Abstract

   Soft and hard tissues possess distinct biological properties.
   Integrating the soft‐hard interface is difficult due to the inherent
   non‐osteogenesis of soft tissue, especially of anterior cruciate
   ligament and rotator cuff reconstruction. This property makes it
   difficult for tendons to be mineralized and integrated with bone in
   vivo. To overcome this challenge, a biomimetic mineralization strategy
   is employed to engineer mineralized tendons. The strategy involved
   infiltrating amorphous calcium phosphate precursors into collagen
   fibrils, resulting in hydroxyapatite deposition along the c‐axis. The
   mineralized tendon presented characteristics similar to bone tissue and
   induced osteogenic differentiation of mesenchymal stem cells.
   Additionally, the interface between the newly formed bone and tendon is
   serrated, suggesting a superb integration between the two tissues. This
   strategy allows for biomineralization of tendon collagen and
   replicating the hallmarks of the bone matrix and extracellular niche,
   including nanostructure and inherent osteoinductive properties,
   ultimately facilitating the integration of soft and hard tissues.

   Keywords: Amorphous calcium phosphate, Soft‐hard interface, Tissue
   integration, Tendon‐bone interface, Tendon intrafibrillar
   mineralization
     __________________________________________________________________

   A schematic diagram depicting the process of biomimetic mineralized
   tendon slices fabrication and bone tunnel implantation. The biomimetic
   mineralization of tendon promotes the tendon‐bone interface
   integration.

   graphic file with name ADVS-10-2304216-g004.jpg

1. Introduction

   Regenerating the tendon‐bone junction remains a persistent challenge
   due to its complex hierarchical gradient structure, which comprises the
   tendon, non‐calcified fibrocartilage, calcified fibrocartilage, and
   bone tissue.^[ [48]^1 ^] The gradient transition of these four
   components provides for the smooth transmission of forces produced by
   the muscle to the bone through the tendon, thereby preventing tissue
   rupture caused by stress concentration.^[ [49]^2 ^] However, the
   intricacy of this natural structure makes self‐healing difficult after
   an injury. As a result, surgical intervention is often required for
   patients suffering from tendon‐bone junction rupture.^[ [50]^3 ^]
   Nevertheless, tendons and bones are connected only by fibrous tissue,
   which cannot integrate them, let alone, form a gradient structure.
   Thus, re‐rupture after surgery remains a prevalent concern in the
   clinic.^[ [51]^4 ^]

   Among all tendon‐bone injuries, anterior cruciate ligament (ACL) and
   rotator cuff injuries are the most severe situations, which disrupt
   motor function and necessitate surgical intervention. Reconstructive
   surgery typically involves attaching tendon grafts to the bone by using
   screws and sutures. Although both tendon and bone are comprised of
   collagen, the internal of bone collagen is filled with hydroxyapatite
   (HAP), which significantly enhances its strength and has a greater
   osteogenic capacity compared to the tendon matrix,^[ [52]^5 ^] while
   the tendon will inhibit the osteogenic differentiation of stem cells
   and prevent biomineralization of the matrix in vivo.^[ [53]^6 ^] As a
   result, the tendon and bone may integrate poorly even after a prolonged
   period, possibly leading to both enlargement of the bone tunnel and an
   increased risk of re‐rupture.

   Traditional tissue engineering strategies involve the fabrication of
   gradient scaffolds in vitro and their transplantation to the injury
   site to induce the formation of the hierarchical gradient structure in
   vivo.^[ [54]^7 ^] Li et al. constitute a triphasic silk‐based graft in
   which three regions respectively refer to ligament, cartilage, and bone
   layers. This triphasic ligamentous graft exhibited the enhancement of
   osseointegration.^[ [55]^8 ^] Kim et al. innovatively fabricated a
   human collagen‐based multilayer scaffold that emulates the transition
   from tendon to bone. This scaffold encompasses four distinct layers,
   each characterized by a unique composition gradient. The initial layer
   represents the tendon, primarily comprising collagen, followed by an
   uncalcified fibrocartilage layer composed of collagen and chondroitin
   sulfate. Subsequently, a calcified fibrocartilage layer, enriched with
   collagen and a reduced amount of apatite represent the calcified
   fibrocartilage layer. Finally, the outermost layer serves as bone,
   exhibiting a composition of collagen interwoven with HAP.^[ [56]^9 ^]
   Among the various scaffold designs for tendon‐to‐bone repair, those
   incorporating a multiphasic structure comprising both non‐calcified and
   calcified regions have demonstrated the most promise.^[ [57]^10 ^]
   These biomimetic scaffolds replicate key features of the native
   tendon‐to‐bone attachment, resulting in improved healing and functional
   recovery. However, despite advancements in scaffold design and
   fabrication, the application of these scaffolds remains limited in vivo
   due to the complex processes and mismatched clinical situations.
   Throughout the surgical process of tendon‐bone reconstruction, the
   preservation of tissue integrity is upheld. In this scenario,
   emphasizing the integration of tendon and bone takes precedence over
   the mere reconstruction of tendon‐bone structures. This indicated that
   the gradient strategy was not clinically applicable. Various methods
   have been explored to enhance tendon‐to‐bone healing, including HAP,^[
   [58]^11 ^] bone cement,^[ [59]^12 ^] bone morphogenetic protein
   (BMP),^[ [60]^13 ^] magnesium,^[ [61]^14 ^] and platelet‐derived growth
   factor.^[ [62]^15 ^] These methods have been shown to be effective in
   accelerating the healing process by increasing new bone or
   fibrocartilage formation at the interface. However, they are incapable
   of inducing bone growth into the graft because the osteogenic additives
   are incompetent in promoting the connection between bone and tendon.
   This is due to the fundamental differences in the composition of bone
   and tendon. Collagen fibers are internally mineralized, creating a
   hybrid structure where HAP is deposited within the fibers in bone.
   Because of this unique structure, bone has a natural ability to heal
   completely. If a similar mineralized collagen structure in the tendon
   can be engineered as in bone, it could be possible to transform
   tendon‐to‐bone integration into bone‐to‐bone integration, leading to
   the successful integration of tendon and bone.

