Abstract

   Rotator cuff tears often require surgical reconstruction; however,
   outcomes remain suboptimal, primarily because of the challenges in
   fully restoring the tendon–bone interface (TBI). The normal
   fibrocartilaginous transition zone is often replaced by fibrous scar
   tissue, thereby increasing the risk of retears. A porous hyaluronic
   acid methacrylate hydrogel is developed to encapsulate adipose‐derived
   stem cells (ADSCs) and bone morphogenetic protein‐2 (BMP‐2). Gelatin
   microspheres served as porogens to create micropores within the
   hyaluronic acid methacrylate hydrogel (HMs), and hollow gelatin
   methacryloyl hydrogel microspheres are used to encapsulate BMP‐2
   (GMBs). In vitro experiments confirm substantial proliferation and
   homing of ADSCs within the porous hydrogel system. When the
   ADSCs‐loaded composite hydrogel (HMs/MBs/ADSCs) is applied to a rat
   rotator cuff injury model, it effectively reconstructs the
   fibrocartilage transition zone and promotes TBI healing. Transcriptome
   sequencing is performed to elucidate the mechanisms underlying
   BMP‐2‐induced fibrocartilage reconstruction. Furthermore, integrated in
   vitro sequencing reveals that HMs/MBs/ADSCs modulated the tissue
   microenvironment to enhance tissue regeneration, primarily through
   BMP‐2′s regulation of the paracrine effects of ADSCs. This study
   explores the comprehensive pathway through which HMs/MBs/ADSCs enhance
   tissue regeneration, emphasizing the synergistic effects of BMP‐2 and
   ADSCs, and offers novel insights into the mechanisms underlying TBI
   healing.

   Keywords: ADSCs, BMP‐2, paracrine effects, porous hydrogels,
   tendon‐to‐bone healing
     __________________________________________________________________

   This study designs an injectable porous hydrogel (HMs/MBs/ADSCs)
   integrating hollow microspheres for sustained BMP‐2 release to enhance
   adipose‐derived stem cells' paracrine effect. It accelerates
   tendon‐bone healing in rat rotator cuff injuries via spatiotemporal
   regulation: early AMPK‐mediated mitochondrial protection and late
   TGF‐β/Smad2/3‐driven fibrocartilage regeneration. By integrating
   material design and mechanistic exploration, this study enhances the
   application potential of tissue‐engineered hydrogels.

   graphic file with name ADVS-12-e15923-g002.jpg

1. Introduction

   Rotator cuff tears cause shoulder pain and limited mobility and
   frequently necessitate surgical repair, with outcomes gradually
   improving as medical technology advances.^[ [46]^1 ^] However, 20–70%
   of patients continue to experience postoperative functional limitations
   and retears.^[ [47]^2 ^] The critical site of failure is the
   tendon–bone interface (TBI), which in its natural state consists of a
   four‐layer gradient structure composed of tendon, uncalcified
   fibrocartilage, calcified fibrocartilage, and bone.^[ [48]^3 ^]
   Surgically reconstructed TBI develops fibrovascular scar tissue
   composed primarily of type III collagen, and its limited mechanical
   strength leads to a high rate of retears.^[ [49]^4 ^] Furthermore, the
   number of cells at the interface gradually diminishes, and organized
   collagen fibers fail to form, thereby preventing restoration of the
   characteristic fibrocartilage transition zone.^[ [50]^5 ^] Achieving
   effective integration in patients with TBI remains challenging.
   Therefore, researchers have explored various tissue engineering
   strategies to enhance rotator cuff healing.^[ [51]^6 ^]

   The complex native structure of the TBI makes it difficult to establish
   a suitable biochemical microenvironment without external intervention
   after the interface is damaged. Therefore, the design of appropriate
   microenvironmental guidance materials is essential for facilitating
   regenerative cell migration, proliferation, and signal transduction.^[
   [52]^7 ^] Cell interactions on flat biomaterial surfaces have commonly
   been studied in vitro.^[ [53]^8 ^] However, cells attached to 2D
   structures do not realistically reproduce the extracellular matrix
   (ECM) environment.^[ [54]^9 ^] To address these limitations and
   accommodate defects with irregular shapes, injectable biomaterials have
   attracted considerable attention. Hydrogels have been widely used
   because their prepolymeric forms closely mimic the physicochemical
   properties of ECM. Hydrogels offer a 3D platform for cell encapsulation
   and the incorporation and release of bioactive ingredients.^[ [55]^10
   ^] Additionally, the influence of hydrogels on cell behavior is
   multifaceted, and optimizing their internal structure can enhance cell
   compatibility, tissue integration, and host responses.^[ [56]^11 ^]
   Among these, increasing porosity or pore size often enhances ECM
   secretion, cell infiltration, tissue growth, and molecular delivery.
   Because of the size limitations of hydrogels, researchers have
   developed several methods for creating macroporous hydrogels, including
   gas foaming, freeze‐drying, and particle leaching.^[ [57]^12 ^]
   However, most methods require cell seeding after hydrogel formation,
   which hinders uniform 3D cell encapsulation owing to challenges in
   controlling cell distribution. Therefore, it is necessary to use a
   cost‐effective and highly biocompatible porogen to construct porous
   hydrogel systems that provide a more suitable 3D microenvironment for
   cells.

   From the tissue microenvironment perspective, rotator cuff injuries are
   often accompanied by microenvironmental damage, resulting in excessive
   production of reactive oxygen species (ROS).^[ [58]^13 ^] Concurrently,
   the prolonged accumulation of ROS increases the secretion of
   proinflammatory cytokines, thereby maintaining a chronic inflammatory
   cycle.^[ [59]^14 ^] Stem cells and bioactive factors are widely
   acknowledged as crucial for inhibiting oxidative stress and reducing
   ROS production.^[ [60]^15 ^] The therapeutic effects of stem cells rely
   primarily on their paracrine functions. Paracrine factors or exosomes
   derived from stem cells can migrate to inflamed sites and regulate the
   activity and function of target cells via intracellular
   communication.^[ [61]^16 ^] The biological activity of stem cells can
   downregulate the NF‐κB pathway, reduce inflammation, and promote TBI
   healing.^[ [62]^17 ^] Among these, adipose‐derived stem cells (ADSCs),
   which are known for their accessibility and potent paracrine effects,
   have become a key focus of regenerative therapies.^[ [63]^18 ^]
   Therefore, using ADSCs as seed cells may enhance the biological
   functionality of tissue‐engineered hydrogels.

   In addition to pathogen clearance and inflammation control, the
   integration of growth factors is crucial for promoting tissue repair.
   Growth factors are essential regulators of cellular processes. They
   regulate the paracrine activity of exogenous stem cells and manipulate
   endogenous stem cells to activate the inherent natural repair
   mechanisms.^[ [64]^19 ^] Various growth factors have been investigated
   for TBI, such as platelet‐derived growth factor, vascular endothelial
   growth factor, and bone morphogenetic protein. However, most growth
   factors are unsuitable for long‐term tissue repair.^[ [65]^20 ^] Bone
   morphogenetic protein 2 (BMP‐2) has been used clinically because of its
   strong bone‐inducing capability.^[ [66]^21 ^] BMP‐2 induces
   chondrogenic differentiation and enhances the synthesis of fibrous
   cartilage ECM components, proteoglycans, and type II collagen.^[
   [67]^22 ^] Therefore, BMP‐2 may serve as a key signaling molecule for
   enhancing the paracrine activity of stem cells and reconstructing the
   fibrous cartilage transition zone. Numerous studies have reported the
   positive effects of BMP‐2 on TBI healing and clarified the related
   mechanisms.^[ [68]^23 ^] However, no studies have elucidated the
   mechanism by which BMP‐2 coordinates with stem cells throughout the
   transition from initial trauma to later repair, which may represent a
   novel breakthrough in TBI healing.

