Mast cell deletion inhibits pathological endochondral osteogenesis in the posttraumatic mouse model. To explore the role of mast cells in traumatic HO, tenotomy was performed on mast cell–deficient (KitW-sh/W-sh) and WT (C57BL/6J) mice. μCT 3D reconstructions disclosed a marked decrease in HO among the KitW-sh/W-sh mice compared with the C57BL/6J mice (Figure 1A and Supplemental Figure 6; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.179759DS1). Specifically, 8, 12, and 24 weeks after tenotomy, heterotopic bone volume (BV) and bone surface area (BA) in the injured tendon tissue of KitW-sh/W-sh mice were markedly lower (Figure 1, H and I). Masson’s trichrome (Masson’s) staining indicated neosynthetic pathological collagen (NPC), while Safranin O and Fast Green (SOFG) staining revealed cartilage matrix (CM). Both NPC and CM formed during the cartilaginous phase of endochondral ossification (13). At 4 weeks after resection in C57BL/6J mice, NPC and CM were most intense, with the mature bone matrix, indicated by red in Masson’s and green in SOFG staining, progressively intensifying by 8 weeks and becoming dominant by 12 weeks (Figure 1, B, C, J, and K). In KitW-sh/W-sh mice, normal tendon tissue remained near the calcaneus at 4 weeks, with markedly lower ectopic osteochondral and mature bone areas at 8 and 12 weeks compared with C57BL/6J mice (Figure 1, B, C, J, and K). While μCT scans showed no statistical difference between groups at 4 weeks after injury, histological staining revealed closer resemblance to normal tendon morphology at the injury site in KitW-sh/W-sh mice (Figure 1, A–C).

Mast cell deficiency inhibits pathological endochondral osteogenesis. (A) Representative μCT 3D modeling images of Achilles tendon (sagittal view) in the indicated groups 4 weeks (cartilaginous phase), 8 weeks (ossification phase), or 12 weeks (osseous phase) after tenotomy. Red dashed ovals represent ectopic bones. Scale bar: 2 mm. (B and C) Representative images for Masson’s staining (cartilage [blue], heterotopic bone and Achilles tendon [red]) and SOFG staining (cartilage [red], heterotopic bone and Achilles tendon [green]) of injured tendon sections in the indicated groups after tenotomy. Scale bar: 5 μm. (D–G) Representative IF staining for SOX9 (red), COL2A1 (green), RUNX2 (red), and OCN (green) of injured tendon sections in the indicated groups after tenotomy, with DAPI counterstaining (blue). Scale bar: 5 μm. (H and I) Representative quantification of A and Supplemental Figure 6, showing ectopic bone volume (BV) and bone surface area (BA) in the indicated groups after tenotomy. n = at least 8 biological replicates. (J and K) Representative quantification of A and B, showing the percentage of neosynthetic pathological collagen (NPC) and cartilage matrix (CM) in injured tendon sections from the indicated groups after tenotomy. n = 5 biological replicates. (L–O) Representative quantification of D–G, showing the positive areas of SOX9, COL2A1, RUNX2, and OCN in injured tendon sections from the indicated groups after tenotomy. n = 5 biological replicates. All data are representative of 2 independent experiments. Data were shown as mean ± SD and compared with 2-tailed unpaired Student’s t test (H and I) or 2-way ANOVA with Šídák’s multiple comparisons test (J–O).

To further elucidate the pathological characteristics of HO following mast cell–specific deletion, we performed immunofluorescence (IF) staining to assess the expressions of the osteochondral marker proteins SOX9, COL2A1, RUNX2, and OCN in the injured tissues. SOX9 (Figure 1, D and L) and COL2A1 (Figure 1, E and M) expression peaked in C57BL/6J mice at 4 weeks after resection and gradually declined thereafter. In contrast, their expression was lowest at 4 weeks after resection, followed by an increasing trend in KitW-sh/W-sh mice. RUNX2 (Figure 1, F and N) expression was highest in C57BL/6J mice at 4 weeks after resection and exhibited a decreasing trend at 8 and 12 weeks. However, RUNX2 expression was nearly absent in KitW-sh/W-sh mice 4 weeks after injury and gradually increased at 8 and 12 weeks. In C57BL/6J mice, OCN (Figure 1, G and O) expression was markedly increased at various time points after resection. On the contrary, KitW-sh/W-sh mice exhibited low expression at 4 weeks after injury and consistently lower expression than C57BL/6J mice from 8 to 12 weeks. In summary, these results indicate that the deletion of mast cell inhibits ectopic bone maturation, impedes the process of endochondral ossification in HO, and attenuates the extent of HO in posttraumatic tendon tissue.

