Chondrogenic niche hydrogel microspheres facilitate cartilage regeneration in osteoarthritis

Abstract

Tissue regeneration is a central frontier in biomedicine, yet articular cartilage defect repair in osteoarthritis (OA) remains a formidable challenge. Although mesenchymal stem cell (MSC)-based therapies show great potential for cartilage regeneration, their clinical translation is hindered by sequential barriers: low cell retention, oxidative stress-induced apoptosis, and inefficient MSC homing to defect sites and subsequent integration. Herein, we fabricate chondrogenic niche hydrogel microspheres (Chonichspheres) via microfluidics. These HMs are composed of gelatin methacryloyl (GelMA)/aldehyde-hyaluronic acid methacrylate (AHAMA) composite matrices loaded with amino fullerenes (AF) and transforming growth factor-β3 (TGF-β3). Chonichspheres exert four synergistic functions: GelMA acts as a structural scaffold to promote MSC adhesion; AF exert sustained antioxidant effects to regulate redox homeostasis in MSCs and OA chondrocytes; AHAMA enables precise targeted homing and tissue integration; and TGF-β3 induces MSC chondrogenic differentiation. Validated by an HM-adapted custom microphysiological system (MPS) and in vivo experiments, Chonichspheres activate the integrin–PI3K–AKT–mTOR axis, protecting MSCs and facilitating chondrogenic differentiation under OA-mimicking dynamic culture conditions. By synergizing active covalent tissue integration with durable, non-sacrificial antioxidant defense, this programmatic platform provides a robust precision regenerative strategy for OA cartilage repair.

Graphical abstract

Highlights

• •Chonichspheres integrate ROS scavenging, MSC adhesion, cartilage anchoring and chondrogenic induction.
• •Chonichspheres protect osteoarthritic chondrocytes from oxidative stress and preserve mitochondrial function.
• •Chonichspheres activate the integrin–PI3K–AKT–mTOR axis to promote MSC adhesion, as well as enhance cell survival and chondrogenic differentiation.
• •This microsphere-microphysiological platform enables precise OA therapy with strong translational potential.

1. Introduction

Tissue regeneration is a core objective in modern biomedicine [1,2]. With robust multipotency and regenerative activity, mesenchymal stem cells (MSCs) form the basis for tackling complex tissue repair [3,4]. Among degenerative disorders, articular cartilage repair in osteoarthritis (OA) poses a typical challenge [5,6]. Due to the absence of blood vessels, nerves, and lymphatics, cartilage exhibits extremely limited self-healing capacity, representing a common bottleneck in MSC-based regeneration [7,8]. Despite promising preclinical outcomes [9], clinical translation of MSC therapy remains hindered by sequential barriers [10], including insufficient cell loading, mechanical fragility during administration, and impaired cell viability under oxidative stress. These issues compromise MSC therapeutic efficacy and lead to inconsistent results, mainly due to inefficient homing and disrupted homeostasis between exogenous MSCs and host chondrocytes. Moreover, conventional models fail to recapitulate the complex OA pathological microenvironment, highlighting the demand for advanced biomimetic platforms. Therefore, constructing a multifunctional system to facilitate cartilage regeneration is urgently required [11,12].

Biological scaffolds provide essential support for cell loading, delivery and functional modulation [[13], [14], [15]]. Hydrogel microspheres (HMs) are promising carriers due to their favorable physicochemical properties [16,17]. Their injectability allows minimally invasive, site-specific delivery for irregular cartilage defects, meeting clinical OA treatment demands [[18], [19], [20]]. The high surface-to-volume ratio improves MSC loading and distribution, supporting sustained regeneration [21,22]. Additionally, the spherical structure and elastic matrix of HMs serve as a physical barrier, regulating local fluid flow and reducing shear stress. This protects MSCs from mechanical damage and anoikis during injection, laying the foundation for safe transplantation [[23], [24], [25]]. However, bare HMs lack bioactive sites for stable MSC adhesion, which is essential for subsequent proliferation and differentiation. To address this, we introduced gelatin methacryloyl (GelMA). As a modified gelatin, GelMA retains RGD peptide sequences that act as adhesion sites [26,27]. These motifs activate integrins to mediate cytoskeletal rearrangement, promoting firm MSC attachment and spreading, thus resolving the cell-loading bottleneck and supporting downstream cellular behaviors [28]. Furthermore, oxidative stress in the OA microenvironment severely impairs MSC survival and therapeutic efficacy [29]. Excessive ROS accumulation disrupts mitochondrial homeostasis, inhibits MSC differentiation, and damages chondrocytes, triggering a vicious cycle of injury, oxidation, cellular dysfunction and repair failure [[29], [30], [31]]. While conventional well-established antioxidants neutralize ROS through sacrificial stoichiometric reactions leading to rapid depletion, AF offers a distinct advantage [32,33]. Functioning as robust “radical sponges”, they utilize their unique carbon cage structure to physically sequester and chemically eliminate ROS through synergistic radical addition and electron transfer cycling, without being chemically consumed. This non-sacrificial property breaks the dose-dependency bottleneck of traditional agents, enabling the durable, long-term antioxidant shielding that is crucial for the prolonged process of cartilage regeneration [[34], [35], [36]]. Cationic AF also target negatively charged OA microenvironments—including cartilage matrix, cell glycocalyx and mitochondrial membranes—via electrostatic interactions, achieving site-specific enrichment and subcellular accumulation.

Effective MSC homing and seamless integration with host cartilage determine long-term regeneration [7,37]. However, the lubricated cartilage surface impedes graft anchorage, and synovial fluid (SF) shear further causes implant detachment [38]. Previous studies have shown that damaged OA cartilage exposes abundant free amino groups, providing specific anchoring sites [39]. Accordingly, we functionalized HMs with aldehyde-hyaluronic acid methacrylate (AHAMA), an extracellular matrix (ECM)-mimetic component with excellent biocompatibility and biodegradability. Under physiological conditions, aldehyde groups on AHAMA rapidly form Schiff base linkages with exposed amino groups on injured cartilage, overcoming its anti-adhesive nature and enabling precise microscale anchoring and in situ retention of HMs [40]. This facilitates targeted docking and robust integration of MSCs within cartilage defects [39,41]. Meanwhile, microsphere-loaded transforming growth factor-β3 (TGF-β3) is released in a sustained and controlled manner to direct chondrogenic differentiation of MSCs [42]. Such biochemical induction promotes functional coupling between exogenous MSCs and host chondrocytes, supporting histological continuity and unobstructed intercellular crosstalk. These synergistic effects collectively ensure the long-term stability and functional durability of regenerated cartilage [43].

Herein, we fabricated a multifunctional biomimetic niche platform termed chondrogenic niche hydrogel microspheres (Chonichspheres) via microfluidic technology. This composite microsphere system consists of a GelMA/AHAMA (GAM) hydrogel matrix, with AF covalently immobilized through dynamic Schiff base linkages and TGF-β3 efficiently encapsulated as a chondrogenic induction factor. The cationic AF serve as an electrostatic bridge to efficiently sequester anionic TGF-β3, enabling high loading efficiency and sustained, controlled release to construct a tunable biomimetic chondrogenic microenvironment. MSCs were further immobilized on the microspheres to establish Chonichspheres@MSCs. Using OA cartilage repair as a model, this system exerts synergistic functions: GelMA supports cell adhesion; AF maintain redox homeostasis; AHAMA facilitates cartilage integration; TGF-β3 directs chondrogenic differentiation. Therapeutic efficacy was validated in an HM-adapted custom microphysiological system (MPS) that recapitulates the OA microenvironment, as well as through comprehensive in vivo assessments (Fig. S1) [44]. We found that the integrin–PI3K axis is markedly inhibited in OA; Chonichspheres reverse this suppression via activating integrin–PI3K–AKT–mTOR signaling, enhancing MSC survival and chondrogenic differentiation under OA-mimicking dynamic culture conditions for functional repair. By coupling targeted ECM anchoring with catalytic radical interception, this programmatic HM platform precisely counteracts the complex pathological barriers of OA. Beyond OA, it shows broad potential for skin, bone and tendon regeneration, illustrating the value of interdisciplinary biomaterial design (Fig. 1).

Fig. 1.