   In this study, we selected ADSCs as the seed cells. To establish a
   suitable microenvironment for ADSCs, we utilized the thermosensitivity
   of gelatin microspheres (GMs) as porogens and incorporated them into
   hyaluronic acid methacrylate (HAMA) to form micropores, thereby
   creating 3D porous hydrogel structures (HMs). Furthermore, gelatin
   methacryloyl (GelMA) hydrogel microspheres (GMMs) were employed to
   efficiently load BMP‐2. Briefly, ADSCs were mixed with HMs, and GMMs
   loaded with BMP‐2 (GMBs) were incorporated to develop an injectable 3D
   porous hydrogel cell‐loaded construct capable of sustaining BMP‐2
   release (HMs/MBs/ADSCs, Figure [69] 1A ). This construct was
   systematically characterized and applied to a rat rotator cuff injury
   model (Figure [70]1B). To explore the mechanism of this construct in
   the repair of the rotator cuff, in vivo and in vitro dual sequencing
   was performed to elucidate the signaling pathways through which BMP‐2
   collaborates with ADSCs to mediate TBI repair (Figure [71]1C). This
   study aimed to facilitate TBI healing by ensuring stable BMP‐2 delivery
   and continuous microenvironment adaptation while providing novel
   insights into tissue repair pathways.

Figure 1.

   Figure 1
   [72]Open in a new tab

   Schematic illustration of the design and application of HMs/MBs/ADSCs.
   A) Fabrication of the HMs/MBs/ADSCs hydrogel. After cross‐linking,
   micropores formed under body temperature conditions. Cells migrated to
   the GMBs and pore edges, eventually attaching to the GMBs and
   aggregating into clusters. B) Establishment of a rat rotator cuff
   injury model. C) Schematic representation of the mechanism mediating
   the reconstruction of the tendon‐to‐bone interface after HMs/MBs/ADSCs
   implantation.

2. Results

2.1. Preparation and Characterization of Microspheres

   Both types of microspheres were prepared using a double‐emulsion
   method. GMs were used as porogens to create pores, whereas GMMs were
   used to construct a sustained‐release system. The proton nuclear
   magnetic resonance (¹H NMR) spectra of GelMA and gelatin are shown in
   Figure [73]S1 (Supporting Information). Subsequently, microspheres with
   diameters of 100 to 150 µm were selected using a sieve. The morphology
   of the screened GMs was similar to that of the GMMs, both before and
   after crosslinking, with regular and roughly spherical shapes
   (Figure [74] 2A). After swelling in water, the diameter of the
   microspheres was 150–200 µm (Figure [75]2B). Moreover, GMs
   (1100.48 ± 95.05%) and GMMs before crosslinking (1300.04 ± 38.85%)
   exhibited a higher swelling rate than that of GMMs after crosslinking
   (695.07 ± 50.26%) (Figure [76]2C), which may be because of the tighter
   internal network structure of GMMs post‐crosslinking. In addition, the
   degree of crosslinking was indirectly evaluated by calculating the
   swelling ratio. The swelling ratio of the crosslinked GMMs was
   calculated to be 7.95 ± 0.41, whereas that of the non‐crosslinked GMMs
   was 14 ± 0.31. The swelling ratio decreased by approximately twofold
   after crosslinking, indicating a significant increase in the degree of
   crosslinking. Owing to temperature sensitivity, GMs were almost
   completely degraded within 40 min, whereas the degradation rate of the
   crosslinked GMMs was 43 ± 2.76% after 28 days (Figure [77]2D). To
   optimize BMP‐2 release, we compared two methods of BMP‐2 incorporation
   into GMMs: crosslinking before loading (GMBs1) and crosslinking after
   loading (GMBs2), and we found that GMBs2 resulted in a slower overall
   release. Generally, the release was faster during the first week, with
   a cumulative release of 42.53 ± 3.6% for GMBs2. The release then
   gradually plateaued, reaching a cumulative release rate of
   54.14 ± 3.68% by day 28. To confirm residual BMP‐2 in the microspheres,
   complete lysis was performed using GelMA lysis buffer, resulting in the
   release of 30.61 ± 2.2% from GMBs2, compared to only 3 ± 0.23% from
   GMBs1 (Figure [78]2E).

Figure 2.

   Figure 2
   [79]Open in a new tab

   Characterization of microspheres and hydrogels. A) SEM surface images
   of different microspheres. B) Wet images of different microspheres
   under a microscope. C) 24‐hour swelling rate of microspheres. D)
   Post‐crosslinking degradation behavior of GMs and GMMs. E) In vitro
   BMP‐2 release from GMMs with different loading methods. F) Macroscopic
   view of HMs/MBs hydrogel before and after crosslinking. G) SEM
   cross‐sectional images of different hydrogels. H) Degradation behavior
   of hydrogels over time. I) In vitro release of BMP‐2 coated with
   different hydrogels. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001.

2.2. Construction of HMs/MBs Hydrogel Injection System

   The ¹H NMR and Fourier‐transform infrared spectroscopy spectra of HAMA
   (the continuous phase of the hydrogel) and hyaluronic acid (HA) are
   shown in Figure [80]S2 (Supporting Information). The characteristic
   peak of methacrylate (C═C bond) at 1700 cm⁻¹ confirmed its successful
   incorporation into HA. To determine the optimal HAMA concentration, we
   compared its rheological properties at concentrations of 1%, 1.5%, and
   2%. The 2% concentration exhibited high viscosity while retaining the
   shear‐thinning behavior. Additionally, the 2% concentration exhibited a
   high storage modulus (G′) after UV crosslinking at a constant amplitude
   (γ) and frequency. Furthermore, it exhibited an improved elastic
   modulus (Figure [81]S3, Supporting Information). Regarding
   cytocompatibility, no significant differences were observed in the
   CCK‐8 assays or cell viability among the three HAMA concentrations, and
   all demonstrated good biocompatibility (Figure [82]S4, Supporting
   Information). Based on these findings, 2% HAMA was selected for
   subsequent experiments. As shown in Figure [83]2F, the HMs/MBs hydrogel
   precursor is formed by mixing GMs and GMBs into HAMA. The precursor was
   extruded from the syringe and formed into various shapes, transitioning
   from sol to gel upon UV crosslinking. By comparing the rheological
   differences between HAMA and GMs or GMBs, we found that they did not
   affect the shear‐thinning behavior of HAMA (Figure [84]S5A, Supporting
   Information). The modulus after UV crosslinking exhibited a similar
   trend: rapid curing with a sharp increase in the storage modulus, which
   stabilized after complete crosslinking (Figure [85]S5B, Supporting
   Information).

2.3. Characterization of HMs/MBs Porous Hydrogel

   Upon UV exposure, the injected hydrogel crosslinked, with HAMA
   exhibiting a uniform pore size of 50 to 100 µm. The addition of GMs
   increased the leaching pore size to ≈200 µm. Embedded GMBs were
   observed via scanning electron microscopy (SEM) (Figure [86]2G) of
   HMs/MBs without affecting the leaching pores. As shown in
   Figure [87]2H, the porous hydrogels (HMs) degraded at a faster rate,
   with an initial rapid phase followed by a slower phase; the degradation
   of HMs/MBs reached 69.97 ± 6.56% by day 28. As shown in Figure [88]2I,
   direct loading of BMP‐2 into HAMA (HB) resulted in a burst release,
   whereas controlled release was achieved through GMMs encapsulation
   (HMBs), with a cumulative release of 53.76 ± 3.92% by day 28. The pores
   formed by GMs had minimal impact on BMP‐2 release, with a cumulative
   release of 55.82 ± 2.76% for HMs/MBs. From a rheological perspective,
   the addition of GMs and GMMs had a minimal effect on the storage
   modulus at constant frequency and strain (Figure [89] 3A; Figure
   [90]S5C, Supporting Information). However, the stress–strain curve
   showed that under a strain of 45%, the three groups maintained a
   certain toughness (Figure [91]S5D, Supporting Information). At 15%
   strain, the compression modulus was 24.7 ± 1.84 kPa for HAMA,
   13.14 ± 1.96 kPa for HMs, and 15.23 ± 2.47 kPa for HMs/MBs (Figure
   [92]S5E, Supporting Information). The pores leached from the GMs
   increased both the rate and extent of hydrogel swelling. HAMA
   (1647.45 ± 153.33%) reached swelling equilibrium in 30 min, whereas HMs
   (2315.08 ± 122.68%) and HMs/MBs (2125.05 ± 124.96%) reached equilibrium
   in 90 min (Figure [93]S6, Supporting Information).

Figure 3.