Mast cell activation induced by soft tissue trauma involves ectopic osteoblast activity during HO progression. To ascertain the consequences of mast cell deficiency on the inflammatory response associated with traumatic HO, we examined the expression of IL1B and TNFA in the damaged tissues. A notable decrease in IL-1β expression was observed at 4, 8, and 12 weeks after injury in the KitW-sh/W-sh mice (Figure 2, A and H). Likewise, TNFA expression displayed a marked reduction at 4 and 8 weeks but with no statistically significant difference observed at 12 weeks (Figure 2, B and I). Mast cells not only release proinflammatory mediators but also orchestrate the recruitment of peripheral immune cells to the site of inflammation (22). Thus, we further examined the extent of inflammatory cell infiltration within the damaged tissue by H&E staining. Consistent with our predictions, mast cell deletion led to a marked decrease in inflammatory cell infiltration. The peak inflammatory cell count in C57BL/6J mice was recorded at 8 weeks after injury (Figure 2, C and J). However, KitW-sh/W-sh mice demonstrated markedly lower inflammatory cell numbers at 4 and 8 weeks compared with C57BL/6J mice (Figure 2, C and J). These observations suggest that mast cell deletion mitigates the local inflammatory response in traumatized tissues.

Mast cell activation–induced inflammation correlates with ectopic bone accumulation. (A and B) Representative IF images for IL1B (red) and TNFA (green) of injured tendon sections in the indicated groups after tenotomy, with DAPI counterstaining (blue). Scale bar: 5 μm. (C–E) Representative images for H&E staining TB staining (mast cells [violet]), and IHC staining of CAM1 in injured tendon sections from the indicated groups after tenotomy. Scale bar: 5 μm. (F) Representative images for alkaline phosphatase (ALP [dark purple]) staining of injured tendon sections in the indicated groups after tenotomy. Scale bar: 5 μm (G) Representative CT cross-section of ectopic bone. Black asterisks and red dashed circles represent the tibial and ectopic bone hyperintense images, respectively. (H and I) Quantification of areas with positive staining for IL1B and TNFA in A and B. n = 5 biological replicates. (J–L) Quantification of cells with positive staining for H&E (J), TB (K), and CAM1 (L) in C–E. n = 5 biological replicates. (M) Quantification of areas with positive staining for ALP in F. n = 5 biological replicates. (N and O) Quantification of G, showing ectopic bone mineral content (BMC) and mineral density (BMD) in the indicated groups after tenotomy. n = at least 8 biological replicates. All data are representative of 2 independent experiments. Data were shown as mean ± SD and compared with 2-tailed unpaired Student’s t test.

Tissue toluidine blue (TB) staining showed the presence of mast cells in the traumatized tendon tissue of C57BL/6J mice at 4, 8, and 12 weeks, with a peak cell count observed at 8 weeks after injury (Figure 2, D and K). Mast cell activation status was characterized by IHC staining for the activated mast cell marker chymase (CAM1; encoded by Cam1), revealing similar TB staining results in the injured tendon tissue of C57BL/6J mice (Figure 2, E and L). Conversely, the affected tendon tissue of KitW-sh/W-sh mice exhibited no expression of mast cells (Figure 2, D and E). The results of alkaline phosphatase (ALP) staining paralleled the aforementioned staining trend (Figure 2, F and M), which seemed to suggest that the 8-week time point after tissue trauma represented the peak period of HO. Further μCT analysis showed that both the heterotopic bone mineral density (BMD) and bone mineral content (BMC) in KitW-sh/W-sh mice were markedly lower than in C57BL/6J mice at corresponding time points after injury (Figure 2, G, N, and O). Collectively, these findings imply that mast cell activation induced by tissue trauma may regulate osteoblast activity, subsequently contributing to abnormal soft tissue ossification.