Schematic illustration of the Chonichspheres for remodeling the pathological microenvironment and promoting cartilage regeneration in OA. (A) The fabrication involves the microfluidic generation of GelMA/AHAMA composite microspheres encapsulating AF and TGF-β3. (B) Surface seeding of MSCs constructs the bio-hybrid Chonichspheres@MSCs. The pathological MPS—featuring co-cultured OA chondrocytes/synoviocytes and dynamic fluid flow—is established to recapitulate the inflammatory intra-articular microenvironment. (C) Targeted Integration: Upon injection, the aldehyde groups on AHAMA form Schiff bases with exposed amino groups on the damaged cartilage matrix, ensuring precise homing and robust physical anchoring. (D) Intracellular Mechanistic Synergy: The system overcomes sequential regenerative barriers via two coordinated pathways: (i) Mitochondrial Rescue: Internalized AF acts as a “radical sponge” to scavenge ROS, relieving oxidative suppression to restore mitochondrial membrane potential (ΔΨm) and metabolic homeostasis (ATP production); and (ii) Chondrogenic Induction: The activation of the ECM-mediated integrin–PI3K–AKT–mTOR axis and the TGF-β3/Smad signaling pathway acts synergistically. This coordinated regulation promotes anabolic gene expression (COL2A1, ACAN, SOX9) while suppressing catabolism (MMP13), ultimately realizing functional cartilage regeneration.

2. Results

2.1. Abnormal accumulation of ROS in the pathological microenvironment of human OA

OA is a chronic degenerative disorder primarily characterized by articular cartilage degradation and synovial inflammation [45,46]. The abnormal accumulation of ROS within the OA microenvironment is a pivotal driver of these pathological processes [47]. To establish clinical rationale for our therapeutic design, we first investigated the ROS expression patterns in the OA pathological microenvironment. As illustrated in Fig. S2A, clinical specimens were obtained from patients undergoing total knee arthroplasty (TKA) or arthroscopic surgery. Specifically, eroded cartilage, synovial tissue, and SF from TKA patients were utilized as the OA group, while adjacent non-eroded cartilage served as the Normal group. Additionally, non-OA synovium and SF collected during arthroscopic procedures were utilized as the corresponding Normal group (Table S1). Initially, primary cells were successfully isolated and identified: primary chondrocytes derived from the adjacent unworn (Nor-Chs) and worn (OA-Chs) articular cartilage of TKA patients maintained their characteristic phenotype, as visualized via representative brightfield microscopy (Fig. S2B). Meanwhile, immunofluorescence staining confirmed that primary synovial fibroblasts from both non-OA (Nor-FLSs) and OA (OA-FLSs) tissues exhibited high expression of Vimentin and F-actin, verifying their identity as fibroblast-like synoviocytes (Fig. S2C).

Subsequently, intracellular ROS levels were quantified across multiple dimensions. Confocal laser scanning microscopy (CLSM) revealed a robust green fluorescence burst (DCFH-DA probe) within OA-Chs compared to healthy controls (Nor-Chs). Quantitative analysis indicated an approximately 2.7-fold increase in intracellular ROS levels (Fig. S2D and E). A similar trend was observed in the synovial compartment: OA-FLSs displayed elevated intracellular ROS levels consistent with chondrocytes, with a 3.0-fold increase compared to controls, suggesting aberrant ROS metabolism in OA synoviocytes (Fig. S2F and G). Flow cytometry further corroborated these findings, showing a significant rightward shift in the fluorescence peak for OA-Chs and OA-FLSs, indicating a substantial pathological accumulation of ROS (Fig. S2H and I). To obtain direct evidence of extracellular free radicals, we employed electron paramagnetic resonance (EPR) spectroscopy to analyze the SF. The results demonstrated characteristic high-amplitude signals in OA patient synovial fluid (OA-SF), directly confirming a high concentration of free radicals within the pathological joint space (Fig. S2J).

To further delineate the heterogeneous landscape of oxidative stress underlying OA pathogenesis at a single-cell resolution, we retrieved publicly available single-cell RNA sequencing (scRNA-seq) datasets of cartilage and synovial tissues from both healthy individuals and OA patients via the Gene Expression Omnibus (GEO) database. Following stringent quality control procedures to exclude low-quality cells, we curated a comprehensive single-cell transcriptomic atlas comprising 65,546 chondrocytes and 30,072 synovial cells. Subsequent dimensionality reduction and unsupervised clustering of the synovial microenvironment resolved distinct cellular subpopulations, encompassing FLSs, fibroblasts, endothelial cells, macrophages, mural cells, T cells, and B cells.

Uniform Manifold Approximation and Projection (UMAP) analysis revealed marked transcriptomic shifts in the cellular architecture between normal and OA cartilage (Fig. S2K). Targeted module scoring of the ROS signaling pathway demonstrated a highly significant enrichment in OA chondrocytes compared to healthy controls (P < 0.001), directly corroborating our in vitro observations of intrinsic oxidative stress at the transcriptional level (Fig. S2L). Moreover, high-dimensional mapping of the synovial microenvironment identified diverse cellular subsets (Fig. S2M). Notably, fibroblast-like synoviocytes (FLSs) and fibroblasts, which constitute the core structural framework of the synovium, were heavily scrutinized given their central role in driving OA pathogenesis. Crucially, evaluating ROS pathway activity across these distinct clusters unveiled a ubiquitous oxidative burden within the OA joint. Consistent with our in vitro CLSM and flow cytometry findings, these core structural cells (FLSs and fibroblasts) exhibited significantly elevated ROS pathway scores in the OA state, alongside immune populations such as macrophages and T cells (Fig. S2N).

The pronounced ROS accumulation in FLSs at the single-cell transcriptomic level firmly validates their aberrant metabolic state. Collectively, multi-dimensional evidence from clinical specimens and single-cell transcriptomics confirms that pathological ROS accumulation in both chondrocytes and the core synovial structural cells is a hallmark of the OA microenvironment. This excessive ROS disrupts redox homeostasis and induces persistent oxidative stress, fueling a vicious cycle of “damage-inflammation-repair failure”. Consequently, the development of engineered microspheres capable of scavenging excessive ROS is a critical prerequisite for achieving cartilage regeneration.

2.2. Characterization of AF, AHAMA, GAM@AF, and Chonichspheres

To lay a solid material foundation for constructing the regenerative niche, we first verified the successful synthesis and functionalization of all core components. We prepared AF-loaded GelMA/AHAMA composite HMs (named GAM@AF) via microfluidic technology using AF, GelMA and AHAMA. Finally, Chonichspheres were constructed by incubation with TGF-β3 (Fig. 2A). Lyophilized AF appeared as a tan-brown powder (Fig. S3A), and its aqueous solutions demonstrated exceptional stability and uniform dispersibility across various biological fluids (Fig. S3B). Crucially, the amino modification not only enhances water solubility but also imparts a positive surface charge to the fullerenes. Transmission electron microscopy (TEM) revealed that AF exists as nanoparticles (Fig. 2Ba, b) with a mean diameter of 52.84 ± 19.55 nm (Fig. 2C) and a zeta potential of 28.89 ± 12.32 mV (Fig. 2D). The positive surface charge of these nanoparticles facilitates binding with negatively charged glycosaminoglycan chains in cartilage, enhancing cartilage targeting [8]. Subsequent Energy-Dispersive X-ray Spectroscopy (EDS) confirmed a homogeneous distribution of carbon and nitrogen within the nanoparticles (Fig. 2Bc-f), with a carbon mass fraction of 97.70 wt% and nitrogen of 2.30 wt% (Fig. 2E), further validating the successful synthesis of AF. Having confirmed its physicochemical properties, we next evaluated the cellular uptake and intracellular distribution profile of AF in MSCs to elucidate its protective mechanism. Confocal microscopy revealed that FITC-conjugated AF was efficiently internalized by the cells, whereas the untreated control group exhibited negligible green background fluorescence, eliminating the interference of cellular autofluorescence and channel crosstalk. Furthermore, corresponding quantitative fluorescence analysis statistically supported these robust cellular uptake results. More importantly, the intracellular fluorescence signal of FITC-AF exhibited significant colocalization with the mitochondria-specific probe (Fig. S4). This direct morphological evidence indicates that internalized AF preferentially accumulates within the mitochondrial compartment, providing robust experimental support for the mechanism by which AF exert their localized antioxidant effects to protect mitochondrial structure and function. Simultaneously, dual-functionalized AHAMA was successfully synthesized to provide targeted anchoring sites, which was verified by ¹H NMR spectroscopy in D₂O (Fig. S5). The AHA spectrum (green) exhibited a characteristic aldehyde proton peak at δ = 5.1 ppm. The AHAMA spectrum (purple) retained this aldehyde signal while displaying new double-bond proton peaks at δ = 5.6 and 6.0 ppm, confirming the successful dual-functionalization.

Fig. 2.