   Figure 3
   [94]Open in a new tab

   Dynamic modulus of porous hydrogels and effects of the 3D environment
   of different hydrogels on ADSCs. A) Dynamic modulus from 10^−1 to 10 ω
   at a constant oscillation amplitude (γ) of 1%. B) DNA quantification of
   ADSCs encapsulated within different hydrogels. C) Confocal images of
   live/dead staining of ADSCs coated in different hydrogels. D) Confocal
   images of nuclear and cytoskeletal staining of ADSCs coated in
   different hydrogels. E) Safranin O staining of ADSCs encapsulated in
   different hydrogels. F) Masson's trichrome staining of ADSCs
   encapsulated in different hydrogels. GMBs indicated by yellow arrows;
   ADSCs indicated by black arrows. n = 3.*p < 0.05, **p < 0.01,
   ***p < 0.001.

   To investigate the 3D microenvironment created by HMs/MBs, ADSCs were
   uniformly mixed into the precursor before gelation, thereby being
   encapsulated within the hydrogel. DNA was extracted from the
   ADSCs‐encapsulated hydrogel after dissociation on days 1, 7, and 14. As
   shown in Figure [95]3B, there were no significant differences among the
   three groups on day 1; however, the HMs/MBs group showed significantly
   high levels on days 7 and 14. The confocal microscopy results of
   live/dead staining revealed that the proliferation rate of ADSCs in the
   HMs and HMs/MBs groups was higher than that in the HAMA group.
   Additionally, cell aggregation was observed on day 7, and significant
   aggregation was observed on day 14. By day 14, the aggregation in the
   HMs/MBs group was the most pronounced, forming cell clusters. This
   suggested that the addition of GMs provided space for cell
   proliferation, whereas BMP‐2 further promoted cell proliferation and
   aggregation (Figure [96]3C). Confocal images of nuclear and
   cytoskeletal staining yielded results similar to those of the live/dead
   staining. The difference lies in the fact that DAPI and actin stain
   GMBs, thereby labeling the GMBs (Figure [97]3D; Figure [98]S7,
   Supporting Information). It was more clearly observed that ADSCs
   aggregated around the GMBs, indicating that BMP‐2 might have played a
   role in attracting cell migration. Cryosections of the ADSCs‐loaded
   hydrogel constructs incubated for 14 days were prepared and stained
   with safranin O and Masson's trichrome. Compared with the cell‐free
   hydrogels (Figure [99]S8, Supporting Information), ADSCs within the
   hydrogels (indicated by black arrows) and GMBs in the HMs/MBs group
   (indicated by yellow arrows) were clearly visible (Figure [100]3E,F).
   Owing to the addition of the pore‐forming GMs, the pore size of the
   hydrogels in the HMs and HMs/MBs groups was slightly larger than that
   in the HAMA group, as seen in the safranin O images. However, this
   difference was not highly pronounced, likely because some extracellular
   matrices secreted by the cells filled the internal network of the
   hydrogels by day 14.

2.4. In Vivo Biosafety Assessment of HMs/MBs/ADSCs Hydrogels: Cell Tracking,
Degradation Behavior, and Short‐Term Inflammatory Response

   At 2 weeks post‐surgery, H&E staining of the heart, liver, spleen,
   lungs, and kidneys of the rats indicated no adverse effects on major
   organs in either the experimental or control groups (Figure [101]S10,
   Supporting Information). Additionally, the blood routine and
   liver/kidney function tests of the HMs/MBs and HMs/MBs/ADSCs hydrogel
   groups showed no statistically significant differences compared with
   those of the blank control group (Figure [102]S11, Supporting
   Information). Therefore, the in vivo implantation of HMs/MBs/ADSCs
   hydrogels is biologically safe. From the in vivo subcutaneous
   degradation experiments, it was observed that neither hydrogel group
   fully degraded after 4 weeks. However, the HMs/MBs/ADSCs hydrogel
   degraded faster than the HMs/MBs hydrogel (Figure [103]S12, Supporting
   Information). The retention of ADSCs is crucial to their paracrine
   effects. Small‐animal in vivo imaging results demonstrated that ADSCs
   encapsulated in the HMs/MBs hydrogels were retained for at least
   6 weeks post‐surgery, although the fluorescence intensity decreased
   over time (Figure [104]S13, Supporting Information). At 2 weeks
   post‐surgery, CD68 and CD163 immunofluorescence staining was performed
   on tissues from the three groups to identify M1 and M2 macrophages
   (Mφ), respectively (Figure [105]S14, Supporting Information). The study
   found that hydrogel groups with better tissue regeneration and
   immunomodulatory capabilities have the potential to polarize Mφ into
   the pro‐regenerative M2 phenotype.^[ [106]^24 ^] The control group
   exhibited the highest number of M1 Mφ and the lowest number of M2 Mφ,
   indicating an acute inflammatory response in the surgical area. By
   contrast, the hydrogel groups demonstrated varying degrees of
   immunomodulatory ability, with reduced levels of acute inflammation
   compared with those of the control group. The HMs/MBs/ADSCs group
   showed more M2 Mφ and fewer M1 Mφ than those of the HMs/MBs group.

2.5. Biomechanical and Bone Repair Effects of HMs/MBs/ADSCs in a Rat Rotator
Cuff Tear Model

   Rats were assigned to three groups as illustrated in Figure [107] 4A.
   Biomechanical tests were performed on the regenerated supraspinatus
   tendon–humerus complexes at 6 and 12 weeks (Figure [108]4C). No
   significant differences were observed in the cross‐sectional areas of
   the supraspinatus tendon (Figure [109]4D). At 6 weeks, the maximum
   failure load was significantly higher in the HMs/MBs/ADSCs
   (21.46 ± 1.76 N) and HMs/MBs (15.05 ± 1.73 N) groups compared with that
   in the blank group (9.48 ± 1.04 N). Stiffness (11.16 ± 1.2 N mm^−1) and
   stress (3.1 ± 0.14 mm⁻¹) in the HMs/MBs/ADSCs group were also higher
   than in the blank group (stiffness: HMs/MBs, 8.28 ± 1.18 N mm^−1;
   blank, 4.74 ± 1.89 N mm^−1; stress: HMs/MBs, 2.16 ± 0.16 mm⁻¹, blank,
   1.35 ± 0.08 mm⁻¹). At 12 weeks, the trend in maximum failure load was
   similar to that at 6 weeks, with stiffness and stress significantly
   higher in both experimental groups than in the blank group (Ultimate
   load to failure: HMs/MBs/ADSCs, 32.48 ± 1.89 N; HMs/MBs,
   27.44 ± 1.31 N; blank, 20.54 ± 3.3 N; stiffness: HMs/MBs/ADSCs,
   22.08 ± 2.33 N mm^−1; HMs/MBs, 13.63 ± 2.71 N mm^−1; blank,
   9.49 ± 1.45 N mm^−1; stress: HMs/MBs/ADSCs, 4.4 ± 0.15 mm⁻¹; HMs/MBs,
   3.67 ± 0.07 mm⁻¹; blank, 2.73 ± 0.35 mm⁻¹). The HMs/MBs/ADSCs group
   exhibited the best overall biomechanical properties, more than twice
   that of the blank group. At 12 weeks post‐surgery, new bone formation
   in the humeral head was further evaluated using micro‐CT
   (Figure [110]4E); 3D reconstruction indicated that new bone
   regeneration in the HMs/MBs/ADSCs group was more substantial and
   complete. Bone volume/total volume (BV/TV), bone mineral density (BMD),
   trabecular number (Tb.N), and trabecular separation (Tb.Sp) were
   semi‐quantitatively analyzed (Figure [111]4F). At 12 weeks, BV/TV and
   BMD in the experimental groups were significantly higher than in the
   blank group (BV/TV [%]: HMs/MBs/ADSCs, 44.17 ± 2.3; HMs/MBs,
   34.61 ± 2.65; blank, 22.44 ± 2.45; BMD (g cm^− ^3): HMs/MBs/ADSCs,
   0.48 ± 0.02; HMs/MBs, 0.39 ± 0.02; blank, 0.27 ± 0.02). Additionally,
   compared with observations in the blank group, Tb.Sp in HMs/MBs/ADSCs
   was lower, whereas Tb.N was higher (Tb.Sp (mm): HMs/MBs/ADSCs,
   0.4 ± 0.03; HMs/MBs, 0.44 ± 0.06; blank, 0.55 ± 0.04; Tb.N (mm⁻¹):
   HMs/MBs/ADSCs, 1.95 ± 0.29, HMs/MBs, 1.87 ± 0.23, blank, 1.32 ± 0.17).