   To address these challenges, we propose the biomimetic mineralization
   of tendon collagen to transform soft tissue into hard tissue for
   enhanced tendon‐bone integration. The interplay between an organic
   matrix or template and an inorganic mineral phase in biomineralization
   gives rise to extraordinary properties. Bone, for example, derives its
   load‐bearing properties from HAP platelets hierarchically deposited in
   a collagen matrix. In the early stage of biomineralization, the
   osteoblasts secrete matrix vesicles gathering calcium and phosphate,
   work as precursors of HAP–amorphous calcium phosphate (ACP)—and finally
   deposit into collagen. Biomimetic mineralization refers to a strategy
   driving ACP to infiltrate recombinant collagen and nucleation in the
   inner space in vitro, which mimics the natural mineralization process
   in the body.^[ [63]^16 ^] The deposition of mineral ions within
   collagen eventually forms a bionic scaffold that closely resembles the
   natural bone nanostructure.

   In this study, tendon tissue was immersed in a biomimetic
   mineralization system filled with ACP. The ACP precursors infiltrated
   the collagen fibers of the tendon and sequentially crystallized to
   realize the tendon collagen biomineralization. This formed an
   inorganic–organic composite with properties similar to those of native
   bone. As an osteogenic implant, the mineralized tendon induced
   osteogenesis of mesenchymal stem cells (MSCs) and narrowed the bone
   tunnel (Figure [64] 1 ). This pattern differs from the chaotic
   deposition of minerals produced by HAP coating or simulated body fluid
   immersion and significantly improves the mechanical properties of
   collagen fibers as well as osteogenic induction ability.^[ [65]^17 ^]
   Overall, this study demonstrates a promising approach for the
   integration of the tendon‐bone interface.

Figure 1.

   Figure 1
   [66]Open in a new tab

   A schematic diagram depicting the process of biomimetic mineralized
   tendon slices fabrication and bone tunnel implantation.

2. Results

2.1. Characterization of Biomimetic Mineralized Tendon

   As the arrangement of tendon collagen fibers was extremely compact, the
   tendon was sectioned to make ACP infiltrate into the acellular matrix.
   Then the tendon slices were mineralized for 4 days, as a longer
   processing time would not bring a significant effect improvement. We
   performed thermogravimetric analysis (TGA) on normal tendon (N‐tendon)
   and mineralized tendon (M‐tendon). When the temperature was
   continuously heated to 800 °C, the mass of the two groups of materials
   stopped decreasing (Figure [67] 2A). The remaining weight was recorded
   as inorganic minerals, as the lost weight came mainly from the burning
   of organic components. The result showed that the minerals accounted
   for ≈27% of the total weight in M‐tendon while 4% in N‐tendon
   (Figure [68]2B). The mineral content of M‐tendon accounted for half the
   natural bone (50–60%) after 4 days of mineralization. The EDS in X‐ray
   detection and analysis further proved successful mineralization of the
   tendon (Figure [69]2C,D). The peak of calcium and phosphate appeared,
   and the ratio of these two elements was ≈1.64. X‐ray diffraction was
   applied to assess the nucleation of HAP within collagen, and
   characteristic diffraction peaks (002), (211) of HAP appeared after
   mineralization (Figure [70]2E). The Fourier transform infrared
   spectroscopy (FTIR) result showed that tendon collagen had
   characteristic amide I, II, and III bonds, and the peaks were located
   at 1655, 1560, and 1240 cm^−1, respectively. The mineralized tendon not
   only presented the characteristic groups of collagen but also had the
   characteristics of HAP. Specifically, the band at 570 cm^−1 was
   attributed to the PO[4] ^3− ν4 bending modes, and the band at 1055
   cm^−1 was attributed to the PO[4] ^3− ν3 antisymmetric stretching
   modes. Due to the crystallization of HAP, the former peak was split
   into two small peaks (Figure [71]2F). Alizarin red S (ARS) staining and
   Von Kossa staining showed that the M‐tendon group formed an insoluble
   dark orange complex with ARS, or replaced silver ions and further
   reduce silver ions to black metallic silver when exposed to intensive
   light, while the collagen matrix in normal tendons was little stained
   with the two dyes (Figure [72]2G,J). This demonstrated a large amount
   of calcium deposited within the M‐tendon. The biomineralization process
   improved the mechanical properties of the tendon dramatically. As
   nanoindentation testing elucidated, Young's modulus and hardness of the
   M‐tendon were several times higher than those of the N‐tendon
   (Figure [73]2K,L).

Figure 2.

   Figure 2
   [74]Open in a new tab

   Characteristics of normal and mineralized tendon slices. A)
   Thermogravimetric analysis (TGA). B) The mineral content of normal and
   mineralized tendon slices (n = 3). C, D) Quantitative analysis of
   elements. E, F) The chemical and physical characteristics of normal and
   mineralized tendon slices. G, H) ARS staining and quantitative analysis
   of normal and mineralized tendon slices. Scale bar = 100 µm. I, J) Von
   Kossa staining and quantitative analysis of normal and mineralized
   tendon slices. Scale bar = 100 µm. K,L) The modulus and hardness of
   normal and mineralized tendon slices n = 3. M, N) SEM and TEM of normal
   and mineralized tendon slices. Scale bar = 200 nm. O) STORM image of
   the mineralized collagen fibrils suggests that the HAP is in the
   collagen fibrils. Scale bar = 200 nm.

   Then, the deposition of HAP and the mode of mineralization were
   observed. Scanning electron microscopy (SEM) results showed that the
   morphology of the M‐tendon was significantly changed when compared with
   that of the N‐tendon (Figure [75]2M). The surface of collagen fibrils
   was coated by HAP, and the D‐period disappeared. This demonstrated that
   collagen fibrils induced the depositing of HAP. During preparation for
   transmission electron microscopy (TEM), the samples were not stained
   with osmic acid or uranium acetate, so the contrast of collagen fibers
   of N‐tendon was low in the TEM field, and only the outline of collagen
   could be seen faintly. In contrast, mineralized collagen was internally
   deposited with a large amount of HAP, which, with its strong electron
   scattering ability, formed the dark area. We also noticed that there
   were blank areas within the field of view of the image, which was
   evidence of moderate mineralization (Figure [76]2N). We then used
   fluorescent dyes to label collagen fibers and calcium, respectively.
   Through STORM microscope imaging and 3D reconstruction, it was clearly
   observed that calcium deposition was in the interior of collagen rather
   than on the collagen surface, indicating the success of intrafibrillar
   mineralization (Figure [77]2O). The mineralization depth was further
   characterized through the energy dispersion spectrum (EDS). Given the
   scarcity of minerals in normal tendon slices, the feeble signals of
   calcium and phosphorus pervade the entire field (Figure [78]S1A,
   Supporting Information). Quantitative analysis reveals a lack of
   noteworthy mineral augmentation at the cross section (Figure [79]S1B,
   Supporting Information). Conversely, within the mineralized tendon
   slices, the signals of calcium and phosphorus gather conspicuously at
   the crosssection, displaying a noteworthy surge in signal intensity
   (Figure [80]S1C,D, Supporting Information). These results suggested a
   full‐thickness mineralization of the matrix.