Synthesis, structural characterization, and drug loading of Chonichspheres. (A) Schematic illustration of the microfluidic fabrication of GelMA/AHAMA@AF (GAM@AF) microspheres, followed by photocrosslinking and TGF-β3 loading. (B) Characterization of AF nanoparticles: (a, b) TEM images and (c–f) STEM-EDS mapping displaying the distribution of Carbon (C, red) and Nitrogen (N, green), with the merged image (f) demonstrating elemental co-localization. (C) Hydration particle size distribution and (D) Zeta potential analysis of AF nanoparticles. (E) Quantitative elemental analysis (wt.%) based on the EDS spectrum. (F) Optical microscopy images displaying: (a) the microfluidic droplet generation process, (b) monodispersed GAM@AF microspheres, and (c) the morphology of a single microsphere. (G) Size distribution histogram of GAM@AF microspheres with Gaussian fitting. (H) FTIR spectra of AHAMA, GelMA, and the composite GAM hydrogel. (I, J) Macroscopic photographs (a) and SEM images (b, c) of (I) blank GAM microspheres (white) and (J) GAM@AF microspheres (brownish-yellow), highlighting the porous microstructure and nanoparticle integration. (K) SEM morphology (a) and corresponding EDS elemental mapping (b–d) of a single GAM@AF microsphere, displaying the uniform distribution of Carbon (C, red), Nitrogen (N, yellow), and Oxygen (O, green). (L) Representative confocal fluorescence images of a single Chonichsphere, displaying serial optical sections (a–f) and the corresponding Z-axis maximum intensity projection (MIP) (g), confirming the uniform co-loading of TGF-β3 (red) and AF (green). Scale bars: 100 nm (B–a), 50 nm (B-b, f), 50 μm (F, I-b, J-b, K, L), 2 μm (I-c, J-c). Data in c, d, and g are presented as mean ± SD (n = 3 independent experiments).

Following the establishment of these functional components, a microfluidic fabrication workflow was employed to construct highly uniform composite HMs, which we refer to as Chonichspheres (Fig. S6A–C, Video. S1). The resulting GAM@AF HMs exhibited a highly regular spherical morphology (Fig. 2Fa) and uniform dispersion (Fig. 2Fb). High-magnification imaging revealed a smooth surface and dense internal structure (Fig. 2Fc). Size distribution analysis showed a primary peak near 220 μm, with a mean diameter of 220.6 ± 9.71 μm and a narrow Polydispersity Index (PDI <0.05), confirming excellent size uniformity (Fig. 2G). Fourier-transform infrared (FTIR) spectroscopy demonstrated the simultaneous presence of carbonyl peaks from AHAMA (1725 cm⁻¹, acryloyloxy C=O) and amide peaks from GelMA (1650 cm⁻¹, C=O/N-H), confirming the successful formation of the composite matrix (Fig. 2H). Further morphological analysis via scanning electron microscopy (SEM) of lyophilized HMs (Fig. 2Ia, Ja) revealed a robust three-dimensional porous network in both GAM and GAM@AF groups (Fig. 2Ib, c and Fig. 2Jb, c). Notably, high-magnification SEM showed the successful entrapment of AF nanoparticles within the GAM@AF network (Fig. 2Jc), with EDS mapping confirming uniform AF distribution (Fig. 2K). Confocal microscopy of FITC-labeled AF within the HMs further corroborated the high homogeneity (Fig. S7A). The loading efficiency of AF was calculated to be 82.67 ± 6.51% (Fig. S7B).

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.bioactmat.2026.07.024

The following are the Supplementary data related to this article.

Based on previous literature, confocal microscopy of FITC-conjugated GelMA HMs confirmed a homogeneous distribution of the polymer network, which inherently contains abundant RGD cell-adhesion sequences (Fig. S8A) [48]. To further evaluate their microstructural morphology, SEM was performed on neat AHAMA HMs, neat GelMA HMs, and AHAMA/GelMA composite HMs. These morphological observations confirmed the successful physical integration of AHAMA and GelMA, resulting in the fabrication of structurally stable composite carriers (Fig. S8B). Finally, we evaluated the therapeutic payload capacity and sustained release profile of the ultimate Chonichspheres. Confocal imaging confirmed that AF and TGF-β3 were successfully co-incorporated into the GelMA/AHAMA (1:2 ratio) matrix, demonstrating excellent dual-drug and cell loading performance (Fig. 2L and Fig. S9). In addition, quantitative ELISA analysis revealed a TGF-β3 loading efficiency of 91.17 ± 4.14%, which is primarily attributed to the electrostatic attraction between the positively charged AF and the growth factor (Fig. S10).

In vitro release assays demonstrated a biphasic release pattern of TGF-β3 from Chonichspheres, characterized by a rapid initial burst within the first three days followed by sustained slow release, which was nearly completed by day 28. In contrast, Schiff base interactions between amino groups of AF and aldehyde groups of AHAMA restricted AF release, with only 55.9% of AF liberated after 28 days (Fig. S11A). Zeta potential measurement was performed to clarify the loading mechanism of TGF-β3. Consistent with its theoretical isoelectric point (pI = 6.5−7.1), TGF-β3 exhibited a negative surface charge of −8.98 ± 1.45 mV under physiological conditions (pH = 7.4) (Fig. S11B). By comparison, AF nanoparticles displayed a strongly positive surface charge (28.89 ± 12.32 mV). The distinct electrostatic potential difference confirms that electrostatic attraction dominates the high loading efficiency and sustained release behavior of TGF-β3 within Chonichspheres. According to previous studies, a TGF-β3 loading concentration of 300 ng/mL was adopted as the optimal dosage for initiating stable chondrogenic differentiation of MSCs [22]. A SMAD luciferase reporter assay was further conducted to verify the bioactivity of continuously released TGF-β3. The results showed that the released TGF-β3 effectively and persistently activated SMAD signaling over a 14-day culture period, and the activation level increased steadily, which was consistent with the cumulative release profile (Fig. S11C). These results indicate that the Chonichsphere system enables controllable factor release based on electrostatic interactions and maintains the biological activity of loaded TGF-β3 for chondrogenic induction. Consequently, this system demonstrates strong potential as a reliable, long-term local delivery platform for therapeutic factors. The HMs exhibited favorable injectability, as validated by macroscopic injection operation (Video. S2). Similar to typical granular hydrogels, Chonichspheres displayed prominent shear-thinning performance under cyclic step-strain loading. Upon removal of high shear deformation, the storage modulus rapidly recovered to its original plateau, demonstrating fully reversible inter-particle network reconstruction without permanent structural collapse (Fig. S11D). Uniaxial mechanical compression tests based on the linear elastic region (0–10% strain) of stress–strain curves quantified the compressive modulus of this injectable Chonichsphere-based cartilage-repair hydrogel as 10.96 kPa (Fig. S11E), a value well matched to the stiffness range of native cartilage pericellular matrix. In vitro enzymatic degradation evaluation in an OA-mimicking pathological microenvironment was performed to verify the structural durability of Chonichspheres. Morphological observation and quantitative weight loss analysis over 35 days of degradation demonstrated that the HMs maintained intact structural integrity for more than four weeks (Fig. S11F–G), providing sufficient mechanical support during the early stage of chondrogenesis and subsequent neocartilage ECM deposition [49].

2.3. Biocompatibility and in vivo retention of Chonichspheres@MSCs in mice

2.3.1. Biocompatibility of Chonichspheres@MSCs

The fabrication workflow for Chonichspheres@MSCs composite microspheres is outlined in Fig. 3A. To comprehensively evaluate the biocompatibility of the base matrix and rule out any potential cytotoxicity induced by growth factor loading, we first assessed the viability of MSCs encapsulated within the base hydrogel (GAM@MSCs) and the TGF-β3-loaded hydrogel (GAM@MSCs@TGF-β3). Live/Dead staining over a 14-day culture period revealed that MSCs in both groups maintained exceptionally high viability, with nearly all cells exhibiting green fluorescence and negligible dead cells (Fig. S12A and B). Quantitative analysis of the relative fluorescence intensity demonstrated a significant and continuous increase over time in both groups, indicating robust cell proliferation within the 3D network (Fig. S12C and D). These results firmly establish that the GAM HMs serve as an excellent biomimetic ECM that supports cell survival and proliferation, and the incorporation of TGF-β3 does not impair cellular viability. Following this baseline validation, to evaluate the microspheres as cytocompatible carriers, we assessed the attachment and colonization of MSCs. Live/Dead staining demonstrated exceptional viability throughout the 14-day culture period (Fig. 3B–D). Chonichspheres actively promoted proliferation, characterized by a progressive increase in cell density. Because cell spreading regulates paracrine function and differentiation, we visualized cytoskeletal organization (Phalloidin/DAPI). By Day 1, MSCs displayed extensive focal adhesions and a well-spread morphology; by Day 14, they formed a confluent monolayer enveloping the microspheres (Fig. 3C–E). SEM confirmed this robust anchorage, showing pseudopodial extension into the biomimetic GelMA network (Fig. 3F and G). High oxidative stress characterizes the OA microenvironment [29]. We therefore assessed MSC adhesion under simulated pathological conditions. Chonichspheres composites rescued H₂O₂-induced adhesion dysfunction, promoting uniform cytoskeletal spreading and enhancing adhesion efficiency. This performance mimics the mechanical support of the native ECM. Notably, this stable anchorage persists even under severe oxidative stress (Fig. S13), providing a niche essential for preventing anoikis.