Figure 4.

   Figure 4
   [112]Open in a new tab

   Biomechanical properties and micro‐CT analysis. A) Diagram illustrating
   of the animal experimental design. B) Location of supraspinatus tendon.
   C) Stretching diagram of supraspinatus tendon‐humeral tissue. D)
   Tensile test analysis of ultimate load to failure, stiffness,
   cross‐sectional area, and stress. n = 5. E) Micro‐CT scans of the
   proximal humerus. F) Quantitative micro‐CT analysis of BV/TV, BMD,
   Tb.Sp, and Tb.N. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001.

2.6. Fibrocartilage and Tendon Regeneration Induced by HMs/MBs/ADSCs In Vivo

   According to H&E staining (Figure [113] 5A) and Masson's trichrome
   staining (Figure [114]S15, Supporting Information), the blank group
   exhibited higher cell density and a more disorganized cell arrangement
   than those of the two experimental groups at 6 weeks. Tendon collagen
   fibers in the blank group appeared disorganized and did not align
   toward the bone. By contrast, the collagen fibers in the experimental
   groups showed an initial alignment, with the HMs/MBs/ADSCs group
   demonstrating a slightly more organized arrangement than that of the
   HMs/MBs group. At 12 weeks, the cell density decreased across all
   groups; however, collagen fibers in the blank group remained
   disorganized, hindering tendon regeneration. Collagen fibers were
   organized in the experimental groups; however, the distal end of the
   HMs/MBs supraspinatus tendon was slightly disorganized, whereas the
   HMs/MBs/ADSCs group exhibited a highly organized arrangement, leading
   to superior tendon regeneration. Therefore, the histological scores of
   the HMs/MBs/ADSCs group were significantly higher than those of the
   other groups (Figure [115]5B). Fibrocartilage regeneration was
   evaluated by safranin fast green staining (Figure [116]5C,D) and
   immunohistochemical staining for COL I and COL II (Figure [117]5E–H).
   At 6 and 12 weeks, tendon–bone fusion in the blank group remained poor,
   with a distinct separation at the boundary. Fibrocartilage‐like tissue
   formation was evident in both experimental groups at 6 weeks, with
   interface fusion beginning. Chondrocytes in the HMs/MBs/ADSCs group
   were arranged more neatly than those in the HMs/MBs group. At 12 weeks,
   chondrocytes in the HMs/MBs/ADSCs group exhibited a more regular
   arrangement and achieved the best interface fusion. Consequently, COL I
   and COL II expression levels were significantly high in the
   HMs/MBs/ADSCs group. Picrosirius red staining further confirmed the
   formation of more mature regenerated tendons in the HMs/MBs/ADSCs group
   at both 6 and 12 weeks, characterized by high collagen type I and low
   collagen type III contents (Figure [118]5I,J; Figure [119]S16,
   Supporting Information).

Figure 5.

   Figure 5
   [120]Open in a new tab

   Morphological analysis of newly formed TBI at 6 and 12 weeks after
   rotator cuff repair. A) H&E staining of TBI. B) Histological score of
   TBI. C) Safranin fast green staining of TBI. D) Semi‐quantitative
   analysis of the metachromatic areas of safranin fast green staining.
   E–H) Immunohistochemical staining images and semi‐quantitative analysis
   of COL I and COL II. I) Picrosirius red staining of TBI. J)
   Semi‐quantitative analysis of collagen composition after 12 weeks in
   Picrosirius red staining. T: tendon; I: interface; B: bone. n = 3.
   *p < 0.05, **p < 0.01, ***p < 0.001.

2.7. Mechanism by Which HMs/MBs and HMs/MBs/ADSCs Promote TBI Regeneration
Explored by In Vivo Transcriptome Sequencing

   To elucidate the molecular mechanisms by which HMs/MBs/ADSCs regulate
   TBI healing, we employed RNA sequencing to systematically analyze
   relevant gene expression profiles. Principal component analysis (PCA)
   was first conducted to assess homogeneity and heterogeneity among the
   experimental groups (Figure [121] 6A). The results showed that PC1 and
   PC2 accounted for 42.5% and 11.3% of the total variance, respectively.
   The blank and treatment groups (HMs/MBs and HMs/MBs/ADSCs) exhibited
   significant separation along the PC1 axis, indicating that the
   treatment conditions had a substantial impact on gene expression
   profiles. Additionally, the distribution differences between the
   HMs/MBs and HMs/MBs/ADSCs groups along the PC1 and PC2 axes suggest
   that the incorporation of ADSCs may exert additional regulatory effects
   on gene expression. No significant outliers were observed, confirming
   the reliability of the data for subsequent analyses. Based on
   differential expression analysis, volcano plot results (Figure
   [122]S17A,B, Supporting Information) revealed that, compared with the
   blank group, the HMs/MBs group had 4780 significantly upregulated and
   3048 significantly downregulated genes, whereas the HMs/MBs/ADSCs group
   exhibited 7512 upregulated and 6920 downregulated genes. Further
   heatmap analysis of genes related to TBI healing (Figure [123]6B,C)
   demonstrated that, compared with the blank group, both treatment groups
   exhibited a significant upregulation trend in genes associated with
   osteochondral and tendon healing. To explore potential signaling
   pathways, the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway
   enrichment analysis was performed (Figure [124]6D,F). The results
   indicated that the transforming growth factor‐beta (TGF‐β) signaling
   pathway was significantly upregulated in both the HMs/MBs and
   HMs/MBs/ADSCs groups compared with that in the blank group. Combined
   with gene heatmap analysis, a notable increase in Smad3 expression was
   observed, suggesting that the TGF‐β/Smad2/3 signaling pathway may play
   a critical role in this process. This finding was consistent with Gene
   Ontology functional enrichment analysis, further supporting the
   importance of the TGF‐β/Smad2/3 pathway in TBI healing
   (Figure [125]6E,G). To validate these findings, immunofluorescence
   staining and western blotting were used to detect key molecules of the
   TGF‐β/Smad2/3 signaling pathway in tissue samples. Immunofluorescence
   results (Figure [126]6H,K) showed that, compared with that in the blank
   group, p‐Smad2/3 expression was significantly enhanced in both the
   HMs/MBs and HMs/MBs/ADSCs groups, with the HMs/MBs/ADSCs group
   exhibiting a more pronounced effect. Western blot results were
   consistent with the immunofluorescence findings (Figure [127]6I,J),
   demonstrating a marked upregulation of p‐Smad2/3 protein levels in the
   HMs/MBs (1.83, p < 0.01) and HMs/MBs/ADSCs (2.55, p < 0.001) groups
   compared with those in the blank group.

Figure 6.

   Figure 6
   [128]Open in a new tab

   Transcriptome sequencing and related pathway analysis of tissue healing
   regulated in vivo by composite hydrogels with or without cells. A) PCA
   of all samples in the Blank, HMs/MBs, and HMs/MBs/ADSCs groups. B,C)
   Heatmaps of genes associated with tissue repair. D,F) KEGG enrichment
   analysis of DEGs. E,G) GO enrichment analysis of DEGs. H,K)
   Immunofluorescence staining and semi‐quantitative analysis for
   p‐Smad2/3. I,J) Western blot and semi‐quantitative analysis of Smad2/3
   and p‐Smad2/3. n = 3. *p < 0.05, **p < 0.01, ***p < 0.001.