   The above results demonstrated the full‐thickness mineralization of the
   tendon slice, characterized by intrafibrillar mineralization. The
   inorganic mineral (hydroxyapatite – HAP) deposits within the organic
   matrix (collagen) in an orderly manner. This harmonious hybridization
   of inorganic and organic components significantly enhances the
   mechanical properties of the matrix.

2.2. Biological Properties of Moderately Mineralized Tendon

   Further, we inoculated C3H10T1/2 cells (mouse pluripotent mMSCs line)
   on the surface of N‐tendon and M‐tendon. After 7 days of culture, dead
   cells were rarely seen under the fluorescence microscope in either
   group (Figure [81] 3A). Quantitative analysis was carried out by a
   fluorescence microplate reader. The ratio of OD535 and OD490
   represented the ratio of living cells to dead cells. The results showed
   no significant difference between the two groups (Figure [82]3B). At
   the same time, we used the CCK8 method to track the proliferation of
   cells for 1, 3, 5, and 7 days. There was no significant difference in
   the rate of cell proliferation (Figure [83]3C). These results showed
   that tendon mineralization did not induce cell death or inhibit cell
   proliferation.

Figure 3.

   Figure 3
   [84]Open in a new tab

   The proliferation and osteogenic differentiation ability of cells on
   the normal and mineralized tendon slices. A, B) Live/dead cells
   staining and fluorescence intensity ratio of C3H/10T1/2 after culture
   on the normal and mineralized tendon slices n = 3. Scale bar = 200 µm.
   C) The proliferation of C3H/10T1/2 measured by CCK‐8 n = 6. D, E) The
   morphology of C3H/10T1/2 was observed under SEM and fluorescence
   microscope, respectively. Scale bar = 20 µm in (D) and 50 µm in (E). F,
   G) Cell area and circularity of C3H/10T1/2 n = 9. H) ALP staining.
   Scale bar = 100 nm. I) Immunofluorescent staining of RUNX2 and BMP2.
   Scale bar = 50 µm. J) mRNA expression levels of osteogenic markers of
   C3H/10T1/2 culture in the normal and mineralized tendon slices,
   respectively, n = 3.

   After the cells were inoculated on the tendon surface for 12 h, the
   specimens were collected and photographed by SEM with cytoskeleton
   staining. It was observed that the cell morphology on the surface of
   the non‐mineralized tendon was slender but was stretched on the surface
   of the mineralized tendon (Figure [85]3D,E). Statistical analysis
   showed that cell area and circularity in the M‐tendon group were larger
   than in N‐tendon (Figure [86]3F,G). The cell morphology might be
   closely related to its differentiation tendency. Therefore, we further
   induced and detected the osteogenesis ability of C3H cells on the
   surface of normal tendons and mineralized tendons. After 7 days of
   culturing with induction, stem cells grown on the mineralized tendon
   showed stronger ALP activity (Figure [87]3H). Similarly,
   immunofluorescence staining results indicated that stem cells on the
   surface of mineralized tendon expressed higher levels of osteogenic
   differentiation‐related factors RUNX2 and BMP2 (Figure [88]3I).
   Finally, we detected the mRNA expression level of cells. qPCR results
   showed that mineralized tendon better promoted the expression of Alp
   and Bmp2 genes in stem cells at 7 days, while the expression of
   osteogenic‐related genes including Alp, Bmp2, Runx2, and Ocn was
   increased at 14 days (Figure [89]3J). In conclusion, M‐tendon exhibited
   biosafety and displayed the ability to induce osteogenic
   differentiation of pluripotent MSCs.

2.2.1. Osteogenic Differentiation Mechanism of Mineralized Tendon by
Transcriptomic Analysis

   The comprehensive role of M‐tendon and N‐tendon in mediating cell
   behavior was then analyzed by conducting RNA‐sequencing (RNA‐seq)
   studies on the cultured cells (Figure [90] 4 ). The gene sets of mice
   were employed as a reference. We found 355 genes downregulated and 70
   genes upregulated in comparing M‐tendon with N‐tendon (fold change ≥2;
   q value <0.05) (Figure [91]4A). When comparing gene expression related
   to osteogenic differentiation and tendon differentiation, we observed
   that genes including Alpl, Ostn, Sox9, Runx2, Nog, Bglap, Bmp2, Smad1,
   Smad5, and Sp7 were highly expressed in the M‐tendon group, while genes
   including Nfatc4, Bgn, Eln, Thbs4, Tnmd, Fmod, Egr1, Egr2, Tnc, and Scx
   were mainly expressed in the N‐tendon group (Figure [92]4B). Gene
   Ontology (GO) analysis revealed that the expression level of the
   biological process and signaling pathway related to lipid metabolism
   was significantly downregulated in the M‐tendon group
   (Figure [93]4C,D). We thought that this was because the activation of
   lipid metabolism required a soft matrix, so there would be
   corresponding down‐regulation after tendon mineralization. In the next
   step, we investigated the interactions between proteins by analyzing
   the key subnetwork, which includes genes like Dhcr24 and fdft1
   (Figure [94]4E). It was conducted further enrichment analysis of the
   genes in this key subnetwork and found that they were also associated
   with lipid metabolism (Figure [95]4F). Within this network, fatty acid
   synthase (FASN) had high connectivity within the module
   (Figure [96]4E). Our literature search revealed that some studies have
   reported that the downregulation of FASN could inhibit the lipogenic
   differentiation of stem cells and enhance their osteogenic
   differentiation potential.^[ [97]^18 ^] FASN serves as a pivotal enzyme
   in the intricate process of de novo biosynthesis of fatty acids and is
   expressed in adipocytes and osteoblasts. Notably, FASN participates in
   the crosstalk between adipose and bone tissue metabolism.^[ [98]^19 ^]
   FASN was proved to be up‐regulated during adipogenic differentiation of
   MSCs.^[ [99]^20 ^] In a study on the regulation of MSCs osteogenic
   differentiation, betaine showed a significant down‐regulation effect on
   genes associated with lipogenesis such as PPARγ, CEBPα, and FASN.^[
   [100]^21 ^] The TGF‐β signaling pathway in osteoclastogenesis and
   immune dysregulation, as well as the anomalous accumulation of FASN in
   Langerhans cells, have been proven to be associated with osteolytic
   Langerhans cell histiocytosis.^[ [101]^22 ^] Further studies on FASN
   hold the potential to facilitate the osteogenic differentiation of stem
   cells and promote bone tissue regeneration. This finding supports our
   conjectural hypothesis that the mineralization of tendons could promote
   the osteogenic differentiation of stem cells. Because of the inherent
   limitations of differentially expressed gene analysis, we performed
   gene set enrichment analysis (GSEA) on all genes. The results showed
   that tendon‐related genes were downregulated in the mineralized tendon
   group while osteoblast differentiation genes were upregulated.
   Specifically, the TGF‐β signaling pathway and the WNT‐β‐catenin
   signaling pathway were both notably upregulated (Figure [102]4G). TGF‐β
   and WNT‐β‐catenin signaling are recognized as crucial regulators of
   osteogenic differentiation in bone regeneration. Overall, these in
   vitro findings demonstrated that mineralized tendon induced the
   osteogenesis of cells by promoting the expression of osteogenic genes
   and initiated related signaling pathways.