Fig. 3.

Biocompatibility and long-term intra-articular retention of Chonichspheres@MSCs. (A) Schematic illustration of the co-culture protocol for fabricating MSC-loaded Chonichspheres@MSCs. (B) Representative three-dimensional Live/Dead staining images of MSCs cultured on microspheres for 1, 7, and 14 days (live cells: green; dead cells: red). (C) Cytoskeletal morphology visualized by Phalloidin (red) and DAPI (blue) staining, demonstrating progressive cell spreading. (D, E) Quantitative assessment of cell viability based on fluorescence intensity (D) and the temporal analysis of relative cytoskeletal density (E). (F, G) SEM characterization of cell–matrix interactions. (F) Quantification of the cell coverage area on the microsphere surface. (G) Pseudo-colored SEM images showing robust adhesion and colonization of MSCs (colored red) on the Chonichspheres surface (colored cyan). (H) Representative bioluminescence images of mice receiving free luciferase-labeled MSCs or Chonichspheres@luciferase-labeled MSCs at Days 0, 7, 14, and 21 post-injection. The color scale bar indicates luminescence counts. (I) Quantitative analysis of relative bioluminescence intensity of luciferase-labeled MSCs in vivo, normalized to Day 0. The signal of free MSCs rapidly decreased over time, while Chonichspheres@luciferase-labeled MSCs exhibited significantly prolonged retention at the injection site. Scale bars: 50 μm (B, C); 50 μm and 10 μm (G). Each data point indicates the mean fluorescence intensity of one independent biological replicate, and a total of three independent biological replicates were performed. At least five microspheres were analyzed in each group per replicate (n = 3). Data are presented as mean ± SD. ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively, compared to the indicated groups; ns indicates no significant difference.

2.3.2. In vitro biocompatibility evaluation of Chonichspheres toward chondrocytes

To comprehensively evaluate the biosafety of Chonichspheres in cartilage repair, the human C28/I2 chondrocyte cell line was utilized in this study to conduct in vitro biocompatibility assays via Live/Dead double staining and Cell Counting Kit-8 (CCK-8) tests. The experiment included three groups: the GAM group, the GAM@TGF group, and the Chonichspheres group, with samples characterized on days 1, 3, and 5 of co-culture (Fig. S14). The Live/Dead staining results (Fig. S14 A) revealed that throughout the continuous 5-day culture period, cells in all three groups predominantly exhibited green fluorescence (indicative of live cells), with negligible red fluorescence (indicative of dead cells) observed. With prolonged culture time, the number of cells in each group steadily increased, and the cells displayed a well-spread morphology. Notably, no time-dependent apoptosis was observed in the Chonichspheres group, indicating the absence of time-dependent cytotoxicity. Quantitative analysis of live cells (Fig. S14 B) and CCK-8 proliferation curves (Fig. S14C) further confirmed that at identical time points, there were no statistically significant differences in the viable cell count or the absorbance values (at 450 nm) between the Chonichspheres group and the GAM or GAM@TGF groups (P > 0.05). Furthermore, the C28/I2 cells in all groups exhibited a normal exponential proliferation trend over the culture duration. These results fully demonstrate that Chonichspheres possess excellent biocompatibility and do not inhibit the physiological growth and proliferation of the C28/I2 chondrocyte cell line.

2.3.3. Chonichspheres@MSCs prolong the retention of MSCs within the joint cavity in DMM mice

The rapid clearance of free MSCs from the intra-articular space represents a critical bottleneck limiting the efficacy of current OA therapies. To evaluate whether Chonichspheres carriers improve engraftment, we tracked the retention of luciferase-labeled MSCs following intra-articular injection. In vivo bioluminescence imaging (IVIS) showed comparable signal intensities between groups at Day 0, but divergent retention starting from Day 7. Bioluminescence signals from free MSCs decayed rapidly, becoming almost negligible by Day 21. Conversely, Chonichspheres@MSCs sustained significantly higher relative bioluminescence through Day 21 (Fig. 3H and I), suggesting that the Chonichspheres anchor MSCs against SF turnover. Histology of major organs (heart, liver, spleen, lung, kidney) showed no abnormalities, confirming systemic biosafety (Fig. S15). Collectively, these data demonstrate that Chonichspheres not only support MSC viability in vitro but also significantly extend their intra-articular therapeutic window, establishing them as an ideal delivery vehicle.

2.4. Design and biomechanical validation of the biomimetic OA MPS

To evaluate the cytoprotection and regenerative potential of microcarriers under physiologically relevant conditions, we engineered a MPS that integrates fluid dynamics with intercellular communication [50,51]. The device comprises tandem upstream (cartilage/synovium) and downstream (MSC) chambers, mimicking dynamic SF circulation [52,53]. Experimental groups included a healthy control and three OA-pathological conditions (Blank, GAM@MSCs@TGF-β3, Chonichspheres@MSCs). To replicate the inflammatory and oxidative OA microenvironment, pathological groups incorporated OA-derived cells and 10% (v/v) osteoarthritic SF. To quantitatively evaluate the biomechanical microenvironment within the MPS, three-dimensional computational fluid dynamics (CFD) simulations were performed using COMSOL Multiphysics. Discretization of the computational domain was achieved using an unstructured tetrahedral mesh, with localized refinement at the fluid-solid interfaces of the Chonichspheres array to ensure computational accuracy (Fig. S16A and B). Under a constant experimental perfusion rate of 10 μL/min, the COMSOL-derived cross-sectional velocity field exhibited a stable laminar flow regime with a maximum velocity of 1.14 × 10⁻³ m/s (Fig. S16C). Furthermore, the simulation mapped the wall shear stress (WSS) exerted on the surfaces of the Chonichspheres. The maximum surface shear stress was calculated to be 3.87 × 10⁻⁴ Pa (Fig. S16D). This highly biomimetic and low-shear microenvironment establishes a robust and reliable platform for subsequent biological assessments [54,55].

2.5. Chonichspheres scavenge ROS and remodel mitochondrial homeostasis in MSCs

In OA, disrupted redox homeostasis transforms ROS from physiological signals into cytotoxic agents [56]. This surplus overwhelms endogenous defenses (e.g., superoxide dismutase, glutathione peroxidase) [57], inducing mitochondrial dysfunction and impairing the viability of engrafted MSCs [58]. Consequently, equipping microcarriers with efficient ROS-scavenging capabilities is essential for effective therapy [59]. By overcoming the static limitations of conventional culture, this MPS enabled precise assessment of Chonichspheres-mediated mitochondrial remodeling (Fig. 4A). Using this platform, we visualized intracellular ROS in MSCs via DCFH-DA staining and confocal microscopy. The Chonichspheres@MSCs group displayed significantly lower fluorescence intensity than the Blank group, comparable to healthy Controls (Fig. 4B). This reduction indicates that sustained AF release effectively scavenges intracellular ROS (Fig. 4C). Elevated ROS levels within the microenvironment are known to precipitate oxidative phosphorylation dysfunction and mitochondrial impairment, leading to compromised ATP synthesis and suppressed cellular energy metabolism [60,61]. We next quantified ATP production. The Chonichspheres system restored ATP generation, with the Chonichspheres@MSCs group showing superior recovery compared to Blanks (Fig. 4D). These metabolic data align with the ROS findings, suggesting AF preserves mitochondrial respiration and energy metabolism under high oxidative stress.

Fig. 4.

Restoration of mitochondrial homeostasis and ROS scavenging in MSCs within a biomimetic OA MPS. (A) Schematic illustration of the dynamic MPS designed to mimic the upstream OA pathological microenvironment (co-culture of OA chondrocytes and synoviocytes under fluid flow) and evaluate the cytoprotective effects of composite microspheres on downstream MSCs. (B, C) Representative confocal fluorescence images (B) and quantitative analysis (C) of intracellular ROS levels in MSCs indicated by DCFH-DA staining (green). Nuclei were counterstained with Hoechst 33342 (blue). (D) Quantitative analysis of intracellular ATP production, indicating the recovery of cellular bioenergetic metabolism. (E, F) Assessment of mitochondrial membrane potential (ΔΨm) using JC-1 staining. (E) Representative images showing JC-1 aggregates (red, indicating healthy high potential) and monomers (green, indicating depolarized low potential). (F) Quantitative analysis of the red/green fluorescence ratio. (G, H) Representative images (G) and quantitative analysis (H) of mitochondrial-specific superoxide (mtROS) levels labeled with the MitoSOX Red probe. Each data point indicates the mean fluorescence intensity of one independent biological replicate, and a total of three independent biological replicates were performed. At least five microspheres were analyzed in each group per replicate. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test. ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively, compared to the indicated groups; ns indicates no significant difference. Scale bars, 50 μm.

