Abstract

   Pathological new bone formation is a critical feature of the
   progression of ankylosing spondylitis (AS), and spine ankylosis is a
   distinctive feature of this condition. Ligaments are the primary
   regions of pathological new bone formation in AS. Here, it is
   demonstrated that ligament tissue‐derived extracellular vesicles (EVs)
   and their interleukin‐17A (IL‐17A) cargo mediate the communication
   between the tissue and other cells. The investigation revealed that
   IL‐17A in EVs can activate the JAK‐STAT3 pathway, thereby stimulating
   the expression of MMP14 in AS ligament. Overexpression of MMP14 can
   lead to changes in the cytoskeleton and mechanical signaling of
   mesenchymal stem cells and other cells. These alterations in cellular
   cytoskeleton and mechanical signaling at ligament sites in patients
   with AS or in stem cells treated with EVs can result in pathological
   new bone formation. Finally, inhibiting IL‐17A activity and EV
   endocytosis effectively controlled inflammation and pathological new
   bone formation. Overall, these data suggest that ligament‐derived EVs
   and the enclosed IL‐17A have a potential role in driving pathological
   new bone formation in AS, and targeting EVs may therefore emerge as a
   novel approach to delaying ectopic ossification in AS.

   Keywords: ankylosing spondylitis, extracellular vesicles,
   interleukin‐17A, pathological new bone formation
     __________________________________________________________________

   Extracellular vesicles (EVs) derived from ligament tissue mediate
   inter‐tissue communication between ligament and bone in ankylosing
   spondylitis (AS). EVs carrying interleukin‐17A promote osteogenic
   differentiation of AS mesenchymal stem cells by altering the cellular
   cytoskeleton and mechanical signaling. This study provides the first
   evidence of tissue EVs contributing to pathological ossification in AS,
   offering potential avenues for targeted therapies.

   graphic file with name ADVS-11-2406876-g009.jpg

1. Introduction

   Ankylosing spondylitis (AS) is a chronic inflammatory disorder
   characterized by a distinct predilection for the sacroiliac, spinal,
   and hip joints,^[ [40]^1 ^] with chronic inflammation and pathological
   fusion of the spine being a particularly distinctive feature. In the
   early stages of the disease, inflammatory pain predominates, while in
   the progressive phase, the spine undergoes a gradual transformation to
   attain a “bamboo‐like” morphology, resulting in progressive loss of
   functional mobility and eventually culminating in spinal ankylosing and
   deformity, necessitating surgical intervention as the sole method of
   treatment.^[ [41]^2 ^]

   Current pharmacotherapeutic interventions for AS primarily focus on
   symptom control. Non‐steroidal anti‐inflammatory drugs (NSAIDs), tumor
   necrosis factor‐alpha (TNF‐α) inhibitors, interleukin‐17 (IL‐17)
   inhibitors, and more recently, Janus kinase (JAK) inhibitors are
   regarded as the mainstay drugs in the early therapeutic “golden window”
   in AS.^[ [42]^3 ^] However, despite their effectiveness in controlling
   inflammation, alleviating symptoms, and slowing disease progression,
   the majority of patients do not receive satisfactory relief with regard
   to pathological ossification after treatment.^[ [43]^4 ^] Moreover,
   details of the underlying mechanisms of pathological ossification in AS
   remain unclear, and the intricate interplay between pathological new
   bone formation and inflammation remains a critical unresolved puzzle in
   this context. Undoubtedly, TNF is not the sole factor implicated in
   driving pathological osteogenesis.^[ [44]^5 ^] Whereas both
   inflammation and mechanical stress have been reported to activate
   signaling pathways, such as the Hippo, Hedgehog, bone morphogenetic
   protein (BMP), and Wnt signaling pathways, to contribute to spinal
   fusion in patients with AS.^[ [45]^6 ^] However, a precise
   understanding of the cellular and molecular mechanisms involved, as
   well as of whether pro‐inflammatory cytokines coordinate
   inflammation‐driven bone remodeling, remains elusive.

   Mesenchymal stem cells (MSCs) and osteoprogenitor cells are considered
   responsive entities committed to osteogenesis within the
   microenvironment conducive to bone formation.^[ [46]^7 ^] Notably, both
   differentiated and recruited MSCs are believed to contribute to
   pathological ossification,^[ [47]^8 ^] but despite some evidence
   regarding the potential origins of these cells, considerable
   controversy remains regarding their migration to new bone formation
   sites and the mechanisms through which they affect pathological
   ossification. Extracellular vesicles (EVs) serve as crucial
   communication mediators between cells, tissues, and organs, and play
   crucial roles in several biological processes. Tissue‐derived EVs are
   nanoscale vesicles that are present in the tissue interstitium and are
   secreted by various cells. They are known to transport a diverse array
   of signaling molecules, including proteins, mRNA, microRNA, and lipids,
   and participate in numerous biological processes across almost all
   organisms.^[ [48]^9 ^] In AS, immunological synapse vesicle transfer
   mediates cell‐to‐cell signaling in T‐cell‐mediated diseases.^[ [49]^10
   ^] Aberrant expression of plasma‐derived EV miRNA in patients may
   contribute to pathogenic mechanisms in AS by disrupting the homeostasis
   of effector T cells and Treg cells.^[ [50]^11 ^] Ligament sites serve
   as focal regions for pathological ossification in AS and generate
   various bioactive mediators that communicate synergistically with
   peripheral organs and cells. However, the molecular mechanisms
   affecting EVs derived from AS ligament and the nature of their
   involvement in the pathogenesis of ossification remain unclear.

   In our previous study, we observed immune hyperactivation in the
   ligament attachments of spinal joints in patients with AS.
   Specifically, RNA sequencing (RNA‐seq) data and bioinformatics target
   prediction and validation were utilized to elucidate the activation of
   inflammatory processes, upregulation of osteogenic markers, and their
   association with helper T cells (Th cells) and macrophages in ligament
   tissue of patients with AS. Utilizing Olink technology for quantitative
   protein analysis of AS ligament tissue EVs (LTEVs), we demonstrated the
   enrichment of interleukin‐17A (IL‐17A) in AS‐LTEVs compared to that in
   EVs from healthy donor (HD) ligament tissue. AS‐LTEVs and their IL‐17A
   cargo have been demonstrated to promote pathological ossification and
   disease progression by altering the cytoskeleton and mechanical
   signaling in stem cells. These findings underscore the need to decipher
   the interplay between EVs and pathological ossification in AS.

2. Results

2.1. Pathological Ossification and Inflammatory Activation in Uncalcified
Ligaments from Patients with AS

   To investigate the inflammatory hyperactivation in the ligament, spinal
   ligament tissue was collected from patients with AS and age‐ and
   sex‐matched controls who had undergone osteotomy surgery (Figure
   [51]1A). All tissues used in the experiments were non‐calcified.
   Safranin O/Fast Green (SOFG) and hematoxylin‐eosin (HE) staining showed
   that spinal ligament from patients with AS exhibited osteogenic
   differentiation and a disorganized arrangement of collagen fibers
   (Figure [52]1B). Immunofluorescence and immunohistochemical analyses
   were used to investigate the localization and expression of osteogenic
   proteins and inflammatory markers in AS ligament tissue and
   demonstrated that Runt‐related transcription factor 2 (RUNX2), CD68,
   and platelet‐derived growth factor receptor alpha (PDGFRα) were
   aberrantly upregulated in AS patient tissue (Figure [53]1C,D,F).
   RNA‐seq analyses of the differentially expressed genes in the tissue
   showed significant enrichment of osteogenic and inflammation‐related
   Gene Ontology (GO) biological process terms, including “regulation of
   osteoblast differentiation” (GO:0045667) and “cytokine‐mediated
   signaling pathway” (GO:0019221) (Figure [54]1E). Spinal ligament tissue
   level of C‐X‐C motif chemokine 12 (CXCL12), a stem cell homing factor,
   was also upregulated in tissue from patients with AS compared to that
   from non‐AS patients (Figure [55]1G). Furthermore, the level of PDGFRα
   was positively correlated with those of markers for AS‐related
   osteoblast differentiation (RUNX2 and CXCL12) and inflammation (CD68)
   (Figure [56]1H). These results indicate that inflammatory processes as
   well as osteogenic differentiation of MSCs are activated in AS
   ligament.