Figure 4.

   Figure 4
   [103]Open in a new tab

   RNA‐seq analysis of the cells cultured on normal and mineralized tendon
   slices. A) Volcano plot of gene expression (M‐tendon versus N‐tendon;
   fold change, ≥2; q value < 0.05). B) Heat map of osteogenesis and
   tenogenesis related genes. C, D) GO analysis of differentially
   expressed genes. E) Protein–protein interaction network diagram of
   differentially expressed genes. F) Enrichment analysis of (E). G) GSEA
   analysis of all genes in M‐tendon and N‐tendon groups. NES, normalized
   enrichment score; FDR, false discovery rate.

2.3. Mineralized Tendon Promotes Tendon‐Bone Regeneration through Tissue
Fusion

   To test the osteogenesis ability of M‐tendon in vivo, we curled the
   tendon slice into a strip and embedded it subcutaneously in the mouse.
   After 4 weeks, micro‐CT results showed that there was a strip of high
   signal shadow on the right side of the spine in the experimental group,
   while there was no signal at the same position in the control group
   (Figure [104] 5A). The bone volume and bone volume fraction of the
   experimental group were much higher than those of the control group
   (Figure [105]5B,C). The staining of tissue sections, including HE and
   MASSON, showed that the matrix arranged in the control group was more
   parallel and orderly (Figure [106]5D,E). The sample from the N‐tendon
   group expressed little osteogenic differentiation‐related proteins BMP2
   and RUNX2. In the M‐tendon group, BMP2 and RUNX2 were significantly
   upregulated, and the content of the collagen matrix was a bit like the
   bone matrix (Figure [107]5F). Meanwhile, more positive areas of MKX and
   TNMD were observed, which suggested a stronger tenogenic
   differentiation property (Figure [108]5H). Sirius Red staining
   demonstrated that the specimens from the M‐tendon group had poorly
   organized tissue and collagen fiber alignment as compared with the
   control specimens. This indicated a transformation of bone‐like tissue
   (Figure [109]5G). The tartrate‐resistant acid phosphatase (TRAP)
   staining revealed recruitment of osteoclasts (Figure [110]S2,
   Supporting Information). These results proved that mineralized tendons
   had the characteristics of bone induction.

Figure 5.

   Figure 5
   [111]Open in a new tab

   Biocompatibility and ectopic ossification of normal and mineralized
   tendon slices in vivo. A–C) Representative micro‐CT images and
   quantitative analysis of results. D,E) HE and MASSON staining of
   samples. Scale bar = 50 µm. F) Immunofluorescent staining of RUNX2 and
   BMP2. Scale bar = 50 µm. G) Sirius Red staining imaged under
   polarization microscopy. Scale bar = 50 µm. H) Immunofluorescent
   staining of MKX and TNMD. Scale bar = 50 µm.

   To further investigate the bone induction property and potential
   clinical application, we established a femoral condyle drilling model
   of mice (Figure [112] 6A). The normal or mineralized tendon slice was
   rolled into a round strip and passed through the bone tunnel. After 4
   weeks, micro‐CT imaging analysis showed that the bone tunnel in the
   M‐tendon group was significantly narrowed, while the bone tunnel size
   in the N‐tendon group was little changed (Figure [113]6B). The newly
   formed bone mass was evaluated by analyzing the bone volume/total
   volume (BV/TV) fraction. The mineralized tendon induced formation
   around the implant with a significantly higher bone mass, and good
   integration of neo‐bone (orange area) and implants (grey area) was
   observed (Figure [114]6B,C). To characterized the spatial form of bone
   trabecular, trabecular thickness (Tb.Th) and trabecular space (Tb.Sp)
   were analyzed in the bone tunnel of the three groups. Tb.Th in the
   M‐tendon group showed the highest value after 4 weeks of implantation,
   and decreased trabecular spacing implied a more complete bone structure
   (Figure [115]6C). Furthermore, higher values of Tb.Th and the lower
   values of Tb.Sp indicated more active bone anabolism as opposed to
   catabolism.

Figure 6.

   Figure 6
   [116]Open in a new tab

   Micro‐CT analysis after graft implantation in a tendon‐bone integration
   model. A) Scheme of operation procedure. B) The horizontal section,
   coronal section and vertical section bone tunnels were measured, and
   the 3D images were reconstructed. C) Quantitative analysis of neo‐bone
   in bone tunnel zone.