Mast cell crosstalk with NGF influences the pathogenesis of traumatic HO. The inflammatory response induced by bacterial infections is a common trigger for the development of traumatic HO (23, 24). Bacterial-derived LPS is likely to be the key substance leading to the progression of traumatic HO (25). Since mast cells are among the first immune cells to respond to antigens during trauma (9), we initially analyzed the differentially expressed genes (DEGs) in bone marrow–derived mast cells (BMMCs) following LPS treatment in vitro (Figure 3A) (26). Following this, we noted a marked upregulation of Ntrk1 (the gene encoding TrkA), while Ngf (the gene encoding NGF) was not detected in mast cells (Figure 3B). Western blotting further confirmed this finding (Figure 3C). These results suggest that the activation of BMMCs might be regulated by NGF under LPS prestimulation. To test the conjecture, we costimulated BMMCs with LPS at 100 ng/mL and varying concentrations of recombinant murine NGF (rmNGF). Cellular TB staining revealed that treatment with LPS or rmNGF alone did not affect BMMCs (Figure 3H). However, when exposed to a concentration of rmNGF ≥ 10 ng/mL in the presence of LPS (100 ng/mL), BMMCs exhibited cell surface wrinkling and contraction in their morphology (Figure 3H). This characteristic change in cell morphology is known as degranulation (27) and is an indicator of mast cell activation (Figure 3D).

NGF activates mast cell degranulation in traumatized tissue. (A) Schematic representation of microarray analysis. (B) Volcano plot of upregulated (red) and downregulated (yellow) DEGs in BMMCs treated as shown in A from the dataset GSE64287. Expression of the Ntrk1 (red box) is significantly increased (Padj < 0.05, log2FC > 2). (C) Western blot analysis and densitometric quantification (right) of NGF and TrkA in BMMCs treated as shown in A, with β-actin as the loading control. n = 4 biological replicates. (D) Diagram depicting the degranulation of mast cells coactivated by NGF and LPS. (E) Representative IF double-staining images for FCER1A+ (green)/TrkA+ (red), KIT+ (green)/TrkA+ (red), and CAM1+ (green)/TrkA+ (red) cells of injured tendon sections in the indicated group 8 weeks after tenotomy, with DAPI counterstaining (blue). Yellow arrows indicate mast cells. Scale bar: 5 μm (left) and 1 μm (right). (F and G) BMMCs were separately challenged for 1 hour with rmNGF (100 ng/mL), LPS (100 ng/mL), rmNGF (1 ng/mL) + LPS (100 ng/mL), rmNGF (10 ng/mL) + LPS (100 ng/mL), or rmNGF (100 ng/mL) + LPS (100 ng/mL). (F) Histamine (HA) levels and (G) β-hexosaminidase (Hex) release ratios were measured by ELISA after the above treatment. n = 5 biological replicates. (H) Under the conditions previously mentioned (F and G), TB staining and quantification (right) of degranulated cells among BMMCs were performed. Black arrows indicate resting mast cells, and red arrows indicate degranulated mast cells. Scale bar: 5 μm (left) and 2 μm (right). n = 5 biological replicates. Data are representative of 2 independent experiments (C and E–H). Data were shown as mean ± SD and compared with 2-tailed unpaired Student’s t test (C) or 1-way ANOVA with Tukey’s multiple-comparison test (F–H).

Upon degranulation, mast cells release various inflammatory substances and cytokines (28). To further validate the ability of NGF to induce mast cell degranulation, we collected supernatants from BMMCs treated with both rmNGF and LPS and measured histamine (HA) and β-hexosaminidase (Hex) levels by ELISA. BMMCs exhibited increased release of degranulation markers (HA and Hex) in response to increasing rmNGF concentrations during LPS costimulation (Figure 3, F and G). Furthermore, IF double-staining results revealed coexpression of TrkA with the mast cell–specific markers high-affinity immunoglobulin E receptor alpha chain (FCER1A) and cellular homolog of the feline sarcoma viral oncogene kit (KIT), as well as with CAM1 within the damaged tendon tissue (Figure 3E). Taken together, these findings suggest that, upon LPS stimulation, NGF/TrkA signaling activates mast cells and eventually is involved in the development of traumatic HO.

NGF regulates the abnormal osteochondral differentiation of tissue-specific stem cells through the activation of mast cells. To further delineate the potential role of NGF in promoting traumatic HO, we administered 2 agents via i.p. injection to C57BL/6J mice with tenotomy: rmNGF and the TrkA-specific receptor inhibitor GW441756 (GW) (Figure 4A). After 8 weeks of injections, μCT scans showed a marked increase in ectopic BV in the rmNGF group compared with the control group and a marked decrease in the GW group (Figure 4B). Histological staining analysis of the ectopic bone marrow cavity area demonstrated results consistent with the μCT results (Figure 4, C and D). IF staining analysis of SOX9, COL2A1, RUNX2, and OCN revealed that NGF/TrkA signaling markedly promoted the pathological process of endochondral osteogenesis in traumatic HO (Figure 4, E–H). Chondrogenic differentiation of mesenchymal stem cells (MSCs) or precursor cells is considered a crucial step in the formation of HO (1, 4). In this regard, we induced chondrogenic differentiation in mouse tendon-derived stem cells (TDSCs) and performed costimulation with rmNGF or GW. Unexpectedly, both cellular staining (Figure 4, I and J) and Western blotting (Figure 4, K and L) results demonstrate that the addition of rmNGF or GW did not influence the chondrogenic differentiation of TDSCs.