Figure 1.

   Figure 1
   [57]Open in a new tab

   Pathological ossification and inflammatory activation in non‐ossified
   ligaments from patients with AS. A). µCT images show the 3D
   reconstruction and spinal ligament tissue from patients with AS. The
   red pentagram indicates uncalcified ligament. B). SOFG and HE staining
   of the AS patient ligament sections showing osteogenic differentiation
   and disorganized arrangement of collagen fibers. C,D). Representative
   immunohistochemistry and immunofluorescence staining images showing
   over‐expression of RUNX2, CD68, and PDGFRα in AS spinal ligament
   tissue. E). Heatmap of differentially upregulated genes, regulation of
   osteoblast differentiation (GO:0045667), and cytokine‐mediated
   signaling pathway (GO:0019221). F). Quantitative analysis of the
   immunofluorescence area occupied by stained regions in the different
   groups in (D). G). Concentration of CXCL12 in AS ligament tissue
   measured using ELISA. H). Pearson's correlation analysis of PDGFRα
   expression and the expression of markers of AS‐related osteoblast
   differentiation (RUNX2 and CXCL12) and inflammation (CD68) based on the
   ELISA and immunofluorescence staining results. I). Left: experimental
   scheme of AS or HD ligament tissue co‐culture with AS‐MSCs. Middle: ARS
   staining of AS‐MSCs after 21 days of osteogenic induction. Right:
   quantitative analysis of the ARS staining (absorption, 562 nm). n = 5
   per group. J). Experimental scheme of AS or HD ligament tissue
   co‐culture with AS‐BMMCs. Concentration of inflammation factors after
   co‐culturing AS or HD ligament tissue with AS‐BMMCs measured using
   ELISA. IL, interspinous ligament; SL, supraspinous ligament; SP,
   spinous process; HE, hematoxylin‐eosin staining; SOFG, Safranin O–Fast
   Green staining.

   MSCs and bone marrow mononuclear cells (BMMCs) are crucial for
   osteogenesis in ligaments in AS; therefore, we further investigated the
   possible effects of AS ligament changes on MSCs and BMMCs.^[ [58]^12 ^]
   Both MSCs and BMMCs were obtained from bone marrow tissue isolated from
   bone sections via centrifugation; the results of analyses to identify
   the markers of MSCs and their growth curve in the culture medium are
   shown in Figure [59]S1 (Supporting Information). Subsequently, we
   homogenized the ligament tissue and co‐cultured it with these cells. As
   evidenced by Alizarin Red S (ARS) staining, matrix mineralization of
   AS‐MSCs was enhanced by AS ligament tissue treatment (Figure [60]1I).
   Additionally, enzyme‐linked immunosorbent assay (ELISA) results
   revealed that AS ligament tissue induced increased secretion of the
   pro‐inflammatory cytokines TNF‐α, IL‐6, and IL‐1β by BMMCs
   (Figure [61]1J). These results indicate that AS ligament tissue can
   induce the osteogenic differentiation of MSCs and stimulate BMMCs to
   secrete inflammatory factors. Collectively, the histological
   experiments and RNA‐seq analyses demonstrated that both inflammatory
   activation and osteogenic differentiation of MSCs occur within
   ligaments during AS progression. This interesting result piqued our
   curiosity about how the disease affects ligament tissue behavior,
   signaling pathways, and the overall ossification response.

2.2. Isolation and Characterization of EVs from AS Ligament Tissues

   In order to investigate how intercellular communication and
   interactions in ligaments are affected in AS, we used the method
   recommended by the Minimal Information for Studies of Extracellular
   Vesicles (MISEV) guidelines for the characterization and functional
   studies of EVs.^[ [62]^13 ^] AS ligament tissue was extracted from
   non‐calcified regions surrounding ectopic ossification in ligaments
   (Figure [63]2A), and transmission electron microscopy (TEM) analysis
   revealed the presence of vesicles in spinal ligament. Specifically,
   vesicular structures of various sizes but morphologies similar to those
   of EVs were observed in spinal fibroblast spaces (Figure [64]2B).

Figure 2.

   Figure 2
   [65]Open in a new tab

   Isolation and characterization of extracellular vesicles from AS
   ligament tissue. A). Illustration of human spinal ligament tissue
   collection. Yellow regions indicate pathological new bone. B). TEM was
   utilized for the examination of extracellular vesicles within the
   interstitial spaces in spinal ligament tissue. C). Experiment diagram
   of extracellular vesicle isolation procedure by sequential
   centrifugation from ligament tissue sections. The right panel shows
   typical TEM images of EVs from two types of tissues. D). Representative
   images of NTA of isolated AS‐ and non‐AS LTEVs. E). Western blotting
   analysis of EV markers (Alix, CD63, TSG101, CD9). F). Uptake of
   PKH26‐labeled EVs by MSCs and BMMCs. Scale bar, 20 and 5 µm. G).
   Representative IVIS images of nude mouse over 14 days after single
   intra‐ankle joint injection of PKH26‐labeled EVs. Quantitative analysis
   of normalized time‐course fluorescent radiant efficiency in nude mouse
   ankle joints over 14 days (n = 3 per group). TEM, transmission electron
   microscopy; NTA, nanoparticle tracking analysis.

   TEM analysis revealed that the LTEVs had a typical EV morphology with a
   cup‐shaped or vesicular structure (Figure [66]2C, right). To further
   study these vesicular structures, the isolated LTEVs were identified
   based on size using nanoparticle tracking analysis (NTA), which showed
   a mean LTEV size of 140.4 ± 46.7 and 141.5 ± 49.8 nm (Figure [67]2D).
   The NTA and TEM results indicated no significant differences in the
   size and morphology between the two types of LTEVs. Furthermore,
   western blotting analysis confirmed that the collected LTEVs expressed
   several EV marker proteins, including Alix, CD63, CD9, and TSG101
   (Figure [68]2E). To examine whether the collected EVs could be taken up
   by other cells, we performed an in vitro experiment using PKH26‐labeled
   EVs. The results showed strong PKH26 dye signal in the cytoplasm of
   AS‐MSCs and AS‐BMMCs (Figure [69]2F). In vivo near‐infrared
   fluorescence imaging analysis of the biodistribution and biosafety of
   the EVs revealed maximal fluorescence signals in the ankle joints and
   paws of healthy mice after PKH26‐EV injection (Figure [70]2G).
   Representative images of HE staining of major organs (heart, liver,
   spleen, lungs, and kidneys) after 4‐week serial injections (once
   weekly) of HD‐LTEVs and AS‐LTEVs are shown in Figure [71]S2 (Supporting
   Information). These results conclusively illustrate that LTEVs actively
   secreted by ligament tissue can mediate inter‐tissue communication
   between the ligament and other cells.

2.3. AS‐LTEVs Mediated Inter‐Tissue Communication and Osteogenic
Differentiation between Ligament Tissues and MSCs

   Communication between ligaments and cells has been demonstrated in
   preceding experiments. To determine whether AS‐LTEVs can also induce
   the osteogenic differentiation of MSCs like AS ligament tissue, AS‐MSCs
   were co‐cultured with LTEVs to detection of optimal concentration for
   EV function (Figure [72]S3, Supporting Information).

   To verify the effectiveness of LTEVs in promoting osteogenesis in vivo,
   we transplanted AS‐MSCs with LTEVs into nude mice and then performed
   used an immunohistochemical assay (Figure [73]3A) to show that a much
   higher number of RUNX2+ cells was injected with LTEVs in the AS group
   (Figure [74]3B; Figure [75]S4A, Supporting Information).
   Immunofluorescence staining of AS‐LTEVs co‐cultured with AS‐MSCs in
   vitro demonstrated that AS‐LTEV intervention increased the expression
   of RUNX2 in AS‐MSCs (Figure [76]3C; Figure [77]S4B, Supporting
   Information). Assessment of alkaline phosphatase (ALP) activity
   (Figure [78]3D), ARS staining (Figure [79]3E), western blotting
   analysis (Figure [80]3F,G), and RT‐qPCR analysis (Figure [81]3H) of
   osteogenic master genes confirmed the positive impact of AS‐LTEVs on
   the osteogenic differentiation of MSCs. Notably, AS‐LTEVs significantly
   enhanced osteogenic differentiation as early as between 24 and 48 h,
   and this effect persisted even at 7 days after induction, indicating a
   sustained effect of LTEV uptake and activity (Figure [82]3I–M). When
   ligament cells obtained by treating ligament tissue collagenase I was
   co‐cultured with LTEVs, we observed that AS‐LTEVs can also promote the
   osteogenesis of AS ligament fibroblasts (Figure [83]S5, Supporting
   Information). These results demonstrate that AS‐LTEVs can promote the
   osteogenic differentiation of both MSCs and ligament cells.