As established previously, Chonichspheres efficiently scavenge ROS and support oxidative phosphorylation, we next investigated mitochondrial protection. Since mitochondria are the primary intracellular source of ROS and are vulnerable to OA-associated oxidative stress, preserving their homeostasis is critical [62]. We therefore utilized the MPS to assess mitochondrial integrity across four experimental cohorts: Control, Blank, GAM@MSCs@TGF-β3, Chonichspheres@MSCs. To elucidate the protective mechanism, we assessed mitochondrial homeostasis. Mitochondrial membrane potential (ΔΨm) was visualized using JC-1 staining. Controls exhibited healthy ΔΨm, characterized by intense red fluorescence (J-aggregates) and a high red/green ratio (Fig. 4E). Conversely, the Blank group displayed a precipitous drop in ΔΨm, with the red/green ratio plummeting to approximately 16% of the Control levels. This collapse in mitochondrial function confirms that the MPS successfully recapitulated the oxidative stress of the OA microenvironment, thereby inducing substantial intracellular ROS accumulation and subsequent mitochondrial dysfunction. The GAM@MSCs@TGF-β3 group showed a partial recovery, maintaining a red/green ratio at approximately 46% of the Control. This mitigation may be attributed to the inherent antioxidant properties or paracrine effects of TGF-β3 under such conditions, although further investigation is warranted to fully elucidate this mechanism [22]. Notably, the Chonichspheres@MSCs system restored ΔΨm approximately to 78% of Control levels, indicating effective mitochondrial repair (Fig. 4F). We also quantified mitochondrial superoxide (mtROS) using MitoSOX Red. While TGF-β3 provided partial reduction, AF-loaded microspheres demonstrated superior scavenging, disrupting the cycle of “ROS accumulation–mitochondrial damage” (Fig. 4G and H).

Collectively, the dynamic circulatory simulations enabled by the MPS substantiate the role of AF as a potent “free radical sponge”, capable of the cyclical and high-efficiency scavenging of excessive ROS within the pathological milieu. Crucially, the Chonichspheres@MSCs system not only ameliorates the oxidative state but, by preserving mitochondrial homeostasis, facilitates a permissive microenvironment essential for the subsequent chondrogenic differentiation of MSCs.

2.6. Targeted integration and mitochondrial homeostasis remodeling by Chonichspheres@MSCs for cartilage repair

To evaluate the adhesive capacity of GAM microspheres to OA cartilage, ex vivo experiments were performed using polished porcine knee cartilage. GelMA (green), AHAMA (red), and GAM (blue) microspheres were utilized for visualization. After rinsing with PBS, most of the green GelMA microspheres were detached from the cartilage surface, indicating insufficient binding energy in a purely charge-dependent interaction mode. In contrast, the red AHAMA and blue GAM microspheres remained firmly anchored to the cartilage surface. This enhanced tissue-anchorage capability, likely mediated by the formation of covalent Schiff base bonds between aldehyde groups on the microspheres and amines in the cartilage ECM, serves as a foundation for the Chonichspheres system [39]. By integrating robust interfacial binding with potent antioxidant and chondro-inductive functionalities, the Chonichspheres system ensures precise and sustained retention within the OA niche, establishing a stable therapeutic interface that maximizes the effects of ROS scavenging and TGF-β3-mediated repair (Fig. S17).

Having established this robust tissue-material interface, we further investigated whether the previously demonstrated cytoprotective effects of GAM@AF HMs on loaded MSCs could extend to the surrounding host tissue via paracrine signaling. Since OA-associated ROS drives mitochondrial dysfunction in chondrocytes [63], we assessed paracrine protection using a Transwell co-culture model (Fig. S18A). MSC-laden HMs (apical chamber) were co-cultured with chondrocytes challenged by H₂O₂ (basolateral chamber). Groups were set as follows: Control (GAM@MSCs), Blank (GAM@MSCs + H₂O₂), GAM@MSCs@TGF-β3 (GAM@MSCs@TGF-β3 + H₂O₂), and Chonichspheres@MSCs (Chonichspheres@MSCs + H₂O₂). Super-resolution imaging (SIM) of MitoTracker-stained chondrocytes revealed that while Controls maintained elongated networks, the Blank group exhibited fragmented, punctate morphology indicative of severe dysfunction (Fig. S18B). Quantitatively, mitochondrial branch length in the Blank group dropped to 0.16-fold of Controls. The GAM@MSCs@TGF-β3 group showed intermediate recovery (mixed spherical/short rods; 0.40-fold). Notably, Chonichspheres@MSCs restored a healthy, filamentous phenotype, recovering branch length to 0.74-fold of Controls (Fig. S18C). These data confirm the system's capacity to reverse oxidative mitochondrial fragmentation.

Beyond the morphological restoration of the mitochondrial network, to quantify functional recovery, we monitored membrane potential (ΔΨm) using JC-1 staining. Controls maintained physiological ΔΨm, indicated by a high red/green fluorescence ratio (Fig. S18D). In contrast, the Blank group exhibited severe depolarization, with the ratio dropping to 0.12-fold of Controls. The GAM@MSCs@TGF-β3 group showed partial recovery (0.41-fold), likely due to enhanced MSC paracrine signaling protecting basolateral chondrocytes [64]. Notably, Chonichspheres@MSCs restored the ratio to 0.71-fold. These consistent functional and morphological results confirm that AF-loaded microspheres rescue compromised host chondrocytes via microenvironmental remodeling, achieving “bidirectional synergistic repair” (Fig. S18E).

2.7. Chonichspheres promote chondrogenic differentiation of MSCs in a mimetic OA microenvironment

In the OA microenvironment, elevated ROS levels drive mitochondrial dysfunction [65]. This oxidative stress impairs the proliferation and differentiation of MSCs, exacerbating cartilage degeneration [66]. We demonstrate that Chonichspheres efficiently scavenge ROS, protecting adherent MSCs. Notably, this therapeutic benefit extends to the host tissue: by restoring mitochondrial membrane potential in OA chondrocytes, the system reverses oxidative dysfunction and promotes tissue repair.

Building upon these findings, using the MPS to simulate the OA microenvironment, we evaluated the rescue of MSC chondrogenic potential (Fig. 5A). MSC-laden microspheres were cultured under dynamic circulation for 14 days, and differentiation was assessed via cartilage-specific markers SOX9 and COL2A1. 3D immunofluorescence showed markedly reduced fluorescence in the OA-mimicking Blank group compared to healthy Controls (Fig. 5B–E). The suppression of SOX9 and COL2A1 confirms that high ROS levels inhibit MSC lineage commitment. While GAM@MSCs@TGF-β3 microspheres showed partial recovery, the Chonichspheres group significantly enhanced fluorescence intensity, indicating robust matrix synthesis. This observation underscores that the incorporation of AF effectively scavenges excessive ROS from the niche, thereby establishing a “safe,” low-oxidative-stress microenvironment essential for maximizing the inductive potency of TGF-β3 [67].

Fig. 5.

Chonichspheres facilitate chondrogenic differentiation of MSCs under simulated OA conditions. (A) Schematic illustration of the experimental workflow utilizing the microfluidic MPS-based 3D co-culture system to assess chondrogenic differentiation. (B, C) Representative 3D immunofluorescence images and merged views showing the expression of (B) Collagen Type II (COL2A1) and (C) SOX9 proteins. (D, E) Quantitative analysis of the relative fluorescence intensity for (D) COL2A1 and (E) SOX9. (F–I) RT-qPCR analysis of the relative mRNA expression levels of the chondrogenic markers (F)COL2A1, (G)SOX9, and (H)ACAN, and the catabolic marker (I)MMP13. (J) Evaluation of the paracrine protective effects of Chonichspheres on endogenous chondrocytes under H₂O₂-induced oxidative stress. Representative Toluidine Blue and Safranin O staining images displaying the ECM deposition of chondrocytes across different groups in the Transwell system. Each data point indicates the mean fluorescence intensity of one independent biological replicate, and a total of three independent biological replicates were performed. At least five microspheres were analyzed in each group per replicate. Data are presented as mean ± SD (n = 3). Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test. ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively, compared to the indicated groups; ns indicates no significant difference. Scale bars: 50 μm and 10 μm (B, C); 100 μm (J).