NGF promotes traumatic HO progression without enhancing chondrogenic differentiation of tissue-resident stem cells. (A) Schematic representation of saline, rmNGF, or GW administration and traumatic HO induction. (B) Representative μCT 3D modeling images and quantification (right) of ectopic BV in Achilles tendon from the indicated groups 8 weeks after tenotomy. Red dashed ovals represent ectopic bones. Scale bar: 2 mm. n = at least 6 biological replicates. (C and D) Representative images for SOFG and Masson’s staining, along with the quantification (right) of bone marrow (BM) areas in injured tendon sections from the indicated groups 8 weeks after tenotomy. The BM areas are expressed as a percentage of the total tendon area. Scale bar: 5 μm. n = 5 biological replicates. (E–H) Representative IF images quantification (right) for SOX9 (red), COL2A1 (green), RUNX2 (red), and OCN (green) of injured tendon sections in the indicated groups after tenotomy, with DAPI counterstaining (blue). Positive staining areas of these proteins were quantified. Scale bar: 5 μm. n = 5 biological replicates. (I and J) TB staining was performed on TDSCs challenged with rmNGF (1, 10, 100 ng/mL), GW (1 μM), or rmNGF (100 ng/mL) + GW (1 μM) in chondrogenic culture for 14 days, compared with control (Con) group and chondrogenesis-alone (Ch) group. (K and L) Under the aforementioned conditions (I and J), Western blotting and densitometric quantification (right) of COL2A1 and SOX9 were performed, with β-actin as the loading control. n = 3 biological replicates. Data are representative of 2 independent experiments (B–L). Data were shown as mean ± SD and compared with 1-way ANOVA with Tukey’s multiple-comparison test.

To address this discrepancy, we examined the expression of NGF and TrkA in the traumatized tendon tissue. Although IF staining revealed abundant expression of NGF in the damaged tissues (Supplemental Figure 1, A and C), TrkA expression was primarily localized to s.c. mucosa and bone marrow cells (Supplemental Figure 1, B and D). Both IF double-staining results (Supplemental Figure 1, E and F) and Western blotting (Supplemental Figure 1, G and H) demonstrate the absence of TrkA expression in TDSCs during the pathogenesis of HO. Thus, NGF cannot effectively couple with TrkA to drive the pathological differentiation of tissue-specific stem cells. Remarkably, the s.c. mucosa and bone marrow cavity are areas known for a high concentration of mast cells (29). Simultaneously, single-cell RNA-Seq (scRNA-Seq) data from the traumatized tissue confirmed that mast cells were the exclusive immune cells expressing TrkA (Supplemental Figure 1I). Given our findings of the activation potential of NGF/TrkA signaling in mast cells, we hypothesized that the binding of NGF to the mast cell surface receptor TrkA may lead to excessive immune response activation, ultimately resulting in HO.

Further analysis with IHC staining revealed that i.p. administration of rmNGF in HO-modeled C57BL/6J mice led to a marked increase in IL1B+ and TNFA+ cells, compared with the saline group, while the GW group showed a marked reduction of them (Figure 5, A and B, and Supplemental Figure 7). These results indicate that NGF/TrkA signaling regulates the local inflammatory response in traumatized tissue. Subsequently, TB staining and IF analysis of CAM1+ cells demonstrate that NGF increased the abundance of mast cells in the traumatized tissues and also enhanced their activation (Figure 5, C and D, and Supplemental Figures 7 and 8). Furthermore, i.p. administration of rmNGF was performed in HO-modeled KitW-sh/W-sh mice to observe the formation of HO. Although there was an upward trend in HO after mast cell deletion, there was no statistical difference compared with the saline group (Figure 5E). IF staining of CAM1+ cells also showed the absence of CAM1 expression in both the saline and rmNGF groups of KitW-sh/W-sh mice (Figure 5F). Taken together, these results confirm that NGF/TrkA signaling can upregulate the local inflammatory response through mast cell activation, leading to the pathogenesis of traumatic HO.