2.8. Possible Regulatory Mechanism of BMP‐2 and ADSCs‐Loaded Porous Hydrogel
Explored In Vitro

   During in vivo sequencing analysis, a critical question emerged:
   whether ADSCs play an independent role in this process or function
   synergistically with BMP‐2 in distinct ways. To investigate this
   further, based on Figure [129]3, ADSCs were loaded into HMs and HMs/MBs
   hydrogel precursors. After cross‐linking, HMs/ADSCs and HMs/MBs/ADSCs
   hydrogels were obtained. An in‐depth in vitro RNA sequencing study was
   subsequently conducted. PCA revealed that PC1 and PC2 accounted for
   53.7% and 18.3% of the total variance, respectively (Figure [130] 7A).
   Notably, the HMs/ADSCs and HMs/MBs/ADSCs groups exhibited significant
   separation along the PC1 axis, indicating that MBs had a substantial
   impact on gene expression, whereas variation along the PC2 axis may
   reflect minor differences within or between groups. No significant
   outliers were observed, confirming the high data quality. Volcano plot
   analysis demonstrated that, compared with the HMs/MBs group, the
   HMs/MBs/ADSCs group had 6906 significantly upregulated and 5370
   significantly downregulated genes (Figure [131]S18A, Supporting
   Information). Further heatmap analysis revealed that compared with
   observations in the HMs/ADSCs group, genes related to osteochondral and
   tendon healing were significantly upregulated in the HMs/MBs/ADSCs
   group, whereas inflammation‐related genes were significantly
   downregulated (Figure [132]7B). To explore potential signaling
   pathways, KEGG pathway enrichment analysis was performed, revealing
   that the AMPK pathway was significantly upregulated in the
   HMs/MBs/ADSCs group (Figure [133]7C). Notably, consistent with the in
   vitro results, the AMPK pathway also exhibited an upregulation trend in
   tissue samples from the HMs/MBs/ADSCs group compared with those from
   the HMs/MBs group (Figure [134]7D). These results suggest that the AMPK
   pathway plays a critical role in the synergistic effects of BMP‐2 and
   ADSCs (Figure [135]7E). To validate the in vivo sequencing findings,
   immunofluorescence staining, and western blotting were used to detect
   key molecules of the AMPK signaling pathway in tissue samples.
   Immunofluorescence results showed that p‐AMPK expression was
   significantly enhanced in the HMs/MBs/ADSCs group compared with that in
   the blank and HMs/MBs groups (Figure [136]7F,I). Western blotting
   results further confirmed this observation, demonstrating a marked
   upregulation of p‐AMPK protein levels in the HMs/MBs/ADSCs group (2.41,
   p < 0.001) compared with those in the blank and HMs/MBs groups
   (Figure [137]7G,H).

Figure 7.

   Figure 7
   [138]Open in a new tab

   Secondary analysis of in vivo transcriptomes, pathway validation, and
   in vitro transcriptome sequencing of HMs/ADSCs and HMs/MBs/ADSCs. A)
   PCA of all samples from HMs/ADSCs and HMs/MBs/ADSCs groups. B) Heatmap
   of genes associated with cell differentiation and inflammation. C) KEGG
   enrichment analysis of DEGs in vitro. D) KEGG enrichment analysis of
   DEGs in vivo. E) Diagram of selecting the AMPK signaling pathway as a
   critical pathway. F,I) Immunofluorescence staining and
   semi‐quantitative analysis for p‐AMPK. G,H) Western blot and
   semi‐quantitative analysis of AMPK and p‐AMPK. n = 3. *p < 0.05,
   **p < 0.01, ***p < 0.001.

2.9. Validation of Mechanisms by Which HMs/MBs/ADSCs Regulate Bone
Marrow‐Derived Stem Cells (BMSCs) Function through Paracrine Effects

   AMPK can mitigate mitochondrial damage caused by ROS and reduce ROS
   levels, thereby exerting antioxidant effects.^[ [139]^25 ^] Therefore,
   it is hypothesized that ADSCs activate the AMPK pathway via paracrine
   signaling, thereby preventing oxidation. To simulate the in vivo
   microenvironment and verify the expression of the relevant pathways, an
   in vitro oxidative stress model was constructed. BMSCs, the most
   enriched cells at the interface, were selected as host cells
   (Figure [140] 8A). To further validate this experimental condition,
   CCK‐8 assay results showed that after 24 h of H₂O₂ stimulation, BMSCs
   viability showed no statistically significant difference compared with
   that in the control group. Additionally, after switching to a complete
   culture medium, the cells were still able to proliferate rapidly,
   although differences in proliferation were observed compared with the
   proliferation in the control group (Figure [141]S19A, Supporting
   Information). The live/dead assay results indicated that the morphology
   of the stimulated cells was slightly different from that of the control
   group, with almost no significant number of dead cells (Figure
   [142]S19B, Supporting Information). The AMPK pathway was verified using
   western blotting (Figure [143]8B,C) and immunofluorescence
   (Figure [144]8F; Figure [145]S20A, Supporting Information). Both showed
   consistent results, with significantly higher p‐AMPK protein expression
   in the H₂O₂+HMs/MBs/ADSCs and H₂O₂+HMs/ADSCs groups than in the control
   and H₂O₂ groups. This effect was more pronounced in the
   H₂O₂+HMs/MBs/ADSCs group (H₂O₂+HMs/MBs/ADSCs versus H₂O₂+HMs/ADSCs in
   WB: 3.46 vs 2.07, p < 0.001). Additionally, the in vitro expression of
   the TGF‐β signaling pathway was further validated in vivo. Western
   blotting (Figure [146]8D,E) and immunofluorescence (Figure [147]8G;
   Figure [148]S20B, Supporting Information) demonstrated that
   HMs/MBs/ADSCs significantly increased p‐Smad2/3 protein expression
   (H₂O₂+HMs/MBs/ADSCs versus Control in WB: 1.87, p < 0.001), whereas the
   H₂O₂ group showed a significant decrease (H₂O₂ vs Control in WB: 0.57,
   p < 0.01). Subsequently, the specific paracrine factors that activate
   both pathways were investigated. Transcriptome data were further
   analyzed by differentially expressed genes, combined with a literature
   review, identifying two factors: adiponectin (APN) and TGF‐β3
   (Figure [149]7B; Figure [150]S18A, Supporting Information). ELISA
   results indicated significantly higher APN concentrations in the
   H₂O₂+HMs/MBs/ADSCs group (327.78 ± 41.55 pg mL^−1) compared with the
   other groups (APN (pg mL^−1): H₂O₂+HMs/ADSCs: 74.49 ± 2, H₂O₂:
   49.67 ± 3.84, Control: 43.11 ± 3.58) (Figure [151]8H). TGF‐β3
   concentrations significantly increased in both the H₂O₂+HMs/MBs/ADSCs
   (107.47 ± 6.25 pg mL^−1) and H₂O₂+HMs/ADSCs (56.25 ± 8.48 pg mL^−1)
   groups, compared with those in the H₂O₂ (below the detection limit) and
   Control (9.62 ± 3.97 pg mL^−1) groups (Figure [152]8I).

Figure 8.

   Figure 8
   [153]Open in a new tab

   Validation of the mechanism of how HMs/MBs/ADSCs regulate BMSCs
   function through paracrine effects. A) Schematic illustration of
   experimental design. B,C) Western blot and semi‐quantitative analysis
   of AMPK and p‐AMPK. D,E) Western blot and semi‐quantitative analysis of
   Smad2/3 and p‐Smad2/3. F,G) Immunofluorescence staining of p‐AMPK and
   p‐Smad2/3. H) The concentration of APN was measured by ELISA. I) The
   concentration of TGF‐β3 was measured by ELISA. n = 3. *p < 0.05, **p <
   0.01, ***p < 0.001, when compared with control group. ^# p < 0.05, ^##
   p < 0.01, ^### p < 0.001, when compared with H[2]O[2] group. ^& p <
   0.05, ^&& p < 0.01, ^&&& p < 0.001, when compared with
   H[2]O[2]+HMs/ADSCs group.