Quantitative real-time PCR (RT-qPCR) (Fig. 5F–I) corroborated these findings. Chonichspheres treatment significantly upregulated chondrogenic markers (COL2A1, SOX9, ACAN) while suppressing catabolic MMP13, outperforming the TGF-β3-only group. These data confirm that the system leverages a synergistic “antioxidant-differentiation” mechanism to drive MSCs toward a hyaline phenotype within the pathological OA microenvironment [68]. To further elucidate the unique advantages of AF over conventional antioxidants in establishing a permissive “window” for TGF-β3 action in this mechanism, we introduced N-acetylcysteine (NAC) as a parallel control. It is well-established that the efficacy of classic antioxidants like NAC is highly concentration-dependent, and higher doses yield robust ROS scavenging. However, considering the payload capacity limits of long-term single-injection HMs, we evaluated them at an equivalent, constrained low mass concentration (400 ng/mL). Assessed via the ABTS total antioxidant capacity assay, the non-sacrificial, carbon-cage structured AF exhibited significantly stronger antioxidant capacity than the NAC at this specifically limited dose (Fig. S19A). Consequently, when MSCs were similarly loaded within NAC-loaded HMs (GAM@TGF-β3@NAC@MSCs), this equivalent dose of NAC failed to sustainably clear ROS over the 14-day culture period. As a result, downstream TGF-β3 signaling remained continuously suppressed by the pathological microenvironment, evidenced by the reduced expression of the cartilage-specific protein SOX9, whereas the AF-integrated Chonichspheres@MSCs exhibited widespread and robust SOX9 expression within the cellular spheroids (Fig. S19B). Rather than implying an intrinsic inferiority of NAC under optimal conditions, these findings demonstrate that the core advantage of AF does not lie in a higher absolute maximal capacity, but in breaking the stoichiometric payload bottleneck under dose-restricted conditions. As a potent “radical sponge”, it thoroughly and durably remodels the oxidative microenvironment, effectively protecting and activating the chondrogenic signaling pathway to drive the ECM-mediated integrin–PI3K–AKT–mTOR signaling axis.

Beyond the direct differentiation of loaded MSCs, effective cartilage regeneration crucially relies on the synergy between exogenous implants and endogenous host chondrocytes. To evaluate whether our system could protect these vulnerable host cells, we utilized a Transwell co-culture system (apical Chonichspheres@MSCs; basolateral chondrocytes + H₂O₂) to simulate the hostile oxidative OA niche. While H₂O₂ exposure severely suppressed ECM synthesis (Fig. 5J), the Chonichspheres@MSCs group restored ECM deposition to levels comparable to healthy controls. These findings confirm that the system not only promotes intrinsic differentiation but also ameliorates the extrinsic microenvironment via antioxidant release, preserving the anabolic function of endogenous chondrocytes.

Collectively, these MPS simulations elucidate the distinct synergy of the Chonichspheres@MSCs system. Although both functional components initiate release synchronously, their distinct kinetics drive a complementary therapeutic effect. The initial burst release of TGF-β3 rapidly triggers the chondrogenic lineage commitment of MSCs. In parallel, the prolonged, sustained release of AF provides continuous ROS scavenging throughout the entire repair process. This long-term antioxidant shielding effectively relieves the microenvironmental suppression of the integrin–PI3K–AKT–mTOR axis, ensuring stable signal transduction for differentiation in a hostile niche. The system facilitates dual repair: driving the chondrogenesis of adherent cargo while remodeling the local niche to support endogenous progenitors. This strategy significantly enhances the translational potential of Chonichspheres for OA cartilage regeneration.

2.8. Transcriptomic mechanisms of Chonichspheres in OA therapy

To define the transcriptomic landscape of adherent MSCs under dynamic MPS circulation, we performed RNA-seq on three cohorts: Control (GAM@MSCs), OA (GAM@MSCs + SF), and Treat (Chonichspheres@MSCs + SF). Principal component analysis (PCA) revealed distinct clustering for each group (Fig. 6A). Notably, the Treat group occupied an intermediate position between the OA and Control clusters, suggesting that Chonichspheres intervention partially reverses OA-induced dysregulation, shifting the profile toward a normalized state. Hierarchical clustering of differentially expressed genes (DEGs) confirmed high reproducibility and marked separation between groups, indicating substantial transcriptomic reprogramming by the composite microspheres (Fig. 6B). We identified 754 upregulated and 1800 downregulated genes in the Treat group relative to OA. Volcano plots illustrate this substantial transcriptomic shift (Fig. 6C). Furthermore, the intersection of these DEGs across the Control, OA, and Treat groups highlights the regulatory landscape of Chonichspheres in restoring chondrogenic homeostasis (Fig. S20A–B).

Fig. 6.

Transcriptomic profiling identifies the integrin–PI3K–AKT–mTOR axis as a key therapeutic target. (A) PCA showing distinct transcriptomic segregation among Control, OA, and Treat groups. (B) Hierarchical clustering heatmap of DEGs. (C) Volcano plot of DEGs between Treat and OA groups (754 upregulated, red; 1800 downregulated, blue). (D) KEGG pathway enrichment analysis for Control vs. OA and Treat vs. OA comparisons. (E) Sankey diagram mapping core ROS/chondrogenesis genes to enriched signaling pathways (PI3K–AKT, Focal adhesion, ECM-receptor interaction). (F–K) Gene Ontology (GO) enrichment analysis categorized by Cellular Component (CC: F, G), Molecular Function (MF: H, I), and Biological Process (BP: J, K). Comparisons reveal a functional shift from metabolic stress (e.g., positive TOR signaling in J) in the OA group to restored integrin-mediated signaling and oxidative stress response (in K) following treatment. (L) Gene Set Enrichment Analysis (GSEA) of the PI3K–AKT pathway (Treat vs. OA). (M) KEGG enrichment emphasizing the Focal adhesion, ECM-receptor, PI3K–AKT, and mTOR pathways. (N) Quantitative expression of integrin family genes. ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. Groups: Control (GAM@MSCs), OA (GAM@MSCs + SF), Treat (Chonichspheres@MSCs + SF).

KEGG analysis comparing OA and Controls showed enrichment in cytokine–receptor interaction and PI3K–Akt pathways. In contrast, Treat vs. OA comparisons highlighted PI3K–Akt, ECM–receptor interaction, and focal adhesion. Notably, the PI3K–Akt pathway was enriched in both comparisons, implicating it as a central driver of OA pathogenesis (Fig. 6D). To visualize the interplay between core ROS/chondrogenic genes and enriched pathways, we generated a Sankey diagram combined with a bubble plot. Key DEGs were significantly enriched in crucial signaling pathways, including ECM-receptor interaction, focal adhesion, and PI3K–AKT. Notably, the PI3K–Akt pathway exhibited the highest enrichment (11 associated genes). These data suggest that following the initial ROS scavenging mediated by AF, Chonichspheres exert their downstream therapeutic effects—including the restoration of MSC homeostasis and subsequent chondrogenic differentiation—primarily via PI3K–Akt modulation (Fig. 6E).

To dissect the molecular mechanisms underlying therapeutic efficacy, we performed Gene Ontology (GO) enrichment analysis. First, in the Cellular Component (CC) dimension: compared to the healthy Control group, the OA group exhibited significant enrichment in components related to cell-substrate junction and focal adhesion. This suggests an aberrant remodeling of cell adhesion structures under pathological OA conditions. However, following Chonichspheres intervention, the OA-induced suppression of cell adhesion was significantly attenuated, as evidenced by the robust enrichment of focal adhesion-related components. This trend indicates that Chonichspheres may mitigate the pathological alterations in MSC adhesion structures induced by the OA microenvironment (Fig. 6F and G). Second, in the Molecular Function (MF) dimension (Fig. 6H and I): compared to the Control group, the OA group showed significant enrichment in integrin binding, reflecting a dysregulation of cell adhesion-related molecular functions. In contrast, the Treat group displayed a significantly elevated enrichment in integrin binding functions post-intervention. This suggests that Chonichspheres enhance the binding activity of integrin-related molecules on the cell surface, thereby improving cell adhesion function. Finally, in the biological process category (Fig. 6J and K): Comparison between the control and OA groups revealed the enrichment of processes including regulation of mitochondrial membrane potential, positive regulation of TOR signaling and oxidative phosphorylation in the OA group, suggesting that OA progression is accompanied by metabolic abnormalities. Following intervention in the Treat group (Fig. 6K), integrin-mediated signaling pathway and regulation of cellular response to oxidative stress were significantly enriched, accompanied by activation of the integrin-mediated signaling pathway. These results indicate that Chonichspheres can alleviate OA-induced pathological disorders by potentiating integrin signaling and modulating oxidative stress.