NGF/TrkA signaling exacerbates inflammation associated with mast cell activation in trauma-induced HO. (A and B) Representative IF staining images and quantification (right) of (A) IL1B (red) and (B) TNFA (green) in injured tendon sections treated with either rmNGF or GW groups after tenotomy. Black arrows indicate positive cells. Scale bar: 5 μm. n = 5 biological replicates. (C) Representative TB staining images and quantification (right) of the mast cells in injured tendon sections treated with either rmNGF or GW groups after tenotomy. The total number of mast cells was counted. Black arrows indicate mast cells. Scale bar: 5 μm. n = 5 biological replicates. (D) Representative IF staining images and quantification (right) of CAM1 (green) in injured tendon sections treated with either rmNGF or GW groups after tenotomy. Yellow arrows indicate positive cells. Scale bar: 5 μm. n = 5 biological replicates. (E) Representative μCT 3D modeling images of Achilles tendons (sagittal view) in indicated group after tenotomy and quantification (right) of ectopic BV. The red rectangular dashed box represents the reconstruction image of the ectopic bone. Scale bar: 2 mm. n = at least 6 biological replicates. (F) Representative IF staining images and quantification (right) of CAM1 (green) in Achilles tendon sections of C57BL/6J and KitW-sh/W-sh mice treated with or without rmNGF groups after tenotomy, with DAPI counterstaining (blue). The number of CAM1+ cells was counted. The white dashed line indicates the margin of the Achilles tendon and calcaneus. Yellow arrows indicate activated mast cells. Scale bar: 20 μm. n = 5 biological replicates. All data are representative of 2 independent experiments. Data were shown as mean ± SD and compared with 1-way ANOVA with Tukey’s multiple-comparison test.

Activated mast cells secrete NT3 to regulate HO progression. Cytokines released by mast cells are recognized as pivotal regulators of immune responses and potential drivers of abnormal stem cell differentiation (30). Given the role of mast cells in neuroinflammation (29), we posited that they might also release neurotrophins in response to danger signals. Analysis of the microarray dataset (GSE64287) revealed upregulated 3, 519 DEGs (adjusted P [Padj] < 0.05, log2 fold change [log2FC] > 2) in LPS-treated mast cells (Figure 6A). Apart from substance P, a neuropeptide known to be secreted by mast cells (Supplemental Figure 12), we observed marked expression of Ntf3 (the gene encoding NT3) in BMMCs, but Ntrk3 (the gene encoding TrkC) was not expressed (Figure 6A). These findings were further validated by Western blotting (Figure 6B). To further examine the relation between mast cells and NT3 during traumatic HO, we investigated NT3 expression levels in the serum of C57BL/6J and KitW-sh/W-sh mice at different time points after Achilles tenotomy. There was a markedly elevated level of serum NT3 in C57BL/6J mice compared with the sham-operated group but not in KitW-sh/W-sh mice. (Figure 6, F–H). IF and IHC staining results support the aforementioned findings (Figure 6, C–E, and Supplemental Figure 9, A and B). IF double-staining experiments revealed the colocalization of FCER1A and KIT with NT3 (Figure 6, I and J). In summary, these data provide evidence that mast cells are capable of secreting NT3, which may have an influence on HO.

Mast cell–derived NT3 is present throughout the pathogenesis of HO. (A) Volcano plot analysis of upregulated (red) and downregulated (yellow) DEGs in BMMCs treated as shown in Figure 3A from the dataset GSE64287. Expression of the Ntf3 (red box) is significantly increased (Padj < 0.05, log2FC > 2). (B) Western blotting and densitometric quantification (right) of NT3 and TrkC in BMMCs treated with PBS and LPS for 1 hour. β-Actin was used as the loading control. n = 4 biological replicates. (C–E) Representative IHC (C), IF staining images (D), and quantification (E) of NT3 of injured tendon sections in indicated group 4 weeks, 8 weeks, or 12 weeks after tenotomy. Black and yellow arrows indicate NT3+ cells. Scale bar: 5 μm. n = 5 biological replicates. (F–H) Serum NT3 levels were measured 4 weeks, 8 weeks, and 12 weeks after HO modeling in C57BL/6J and KitW-sh/W-sh mice. n = at least 3 biological replicates. (I and J) Representative IF double-staining images and quantification (right) of FCER1A+ (green)/NT3+ (red) and KIT+ (green)/NT3+ (red) cells of injured tendon sections in indicated group 12 weeks after tenotomy, with DAPI counterstaining (blue). The number of positive cells was counted. Yellow arrows indicate mast cells. Scale bar: 5 μm (left) and 1 μm (right). n = 5 biological replicates. Data are representative of 2 independent experiments (B–J). Data were shown as mean ± SD and compared with 2-tailed unpaired Student’s t test (B, E, I, and J) or 1-way ANOVA with Tukey’s multiple-comparison test (F–H).

