of the PI3K/AKT pathway in the initialization of osteoblastogenesis by BMP2 [32]. In the process of chondrogenesis, PI3K is necessary for chondrocyte differentiation, survival, and hypertrophy [34]. Another study found that activated expression of AKT in transgenic mice promoted chondrocyte differentiation, while knockout of AKT delayed this process [35]. Consistent with prior studies, this research utilized Western blot analysis to demonstrate that CA effectively suppressed the phosphorylation of AKT in the PI3K signaling pathway and enhanced the phosphorylation of PTEN. Consequently, it inhibited the osteogenic differentiation of TDSCs. The inhibitory effect of CA on HO formation is closely related to the downregulation of these key signaling pathways and the inhibition of osteogenic factor expression, which contributes to Achilles tendon healing. These findings offer valuable insights into the functional mechanism of CA in HO mice and can be used as a foundation for further research.

While the primary mechanisms of CA in HO involve activation of Nrf2-mediated antioxidant responses and suppression of the PTEN/AKT pathway, its potential influence on MAPK signaling should not be overlooked. Existing studies, such as those involving matrine, confirm that inhibition of MAPK pathways—particularly ERK and p38—can suppress macrophage polarization and HO progression [36]. Although the current study does not explicitly examine the effect of CA on MAPK signaling in HO, its known anti-inflammatory properties in other models suggest that CA may also modulate MAPK signaling to reduce inflammation and aberrant differentiation [37]. Integrating MAPK pathway analysis in future studies could further elucidate the comprehensive mechanism of CA in preventing HO.

In addition, in both traumatic and genetic HO models, TGF-β is highly activated during inflammatory and osteogenic phases, promoting the recruitment and osteogenic differentiation of mesenchymal stromal/progenitor cells (MSPCs) [38]. CA has been shown to inhibit TGF-β-induced fibrosis and oxidative stress in renal models by modulating Akt and Nrf2 pathways [39]. In the context of HO, CA may also interact with TGF-β signaling indirectly by suppressing Akt phosphorylation—a pathway implicated in TGF-β-mediated fibrogenic and osteogenic responses. Although direct evidence of CA–TGF-β interaction in HO is not yet fully established, the inhibition of Akt by CA likely disrupts downstream TGF-β signaling, thereby reducing oxidative stress and osteo-chondrogenic differentiation.

However, it is important to acknowledge that the present study relied exclusively on a burn plus Achilles tendon transection-induced mouse model of traumatic HO. While this model effectively recapitulates key aspects of post-traumatic HO, it does not fully encompass the diverse etiologies underlying HO in clinical settings, such as BMP-induced ossification, spinal cord injury-associated HO, or genetic forms like fibrodysplasia ossificans progressiva (FOP). Future investigations employing alternative preclinical models—including BMP implantation, neurogenic HO, or genetically modified FOP mice—would be valuable to further validate the broad therapeutic potential of carnosic acid and enhance the translational relevance of our findings. Besides, while our data strongly suggest the involvement of the Nrf2 and PTEN/AKT pathways, future studies employing genetic knockdown or knockout models in combination with CA treatment will be essential to definitively establish causal relationships and rule out contributions from other interconnected signaling pathways.

Conclusion

In summary, this study demonstrates that CA in TDSCs can enhance the effects of antioxidant enzymes and reduce the overall level of intracellular ROS by activating Keap1/Nrf2 signaling. And inhibited the osteogenic differentiation of TDSCs through the PTEN/AKT signaling pathway (Fig. 7). In addition, in vivo experimental results showed that CA could alleviate trauma-induced heterotopic ossification formation through its antioxidant effect. Therefore, these preclinical data suggest that CA may be a promising candidate for further evaluation in the prevention or treatment of traumatic heterotopic ossification.

Fig. 7.

Mechanism of CA inhibition of heterotopic ossification formation

Author contributions

Donglei Wei: Experimental operation, data analysis, visualization, writing - original draft, writing - review and editing. Dezhi Song: Data analysis, writing - review and editing, funding acquisition. Hui Wang, Yuangang Su, Jiamin Liang, and Jiake Xu: Methodological guidance, animal experiments, interpretation of data and resources. Jinmin Zhao: Project administration and funding acquisition. Qian Liu: Supervision, writing - review and editing, funding acquisition.

Funding

This study was funded by the Guangxi Science and Technology Base and Talent Special Project (Grant No. GuikeAD19254003), Guangxi Natural Science Foundation (2025GXNSFAA069714) and the Guangxi Natural Science Foundation (2023GXNSFDA026058). We appreciate the facilities and technical help provided by the Guangxi Key Laboratory of Regenerative Medicine.

Data availability

All data generated or analyzed during this study are included in this article. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Animal Ethics Committee of Guangxi Medical University approved the animal experiment “Research on the Molecular Mechanism of Carnosic Acid Reduces Traumatic Heterotopic Ossification in Mice through Antioxidant Effects (Approval Number: 202307005)” in accordance with the relevant regulations of the school on July 5, 2023. No human cells/samples were used in this study.

Consent for publication

The informed consent was obtained from study participants.

Competing interests

The authors declare no competing interests.