2.10. Evaluation of Antioxidant Effects of HMs/MBs/ADSCs In Vitro

   Mitochondria generate ROS as byproducts of the respiratory chain. While
   moderate levels maintain cellular functions, excessive ROS damage
   mitochondria.^[ [154]^26 ^] To verify the antioxidant effect of
   HMs/MBs/ADSCs, this study constructed an oxidative stress model based
   on ROS upregulation during the inflammatory phase, which induces
   increased cellular ROS production (Figure [155]8A). Free radical
   scavenging in the supernatant revealed that the HMs/MBs/ADSCs
   (64.22 ± 3.5%) and HMs/ADSCs (53.2 ± 1.55%) groups exhibited higher
   DPPH radical scavenging rates compared with those of the H₂O₂ group
   (16.15 ± 2.94%) (Figure [156] 9A). Similarly, the HMs/MBs/ADSCs
   (69.62 ± 1.34%) and HMs/ADSCs (58.86 ± 3.01%) groups showed higher ABTS
   radical scavenging rates than those of the H₂O₂ group (13.24 ± 5.39%)
   (Figure [157]9B). Free radical scavenging assays demonstrated that the
   hydrogels effectively cleared DPPH and ABTS radicals, achieving
   antioxidant effects. Additionally, the concentration of proinflammatory
   factors (interleukin‐1β [IL‐1β] and tumor necrosis factor [TNF]‐α) in
   the supernatant was measured (Figure [158]9C,D). The results indicated
   that the HMs/MBs/ADSCs and HMs/ADSCs groups had lower levels of
   proinflammatory factors compared with those of the H₂O₂ group, with the
   HMs/MBs/ADSCs group showing more pronounced effects (IL‐1β:
   62.94 ± 2.92 pg mL^−1 for HMs/MBs/ADSCs, 139.24 ±  12.04 pg mL^−1 for
   HMs/ADSCs, 353.55 ± 13.06 pg mL^−1 for H₂O₂; TNF‐α:
   20.09 ± 1.79 pg mL^−1 for HMs/MBs/ADSCs, 36.24 ± 4.08 pg mL^−1 for
   HMs/ADSCs, 218.53 ± 16.07 pg mL^−1 for H₂O₂). Total ROS and superoxide
   levels were quantified using DCFH‐DA and MitoSOX reagents,
   respectively. Compared with that in the H₂O₂ group, DCFH‐DA
   fluorescence intensity was markedly reduced in the H₂O₂+HMs/MBs/ADSCs
   (0.14, p < 0.001) and H₂O₂+HMs/ADSCs (0.43, p < 0.001) groups
   (Figure [159]9E,H). Moreover, superoxide levels in the
   H₂O₂+HMs/MBs/ADSCs (0.10, p < 0.001) and H₂O₂+HMs/ADSCs (0.34,
   p < 0.001) groups were significantly reduced (Figure [160]9F,I).
   Therefore, HMs/ADSCs and HMs/MBs/ADSCs hydrogels effectively cleared
   excess ROS. Excess ROS accumulation disrupts mitochondrial structure
   and alters membrane potential. The H₂O₂ group exhibited significantly
   high levels of JC‐1 monomers (Figure [161]9G). Both HMs/MBs/ADSCs
   (0.12, p < 0.001) and HMs/ADSCs (0.35, p < 0.001) effectively
   stabilized ΔΨm, with the effect being more pronounced in the
   HMs/MBs/ADSCs group (p < 0.001) (Figure [162]9J). The H₂O₂ group showed
   a significant decrease in membrane potential, which stabilized after
   hydrogel intervention. Mitochondria are the primary sites for ATP
   production. To indirectly assess mitochondrial function, ATP levels
   were further measured. The results indicated that HMs/MBs/ADSCs
   effectively restored ATP production (H₂O₂+HMs/MBs/ADSCs vs H₂O₂:
   0.72 vs 0.16, p < 0.001) (Figure [163]9K). These results suggest that
   BMP‐2 and ADSCs act synergistically. The AMPK pathway is activated at
   an early stage, leading to a reduction in ROS production. An
   antioxidant effect is achieved in BMSCs, thereby restoring their
   biological functions.

Figure 9.

   Figure 9
   [164]Open in a new tab

   HMs/MBs/ADSCs alleviated the H[2]O[2]‐induced oxidative stress of
   BMSCs. A,B) Scavenging rate of DPPH and ABTS free radical. C,D) The
   concentration of IL‐1β and TNF‐α was measured by ELISA. E)
   Representative DCFH‐DA staining images for intracellular ROS. F)
   Representative MitoSOX staining images. G) Immunofluorescence images of
   mitochondrial membrane potential (ΔΨm) detected by JC‐1 probe. H–J)
   Semiquantitative analysis of the relative fluorescent intensity of
   (E)–(G). K) ATP levels in BMSCs measured by an ATP Assay Kit. n = 3.
   *p < 0.05, **p < 0.01, ***p < 0.001, when compared with control group.
   ^# p < 0.05, ^## p < 0.01, ^### p < 0.001, when compared with H[2]O[2]
   group. ^& p < 0.05, ^&& p < 0.01, ^&&& p < 0.001, when compared with
   H[2]O[2]+HMs/ADSCs group.

2.11. HMs/MBs/ADSCs Enhanced Proliferation and Migration of BMSCs and
Promoted Differentiation of Both BMSCs and Tendon‐Derived Stem Cells (TDSCs)

   A Transwell chamber was used to co‐culture BMSCs with different
   hydrogels to compare cell proliferation. Based on live/dead cell
   staining and CCK‐8 results, no significant differences were observed
   between the groups on the first day. On days 3 and 5, the OD values
   were significantly higher in the HMs/MBs/ADSCs group than in the
   control group, with almost no dead cells and a significantly high cell
   count in the area (Figure [165]S21A,B, Supporting Information). Scratch
   assays and Transwell migration experiments were conducted to evaluate
   the migration capacity of the hydrogels (Figure [166]S22, Supporting
   Information). The scratch assay results demonstrated that the
   HMs/MBs/ADSCs group exhibited greater recovery areas (Figure [167]
   10A,C), whereas crystal violet staining revealed that the HMs/MBs/ADSCs
   group had the highest number of migrating cells (Figure [168]10B,D).
   RT‐qPCR analysis demonstrated enhanced expression of osteogenic genes
   (OPN, OCN, and RUNX2) (Figure [169]10E), chondrogenic genes (COL II,
   SOX9, and ACAN) (Figure [170]10F), and tenogenic genes (COL I, TNMD,
   and SCX) (Figure [171]10G) in the HMs/MBs/ADSCs group.
   Immunofluorescence analysis confirmed the upregulated expression of
   OPN, COL II, and TNMD in the HMs/MBs/ADSCs group (Figure [172]10H–M).

Figure 10.

   Figure 10
   [173]Open in a new tab

   HMs/MBs/ADSCs enhance the migration of BMSCs and facilitate the
   differentiation of both BMSCs and TDSCs. A) Microscopy images of the
   scratch wounds at 0, 12, and 24 h. B) Transwell assay of BMSCs
   migration. C) Quantitative analysis of the recovery area in the scratch
   assay. D) Quantitative analysis of BMSC migration counts in the
   Transwell assay. E) RT‐qPCR analysis of osteogenesis‐related gene
   expression (OPN, OCN, and RUNX2) in BMSCs. F) RT‐qPCR analysis of
   chondrogenesis‐related gene expression (COL II, SOX9, and ACAN) in
   BMSCs. G) RT‐qPCR analysis of tenogenesis‐related gene expression (COL
   I, TNMD, and SCX) in TDSCs. H–J) Immunofluorescence analysis of related
   protein expression (OPN, COL II, and TNMD). K–M) Semi‐quantitative
   analysis of protein expression from immunofluorescence assays. n = 3.
   *p < 0.05, **p < 0.01, ***p < 0.001.