Next, we investigated NT3 expression in damaged tendon tissue using IF staining at the early stage of HO onset. The results reveal a remarkable increase in NT3 expression in C57BL/6J mice after 4 weeks of HO modeling compared with the sham-operated group (Supplemental Figure 2, A and C). Furthermore, IHC staining provided insight into the spatial distribution of TrkC expression in traumatized tissue. Damaged tendon and bone cells in the HO-modeled group exhibited marked TrkC expression, which was absent in the sham-operated group (Supplemental Figure 2, B and E). Similarly, the expression of NT3 and TrkC in HO-modeled tissue proteins markedly increased compared with the sham-operated group (Supplemental Figure 2D). During chondrogenic differentiation induction in TDSCs, Western blotting results demonstrate markedly higher expression of TrkC (Supplemental Figure 2F). However, the expression of NT3 was comparatively lower and not statistically significant (Supplemental Figure 2F). Above findings suggest that TDSCs may undergo direct regulation through paracrine NT3 during pathological differentiation. To test this hypothesis, we constructed lentivirus-carrying (LV-carrying) Ntrk3 knockdown sequences to transfect TDSCs or cocultured with recombinant human NT3 (rhNT3) in a chondrogenic medium. Western blotting and TB staining revealed a reduction in the degree of chondrogenic differentiation of TDSCs in the Ntrk3 knockdown group, whereas the rhNT3-treated group exhibited an increase (Supplemental Figure 2, G–J). In summary, these findings suggest that mast cell–derived NT3 promotes HO by directly regulating the abnormal chondrogenic differentiation of TDSCs.

Lipid A in LPS cooperates with NGF in binding to TrkA and enhances its phosphorylation. The results of the aforementioned experiments show that LPS-treated mast cells secrete NT3 (Figure 6, A and B). TLR4 is a specific receptor for LPS (31), and its activation may influence NT3 expression in mast cells. To clarify whether mast cell-derived NT3 is regulated by activated TLR4, we inhibited TLR4 activity using the TLR4-specific inhibitor Resatorvid (RD). Afterward, we found NT3 expression was not markedly altered in BMMCs (Figure 7A). Further experiments revealed that neither the gene chip data nor the Western blotting results show marked expression of TLR4 (Figure 7, B and C). These results indicate that LPS-induced secretion of NT3 from mast cells cannot be mediated through activation of TLR4 and that LPS must be acting on a different receptor. Notably, the expression of TrkA was markedly upregulated when mast cells were treated with LPS (Figure 3, B and C), suggesting a potential interaction between LPS and TrkA. Therefore, we conducted an in vitro protein competitive binding assay to confirm this potential interaction. The results demonstrate the existence of mutual binding between LPS and TrkA (Figure 7D).

NGF and lipid A in LPS cobind to TrkA, triggering mast cells to secrete NT3 after trauma. (A) Western blotting and densitometric quantification (right) of NT3 in BMMCs after LPS (100 ng/mL) or RD (1 μM) treatment for 1 hour, with β-actin as the loading control, n = 3 biological replicates. (B and C) Representative (B) expression of Tlr4 (Padj > 0.05, log2FC < 1) in the GSE64287 dataset, confirmed by (C) Western blotting. (D) Western blotting and densitometric quantification (right) of LPS competitive binding to His-tagged or Fc-tagged TrkA in vitro (n = 3 biological replicates). (E) Binding model for LPS (yellow) with TrkA (purple), showing key atoms in LPS (blue sticks) and residues in TrkA (green sticks). (F–I) Representative (F) planar chemical structure, (G) lowest free energy 3D structure of LPS, (H) hydrogen bonds between LPS and TrkA in lipid A (red arrows) or PS (black arrow), and (I) schematic of LPS delipidation strategy. (J) Western blotting and densitometric quantification (right) of TrkATyr490 in BMMCs after treatment with LPS, PS, or rmNGF for 1 hour, with β-actin as the loading control, n = 3 biological replicates. (K and L) Western blotting and quantification (right) of NT3 in BMMCs after pretreatment with LPS (100 ng/mL) or PBS, followed by treatment with rmNGF (0–100 ng/mL) or GW (1 μM) for 1 hour, with β-actin as the loading control, n = 4 biological replicates. (M) IF double-staining images and quantification (right) of CAM1+ (green)/NT3+ (red) cells in injured tendon sections, with DAPI counterstaining (blue). Scale bar: 5 μm (left) and 1 μm (right), n = 5 biological replicates. Data are representative of 2 independent experiments (A–D and J–M). Data shown as mean ± SD, compared by Student’s t test (B–D) or 1-way ANOVA (J–M).