Figure 3.

   Figure 3
   [84]Open in a new tab

   AS‐LTEVs mediated osteogenic differentiation and communication between
   ligament tissue and MSCs. A). Schematic of the in vivo osteogenesis
   assay. AS‐MSCs and AS‐ or HD‐LTEVs were transplanted into nude mice.
   B). Left: immunohistochemical staining to compare the osteogenic
   effects of AS‐ and HD‐LTEVs on AS‐MSCs in vivo. Scale bar: 20 µm.
   Right: quantitative analysis of RUNX2‐positive cells in the in vivo
   osteogenesis assay. n = 3 per group. C). Immunofluorescence staining
   for RUNX2 in AS‐MSCs treated with AS‐ or HD‐LTEVs for 48 h. Scale bar:
   20 µm, n = 3 per group. D). ALP staining and quantitative analysis of
   AS‐MSCs after treatment with PBS, AS‐LTEVs, or HD‐LTEVs for 48 h with 7
   days osteogenic induction. n = 5 per group. E). ARS staining and
   quantitative analysis of AS‐MSCs after treatment with PBS, AS‐LTEVs, or
   HD‐LTEVs for 48 h with 21 days of osteogenic induction. n = 5 per
   group. F,G). Western blotting and quantitative analysis of osteogenesis
   transcription factors in AS‐MSCs after treatment with PBS, AS‐LTEVs, or
   HD‐LTEVs for 48 h. n = 3 per group. H). Relative expression of mRNA of
   key signaling transcription factors involved in osteogenesis in AS‐MSCs
   determined by RT‐qPCR after co‐incubation with each LTEVs for 48 h. n =
   3 per group. I–L). The effects of AS‐LTEVs on AS‐MSCs were assessed at
   three time points: 24, and 48, and 7 days, using ARS and ALP staining.
   n = 4 per each group. M). RT‐qPCR analysis of osteogenic markers
   (RUNX2/OCN/Osterix/OPN) of AS‐LTEVs on AS‐MSCs at 24, and48 h, and 7
   days. ALP, alkaline phosphatase; ARS, alizarin red S.

2.4. Proteomic Analysis of AS‐LTEVs using Olink Technology Revealed Elevated
Expression of IL‐17A

   EVs serve as carriers that transport bioactive components into target
   cells. The preceding experiments suggested that AS‐LTEVs contain
   bioactive components that positively regulate bone formation. Proteins,
   as the main cargo of EVs, play a crucial role in regulating EV‐mediated
   intercellular communication in numerous biological processes.

   To identify the AS‐LTEV proteins responsible for osteogenic
   differentiation, AS‐LTEVs and HD‐LTEVs were profiled using Olink
   proteomics (Figure [85]4A, left). Ninety‐two LTEV proteins were
   detected and analyzed; thirteen proteins were found to be upregulated
   in the AS‐LTEV group compared to in the HD‐LTEV group. Principal
   component analysis (PCA) of Olink also confirmed that there were
   differences in the expression profiles of AS‐LTEVs and HD‐LTEVs
   (Figure [86]4B). Meanwhile, the differential gene expression analysis
   of Olink‐related target proteins was performed using ligament RNA‐seq
   (Figure [87]4A, right). RNA‐seq of the ligaments revealed 27
   upregulated genes with an intersection of four upregulated proteins
   found in the Olink proteomics dataset (Figure [88]4C). Volcano plot
   analysis showed that IL‐17A, IL‐22RA1, CCL25, and CXCL11 levels were
   increased in both AS‐LTEVs and AS ligament tissue (Figure [89]4D), and
   “chemokine‐mediated signaling pathway” and “interleukin‐20 binding”
   terms were identified based on GO enrichment analysis (Figure [90]4E).
   To further confirm the proteins most likely responsible for
   pathological ossification in AS ligaments, the 13 upregulated proteins
   were analyzed for trend variations between subjects with and without
   AS. Notably, IL‐17A exhibited a more differential upregulation trend
   (Figure [91]4F), and RT‐qPCR results also confirmed the upregulation of
   IL‐17A expression in AS‐LTEVs (Figure [92]4G).

Figure 4.

   Figure 4
   [93]Open in a new tab

   Proteomic analysis of AS‐LTEVs using Olink technology reveals elevated
   expression of IL‐17A. A). Heatmap showing protein expression in AS‐ and
   HD‐LTEVs. Different colors represent different levels of protein
   expression, with blue to red representing low to high levels of
   expression. B). Olink PCA analysis of HD‐LTEVs and AS‐LTEVs. C).
   RNA‐seq of the ligaments revealed 27 upregulated genes with an
   intersection of four upregulated proteins found in the Olink proteomics
   dataset. D). Volcano plot of upregulated and downregulated proteins in
   the HD‐ and AS‐LTEV groups. E). The top 8 significant pathways were
   enriched from the 92 critical proteins in GO enrichment analysis. F).
   Expression tendency of 13 upregulated proteins between AS‐LTEVs and
   HD‐LTEVs using Olink proteomics analysis G). RT‐qPCR analysis of IL‐17A
   expression in EVs. H). Immunofluorescent staining and quantitative
   analysis of CD4‐ and CCR6‐positive cells in AS and HD spinal ligament.
   I). Heatmap of differentially upregulated genes of Th cell and
   macrophage markers in HD and AS spinal ligament. J). Western blotting
   and quantitative analysis of IL‐17A in AS or HD spinal ligament. K).
   The concentration of IL‐17A in ligament tissue from HDs and patients
   with AS was measured using ELISA. L). Representative IVIS images of
   nude mice over 14 days demonstrating the effect of IL‐17A injection on
   the in vivo distribution of EVs. Quantitative analysis of normalized
   time‐course fluorescent radiant efficiency within nude mouse ankle
   joints over 14 days (n = 3 per group).

   Immunofluorescence staining was performed to confirm the presence of
   conditions conducive for IL‐17A production in AS ligament tissue and
   the expression of Th cell markers. The results showed CCR6 and CD4
   co‐staining in AS ligament, and both of these are commonly used markers
   of Th cells in ligament tissue (Figure [94]4H). RNA‐seq analysis of
   ligament tissue also showed that Th17/22 cell and macrophage
   cell‐specific biomarker levels were upregulated (Figure [95]4I). ELISA
   and western blotting analyses confirmed that the level of IL‐17A in
   spinal ligament tissue was higher in patients with AS than in non‐AS
   patients (Figure [96]4J,K). Finally, IVIS imaging over 14 days
   demonstrated that IL‐17A injection could accelerate the biodistribution
   of AS‐LTEVs in nude mice (Figure [97]4L). These results indicate that
   the overexpression of IL‐17A in AS‐LTEVs could play a key role in the
   progression of pathological ossification in AS.

2.5. AS‐LTEVs Promote Pathological Ossification by Activating the
STAT3/MMP14/Hippo Signaling Pathway

   To explore the downstream signaling pathways of IL‐17A, pathway
   enrichment analysis was conducted and revealed a significant enrichment
   of genes related to the Hippo signaling pathway, extracellular matrix
   (ECM) organization, and positive regulation of protein phosphorylation
   (Figure [98]5A; Figure [99]S6, Supporting Information). IL‐17A can
   activate the JAK/signal transducer and activator of the transcription
   (STAT) pathway by inducing rapid tyrosine phosphorylation of STAT 1, 2,
   3, and 4.^[ [100]^14 ^] Our results demonstrated that AS‐LTEVs can also
   activate the JAK‐STAT3 pathway (Figure [101]S7A, Supporting
   Information). Matrix metalloproteinase (MMP) family proteins are major
   players in ECM organization, and RNA‐seq analysis revealed the
   upregulation of MMP14 gene expression in AS patient ligament
   (Figure [102]1E). Yes‐associated protein (YAP) is the core effector of
   the Hippo signaling pathway, and its phosphorylation negatively
   regulates osteogenesis.^[ [103]^15 ^]

Figure 5.