To delineate the molecular landscape, we profiled PI3K–AKT pathway gene expression (Fig. S20C and E). Gene Set Enrichment Analysis (GSEA) confirmed significant positive enrichment of this pathway in the Treat group relative to OA (NES >1, p < 0.05), validating the intervention's potent regulatory impact (Fig. 6L). Concurrently, KEGG analysis (Fig. 6M) corroborated that ECM-receptor interaction and Focal adhesion pathways were enriched, indicating that adherent cells successfully sense the GAM matrix. Crucially, the concurrent enrichment of PI3K-AKT and mTOR pathways links ECM sensing to intracellular metabolic regulation. It strongly demonstrates that integrin-PI3K-AKT-mTOR acts as the core mechanism whereby Chonichspheres facilitate OA cartilage repair (Figs. S20D, F and S21 A-E) [69]. We further validated this by quantifying the expression of key axis genes (PI3K, AKT1, mTOR, RPS6KB1) in the Treat versus OA cohorts. Quantitative analysis confirmed that Chonichspheres treatment significantly altered the expression of key PI3K–AKT–mTOR pathway genes, including CSF1R, NGFR, WNT5B, WNT2B, and RELN. This validates the transcriptional reprogramming of this axis observed in GSEA (Fig. S21F). Furthermore, core integrin genes (ITGB1, ITGB3, ITGA2, ITGA5) were significantly upregulated in the Treat group (Fig. 6N). This genetic evidence identifies integrin signaling as a critical mediator of the system's therapeutic efficacy.

Collectively, our bioinformatic analyses confirm that Chonichspheres induce profound transcriptomic reprogramming. By driving robust ROS scavenging (intracellular and mitochondrial) and restoring cellular homeostasis, the system effectively relieves the oxidative suppression of the integrin-PI3K-AKT-mTOR axis. The consequent activation of this signaling axis promotes MSC viability and chondrogenic differentiation—a regenerative mechanism consistent with recent literature highlighting the vital role of PI3K-AKT pathway modulation in osteochondral repair [70]. Ultimately, this synergistic cascade culminates in effective OA cartilage repair.

2.9. Chonichspheres regulate MSC survival and chondrogenic differentiation via the integrin-PI3K-AKT-mTOR axis

Guided by the insights from our multi-dimensional bioinformatic analyses, we proceeded to experimentally validate the molecular mechanism by which Chonichspheres preserves MSC viability and promotes chondrogenic differentiation, specifically focusing on the activation of the integrin–PI3K–AKT–mTOR signaling axis. Integrins are a family of transmembrane receptor proteins. Their extracellular domains bind to ECM components such as fibronectin and collagen, while their intracellular domains associate with various cytoskeletal and signaling molecules. In the unbound state, FAK remains inactive in the cytoplasm. Upon engagement with the ECM, integrins undergo conformational changes and cluster at the cell membrane to form focal adhesions (FAs), concurrently recruiting molecules including FAK, paxillin, and talin to these sites. Clustered FAK undergoes autophosphorylation at Tyr397, a critical residue that serves as a hallmark of FAK activation and a core binding motif for the subsequent recruitment of downstream signaling molecules such as Src kinases [71]. The PI3K signaling network acts as a central regulator of metabolism and growth in response to external stimuli, governing nutrient uptake, energy production, proliferation, and differentiation, and thus plays an essential role in MSC fate determination [72].

For further validation, we conducted RT-qPCR, Western blot, and immunofluorescence staining analyses. Together, these methods validated the mRNA and protein expression levels of the integrin-PI3K-AKT-mTOR axis (Fig. 7A). First, we evaluated the mRNA expression levels of key integrin subunits using RT-qPCR. The results demonstrated a significant restoration of integrin-related gene expression in the Chonichspheres@MSCs + SF group relative to the GAM@MSCs + SF group. Specifically, compared with the GAM@MSCs + SF group, the mRNA levels of the integrin α2 (ITGA2), α5 (ITGA5), β1 (ITGB1), and β3 (ITGB3) subunits were markedly upregulated by approximately 3.0-fold, 2.1-fold, 2.5-fold, and 1.8-fold, respectively (Fig. 7B–E). These findings suggest that the therapeutic efficacy of Chonichspheres is primarily mediated through the activation of ITGB1-, ITGB3-, ITGA2-, and ITGA5-dependent integrin signaling pathways [73].

Fig. 7.

Chonichspheres protect MSCs from the OA microenvironment via activation of the integrin–PI3K–AKT–mTOR axis. (A) Schematic illustration of the experimental design and workflow for mechanistic validation. (B–E) RT-qPCR analysis of the integrin subunits ITGA2(B), ITGA5(C), ITGB1(D), and ITGB3(E). (F) Representative Western blot images showing key proteins of the PI3K–AKT–mTOR signaling pathway following pharmacological inhibition. (G–I) Quantitative densitometric analysis of the relative phosphorylation levels of p-PI3K (G), p-AKT (H), and p-mTOR (I). (J–O) Immunofluorescence assessment of pathway activation. Representative 3D images and corresponding quantitative analysis (MFI) of p-FAK (J, K), p-AKT (L, M), and p-mTOR (N, O). Nuclei are stained blue (DAPI), F-actin is stained red (Phalloidin), and target proteins are stained green. Each data point indicates the mean fluorescence intensity of one independent biological replicate, and a total of three independent biological replicates were performed. At least five microspheres were analyzed in each group per replicate. Data are presented as mean ± SD (n = 3). Statistical significance was determined using one-way ANOVA followed by Tukey's post hoc test. ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively.

Consistent with literature [74] and bioinformatic data, we validated the integrin-PI3K-AKT-mTOR axis via Western blot. While total protein levels (PI3K, AKT, mTOR) remained constant across all groups (Fig. S22), relative to the OA group (GAM@MSCs + SF), the Treat group (Chonichspheres@MSCs + SF) exhibited significantly upregulated phosphorylation of PI3K, AKT, and mTOR. Quantitative analysis confirmed this restoration, demonstrating that Chonichspheres activate this axis to counteract the suppressive OA microenvironment. To establish a causal relationship between this signaling axis and the observed therapeutic efficacy, a loss-of-function study was conducted using the specific PI3K inhibitor LY294002. Western blot analysis revealed that while total protein levels remained consistent across all experimental groups, LY294002 administration markedly abrogated the Chonichspheres-induced target phosphorylation (Fig. 7F–I). Furthermore, pharmacological blockade of this pathway effectively reversed the enhancements in chondrogenic differentiation and mitochondrial protection conferred by the Chonichspheres. These functional validations provide direct evidence that the regenerative effects of the Chonichspheres system fundamentally depend on the integrin–PI3K–AKT–mTOR axis, confirming its role as the core mechanistic driver. To further investigate the integrin axis, we assessed phosphorylated focal adhesion kinase (p-FAK) via 3D immunofluorescence. While the OA group (GAM@MSCs + SF) showed weak p-FAK signal, the Treat group (Chonichspheres@MSCs + SF) displayed markedly enhanced fluorescence intensity (Fig. 7J and K). To further validate the activation of FAK at the protein level, we performed Western blot analysis of Chonichspheres-treated MSCs in the presence or absence of the specific FAK inhibitor PF-573228. The results demonstrated that the Chonichspheres intervention significantly upregulated p-FAK levels compared to the OA group (GAM@MSCs + SF), while total FAK expression remained unchanged. Importantly, the application of PF-573228 effectively abrogated the Chonichspheres-induced increase in p-FAK, confirming the direct involvement of FAK activation in the Chonichspheres-mediated signaling pathway (Fig. S23). Consistently, downstream effectors p-AKT and p-mTOR were also evaluated. In the Treat group, both markers exhibited enriched accumulation at the cell membrane and cytoplasm, significantly surpassing the OA group in intensity (Fig. 7L–O). These in situ findings corroborate the Western blot data, confirming robust activation of the integrin-PI3K-AKT-mTOR axis within Chonichspheres.

Collectively, these data demonstrate that Chonichspheres robustly activate the integrin–PI3K–AKT–mTOR pathway in MSCs, as evidenced by the significant upregulation of key mRNA and phosphorylated protein markers. Functioning as an ECM-biomimetic scaffold, Chonichspheres provide adhesion motifs that trigger integrin clustering and FAK autophosphorylation. The robust ROS-scavenging capability of AF effectively eliminates mitochondrial superoxide and restores the broader bioenergetic function of MSCs within the OA microenvironment, as evidenced by reduced MitoSOX levels and an elevated JC-1 ratio. This comprehensive restoration of mitochondrial homeostasis fundamentally relieves intracellular oxidative suppression, thereby successfully propagating downstream signaling via the PI3K–AKT–mTOR axis. Concurrently, Chonichspheres promote endogenous ATP production in MSCs, supplying the bioenergetic demands for chondrogenesis and anabolism [66]. This metabolic restoration synergizes with the TGF-β3/Smad signaling pathway to actively drive MSC differentiation toward hyaline cartilage. Consequently, the integrin–PI3K–AKT–mTOR axis activated by Chonichspheres constitutes a pivotal mechanism that synergistically couples the preservation of mitochondrial homeostasis with robust chondrogenesis, ultimately underpinning successful OA repair.