3. Discussion

   Tissue‐engineered hydrogels represent a promising solution for TBI
   treatment, as they inherently integrate ECM, stem cells, and
   osteochondral‐inducing components. Moreover, hydrogels can be injected
   into irregular TBI defect sites using minimally invasive strategies.
   Macroporous hydrogels have garnered increasing attention for use in
   tissue regeneration owing to their superior structural properties.^[
   [174]^27 ^] In this study, the thermosensitive properties of GMs were
   exploited to create micropores in a hydrogel, thereby constructing a 3D
   porous hydrogel system. Based on previous research by the group,^[
   [175]^28 ^] the thickness of the tissue layer formed by cell growth is
   ≈50–100 µm. Accordingly, the diameter of the microspheres was designed
   to be 1.5 or 2 times this thickness to ensure that the internal space
   of the microspheres was mostly filled after cell growth. When
   microspheres are filled with cellular tissue, adjacent microspheres may
   connect and eventually form a complete tissue structure. Therefore, GMs
   with a dry‐state diameter of 100 to 150 µm and hydrated‐state diameter
   of 150 to 200 µm were constructed to provide a 3D spatial interface,
   promoting cell growth and tissue formation within the microspheres. The
   hydrogel exhibited a high swelling ratio, which further increased with
   the addition of GMs. From the perspective of polymer swelling, the HAMA
   hydrogel used in this study contained numerous hydrophilic groups. In
   addition, the incorporation of GMs created more porous structures
   within the system. These pores provided additional channels for water
   molecules to penetrate and significantly enhanced the swelling process.
   This mechanism ensures that the cells within the hydrogel are supplied
   with sufficient water and nutrients, thereby maintaining normal
   cellular metabolism. As shown in Figure [176]3B–D, ADSCs encapsulated
   in porous HAMA exhibited significantly enhanced proliferation over time
   compared with those in non‐porous HAMA, with a tendency to aggregate.
   This finding indicated that the porous platform facilitated cell homing
   and enhanced cell proliferation. Encapsulating chondrocytes in porous
   GelMA hydrogels facilitates chondrocyte proliferation and promotes the
   secretion of ECM components, including type II collagen, type I
   collagen, and fibronectin, after 28 days.^[ [177]^29 ^] Similarly,
   Chang et al. utilized GMs as porogens to fabricate a 3D porous sodium
   alginate hydrogel that enhanced the proliferation and protein secretion
   of cancer cells.^[ [178]^30 ^]

   To further regulate cellular functions and sustain paracrine activity,
   BMP‐2 was incorporated into a porous platform. The effective release of
   BMP‐2 is crucial for tissue repair. Current studies indicate that the
   covalent attachment of growth factors to carriers is commonly
   accomplished through chemical reactions involving carboxyl, amine, or
   cysteine functional groups, facilitating long‐term release.^[ [179]^31
   ^] However, chemical conjugation often compromises the biological
   activity of growth factors. From a clinical translation perspective,
   scaffolds preloaded with specific growth factors are difficult to
   preserve and exhibit limited applicability. The gentle adsorption‐based
   loading of BMP‐2 after drying effectively preserved its biological
   activity without disrupting its complex and ordered spatial structure.
   Most studies have performed cross‐linking prior to immersion in a
   growth factor or incorporated it during preparation, often resulting in
   significant losses.^[ [180]^32 ^] As the swelling rate of the
   microspheres before crosslinking was higher than that after
   crosslinking (Figure [181]2C), more factors could potentially be
   encapsulated by immersing them in the BMP‐2 solution before
   crosslinking. These results demonstrated that the sustained‐release
   effect was superior (Figure [182]2E). Moreover, more BMP‐2 was retained
   after GMM lysis, suggesting that this loading method preserved the
   biological activity of BMP‐2. This approach likely acts as a protective
   shield for simply adsorbed BMP‐2, reducing the initial burst release
   rate, whereas BMP‐2 adsorbed via electrostatic interactions is released
   only during GMMs degradation. Subsequently, the BMP‐2‐loaded
   microspheres were encapsulated within the porous HAMA hydrogel,
   providing dual protection for BMP‐2 and promoting its controlled
   release (Figure [183]2I). Compared with observations in most studies,
   the overall release rate of BMP‐2 was slower, with a relatively lower
   initial burst release rate.^[ [184]^21 , [185]^33 ^] This release
   profile aligns better with the requirements of the tissue repair
   process. The presence of BMP‐2 enabled ADSCs to establish a
   multifunctional interaction platform within a 3D porous environment.
   ADSCs encapsulated in BMP‐2 exhibited a faster proliferation rate.
   Additionally, the GelMA microspheres attracted ADSCs owing to their
   BMP‐2 activity, leading to cell adhesion on the microspheres. BMP‐2
   interacts closely with the ECM. By binding to integrins on the cell
   surface, BMP‐2 facilitates cell adhesion and migration and exerts
   significant biological functions.^[ [186]^34 ^]

   Among the four‐layer structures of the TBI, fibrocartilage tissue
   exhibits limited self‐repair capacity and is prone to scar tissue
   formation, which increases the risk of re‐tearing. This has
   consistently been the focus of this study. Wang et al. demonstrated
   that BMP‐2 facilitates the repair of osteochondral tissue at the TBI.^[
   [187]^35 ^] Lee et al. injected a collagen gel containing recombinant
   human BMP‐2 into the tendon–bone tunnel interface, resulting in
   fibrocartilage and new bone formation after 6 weeks.^[ [188]^36 ^]
   Similarly, our tissue repair outcomes demonstrated that BMP‐2 was more
   effective in promoting TBI healing in the rat rotator cuff injury model
   than in the suture‐only group. Micro‐CT results showed that the two
   BMP‐2 groups exhibited greater new bone formation, highlighting the
   potent osteoinductive capacity of BMP‐2. Histological analysis revealed
   that the BMP‐2 group achieved interface fusion by 6 weeks, with
   significant fibrocartilage formation evident at 12 weeks. Additionally,
   combined picrosirius red staining and immunohistochemistry of COL I and
   COL II confirmed that the interface in these groups formed
   fibrocartilage primarily composed of COL I and COL II, rather than scar
   tissue dominated by COL III. Additionally, TBI biomechanics
   demonstrated better mechanical strength. In conclusion, BMP‐2 signaling
   plays a critical role in the reconstruction of the fibrocartilage
   transition zone. However, the combination of BMP‐2 and ADSCs
   demonstrated superior TBI healing outcomes, including enhanced bone
   regeneration, mechanical strength, and interface fusion.

   To explore the repair mechanism, we sequenced the samples 6 weeks after
   repair. Interestingly, it was discovered that in both BMP‐2‐loaded
   groups, the pathway promoting tissue repair was the TGF‐β/Smad2/3
   pathway, rather than the classical Smad1/5/8 pathway frequently
   reported in the literature.^[ [189]^37 ^] The TGF‐β/Smad superfamily
   plays a pivotal role in cartilage regeneration. TGF‐β/Smad signaling is
   essential for cartilage generation and ECM maintenance, with multiple
   regulatory branches. Among them, the TGF‐β/Smad2/3 signaling pathway
   promotes cartilage formation while inhibiting chondrocyte
   hypertrophy.^[ [190]^38 ^] The Smad1/5/8 pathway promotes osteogenesis
   and inhibits ectopic ossification but induces chondrocyte
   hypertrophy.^[ [191]^39 ^] Hu et al. demonstrated through
   transcriptomic and proteomic analyses that TGF‐β/Smad1/5/8 inhibits
   osteochondral tissue regeneration.^[ [192]^40 ^] Zhang et al. also
   reported crosstalk between the TGF‐β/Smad and integrin signaling
   pathways, which regulate mesenchymal stem cell chondrogenesis.^[
   [193]^41 ^] Specifically, blocking the TGF‐β/Activin/Nodal signaling
   pathway induces hypertrophy and activates the BMP/Smad1/5/8 and
   integrin signaling pathways. By contrast, inhibiting the integrin–ECM
   interaction suppresses hypertrophy while activating TGF‐β/Smad2/3.
   Thus, under different conditions, Smad2/3 and Smad1/5/8 transition
   between pathways to support cartilage formation. BMP‐2 can upregulate
   or downregulate Smad2/3 under certain conditions.^[ [194]^42 ^]
   However, no related studies have been conducted on bone tissue
   engineering. Chen et al. investigated cartilage regeneration mechanisms
   in osteoarthritis using hyaluronic acid and platelet‐rich plasma,
   finding that they activate TGF‐βRII and downstream Smad2/3 signaling.^[
   [195]^43 ^] Therefore, it is hypothesized that the study mediated by
   porous HAMA hydrogels could stimulate fibrocartilage regeneration by
   BMP‐2 mainly through the TGF‐β/Smad2/3 pathway. Additionally,
   sequencing results revealed that TGF‐β3 was the predominant isoform
   expressed in the TGF‐β family, with significant release detected in the
   supernatants of HMs/MBs/ADSCs and BMSCs co‐culture (Figure [196]8I).
   These findings demonstrate that ADSCs exert their effects through
   paracrine signaling, whereas BMP‐2 enhances their activity, jointly
   inducing robust expression of the TGF‐β/Smad2/3 pathway. TGF‐β3,
   expressed in both cartilage and tendon, activates downstream signaling
   pathways by binding to TGF‐β receptors on the cell membrane.^[ [197]^44
   ^] Upon receptor binding, kinase activity is activated, leading to
   phosphorylation of Smad2 and Smad3. Phosphorylated Smad2 and Smad3 bind
   to Smad4 to form a complex that translocates into the nucleus to
   function as a transcription factor regulating target gene expression
   (Figure [198]1C). TGF‐β3 and Smad2/3 are pivotal in tissue
   reconstruction.^[ [199]^45 ^] They promote fibroblast proliferation and
   collagen synthesis, thereby facilitating tissue repair. Activating the
   TGF‐β/Smad2/3 pathway promotes the expression of cartilage‐ and
   tendon‐related genes.^[ [200]^46 ^]