   The histological results showed the bone tunnel was mainly filled with
   fibroblasts and blood cells as no graft was implanted in the control
   group. The disordered tissue was organized and replaced by fibrous
   tissue gradually. In the N‐tendon group and M‐tendon group, the bone
   tunnel was filled with normal tendons or mineralized tendons, the
   collagen matrix of grafts could be stained with aniline blue in MASSON
   staining (Figure [117] 7B). In particular, the diameter of the bone
   tunnel in the control and N‐tendon groups did not dwindle and the
   interface between the tendon tissue and the native bone tissue was
   arc‐shaped. Conversely, the soft tissue at the center of the
   mineralized tendon was significantly reduced, and the soft tissue was
   intertwined with the neo‐bone (Figure [118]7A,B,D). This zigzag shape
   helped increase the contact area between bone and tendon tissue, so as
   to improve the adhesion of the interface. The mineralized tendon in
   contact with bone has been initiating the remolding and transforming it
   into bone. Potential osteogenesis was assessed by IF staining of BMP2
   (green) and tenogenesis by TNMD (green). Significant positive
   expression of BMP2 was observed in mineralized tendons, which was
   conducive to new bone formation (Figure [119]7E). Accompanied by
   decreased staining of tendon induction marker (TNMD, Figure [120]7F),
   it can be speculated that the biological characteristics of the tendon
   had become bone‐like after mineralization. Similar to the findings from
   subcutaneous implantation, the presence of osteoclasts was identified
   within the bone tunnel grafts (Figure [121]S3, Supporting Information).
   The mineralized tendon slices effectively attract and engage
   osteoclasts, thereby initiating the intricate process of bone
   remodeling and formation. Furthermore, typical cartilage features were
   not detected by Safranin O staining at the interface (Figure [122]7C).
   The results suggested that the M‐tendon transformed into bone tissue
   through intramembrane osteogenesis directly instead of through
   endochondral osteogenesis. In conclusion, the femoral tunnel drilling
   model proved that M‐tendon retained the characteristics of soft tissue
   (the areas not in contact with the primary bone tissue in the middle),
   but it could transform into bone tissue and form a perfect integration
   with the primary bone tissue.

Figure 7.

   Figure 7
   [123]Open in a new tab

   Histological analysis of the tendon‐bone integration process. A–C) HE,
   MASSON, and Safranin O staining of bone tunnel section after 4 weeks of
   implantation. Scale bar = 200 µm in (A) Scale bar = 40 µm in (B,C).
   D–F) Immunofluorescent staining of COL1A1, BMP2, and TNMD in the
   control, normal and mineralized tendon groups after surgery. Scale bar
   = 50 µm.

3. Summary and Discussion

   This study aimed to develop a mineralized tendon implant via ACP
   infiltration in a biomimetic mineralization system. The mineralized
   tendon induced osteogenesis of MSCs and generated a serrated interface
   in vivo. This strategy facilitated neo‐bone growth across the interface
   and facilitated the integration of the tendon‐bone interface.

   The process of healing tendons and ligaments to bones presents a
   significant challenge in clinic due to the intricate structure,
   composition, cell population, and mechanics of the interface.^[
   [124]^23 ^] Tissue engineering has made remarkable progress, and
   various approaches, such as advanced biomaterials, bioactive growth
   factors, and multiple stem cell lineages have been developed to
   facilitate healing at this tissue interface. Biomaterials played the
   most important role in promoting tissue regeneration, they provided
   structural support and appropriate biomechanical and biochemical
   properties that emulate the natural microenvironment.

   Currently, a common strategy involves utilizing multilayer biomaterial
   scaffolds to replace the tendon‐bone interface. However, this strategy
   has yielded limited success in research due to challenges in achieving
   a continuous functionalized gradient and dispersing the mechanics
   properly. The interface between the two components is affected by
   stress concentrations, and the technical requirements remain
   impractical for clinical applications. Therefore, the application of
   structural bionics strategies remains a significant challenge. The
   absence of hierarchical structural regeneration processes underscores
   the urgency for novel approaches to enhance functional integration and
   promote the regeneration of tendon‐bone structures. In cases where the
   scaffold fails to achieve continuous gradient construction in vivo, it
   is essential to assess the necessity of a multilayer scaffold that
   encourages either bone or tendon formation. During the surgical
   reconstruction of the rotator cuff and ACL, the presence of native
   tendon and bone tissue at the interface raises the critical issue of
   whether the transitional side can generate sufficient bonding to unite
   the two tissues.

   The process of bone ingrowth plays a pivotal role in achieving
   integration between the graft and the bone tunnel. By controlling the
   composition, structure, and morphology of the gradient scaffold
   biomaterial, it is possible to influence cell attachment,
   proliferation, and differentiation to achieve optimal tendon‐bone
   interface repair. Upon culturing bone marrow‐derived mesenchymal stem
   cells on a scaffold featuring a mineral gradient, it was observed that
   cells contacted with a matrix exhibiting a higher mineral content
   demonstrated earlier manifestation of osteogenic behavior compared to
   those on an unmineralized matrix.^[ [125]^24 ^] Polyethylene
   terephthalate (PET) exhibits commendable mechanical properties,
   rendering it a viable choice for employment as an artificial ligament
   in ACL reconstruction. Nevertheless, PET encounters challenges with
   tissue ingrowth leading to a relatively high rate of graft failure in
   the long‐term follow‐up. Calcium phosphate coating on the PET surface
   via electrodeposition enhanced the graft‐bone integration.^[ [126]^25
   ^] Biological scaffolds can be composed of biodegradable or synthetic
   materials and can have different shapes and properties to suit a
   variety of repair needs. The combination of biodegradable and
   non‐biodegradable scaffolds provides space for bone tissue ingrowth,
   while the release of BMP‐7 exerts additional impetus toward
   osteogenesis.^[ [127]^26 ^]

   Implant‐induced bone tissue ingrowth is an effective strategy to
   promote interfacial integration.^[ [128]^27 ^] Several studies have
   modified nanofiber scaffolds through spatial control of mineral content
   to not only mimic changes in minerals but also create mechanical
   gradients.^[ [129]^28 ^] The incorporation of minerals facilitates bone
   formation and narrows the bone channels, yet the newly formed bone
   fails to integrate into the implant and establish a solid bond because
   the minerals and implant were merely attached or mixed physically and
   performed their respective functions. The minerals induce bone
   formation and eventually catabolize.^[ [130]^29 ^] If the minerals and
   implant can be organically combined, it would be possible to avoid the
   aforementioned problems.