   Figure 5
   [104]Open in a new tab

   AS‐LTEVs promote pathological ossification by activating the
   STAT3/MMP14/Hippo signaling pathway. A). Pathway enrichment analysis of
   differentially expressed genes in spinal ligaments from patients with
   AS. B). Immunohistochemical staining of phosphorylated YAP (S127) and
   MMP14 in entheseal tissues. Scale bar: 100 µm. C). The concentration of
   MMP14 in ligament tissue from HDs and patients with AS was measured
   using ELISA. D,E). Representative flow cytometry analysis of
   MMP14‐positive cells in HD or AS spinal ligament. F,G). Protein docking
   analysis and co‐immunoprecipitation were used to evaluate the
   associativity of phosphorylated STAT3 and MMP14 with the osteogenic
   differentiation of AS‐MSCs. H). Immunofluorescence image of actin
   polymerization (phalloidin, red) and nuclei (DAPI, blue) in AS‐MSCs
   cultured on MMP14‐coated dishes with or without LAT‐A administration
   for 12 h. Scale bar: 10 µm. I). Immunofluorescence imaging and analysis
   of nuclear translocation of YAP in AS‐MSCs treated with AS‐LTEVs or
   plated on MMP14 for 6 h. Scale bar: 50 µm. n = 5 per group. J,K)
   Western blotting analysis of Hippo pathway‐related proteins in AS‐MSCs
   after co‐incubation with each group for 2 days. L,M). Western blotting
   and quantitative analysis of MMP14 and osteogenic markers. N).
   Influence of MMP14 expression on trilineage differentiation of AS‐MSCs.
   O). AS‐LTEVs enhanced AS‐MSC osteogenesis and migration, and this
   enhancement was inhibited by secukinumab (an IL‐17A inhibitor). LAT‐A,
   latrunculin A; YAP, Yes‐associated protein.

   We used immunohistochemical staining to validate the aberrant
   expression of YAP and MMP14 in AS ligament pathological ossification
   and observed that phosphorylated YAP and MMP14 levels in the spinal
   ligament tissue of patients with AS were inversely associated
   (Figure [105]5B), and ELISA results showed a significant increase in
   MMP14 expression in AS ligament tissue (Figure [106]5C). Furthermore,
   flow cytometry analysis revealed that the AS‐ligament had more
   MMP14‐positive cells than HD‐ligament (Figure [107]5D,E). These results
   suggest that there is a correlation between the activation of the Hippo
   and JAK/STAT3 pathways and that this correlation may mediate
   pathological ossification by affecting the expression of MMP14. To
   further validate the correlation between the JAK/STAT3 pathway and
   MMP14, protein docking analysis was used to assess the possible
   interaction between MMP14 and phosphorylated‐STAT3 (Figure [108]S8A,
   Supporting Information). ColabFold was used to model the protein
   interaction structures and conduct a confidence analysis (Figure
   [109]S8, Supporting Information). Finally, of the five predicted
   models, the model with the highest confidence score was used to
   visualize the protein‐protein interaction (Figure [110]5F; Figure
   [111]S8F, Supporting Information). Co‐immunoprecipitation experiments
   indicated that phosphorylated‐STAT3 can interact with MMP14 (Figure
   [112]S7B, Supporting Information) and that this interaction is enhanced
   in the presence of IL‐17A (Figure [113]5G).

   The Rho family GTPase‐effector pathway is a major regulator of stress
   fiber formation, and MMP14 and AS‐LTEVs were found to activate the
   Rho/Rho‐kinase signaling pathway (Figure [114]S7C, Supporting
   Information). Stress fiber formation is a major contributor to
   cytoskeletal modification and MSC differentiation, whereas actin
   cytoskeleton aggregation and disassembly can affect mechanical signals,
   which in turn can activate the Hippo signaling pathway.^[ [115]^16 ^]
   To confirm the activating effect of MMP14 on the Hippo signaling
   pathway, cells were plated on MMP14‐coated dishes to assess actin
   polymerization (Figure [116]5H, left).

   We observed that MMP14 could promote actin cytoskeleton polymerization
   in AS‐MSCs and that AS‐LTEVs could reverse the actin cytoskeleton
   disassembly in AS‐MSCs when latrunculin A (LAT‐A), a cellular actin
   polymerization inhibitor, was administered (Figure [117]5H, right).
   Furthermore, when cells were treated with MMP14 or AS‐LTEVs, the
   nuclear translocation of YAP increased (Figure [118]5I). Western
   blotting analysis showed that the level of phosphorylated large tumor
   suppressor kinase 1 (LATS1) increased and that of phosphorylated YAP
   decreased in AS‐MSCs after AS‐LTEV treatment (Figure [119]5J,K). These
   findings indicate that AS‐LTEVs contribute to actin cytoskeleton
   polymerization, activation of the Hippo signaling pathway, and the
   subsequent nuclear translocation of YAP.

   Western blotting analysis also revealed that MMP14 upregulation under
   the influence of AS‐LTEVs and the enclosed IL‐17A could stimulate the
   expression of RUNX2 during osteogenic differentiation of AS‐MSCs
   (Figure [120]5L,M). MSCs differentiate into osteoblasts when MMP14
   level is elevated, whereas reduction in MMP14 level promotes their
   differentiation into chondrocytes or adipocytes (Figure [121]5N; Figure
   [122]S9A,B, Supporting Information). As expected, AS‐LTEVs induction
   significantly increased matrix mineralization, and inhibiting the
   endocytosis of AS‐LTEVs could reduce matrix mineralization in AS‐MSCs
   (Figure [123]5O). Moreover, AS‐LTEVs could promote the migration of
   MSCs, potentially facilitating their attachment to ligament sites, and
   inhibiting the endocytosis of AS‐LTEVs could prevent the migration of
   AS‐MSCs (Figure [124]S9C, Supporting Information). These results
   suggest that AS‐LTEVs facilitate AS‐MSC osteogenesis by modulating
   MMP14 to activate the Hippo signaling pathway.

2.6. AS‐LTEVs Facilitate Inflammation Activation and Pathological
Ossification by Delivering IL‐17A In Vivo

   As macrophage polarization plays a crucial role in activating
   inflammation, we hypothesized that AS‐LTEVs would amplify this
   polarization effect. Figure [125]6A depicts a schematic of the device
   and experimental design used for this experiment. After 24 h of
   stimulation, flow cytometry was used to assess the proportion of M1
   macrophages with upregulated CD86 and MHC II expression. The results
   demonstrated that AS‐LTEVs induced an increase in BMMC macrophage (M1)
   polarization; conversely, the opposite effects were observed in the
   HD‐LTEV and other treatment groups (Figure [126]6B; Figure [127]S10A,
   Supporting Information). Due to the amplified M1 macrophage
   polarization, the secretion of pro‐inflammatory factors such as TNF‐α,
   IL‐1β, and IL‐6 also increased (Figure [128]6D–F).

Figure 6.

   Figure 6
   [129]Open in a new tab

   AS‐LTEVs facilitate inflammation activation by delivering IL‐17A in
   vivo. A). Diagrammatic representation of experimental strategies for
   macrophage polarization. B). Representative flow cytometry results of
   M1 macrophages after treatment with PBS, HD‐LTEVs, AS‐LTEVs, AS‐LTEVs
   with secukinumab, AS‐LTEVs with secukinumab, and Dynasore. Percentages
   indicate the proportion of M1 macrophages. n = 3 per group. C).
   Representative flow cytometry results of DCFH‐DA after treatment with
   PBS, HD‐LTEVs, AS‐LTEVs, AS‐LTEVs with secukinumab, AS‐LTEVs with
   secukinumab, and Dynasore. The right panel shows the quantitative
   analysis of the positive cell rate of ROS in the different groups. n =
   4 per group. D–F). ELISA‐based measurement of TNF‐α, IL‐10, and IL‐1β
   levels in supernatants prepared from the culture medium after each of
   the different treatments in (B). n = 5 per group. G). Schematic diagram
   of the experimental schedule for the CAIA mouse model. H). µCT scans
   showing pathological new bone formation in hind paws in different CAIA
   model mice groups. I). HE staining of CAIA model mice paws and ankle
   joints sections showing articular destruction and inflammatory cell
   infiltration. J). Quantitative analysis of structural parameters of new
   bone by µCT analysis. n = 5 per group. K–M). Concentration of IL‐17A,
   TNF‐α, and IL‐6 in the serum of CAIA model mice. N. Therapeutic
   efficacy of the various substances assessed in terms of changes in the
   main symptoms of CAIA: arthritis clinical score, paw thickness, paw
   temperature, and pain‐associated behavior analysis using the von Frey
   filament test. n = 5 per group. DCFH‐DA, dichlorodihydrofluorescein
   diacetate; ROS, reactive oxygen species.