   In vitro experiments demonstrated that ADSCs within the 3D porous
   hydrogel supported by BMP‐2 not only promoted BMSCs proliferation and
   migration but also enhanced osteogenic and chondrogenic gene
   expression. Additionally, tendon‐related gene expression was
   upregulated in TDSCs (Figure [201]10G,J). BMP‐2 enhances matrix
   metalloproteinase expression, facilitating cell–matrix interactions and
   strengthening migratory capacity.^[ [202]^47 ^] For example, in vitro
   experiments showed that MSCs treated with BMP‐2 exhibited enhanced
   migratory activity, potentially providing additional cellular sources
   for tissue regeneration.^[ [203]^48 ^] Thus, loading BMP‐2 into the
   porous hydrogel may activate the TGF‐β pathway, stimulating endogenous
   stem cells to migrate inward and exert biological functions. BMP‐2
   synergizing with ADSCs simultaneously leverages the activities of both
   exogenous and endogenous stem cells, leading to a more pronounced
   pathway activation. To investigate the mechanisms underlying
   ADSCs‐mediated pathway activation, we performed in vitro sequencing.
   The AMPK pathway plays a key role in this process, as its activation
   reverses mitochondrial damage caused by ROS, reduces ROS levels, and
   exerts antioxidant effects.^[ [204]^49 ^] Interestingly, the AMPK
   pathway was also detected during the 6‐week repair stage. Its
   expression indicates healthy mitochondria, ensuring adequate energy
   supply for tissue repair.^[ [205]^50 ^] Zhang et al. investigated
   ADSCs‐derived exosomes in preventing degenerative changes in torn human
   rotator cuff tendons, which may be mediated by AMPK activation, reduced
   proinflammatory cytokine synthesis, and increased collagen synthesis.^[
   [206]^25a ^] Chen et al. synthesized a biomembrane that regulates
   mitochondrial dynamics in dysfunctional MSCs via the AMPK pathway,
   reducing ROS levels and promoting bone integration at the implant
   interface.^[ [207]^51 ^] Additionally, Hu et al. demonstrated that APN,
   a fat‐derived adipokine, reverses mitochondrial dysfunction and
   improves titanium implant integration under diabetic conditions via the
   AMPK pathway.^[ [208]^52 ^] APN possesses strong antioxidant,
   mitochondrial protective, and anti‐diabetic effects. AMPK
   phosphorylation is a key mechanism of APN activity, with most studies
   focusing on the cardiovascular and endocrine systems.^[ [209]^53 ^] APN
   binds to its receptors (AdipoR1 and AdipoR2) to initiate intracellular
   signaling. Interestingly, sequencing revealed significant APN
   upregulation in HMs/MBs/ADSCs, and ELISA detected high APN
   concentrations (Figure [210]8H). Therefore, in the 3D porous hydrogel
   system, ADSCs primarily affect local tissues via paracrine secretion,
   whereas BMP‐2 enhances their paracrine activity, consistent with
   previous findings.^[ [211]^19 ^] Moreover, BMP‐2 amplified ADSC
   paracrine activity and significantly increased phosphorylated AMPK
   expression. Sequencing results also revealed the involvement of fatty
   acid metabolism, a critical pathway for cellular energy production. In
   tissue injury microenvironments, β‐oxidation generates substantial ATP
   to meet the high energy demands of repair and proliferation.^[ [212]^54
   ^] The differentiation potential and paracrine effects of exogenous
   ADSCs may indirectly influence fatty acid synthesis.^[ [213]^55 ^]
   Their paracrine effects involve growth factors and cytokines that
   regulate fatty acid synthesis gene expression. This study demonstrated
   that ADSCs activate the AMPK pathway via APN secretion, further
   influencing fatty acid metabolism.^[ [214]^56 ^] These mechanisms
   collectively promote tissue regeneration and functional recovery.

   In summary, this study successfully achieved multistage synergistic
   regulation of TBI repair by constructing a 3D porous HAMA hydrogel
   system loaded with BMP‐2. At the material design level
   (Figure [215]1A), thermosensitive GMs degraded at body temperature,
   forming a microporous network that provided space for ADSC
   proliferation. Simultaneously, the controlled release properties of
   GMMs ensured sustained BMP‐2 release and guided cell migration, leading
   to the gradual aggregation of ADSCs around the pores and GMMs. The in
   vivo repair process (Figure [216]1B,C) demonstrated a dual protective
   mechanism. During the early oxidative stress phase, BMP‐2 stimulated
   ADSCs to activate the AMPK pathway via APN paracrine secretion,
   clearing ROS and restoring mitochondrial function in BMSCs, thereby
   preserving their biological activity. In the later repair phase, BMP‐2
   not only promoted osteogenic differentiation but also synergized with
   TGF‐β3 secreted by ADSCs to activate the TGF‐β/Smad2/3 pathway,
   inducing BMSC differentiation into chondrocytes and fibrocartilage
   formation, as well as promoting tendon regeneration by TDSCs. This
   process ultimately reconstructed the three‐layer structure of bone,
   cartilage, and tendon. By integrating material design with biological
   regulation, this composite hydrogel system provides an innovative
   therapeutic strategy for TBI repair.

   Our current study has certain limitations. First, the acute rotator
   cuff injury model used in this research differs from the more
   clinically common chronic injury model. Additionally, the hydrogel
   properties were designed primarily for an open surgical environment,
   while arthroscopic surgery is more prevalent in clinical practice.
   Future studies should optimize hydrogel characteristics for the
   specific conditions of arthroscopic surgery. Second, although the ideal
   source of ADSCs should be autologous transplantation, allogeneic ADSCs
   were used in this study due to considerations of resource efficiency
   and the relatively mild immune response in rats. The biocompatibility
   of these cells was confirmed through in vivo biosafety and histological
   evaluations. Finally, since in vitro models cannot fully replicate the
   in vivo environment, a dual‐sequencing approach (in vivo and in vitro)
   was adopted to explore the relevant mechanisms. While these mechanisms
   were validated both in vivo and in vitro, further verification through
   knockout of key genes in the related pathways could enhance the
   robustness of the conclusions.

4. Conclusion

   This study developed an injectable porous hydrogel‐based cell‐loaded
   construct (HMs/MBs/ADSCs). GMMs encapsulated BMP‐2 using a unique
   method to achieve sustained release. Simultaneously, a porous HAMA
   hydrogel system was constructed using GMs, providing a 3D bioactive
   environment for ADSCs and promoting ADSCs proliferation and paracrine
   effects. This construct significantly enhanced TBI healing and
   fibrocartilage transition zone reconstruction in a rat rotator cuff
   injury model. Both in vivo and in vitro analyses demonstrated the
   critical roles of the AMPK and TGF‐β/Smad2/3 signaling pathways in TBI
   repair. Mechanistic studies showed that HMs/MBs/ADSCs leveraged
   paracrine effects to clear ROS and protect mitochondria via the AMPK
   pathway in the early stages, supplying energy for tissue repair. In the
   later stages, the TGF‐β/Smad2/3 pathway was activated to promote
   fibrocartilage and tendon regeneration. By integrating material design
   with mechanistic exploration, this study not only enhances the
   application potential of tissue‐engineered hydrogels but also provides
   new insights for clinical treatment.

Conflict of Interest

   The authors declare no conflict of interest.

Supporting information

   Supporting Information
   [217]ADVS-12-e15923-s001.docx^ (108.9MB, docx)

Acknowledgements