   Biomimetic mineralization technology has demonstrated the ability to
   create organic/inorganic hybrids by engineering minerals and collagen.
   Previous studies have employed the biomimetic mineralization strategy
   for biomimetic bone graft fabrication.^[ [131]^30 ^] In simulating the
   bone formation process, ACP is synthesized in vitro and then
   crystallized within the collagen, extending along the c‐axis that is
   parallel to the long axis of collagen to form mineralized collagen.^[
   [132]^31 ^] Mineralized collagen is the primary composite of bone
   matrix. This makes an effective integration with native bone while
   simultaneously inducing bone formation in vivo. Nevertheless,
   recombinant collagen lacks sufficient mechanical properties to replace
   the function of the tendon. The biomimetic mineralization of native
   tissues on a micrometer scale is still not documented, although the
   scale of ACP is theoretically tiny enough to enter into the interspace
   of tissue fibrils. As tendons have the same collagen type as bone (type
   I), mineralization of tendon tissue via biomimetic mineralization is
   effective in establishing a firm bond at the tendon‐bone interface. The
   tiny ACP permits its infiltration into the collagen surrounding the
   tendon surface. Although ACP nucleation occurs prior to its migration
   to the inner regions of the tendon tissue, the mineralized layer is
   sufficient to form a tightly bound interface with neo‐bone. Within the
   mineralized layer of the tendon, mineralized collagen fibers exhibit
   nanoscale intrafibrillar oriented minerals, akin to the bone, which
   stimulate bone tissue growth while retaining the tendon's mechanical
   properties.

   The stiffness of the extracellular matrix and the presence of calcium
   and phosphate ions have long been known to influence stem cell
   differentiation. The osteogenic properties of calcium and phosphate
   scaffolds are attributed to their ability to regulate extracellular ion
   calcium and phosphorus concentrations in stem cells through
   Ras/Raf/Erk‐dependent signaling pathways or ATP‐adenosine controlled
   mechanisms. However, the addition of calcium and phosphate alone has a
   limited ability to regulate the expression of osteogenesis‐related
   genes (RUNX2, OCN, PDPN, and DMP1). In contrast, biomimetic
   intrafibrillar mineralization of collagen by HAP significantly
   upregulates osteogenesis‐related genes in bone marrow MSCs.^[ [133]^30a
   ^] In this study, mineralized tendons promoted the expression of
   osteogenic genes and initiated related signaling pathways, which
   suggests that intrafibrillar mineralization collagen may be necessary
   to influence bone marrow MSC differentiation in the absence of
   osteoinductive factors.

   Furthermore, the remarkable osteogenic potential of mineralized tendons
   implies that biomimetic mineralization‐based implants could be viably
   employed in the context of bone regeneration therapy. Collagen
   membranes featuring intrafibrillar mineralization demonstrated
   exceptional efficacy in terms of osteoinduction and osteoconduction,
   leading to successful instances of ectopic bone formation within living
   organisms.^[ [134]^32 ^] Notably, mineralized collagen scaffolds
   constructed by nucleic acid‐based biomimetic mineralization systems
   have been substantiated as highly effective in bone tissue engineering
   and regenerative medicine.^[ [135]^30b ^]

   In conclusion, autologous soft tissues represent the quintessential
   reservoir for tissue regeneration, while ACL and rotator cuff
   reconstruction are challenging in clinic, because of the poor
   tendon‐bone interface integration. This study proposed a strategy to
   mineralize the tendon collagen resulting in a bone affinitive
   interface. Mineralized tendon with osteal physicochemical and
   biological properties facilitates the tendon‐bone interface
   integration, although the mineralization degree is less than bone. It
   would be promising to optimize the biomimetic mineralization system to
   improve the mineralization degree and efficiency of soft tissue in the
   future. Nonetheless, we believe our strategy provides an alternative
   procedure for ACL and rotator cuff reconstruction and bone
   regeneration, which would eventually promote clinical outcomes.

4. Experimental Section

Preparation of Decellularized Tendon Slices

   The acellular tendon slices were acquired via a freeze‐thaw procedure.
   Briefly, tendons were harvested from swine forelimbs, and aponeuroses
   were removed. Then, the tendon was cut into 1.5 cm × 1.5 cm pieces. The
   samples were subjected to liquid nitrogen and phosphate buffer saline
   (PBS) at 37 °C for a total of ten cycles. Next, the samples were
   immersed into 1% (v/v) sodium dodecyl sulphate for 1 h and rinsed with
   PBS. The decellularized tendons were embedded in an optimal cutting
   temperature compound, and 80µm thick slices were obtained using a
   freezing microtome (Leica CM1950, Leica, Wetzlar, Germany). The slices
   were washed thrice with PBS and cross‐linked with 0.3 m
   1‐(3‐dimethylaminopropyl)−3‐ethylcarbodiimide hydrochloride/0.06 m
   1‐(3‐dimethylaminopropyl)−3‐ethylcarbodiimide hydrochloride for 4 h.

Biomineralization of Decellularized Tendon Slices

   The mineralizing solution was prepared by mixing equal volumes of
   solution A [3.33 mm CaCl[2], 300 mm NaCl, 480 µg mL^–1 poly‐aspartic
   acid (Mw = 8000–11 000 kDa)] and solution B (19 mm NaH[2]PO[4]).
   Decellularized tendon slices were immersed into more than ten times
   their volume of mineralizing solution and incubated at 37 °C for a
   total of 4 days. The mineralizing solution was refreshed every 12 h to
   facilitate the mineralization process.

Scanning Electron Microscopy and Energy Dispersion Spectrum

   The specimens were fixed with 2.5% glutaraldehyde, then dehydrated with
   an ascending ethanol series (30–100%), critical‐point dried (Hitachi,
   HCP‐2), and sputter‐coated with gold. The morphology and element
   composition of tendon slices were recorded via scanning electron
   microscopy (SEM, Hitachi S‐8010, Japan) at 3 kV and energy dispersion
   spectrum (EDS) at 15 kV, respectively. To characterize the depth of
   mineralization, the elemental distribution in the cross section was
   detected via EDS.

Transmission Electron Microscopy

   The specimens were fixed with 2.5% glutaraldehyde, dehydrated with an
   ascending ethanol series (30–100%), embedded in epoxy resin, sectioned
   to 80 nm thickness using a Leica ultramicrotome (Leica EM UC7), and
   collected on copper grids. Without staining with uranium acetate, the
   ultrastructure of specimens was observed under transmission electron
   microscopy (TEM, Hitachi, H‐7650) at 120 kV.