   Similarly, as the overproduction of reactive oxygen species (ROS) is
   one of the causes of AS inflammation and macrophage activation, the
   ROS‐producing ability of AS‐BMMCs induced by AS‐LTEVs was investigated.
   Flow cytometry analysis revealed that AS‐LTEVs induced ROS
   overproduction in AS‐BMMCs (Figure [130]6C; Figure [131]S10B,
   Supporting Information). These results indicate that AS‐LTEVs could
   activate the inflammatory response of BMMCs derived from bone marrow.

   The collagen antibody‐induced arthritis (CAIA) model is characterized
   by acute inflammatory responses and pathological ossification of the
   metatarsophalangeal joint (Figure [132]S11, Supporting Information). To
   confirm the critical role of AS‐EVs in pathological new bone formation
   in vivo, AS‐LTEVs with or without an IL‐17A antagonist (secukinumab)
   and an endocytosis antagonist (Dynasore) were administered to CAIA
   model mice after the first induction dose (Figure [133]6G). Following
   intervention with AS‐LTEVs in CAIA mice, µCT analysis revealed that
   pathological new bone formation in the hind paws was strengthened.
   However, combination therapy with secukinumab and Dynasore inhibited
   the osteogenic effects of AS‐LTEVs (Figure [134]6H,J). Furthermore,
   ankle and paw inflammatory swelling was augmented by AS‐LTEVs compared
   to that in the CAIA group (Figure [135]6I). ELISA results showed that
   the serum levels of the inflammation factors TNF‐α, IL‐17A, and IL‐6,
   were increased by the administration AS‐LTEVs (Figure [136]6K–M).
   Meanwhile, arthritis clinical score, paw thickness, paw temperature,
   and paw withdrawal threshold increased in CAIA model mice in the
   AS‐LTEV group and decreased in the combination therapy group
   (Figure [137]6N). These results strongly suggest that AS‐LTEVs
   activated inflammation and pathological new bone formation in this AS
   animal models owing to the presence of IL‐17A within the vesicles.

   Spinal ankylosis and hip joint stiffness are the main features of AS
   due to structural damage and bony fusion in axial joints, including the
   spinal, sacroiliac, and knee joints (Figure [138]7A). A
   proteoglycan‐induced spondylitis (PGIS) mouse model exhibiting both
   axial and peripheral inflammation was established to observe the
   pathological changes in spinal ankylosis. µCT analysis showed that
   spinal ankylosis gradually developed at around 24 weeks after model
   induction (Figure [139]S12, Supporting Information), and AS‐LTEVs could
   accelerate the rate of pathological ossification at the 24th week in
   the PGIS model (Figure [140]7B). µCT results also revealed an increase
   in bone volume/ total volume (BV/TV) in the AS‐LTEV group, whereas a
   decrease was observed in the combination therapy group
   (Figure [141]7C). HE and SOFG staining revealed the pathological
   process of new bone formation at the spinal entheseal and disc site
   (Figure [142]7D).

Figure 7.

   Figure 7
   [143]Open in a new tab

   AS‐LTEVs facilitate pathological new bone formation by delivering
   IL‐17A in vivo. A). Radiological images of patients with AS. Red arrows
   indicate pathological new bone formation and ankylosis of the hip and
   spine. B). µCT scans showing pathological new bone formation in PGIS
   model mouse spine. C). Quantitative analysis of structural parameters
   of newly formed bone based on µCT analysis (n = 5 per group). D). HE
   and SOFG staining of new bone from PGIS model mice. E). Representative
   immunofluorescence staining images showing phosphorylated YAP, MMP14,
   and RUNX2 in the spine of PGIS model mice. Scale bar: 200 µm (n = 5 per
   group). F). µCT scanning and HE staining results indicate a markedly
   reduced DHI because of disc bone formation in PGIS model mice. G–I).
   Quantitative analysis of phosphorylated‐YAP, MMP14, and RUNX2 in PGIS
   model mice. J). Therapeutic efficacy of various test substances in the
   PGIS model in terms of changes in spondylitis histological score.

   Meanwhile, the disc height index (DHI) was suppressed in the PGIS
   model, and the inhibition of AS‐LTEV endocytosis restored the DHI
   suppression caused by pathological ossification (Figure [144]7F).
   Consistent with the osteogenic process of AS‐MSCs, immunofluorescence
   staining revealed that MMP14 and phosphorylated YAP were primarily
   co‐stained with RUNX2 in the spine of PGIS model mice (Figure [145]7E).
   Combination therapy could decrease the expression of MMP14 and promote
   YAP phosphorylation (Figure [146]7G–I). The therapeutic efficacies of
   various substances were assessed in terms of the spondylitis
   histological score in the PGIS model (Figure [147]7J). These results
   indicate that the combination therapy of secukinumab and Dynasore
   targeting AS‐LTEVs significantly suppressed pathological new bone
   formation in AS animal models.

3. Discussion

   The correlation between inflammation and pathological new bone
   formation has so far not been clarified.^[ [148]^6c ^] Owing to the
   limited understanding of the molecular mechanisms underlying AS,
   effective therapeutic strategies involving pharmacotherapy or
   preventive interventions remain elusive. This leaves surgical
   intervention as the sole treatment option for patients in advanced
   stages with severe ankylosing deformities.^[ [149]^4 ^] Thus, a better
   understanding of the mechanism of pathological new bone formation in AS
   is needed to further develop nonsurgical treatment approaches (Figure
   [150]8 ).

Figure 8.

   Figure 8
   [151]Open in a new tab

   Signaling pathways through which AS‐LTEVs promote AS‐MSCs osteogenic
   differentiation by inducing alterations in the cellular cytoskeleton
   and mechanical signaling. Schematic diagram illustrating how EVs
   derived from ligament tissue and the enclosed interleukin‐17A mediate
   inter‐tissue communication between ligament tissue and cells. EVs
   stimulate macrophages to produce inflammatory factors and recruit and
   stimulate MSCs for osteogenic differentiation. MT1‐MMP, membrane‐type 1
   matrix metalloproteinase (also known as MMP14).

   Disease‐related EVs can strongly influence the ability of cells to
   proliferate and differentiate.^[ [152]^17 ^] Both EV‐enclosed proteins
   and RNA are vital players in numerous biological and pathological
   processes. EVs derived from peripheral blood have been reported to
   contribute to immune cell activation in different situations and to
   affect disease activity and severity in AS.^[ [153]^11 , [154]^18 ^] In
   addition, EVs derived from spinal ligament cells contribute to the
   ossification of the posterior longitudinal ligament by promoting
   osteogenic differentiation of MSCs and ligaments.^[ [155]^19 ^] Since
   EVs contribute to various inflammatory and ossification conditions and
   AS is a chronic inflammatory disease, it is reasonable to speculate
   that AS‐EVs may contribute to the process of pathological new bone
   formation in AS. In the current study, we were especially intrigued by
   the tissue immunohistochemistry and RNA‐seq analyses results.
   Infiltration of osteogenic cells and expression of inflammatory
   cytokines in ligament tissue samples collected from patients with AS
   indicated that regions of potential pathological new bone are inflamed,
   which is consistent with previous studies.^[ [156]^20 ^] EVs derived
   from uncalcified ligament tissue in AS were extracted. We discerned a
   pronounced augmentation of these vesicles in both osteogenic processes
   and inflammatory responses. Remarkably, upon administration to spine
   and ankle joints of mice, these AS‐LTEVs demonstrated a remarkable
   capacity to amplify pathological ossification. These findings strongly
   suggest that AS‐LTEV‐mediated AS‐MSC osteogenic activity plays a
   critical role in pathological new bone formation in AS.