LPS is a macromolecular antigen primarily composed of lipid and polysaccharide (PS) portions, with the lipid portion consisting of lipid A and the PS portion comprising the O antigen and core PS (32) (Figure 7, F and G). Ligand-protein molecular docking analysis showed that there was a marked mutual binding interaction between LPS and TrkA (Figure 7E). Further docking region analysis revealed that H375, H376, O117, O90, O134 in the lipid A portion of LPS, O125 in the core PS portion, and the N-terminal Pro115, Arg116, Ser139, and Gln141 residues in TrkA were hydrogen bonded to each other (Figure 7, E and H). The above findings indicated that lipid A in LPS was the main component used to bind TrkA, so we chemically removed lipid A from LPS to obtain purified PS molecules (Figure 7I). After coculturing PS, LPS, or rmNGF with mast cells, it was found that delipidated LPS could not induce phosphorylation of TrkA, whereas rmNGF and LPS costimulation markedly activated TrkA (Figure 7J). This suggests that Lipid A in LPS acts as a potential costimulatory factor by cooperating with NGF to bind to TrkA, thereby triggering an increased level of TrkA phosphorylation. This may be the reason why NGF combined with LPS can markedly promote the formation of traumatic HO in vivo (Supplemental Figure 3).

NGF-activated mast cells release NT3 to promote trauma-induced HO in vitro and in vivo. Further cell induction experiments showed that NGF promoted NT3 secretion from mast cells in a dose-dependent manner in the presence of LPS (Figure 7, K and L), confirming the involvement of NGF/TrkA signaling in NT3 secretion by mast cells in vitro. In animal experiments, we investigated the coexpression of CAM1 and NT3 in tendon tissues of HO-modeled C57BL/6J mice following i.p. injections of saline, rmNGF, and GW. We observed a marked increase in CAM1+/NT3+ cell numbers in the rmNGF-injected group and a marked decrease in the GW-injected group, both compared with the saline group (Figure 7M). As a whole, these findings indicate that the activation of mast cells through NGF/TrkA signaling contributes to the high expression of NT3. To clarify the role of NT3 secreted by NGF-activated mast cells in aberrant cell fate, we collected concentrated conditioned medium (cM) from different treatments and cultured TDSCs using it for 14 days (Figure 8A). Cellular TB staining demonstrated that the protein factors secreted by mast cells treated with both rmNGF and LPS markedly enhanced the chondrogenic differentiation of TDSCs (Figure 8B). Western blotting confirmed TB staining results (Figure 8, C–E). Specific knockdown of Ntf3 expression in mast cells markedly inhibited the chondrogenic differentiation of TDSCs (Figure 8, B–E).

NGF-activated mast cells release NT3 to promote traumatic HO. (A) Schematic representation of the acquisition of concentrated conditioned medium (cM). (B–E) Representative TB staining (B), Western blotting (C), and densitometric quantification of COL2A1 and SOX9 (D and E) were performed on TDSCs after 14 days of induction culture, following the experimental procedure described in A. In Western blotting, β-actin was used as the loading control. n = at least 3 biological replicates. (F) Schematic representation of experimental protocol of i.v. and i.d. transfer of BMMCLV, BMMCLV: NGF, BMMCNtf3-LV, and BMMCsiNtf3-LV into KitW-sh/W-sh mice for the generation of HO mouse model. (G and I) Representative IF double-staining images and quantification (I) of NT3+ (red)/CAM1+ (green) cells of injured tendon sections in indicated group 8 weeks after tenotomy, with DAPI counterstaining (blue). The number of positive cells was counted. Yellow arrows indicate mast cells. Scale bar: 5 μm. n = 5 biological replicates. (H and J) Representative μCT 3D modeling images of Achilles tendons (sagittal view) in indicated group after tenotomy and quantification (J) of ectopic BV. Red dashed ovals represent the reconstruction image of the ectopic bone. Scale bar: 2 mm. n = at least 10 biological replicates. Data are representative of 2 independent experiments (B–E and G–J). Data were shown as mean ± SD and compared with 1-way ANOVA with Tukey’s multiple-comparison test.