Thermogravimetric Analysis (TGA)

   TGA was performed with a thermal gravimetric analyzer (DSC1/400,
   TGA/DSC1 1100SF‐MS, METTLER‐TOLEDO, Switzerland) after specimens were
   dried in a drying oven at 65 °C overnight. The specimen was heated from
   25 to 200 °C for dehydration and 200 to 600 °C for organic
   decomposition. The ratio of the final weight and weight loss at 800 °C
   was analyzed.

X‐Ray Diffraction (XRD) and Fourier Transform Infrared Spectroscopy (FTIR)

   The crystal structure of specimens and HAP powder was characterized by
   XRD (D8 Advance, Bruker, Germany) operated at 45 kV and 40 mA. All the
   specimens were air‐dried at 65 °C overnight and pulverized to produce
   fine powders. The recorded patterns were analyzed using JADE6 software
   (Materials Data Incorporated). FTIR (Nicolet iN10, Thermo Scientific,
   USA) was performed to assess their chemical composition. The specimens
   were dried and pulverized as above‐mentioned. After being mixed with
   potassium bromide, the specimens were analyzed with FTIR (Iraffinity‐1,
   Shimadzu, Japan).

Mechanical Properties Assay

   Nanoindentation tests were performed utilizing a Nano‐indenter (G200,
   Agilent Tech.) to characterize the mechanical properties of specimens.
   The quantitative data was calculated based on the standard mathematical
   model employed by the apparatus.

Alizarin Red S Staining and Von Kossa Staining

   Alizarin red S staining (ARS) and Von Kossa were used to identify the
   mineral composition. The tendon slices were rinsed with distilled
   deionized water and incubated in 2% ARS solution for 30 min. For Von
   Kossa staining, the specimens were incubated with 5% silver nitrate
   solution for 15 min under ultraviolet. Then, a 5% sodium thiosulfate
   solution was employed to remove the residual silver nitrate. The
   stained specimens were observed and photographed using microscopy.

3D‐STORM Imaging

   The mineralized tendon slices were labeled with a fluorescent reagent
   by immunofluorescence staining. The slices were first fixed with 4%
   paraformaldehyde. After washing thrice, the samples were blocked with
   1% bovine serum albumin (BSA) for 1 h at room temperature and then
   overnight at 4 °C with primary anti‐collagen I (ab34710). Then, the
   samples were labeled with secondary antibodies for 1 h and calcein
   (Aladdin, C118587) for 30 min. The samples were rinsed thoroughly and
   subsequently immersed in the imaging buffer before STORM imaging. All
   STORM imaging data were acquired by virtue of a Nikon Ti‐E inverted
   microscope (N‐STORM/A1R, Nikon, Japan) and analyzed by Nikon
   NIS‐Elements AR software.

Cell Culture

   MSCs lines (C3H/10T1/2, The Chinese Academy of Sciences, Shanghai) were
   cultured in a growth medium (L‐DMEM supplemented with 10% FBS and 1%
   penicillin/streptomycin). Cell culture was performed at 37 °C and 5%
   CO[2] in a humidified environment. The medium was renewed every 3 days.
   After the designated culture time, the cells were used for further
   study.

Cell Viability Assay

   Prior to the experiments, the slices of both normal and mineralized
   tendons underwent a thorough cleansing process. They were subjected to
   three successive washes with 75% alcohol, lasting 30 min each, in order
   to achieve sterilization and eliminate the residues. Subsequently, the
   sterilized slices were further washed three times with PBS, with each
   wash lasting 15 min, to thoroughly remove the remnants of the 75%
   alcohol. The normal and mineralized tendon slices were then placed in
   each well, respectively, before C3H cells were seeded in 98‐well plates
   at a concentration of 1 × 10^3 cells per well and cultured for 1, 3, 5,
   and 7 days. The medium was replaced with 10% CCK‐8 (Cell Counting
   Kit‐8, Dojindo, CK04) solution at the designated time and incubated for
   1 h. Cell viability was quantified by a microplate reader (Varioskan
   Flash, Thermo Fisher Scientific, USA) at a wavelength of 450 nm.

Live/Dead Cell Staining

   After cultured for 7 days on mineralized slices, cells were
   centrifuged, rinsed with 1 × assay buffer, and resuspended with a
   density of 1×10^5‐1×10^6 cells mL^–1. Measured 200 µL of cell
   suspension was mixed with 100 µL of staining solution (2 µm calcein‐AM
   and 4.5 µm propidium iodide) and incubated at 37 °C for 15 min. Live
   and dead cells were visualized with confocal laser scanning microscopy.
   Their fluorescence intensity was quantified by a microplate reader
   (Thermo Scientific 5 250 040) at 490 nm and 545 nm, respectively.

Cell Morphology

   For immunofluorescent staining, specimens were fixed with 4%
   paraformaldehyde and permeabilized using 0.1% Triton X‐100. The samples
   were further blocked using 1% BSA in PBS for 1 h and stained with Alexa
   Fluor 488‐conjugated phalloidin as well as
   4′,6‐diamidino‐2‐phenylindole. The sample preparation for SEM is
   described above.

Osteogenesis Differentiation

   The C3H cells were seeded on mineralized slices at a density of 5000
   cells cm^–2. After 3 days, the growth medium was replaced with the
   osteogenic medium (growth medium supplemented with 10 mm
   β‐glycerophosphate, 10 nm dexamethasone, 50 µg mL^–1 ascorbic acid) to
   induce the formation of calcium nodules. The osteogenic medium was
   replaced every 3 days. Specimens were harvested for further testing at
   designated induction times.

Quantitative Real‐Time Polymerase Chain Reaction (qPCR) Analysis

   After cells were cultured in osteogenic medium for 7 and 14 days,
   quantitative real‐time PCR (qPCR) analysis was performed to evaluate
   the expression level of osteogenesis‐related genes. Total RNA was
   extracted using an RNA extraction reagent (Takara, 9109) according to
   the protocol. Thereafter, cDNA was synthesized and amplified in a
   reaction system. The qPCR was performed with a SYBR Green Mix (Takara,
   RR420A) in a fluorescence signal detection device (Roche, LightCycler
   480).