   The critical question we aimed to answer through this study was how
   AS‐LTEVs are responsible for the development of pathological new bone
   formation at ligament entheseal sites in AS. We used Olink proteomics
   combined with tissue RNA‐seq analysis to show for the first time that
   IL‐17A is upregulated in LTEVs from patients with AS. Furthermore, we
   confirmed that AS‐LTEVs contributed to potential pathological new bone
   formation in vitro and in the CAIA/PGIS models. Intriguingly, we found
   that the arthritis clinical score decreased with the administration of
   an IL‐17A antagonist in CAIA mice, indicating that AS‐LTEV‐enclosed
   IL‐17A may play a role in maintaining the inflammatory
   microenvironment. Furthermore, this enhanced osteogenic activity and
   serum IL‐17A concentration could be effectively suppressed using
   inhibitors targeting IL‐17A, indicating that IL‐17A delivered from
   AS‐LTEVs plays a crucial role in the pathological new bone formation in
   ligament.

   Bone formation begins with the convergence of MSCs at the site of
   future bone formation. During intramembranous ossification, MSCs
   proliferate and undergo compact condensation, and a subset of these
   MSCs differentiates into pre‐osteoblasts and then osteoblasts.^[
   [157]^21 ^] Some recent studies provide strong evidence for a direct
   and crucial contribution of osteoblasts in direct osteogenesis in AS.^[
   [158]^7 , [159]^12 ^] Several studies have also demonstrated
   pro‐osteoblastogenic effects of IL‐17A. IL‐17A is produced by Th17
   cells, the main T cell subset, and can stimulate osteoblast
   differentiation and migration.^[ [160]^22 ^] IL‐17A has been shown to
   activate MMPs and promote matrix turnover during chronic inflammation,
   leading to cartilage loss in joint tissue.^[ [161]^22 , [162]^23 ^]
   MMP14 plays an important role in ECM remodeling and enhancing cell
   migration.^[ [163]^24 ^] Consistently, we found high MMP14 expression
   in AS ligament tissue, and stimulation of AS‐MSCs with AS‐LTEVs in
   vitro significantly increased MMP14 expression. The downstream
   molecular signaling pathways of MMP14 vary in different cells and
   pathological processes. In this study, we found that aberrant AS‐LTEVs
   and their cargo IL‐17A could activate mechanical signals through the
   upregulation of MMP14 in AS‐MSCs. Previous studies have shown that
   inhibition of actin depolymerization enhances osteoblast
   differentiation and bone formation in human stromal stem cells.
   Cytoskeletal modification mediated by Rho GTPases is a major
   contributor to the differentiation and migration of MSCs.^[ [164]^25 ^]
   Furthermore, the cytoskeleton undergoes differential modifications
   during osteogenesis and adipogenesis.^[ [165]^26 ^] During osteogenic
   differentiation, actin polymerization increases, leading to the
   generation of more stress fibers, and actin bundles are clearly visible
   under a microscope. However, actin polymerization is reduced in
   adipogenic differentiation.^[ [166]^26 , [167]^27 ^] This is consistent
   with our finding that upregulation of MMP14 by AS‐LTEVs mediated the
   promotion of actin polymerization.

   Cytoskeletal modification appears to be an early event guiding the
   differentiation of MSCs.^[ [168]^28 ^] Actin polymerization resulted in
   increased nuclear localization of YAP/TAZ through the activation of the
   Hippo pathway.^[ [169]^16 , [170]^29 ^] YAP shuttles between the
   cytoplasm and nucleus in response to substrate stiffness and other
   environmental cues, and cytoplasmic localization of YAP is associated
   with softer surrounding ECM and adipogenic conditions, whereas YAP
   translocation to the nucleus occurs with a stiffer ECM and osteogenic
   microenvironment.^[ [171]^30 ^] RUNX2 being the first YAP interaction
   partner identified,^[ [172]^31 ^] mechanical signals leads to the
   nuclear translocation of YAP thus promoting the expression of
   interaction partner genes and resulting in accelerated mineralization
   in osteoblasts. This is consistent with our experimental finding that
   activation of the Hippo pathway can be effectively suppressed by
   inhibitors targeting IL‐17A. Taken together, these results indicate
   that aberrant AS‐LTEVs play a vital role in influencing the mechanical
   signals of MMP14, resulting in YAP translocation to the nucleus and
   subsequent enhancement of pathological new bone formation.

   IL‐17A inhibitors such as bimekizumab, secukinumab, and ixekizumab have
   been used in several clinical trials on the treatment of AS as well as
   psoriatic arthritis and certain skin diseases.^[ [173]^32 ^] However,
   although pathological ossification is a long‐term, chronic process
   there has been no long‐term clinical follow‐up assessment of the
   efficacy of IL‐17A inhibitors in inhibiting ossification. Our study
   demonstrated that AS‐LTEVs may serve as a way to mediate the release of
   IL‐17A. Therefore, considering the potential role of EVs in the disease
   process, they could provide a solution to the challenge of preventing
   pathological ossification in AS. Monotherapy alone is unlikely to
   prevent ectopic ossification while controlling inflammation. Hence,
   combination anti‐inflammatory and anti‐ossification therapies have been
   adopted.^[ [174]^33 ^] Dynasore, an inhibitor of the proteins involved
   in endocytosis, inhibited the endocytosis of EVs. We found that
   pathological ectopic ossification and inflammation in the CAIA and PGIS
   models could be significantly inhibited by combined use of IL‐17A
   inhibitors and endocytosis inhibitors. Hence, elucidation of the
   mechanisms underlying pathological ossification and exploration of
   combination therapies for AS may make it possible to fundamentally
   alleviate AS.

   Nonetheless, this study had several limitations. The collection of AS
   tissue samples relies on late‐stage patients with extensive spinal
   fusion, and there may be variations in pathological ossification at
   ligament sites. In addition, although the PGIS and CAIA models share
   similarities with AS and are widely used in AS research, they cannot
   fully replicate its pathophysiology. Therefore, further research to
   establish a more representative animal model of AS is required.
   Moreover, while the focus of this study was on the aberrant activation
   of Hippo in AS‐MSCs, other types of cells and pathways may also be
   activated, and their role in other pathological processes must also be
   investigated.

4. Conclusion

   In summary, we demonstrated that AS‐LTEVs promote pathological new bone
   formation through IL‐17A‐mediated activation of Hippo pathway
   signaling. These findings provide novel mechanistic insights into
   pathological new bone formation and a better understanding of its
   relationship with mechanical strain, which could lead to the
   development of novel therapeutic strategies for pathological new bone
   formation.

5. Experimental Section

Human Tissue

   This study conforms to the Declaration of Helsinki and approved by the
   Ethical Committee of Nanjing Drum Tower Hospital, Affiliated Hospital
   Medical School, Nanjing University (#2011052). Demographic and clinical
   data of patients are shown in Tables [175]S1 and [176]S2 (Supporting
   Information). All patients provided written informed consent. AS
   patients who were unable to maintain a horizontal gaze and comfortable
   extension of the sacroiliac and knee joints in a natural upright
   position met the criteria for deformity correction. Non‐AS patients all
   had spinal fractures and met the criteria for surgical treatment.

Homogenization of Ligament Tissue

   All procedures were performed at 0–4 °C. The non‐ossified ligament
   tissue of interest was excised during surgery, and fat and connective
   tissue were trimmed from the ligament and discarded. The tissue was
   placed in cold homogenization buffer (20 mm HEPES with inhibitor
   cocktails [Table [177]S3, Supporting Information]) and diced into small
   pieces (1 cm cubes) using a knife. Alternatively, the tissue was passed
   twice through a meat grinder. Thereafter, 3–4 volumes of homogenization
   buffer per volume of tissue was added and the mixture was transferred
   into a glass‐Teflon homogenizer. The tissue was homogenized at
   500–1500 rpm, allowing 5–10 s per stroke. Wash the homogenization with
   20 mm HEPES three times to remove inhibitor cocktails. The resulting
   tissue homogenate was used for subsequent co‐culture experiments.