The NT3 neutralizing antibody experiment further confirmed the role of NT3 in promoting the chondrogenic differentiation of TDSCs (Supplemental Figure 5 and Supplemental Tables 3 and 4). In animal experiments, we successfully conducted mast cell reconstitution in KitW-sh/W-sh mice through tail i.v. and i.d. injections of BMMCs (Figure 8F and Supplemental Figure 4, A–C). Following mast cell reconstitution in KitW-sh/W-sh mice, ectopic BV exhibited a notable increase compared with the control group (Supplemental Figure 4D). Moreover, injection of KitW-sh/W-sh mice with BMMCs overexpressing Ntf3 or a mixture of BMMCs and rmNGF resulted in markedly higher BV, compared with the group injected with BMMCs alone (Figure 8, G–J). However, injection of KitW-sh/W-sh mice with BMMCs knockdown Ntf3 showed a marked decrease in ectopic BV (Figure 8, G–J). In conclusion, these experimental findings confirm that NT3 secreted by NGF-regulated mast cell induced the formation of traumatic HO.

Analysis of single-cell data and human tissue samples revealed the presence of NGF-activated mast cells secreting NT3 in the pathological process of HO. Due to our previous study highlighting the importance of NT3 as a crucial link between neurology and immunity in the pathogenesis of traumatic HO, we analyzed the single-cell dataset to identify immune cells capable of expressing Ntf3 (Figure 9, A and B). We found that Ntf3 was expressed by mast cells, monocytes, T lymphocytes, and macrophages, but only mast cells expressed Ntrk1 (Figure 9C). This reinforces the critical role of NGF/TrkA signaling in mast cells for the regulation of NT3 secretion. Subsequent analysis of Ntf3 and Ntrk1 expression in mast cells at different time points (days 0, 3, 7, and 21) following HO modeling revealed the coexpression of both genes on the seventh day after injury. (Figure 9D). This indicates that NT3 secreted by NGF-activated mast cells at this specific time point directly participates in the regulation of abnormal differentiation of TDSCs.

scRNA-Seq analysis and human traumatic tissue assay confirm the involvement of NGF/TrkA signaling and mast cell–derived NT3 in HO. (A) Schematic diagram of the experimental workflow for the scRNA-Seq dataset GSE126060. (B) UMAP plots revealed 12 distinct cell clusters, including mast cells (red dashed line). (C) Bubble plot showing the expression of Ntf3 and Ntrk1 in 5 types of immune cells. (D) Seven days after HO induction, mast cells exhibited a high expression level of Ntf3 and Ntrk1 (red box). (E, F, K, and L) Representative H&E (E), TB staining images (F), and quantification of the total cells (K) and mast cells (L) count in human tendons at 0 and 7 days after trauma, as well as in HO tissues. Black arrows indicate inflammatory cell and mast cell. Scale bar: 5 μm. n = 4 biological replicates. (G, H, M, and N) Representative IHC staining images (G and H) and quantification of NGF, TrkA, and TrkC expression (M and N) in human ligaments at 0 and 7 days after trauma, as well as in HO tissues. Black arrows indicate positive areas. Scale bar: 5 μm. n = 4 biological replicates. (I, J, and O) Representative polychromatic immunofluorescence staining images (I and J) and quantification of TrkC+/TrkA+ (green), CAM1+ (red), and NT3+ (silver) cells (O) in human ligaments at 7 days after trauma, with DAPI counterstaining (blue). The number of colocalized positive cells was counted. Yellow arrows indicate mast cells. Scale bar: 80 μm (left) and 20 μm (right). n = 4 biological replicates. Data are representative of 2 independent experiments (E–O). Data were shown as mean ± SD and compared with 1-way ANOVA with Tukey’s multiple-comparison test (K–M) or 2-tailed unpaired Student’s t test (N and O).

Finally, we performed histological analyses on injured human tissue samples. Staining with H&E (Figure 9, E and K) and TB (Figure 9, F and L) revealed marked infiltration of inflammatory cells and mast cells in the ligament on days 0 and 7 after trauma as well as in the ligament tissue where HO occurred. IHC staining showed that NGF expression increased with time after trauma (Figure 9, G and M). Importantly, marked differences were observed in the expression of TrkA and TrkC within the same tissue region, with TrkA expression being scarce 7 days after injury, while TrkC expression showed a marked increase (Figure 9, H and N). This suggests that NT3/TrkC signaling plays a crucial regulatory role in the aberrant differentiation of tissue-specific stem cells. Moreover, polychromatic IF colocalization analysis of CAM1, NT3, and TrkA demonstrated that activated mast cells express TrkA and secrete NT3 during the pathogenesis of HO in human tissue (Figure 9, I, J, and O). These findings provide further support for our observations.