   The primers used were the following:
   Genes 5′‐3′ Sequence

   Mouse

   Alp

   Forward

   Reverse

   ATCGACGTGATCATGGGTGG

   GAGAGCGAAGGGTCAGTCAG

   Mouse

   Runx2

   Forward

   Reverse

   AATTGCAGGCTTCGTGGTTGAG

   GCTGTATGGTGAGGCTGGTAGG

   Mouse

   Bmp2

   Forward

   Reverse

   GATCACCTCTCTTCCTCAGCCC

   CAACACTAGAAGACAGCGGGT

   Mouse

   Ocn

   Forward

   Reverse

   TGAAGACCGCCTACAAACGC

   ACAGGGAGGATCAAGTCCCG
   [136]Open in a new tab

Transcriptome Sequencing and Data Analysis

   C3H cells were cultured on the surface of N‐tendon or M‐tendon for 14
   days, and then the cells were collected to extract RNA. After
   verification of its integrity, the extracted RNA was
   reverse‐transcribed to create a cDNA library for subsequent sequencing.
   Gene ontology (GO) and KEGG pathway enrichment analysis were performed
   on the data analysis platform provided by BGI (biosys.bgi.com). The
   protein–protein interaction network map and its subsequent gene
   enrichment analysis were produced in string‐db.org and by Cytoscape
   software. To be specific, the genes downregulated in the M‐tendon group
   were first subjected to protein‐protein interaction prediction
   analysis. Subsequently, the resultant files were input into Cytoscape,
   after that the key subnetwork was obtained by utilizing the MCODE
   plugin's functionality. Finally, the genes encompassed within the
   subnetwork were subjected to enrichment analysis.

Animal Experiments

   Animal experiments were approved by the Zhejiang University
   Experimental Animal Welfare and Ethics Committee under Institutional
   Animal Care and Use Committee guidelines (ZJU20210029). Eight‐week‐old
   mice were obtained from a licensed vendor and reared at an
   environmental temperature of 25 ± 2 °C with a dark/light cycle of 12 h.
   The animals were acclimated for 7 days before experiments. All
   operations abided by standard surgical procedures.

In Vivo Biocompatibility and Ectopic Osteogenesis Ability

   Ten mice were randomly assigned into a normal tendon group or a
   mineralized tendon group (five replicates for each group). Freshly
   prepared specimens were sterilized with 75% ethanol before subcutaneous
   implantation. Briefly, mice were anesthetized with pentobarbital. In
   the implantation, the surgical area was prepared and a 6 mm
   longitudinal incision was made at the middle of the dorsal skin. Muscle
   and skin were separated with scissors to create a subcutaneous pocket.
   The acellular tendon slice or mineralized acellular tendon slice was
   rolled into a cylinder with a diameter of 0.8 mm and then implanted
   into the subcutaneous pocket. Each specimen was sutured 5 mm away from
   the incision. After implantation, the incision was sutured, and the
   mice were monitored until recovery from anesthesia.

Bone Tunnel Drilling and Transplantation Model

   Control, acellular tendon, and mineralized acellular tendon groups were
   randomly assigned (five mice for each group). Sterilized grafts were
   passed through 0.8 mm bone tunnels in the distal femur. Briefly, mice
   were anesthetized with pentobarbital. A ≈5 mm incision was created in
   the medial of the distal femur. All the mice underwent medial
   arthrotomy and patella dislocation to expose the distal femur. A bone
   tunnel ≈4 mm in depth and 0.8 mm in diameter was drilled inside‐out in
   the femoral metaphysis by using a dental drill. The bone tunnel was
   irrigated with normal saline to remove bone debris and subsequently
   filled according to the designated group. Muscle, fascia, and skin were
   sutured in turn after repositioning the patellar.

Micro‐Computed Tomography Imaging and Analysis

   After 4 weeks of implantation, the mice were sacrificed for imageology.
   Scans were performed at a voltage of 80 kV and a resolution of 18 µm
   per pixel using micro‐computed tomography (micro‐CT, U‐CT‐XUHR, Milabs,
   Netherlands). The pixels exhibiting a CT value >1400 were designated as
   bone. The images were reconstructed and analyzed via Imalytics
   Preclinical.

Histology, Immunohistochemistry, and Immunofluorescence Assays

   The mice of each group were sacrificed at 4 weeks post‐implantation.
   Implants or distal femurs containing the bone tunnel were harvested and
   fixed with 4% formaldehyde. Samples were decalcified with 10% ethylene
   diamine tetra‐acetic acid thoroughly, dehydrated through an ascending
   graded series of ethanol (50–100%), and embedded in paraffin. The
   embedded samples were sectioned longitudinally at a thickness of 7 µm
   for histological examination. The sections were stained with
   hematoxylin & eosin (HE), Safranin O, and MASSON staining.

   For immunofluorescence and immunohistochemistry analysis, the staining
   procedure included deparaffinization, hydration, antigen retrieval, and
   antibody incubation. Endogenous peroxidase inactivation and
   diaminobenzidine treatment were acquired for immunohistochemistry
   analysis. The following antibodies were used in the study: anti‐BMP2
   (Beyotime, AF0075), anti‐RUNX2 (Sigma, HPA022040), anti‐COL1 (Abcam,
   ab34710), anti‐COL2 (Santa Cruz, sc‐52658), and anti‐TRAP (Santa Cruz,
   sc‐376875). The stained sections were examined under a confocal
   fluorescence microscope (Nikon A1R, Japan) or an optical microscope.

Statistical Analysis

   Quantitative data are presented as mean ± standard deviation (SD).
   Student's t‐test was performed to assess whether statistical
   differences existed between two groups. Multiple comparisons were
   performed with a one‐way analysis of variance (ANOVA) and Tukey's
   post‐test. P values of <0.05 were considered statistically significant.
   The significance level is presented as ^* p < 0.05.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   Y.C., Y.Z., and X.C. contributed equally to this work. Y.C. and X.C.
   designed the project. X.C., W.S., and Z.Y. supervised the project.
   Y.C., Y.Z., and X.C. performed the experiments and analyzed the data.
   Y.C. and Y.Z. wrote the manuscript. Y.C., Y.Z., and X.L. revised the
   manuscript. J.H., B.Z., T.Z., and C.F. conduct animal experiments. W.Y.
   supervised the 3D‐storm imaging. H.P. supervised the biomimetic
   mineralization procedure. All authors discussed and approved the
   results and manuscript.

Supporting information

   Supporting Information
   [137]Click here for additional data file.^ (303.5KB, pdf)

Acknowledgements