Mice Models

   All PGIS mice were derived using 8‐week‐old male BALB/c mice (weight,
   18–24 g). PGIS mice were obtained using standard methods as previously
   described.^[ [178]^34 ^] Briefly, 100 mg proteoglycan emulsified in
   complete Freund's adjuvant (Chondrex, WA, USA) was subcutaneously
   injected into the back of each mouse. On weeks 3 and 6 from the first
   immunization, a booster injection of the same doses of antigen
   emulsified in incomplete Freund's adjuvant (Chondrex, WA, USA) was
   administered. Induction of CAIA was performed by intravenous injection
   of 1.5 mg of a 5‐clone cocktail on day 0, followed by intraperitoneal
   injection of 25–50 µg of LPS on day 3 in susceptible BALB/c mice.^[
   [179]^35 ^] All animal experiments in this study were approved by the
   Animal Care and Use Committee of the Affiliated Drum Tower Hospital of
   Nanjing University Medical School (DWSY‐22022148). The same particle
   dose (1×10^9, 100 µl) with in vitro experiments is administered
   injections once a week in the in vivo experiments to inhibit
   inflammation and ectopic bone formation in animals. All experimental
   procedures conducted on mice strictly adhered to the rules and
   guidelines for the ethical use of animals.

Cell Culture

   Human bone marrow stem cells (MSCs) and bone marrow mononuclear cells
   (BMMCs) were isolated from 10 patients (5 AS patients and 5 non‐AS
   patients, Table [180]S1, Supporting Information), purified by using
   density gradient centrifugation after red cell lysis, and further
   cultured in Dulbecco's modified Eagle's medium (Gibco, USA) containing
   10% fetal bovine serum (Cellmax, China) and 1% penicillin‐streptomycin
   (Gibco, USA) and cultured in a humidified incubator at 37 °C and 5%
   CO[2]. MSCs were induced for trilineage differentiation experiments
   using osteogenic, chondrogenic, and adipogenic differentiation media,
   respectively.

Osteogenic Differentiation Culture

   MSCs were cultured in a complete medium (a‐MEM with 10% fetal bovine
   serum and 1% penicillin and streptomycin). After 2 days, the medium was
   changed, and the MSCs were seeded at a density of 1 × 10^4 cells per
   well in a 24‐well plate and cultured for 24 h. Thereafter, the complete
   medium was replaced with osteogenic differentiation basal medium
   (Cyagen), and LTEVs (1×10^8 particles/ml) were added for 24, and 48 h,
   or 7 days with differentiation basal medium change every 2 days.
   Osteogenic differentiation was assessed using 1% ARS (Cyagen) staining.
   Calcified nodules were eluted with 10% cetylpyridinium chloride, and
   absorbance was measured at 562 nm.

Isolation and Characterization of Ligament EVs

   Thirty patients (20 AS patients and 10 non‐AS patients) osteotomy
   tissue was enrolled to extract EVs. Non‐ossified ligament tissue was
   collected from surgical specimens, washed three times with PBS, cleared
   of attached fat and connective tissue, and then cut into small pieces
   weighing 0.5–20 g. These were then sliced into small pieces of
   thickness 3–4 mm on ice, which were then incubated with collagenase I
   (2 mg ml^−1; Sigma, USA) and deoxyribonuclease I (40 U/ml; Sigma, USA)
   for 2 h at 37 °C. Filtration through a 0.70 µm pore‐size filter
   (Millipore, USA) was used to remove large tissue blocks, and the
   filtrate was then differentially centrifuged at 800 g for 10 min and
   2000 g for 30 min to remove cells and tissue debris. The obtained
   supernatant was then further centrifuged at 16,000 g for 30 min to
   collect large EVs; small EVs were collected by ultracentrifugation at
   118,000 g for 2.5 h. All centrifugations were performed at 4 °C. Large
   EV‐ and small EV‐enriched pellets were mixed and resuspended in PBS.^[
   [181]^36 ^] Ligament‐derived EVs were resuspended in 200 µl of PBS for
   NTA, TEM, and western blotting or used for co‐culture with MSCs or
   BMMCs.

Analysis of LTEVs Distribution

   Purified PKH26‐labelled LTEVs (1×10^9 particles/ml, 100 µl) were
   administered to CAIA model mice to evaluate biodistribution. The mice
   were imaged using the IVIS Spectrum system 1–14 days after injection,
   and the major organs (heart, liver, spleen, lungs, and kidneys) and
   paws were harvested for in vitro and in vivo imaging for 1–2 s
   (excitation, 551 nm; emission, 567 nm). Data were analyzed using IVIS
   Living Image Software).

Macrophages Polarization

   Bone marrow was obtained by centrifugation of bone sections, and
   single‐cell suspensions were prepared after red cell lysis. Mononuclear
   cells were cultured in a complete DMEM medium containing 10% fetal
   bovine serum (Dakewe, China). Mononuclear cells were induced to
   differentiate into macrophages by adding M‐CSF (25 µg ml^−1) to
   complete culture medium. The matured macrophages were seeded and
   replated, followed by the addition of LPS (50 µg ml^−1) and IFN‐γ
   (20 µg ml^−1) to induce polarization towards M1. After 24 h of
   stimulation, flow cytometry was used to detect the proportion of M1
   macrophages with upregulated CD86 and MHCII expression.

Flow Cytometry

   EVs were added at the time of initiation of macrophage polarization and
   incubated for 48 h. The following antibodies were used to detect
   expression of cell surface antigens by flow cytometry analysis as
   previously described: CD11b‐APC (eBioscience, USA), CD68‐FITC
   (eBioscience, USA), CD86‐Percp (eBioscience, USA), and MHCII‐PE
   (eBioscience, USA). BMMCs were blocked and then incubated with
   antibodies for 30 minutes, washed three times, and resuspended in PBS.
   Candidate cells were detected using a BD Biosciences Influx cell sorter
   and analyzed using FlowJo v10.0 software. CD11b and CD68 positive cells
   were identified as macrophages, and the expression levels of CD86 and
   MHCII were detected to assess the M1 polarization proportion of BMMCs.

Intracellular ROS Assay

   The reactive oxygen species (ROS) detection kit (Beyotime, China) was
   used according to the manufacturer's instructions. Briefly, EVs were
   co‐incubated with macrophage for 48 hours. Macrophages were washed
   three times with PBS and then incubated with DCFH‐DA. After incubation
   at 37 °C for 20 min, the cells were observed under a fluorescence
   microscope, or collected for measurement of DCF fluorescence and mean
   fluorescence intensity (MFI) using flow cytometry (BD Accuri C6 Plus,
   BD Bioscience, USA) to assess intracellular ROS levels.

RNA Sequencing and Data Analysis

   Total mRNA was extracted from AS‐ligament tissue or HD‐ligament tissue
   according to the manufacturer's instructions (Thermo Fisher Scientific,
   USA). The NanoDrop ND‐1000 (NanoDrop, USA) was used to control the
   amount and purity of total RNA. The Agilent 2100 (Agilent, USA) was
   used to test the integrity of RNA, and all samples had an RNA Integrity
   Number (RIN number) >7.0 (indicating the quality of the samples was
   appropriate for subsequent mRNA analysis). The purified mRNA was
   fragmented and reverse‐transcribed to create the final cDNA library in
   accordance with the protocol for the mRNA‐Seq sample preparation kit
   (Illumina, San Diego, USA). The cDNA libraries were then sequenced on
   an Illumina Novaseq 6000 (LC‐Bio Technology Co.) according to the
   manufacturer's instructions.

   The volcano plot, heatmap, and KEGG analysis of differentially
   expressed genes (DEGs) were generated with OmicStudio
   ([182]https://www.omicstudio.cn/index) tools with clusters Profiler R
   package. Heatmaps were generated by uploading the FPKM values for the
   three groups on the website. The screening criteria for DEGs were
   |log2FC|≥1 and p<0.05. Functional enrichment analysis, including GO
   functional annotation and KEGG pathway enrichment analysis, was
   conducted for DEGs. Gene ontology functional annotation was performed
   using Metascape^[ [183]^37 ^] to functionally annotate genes.

Olink Proteomics

   EVs proteins were measured using an Olink Inflammation panel (Olink
   Proteomics, Sweden) according to the manufacturer's instructions.
   Briefly, pairs of oligonucleotide‐labeled antibody probes bind to the
   target protein. The addition of DNA polymerase results in a DNA
   polymerization reaction that produces a unique PCR target sequence.
   Subsequently using a microfluidic real‐time PCR instrument (Signature
   Q100, LC‐Bio Technology Co.) The data were then quality controlled and
   normalized using an internal extension control and an inter‐plate
   control to adjust for intra‐ and inter‐run variation.

Quantitative Real‐Time PCR (RT‐qPCR)

   Total RNA was isolated from MSC and ligament tissue using TRIzol
   reagent (Invitrogen, USA) and transcribed into cDNA using a
   PrimeScriptTM RT reagent kit (Vazyme, China) according to the
   protocols. Quantitative real‐time PCR was performed on the
   LightCycler480 PCR System (Roche, Switzerland) using a SYBR Green Q‐PCR
   kit (Vazyme, China). The relative expression levels of each gene were
   analyzed using the 2−^△△Ct method. The forward and reverse primers for
   each gene are shown in Table [184]S4 (Supporting Information).

Coating with Purified ECM Molecules

   MMP14 and RGD peptides (GCGYGRGDSPG, Genscript) were coated on glass
   coverslips or Matrigel using protocols as previously described,
   Briefly, FN and TNC were coated in 0.01% Tween 20‐PBS at 1–2 ug cm^−2
   before saturation with 10 mg ml^−1 heat inactivated BSA/PBS.

Enzyme‐Linked Immunosorbent Assay (ELISA)

   The following ELISA kits were used according to the manufacturer's
   protocol: CXCL12 (Solarbio, China), TNF‐α (Solarbio, China), IL‐1β
   (Solarbio, China), IL‐6 (Solarbio, China), and MMP14 (Thermo
   Scientific, USA).

Western Blot

   Total protein was extracted from cell and tissue samples using RIPA
   buffer (Sigma, USA), supplemented with phosphatase inhibitors (Bimake,
   USA) and protease inhibitors (Proteintech, China). Cytoplasmic and
   nuclear proteins were separated using a cytoplasmic and nuclear
   extraction kit (Proteintech, China). The protein concentration was
   determined using a BCA protein assay kit (Proteintech, China). Standard
   proteins were separated on 7.5‐12.5% SDS‐PAGE gels and transferred onto
   polyvinylidene difluoride membranes (Millipore, USA) by
   electrophoresis. The membranes were blocked with 5% skim milk.
   Subsequently, primary antibodies were incubated with the membranes
   overnight at 4 °C. Then, secondary antibodies (Abcam, UK) were added at
   room temperature for 1 h. Membrane signals were detected using
   immobilized Western chemiluminescent HRP substrate (Millipore, USA).
   Quantitative densitometry analysis of protein expression was performed
   using ImageJ (version 1.80).

Immunofluorescence Staining

   The intestinal pathological tissues fixed in formalin and embedded in
   paraffin were fixed in 4% paraformaldehyde for 15 min. After washing
   the sections with PBS, they were treated with Triton X‐100 for 15 min
   (Solarbio, China) and blocked for 1 hour with 5% goat serum
   (Proteintech, China). Subsequently, the sections were incubated with
   primary antibodies for 24 h. On the following day, the sections were
   treated with Alexa Fluor‐conjugated secondary antibodies for 1 hour,
   followed by DAPI staining (Beyotime, China). The results were observed
   under a confocal microscope (Leica, Germany).

Co‐Immunoprecipitation (Co‐IP)

   Treating MSCs with IL‐17A for 48 hours, cells were lysed using cell
   lysis buffer containing both protease and phosphatase inhibitors to
   extract total cellular proteins. To validate the interaction between
   phosphorylated STAT3 and MMP14, the lysates were then incubated
   overnight at 4 °C with anti‐phosphorylated STAT3 antibody or IgG.
   Subsequently, protein A/G magnetic beads were added and incubated at
   4 °C for 4–6 h. After multiple washes, immunoprecipitated protein
   complexes were separated, and the interaction between
   phosphorylated‐STAT3 and MMP14 was identified through western blot
   analysis.

Protein Docking Analysis

   Soft protein–protein docking was performed between phosphorylated‐STAT3
   and MMP14 to investigate the relationships by using HDOCK SERVER
   ([185]http://hdock.phys.hust.edu.cn).^[ [186]^38 ^] The protein
   structural domains of phosphorylated‐STAT3 and MMP14 were obtained from
   the Protein Data Bank PDB database ([187]http://www.rcsb.org/)^[
   [188]^39 ^] and Uniport ([189]https://www.uniprot.org/).^[ [190]^40 ^]
   Pymol (Version 2.6) and Colab
   ([191]https://colab.research.google.com/github/sokrypton/ColabFold/blob
   /main/AlphaFold2.ipynb)^[ [192]^41 ^] were used to investigate
   protein‐protein interactions and further visual analysis.

Histological Staining

   Tissues were fixed in formalin buffer, decalcified in 0.5 m EDTA,
   dehydrated in graded ethanol, embedded in paraffin, and sectioned. The
   tissue sections were placed on a hot plate and baked at 65 °C for 1 h,
   followed by dewaxing and hydration according to the aforementioned
   procedure. Sections were stained with hematoxylin and eosin (HE) or
   Safranin O/fast green (SOFG) to evaluate general structure and bone
   formation. Immunohistochemical analysis was performed using specific
   antibodies on specimens. The number of positive cells in each area was
   quantified according to the aforementioned regions.

μ‐CT Analysis

   Micro‐CT scanner (mCT80; Scanco Medical AG, Switzerland) was used to
   analyzed human ligament and mice tissue. 2D and 3D images were
   reconstructed using Scanco Medical software. Quantification of
   heterotopic ossification was performed using the parameters bone
   mineral density/ tissue volume (BV/TV) and bone volume (BV).

Arthritis Clinical Score

   The clinical symptoms of the mice were monitored twice a week. The
   severity of joint swelling was scored as follows: 0 for no joint
   swelling, 0.1 for swelling in a single toe joint, 0.5 for mild swelling
   in the wrist or ankle joint, and 1.0 for severe swelling in the wrist
   or ankle joint.^[ [193]^42 ^] Histological score of the spine was
   determined on a scale of 1–4: 1 = few infiltrating immune cells, 2 =
   mild inflammation of the discs or along the vertebrae (0–30% of the
   discs), 3 = inflammation of the discs and/or along the vertebrae
   (30–70% of the discs), and 4 = inflammation of 70% of the discs and
   along the vertebrae.^[ [194]^43 ^] Pain‐associated behavior of
   CAIA‐induced animals was assessed by measuring the mechanically induced
   pain threshold using a series of von Frey filaments as previously
   described.^[ [195]^44 ^]

Statistical Analysis

   Data obtained from experiments in duplicate or triplicate and repeated
   at least three times were represented as mean SD. For quantitative
   analysis, a minimum of three biological replicates were analyzed. The
   normality of the data was assessed using the Shapiro‐Wilk test. The
   significant differences between two samples were compared by a
   two‐tailed nonpaired Student's t‐test. The multiple‐comparison tests
   were performed using one‐way analysis of variance (ANOVA). One‐way
   ANOVA was performed with Levene's test for homogeneity of variance,
   followed by the Bonferroni post hoc test based on the comparison to be
   made and the statistical indication of each test. A p‐value < 0.05 was
   considered statistically significant. All the data were summarized
   using Origin 2019 software (OriginLabs, USA).

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   K.Y.W., J.S.L., C.Y.S., and M.Q. contributed equally to the
   experimental performing, data acquisition and analysis, and manuscript
   drafting. M.Q., M.H.C., and H.D.B. contributed to clinical tissue
   sample collection. K.Y.W., C.Y.S., and M.Q. contributed to data
   analysis and interpretation. K.Y.W. contributed to electron microscopy
   experiments. B.P.Q. and Y.Q. contributed to manuscript revision. All
   authors have read and approved the current version of the manuscript.

Supporting information

   Supporting Information
   [196]ADVS-11-2406876-s001.docx^ (25.4MB, docx)

Acknowledgements