2.10. Behavioral and radiological evaluation of Chonichspheres@MSCs in DMM-induced OA

To evaluate the in vivo therapeutic efficacy of Chonichspheres@MSCs, we established a murine model of OA in 8-week-old C57BL/6 mice via Destabilization of the Medial Meniscus (DMM) surgery combined with treadmill exercise intervention. Subsequently, the microsphere-adherent MSCs were administered via intra-articular injection biweekly for a duration of 8 weeks post-DMM. At the study endpoint, CatWalk gait analysis was employed to assess plantar gait parameters (Fig. 8A). Fig. 8B presents representative images illustrating gait patterns and pressure distribution—specifically Gait screenshots, Gait intensity maps, and both 2D and 3D footprints—across the experimental groups. Quantitative analysis revealed that the Chonichspheres@MSCs group exhibited significantly ameliorated locomotor behavior in the left hind (LH) limb compared to the DMM group. Notably, the treatment resulted in a marked increase in both footprint area and locomotor speed (Fig. 8C and D). Furthermore, to more sensitively evaluate osteoarthritis-associated pain and functional recovery, the weight-bearing ratio of the affected to the unaffected limb was calculated. As shown in Fig. 8E, PBS-treated mice exhibited a severe reduction in this ratio compared to the Sham group, indicative of significant joint pain and subsequent weight-bearing avoidance. In contrast, treatment with Chonichspheres@MSCs robustly restored the weight-bearing balance, achieving a ratio 3.8-fold higher than that of the PBS group. This result directly demonstrates the superior efficacy of Chonichspheres@MSCs in alleviating OA-associated pain and restoring symmetric locomotor function. Collectively, these findings demonstrate that the intra-articular injection of Chonichspheres@MSCs effectively restores DMM-induced gait abnormalities in C57BL/6 mice.

Fig. 8.

Chonichspheres@MSCs ameliorate gait abnormalities and attenuate osteophyte formation in DMM-induced mice. (A) Schematic timeline of the experimental design, including DMM surgery, treadmill exercise, therapeutic intervention, and downstream evaluations. (B) Representative images from the gait analysis, displaying gait screenshots, gait intensity maps, and both 2D and 3D footprints across the experimental groups. (C, D) Quantitative analysis of gait parameters, specifically comparing the Left Hind (LH) print area (cm²) and swing speed (cm/s) among the groups. (E) Quantitative analysis of the affected/unaffected limb weight-bearing ratio among the different groups, serving as an indicator for pain relief and functional recovery. (F) Representative 3D micro-CT reconstructions of knee joints in Anterior-Posterior (AP) and Lateral (LAT) views. (G, H) Quantitative analysis of the osteophyte number and osteophyte volume. Data are presented as mean ± SD (n = 5 biologically independent animals). ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively. NS indicates no statistical significance.

Osteophyte formation is a radiological hallmark of knee OA [75,76]. We assessed this pathology in C57BL/6 mice using micro-CT. 3D reconstruction revealed pronounced osteophytosis in all groups except Sham, with the most severe presentation in PBS controls (Fig. 8F). Quantitative analysis demonstrated that Chonichspheres@MSCs treatment significantly reduced both osteophyte number and volume compared to PBS, GAM@MSCs, and GAM@MSCs@TGF-β3 groups (Fig. 8G). Notably, osteophyte volume in the Chonichspheres@MSCs group was suppressed to just 0.15-fold that of the PBS group (Fig. 8H), indicating a significant amelioration of OA severity in vivo.

2.11. Chonichspheres@MSCs promote cartilage regeneration in DMM-induced OA mice

Following radiological assessment, the mice were euthanized at 8 weeks post-DMM for histological evaluation of the articular cartilage(Fig. 9A). Hematoxylin and eosin (H&E), Safranin O-Fast Green, and double immunofluorescence staining were employed to visualize cartilage surface morphology and structural repair. H&E and Safranin O-Fast Green of Sham controls confirmed healthy cartilage architecture with smooth surfaces, organized chondrocytes, and intense ECM staining (Fig. 9B and C). In contrast, PBS-treated joints displayed severe degeneration, characterized by structural collapse, cellular disorganization, and near-total ECM loss. While GAM@MSCs treatment alleviated erosion, the repair was predominantly fibrocartilaginous, highlighting the limited capacity of MSCs to generate hyaline tissue without specific induction. Conversely, GAM@MSCs@TGF-β3 partially promoted hyaline cartilage regeneration. Notably, the Chonichspheres@MSCs afforded the highest protection, exhibiting near-native tissue structure with only minor superficial abrasion. Quantitative Mankin scores and erosion depth measurements corroborated these findings, confirming that AF-mediated cell preservation maximizes repair efficacy (Fig. 9E and F).

Fig. 9.

Chonichspheres@MSCs promote hyaline cartilage regeneration in DMM-induced OA mice. (A) Schematic of in vivo experimental design. (B) Representative histological images of articular cartilage sections stained with Safranin O-Fast Green, revealing the proteoglycan-rich matrix (red) and underlying bone (green). (C) Representative images stained with Hematoxylin & Eosin, demonstrating the general tissue morphology and repair status. (D) Representative double immunofluorescence staining confirming the hyaline phenotype: COL2A1 (green), Aggrecan (red) and DAPI (blue). (E, F) Quantitative histological analysis of (E) the Mankin score and (F) the depth of cartilage erosion. (G, H) Quantitative analysis of the mean fluorescence intensity for (G) COL2A1 and (H) Aggrecan within the cartilage zone. Data are presented as mean ± SD (n = 5 biologically independent animals). Scale bars: 200 μm and 100 μm. Statistical significance was determined using one-way ANOVA followed by Tukey's post hoc test. ∗, ∗∗, ∗∗∗, and ∗∗∗∗ indicate p < 0.05, p < 0.01, p < 0.001, and p < 0.0001, respectively.

We further assessed cartilage matrix integrity via double immunofluorescence for COL2A1 and aggrecan. PBS-treated joints exhibited severe matrix depletion, retaining only 23.1% of COL2A1 and 14.9% of aggrecan relative to the Sham group. In contrast, the Chonichspheres@MSCs group demonstrated robust restoration. Quantitative analysis confirmed that the therapeutic system recovered COL2A1 and aggrecan expression to 76.8% and 81.1% of healthy levels, respectively, significantly outperforming all other experimental cohorts (Fig. 9D–G, H). In summary, these in vivo findings establish Chonichspheres as a superior biomimetic vehicle that prolongs intra-articular MSC retention to engineer a pro-regenerative microenvironment. Mechanistically, functionalized AF provide dynamic ROS scavenging, shielding MSCs from oxidative stress to preserve mitochondrial homeostasis. Synergistically, controlled TGF-β3 release drives chondrogenic differentiation. This multi-functional strategy effectively repairs cartilage defects in the DMM-induced model, underscoring its significant translational potential for OA therapy.

3. Discussion

The translational failure of existing MSC therapies is frequently attributed to the neglect of the hostile microenvironment during the early transplantation phase—a “holistic barrier” that severely compromises therapeutic efficacy [77]. While various multifunctional biomaterial platforms have been engineered for osteoarthritis, the proposed “Chonichspheres” strategy is distinguished by its multi-stage programmatic synergy. It couples active microscale ECM integration with continuous redox modulation to proactively intervene in OA pathogenesis. First, to facilitate cell adhesion and spreading, the Chonichspheres provide a biomimetic structural foundation that ensures stable MSC loading and robust initial anchorage. Second, providing ROS protection is critical, given that the excessive accumulation of ROS within the OA joint cavity is a primary driver of mitochondrial dysfunction and anoikis [78]. As an inert carbon-based nanomaterial, AF acts as a catalytic radical sponge that reversibly captures ROS via electron transfer cycling, without irreversible molecular degradation or toxic secondary metabolites. This mechanism preserves the mitochondrial membrane potential of both exogenous MSCs and endogenous chondrocytes, and establishes a long-lasting, redox-stable microenvironment. Third, to address integration challenges posed by the natural anti-adhesive barrier of the cartilage surface and achieve targeted anchoring, we exploited the pathological exposure of amino groups on OA-damaged cartilage to introduce AHAMA. Through a Schiff base reaction, this facilitates in situ covalent anchoring of the Chonichspheres to the host tissue. This active targeting strategy surpasses traditional passive physical filling by constructing a robust mechanical fulcrum at the defect site, effectively overcoming the persistent challenges of MSC homing and tissue integration in regenerative medicine [57]. Finally, to effectively guide chondrogenic differentiation, AF and TGF-β3 are released simultaneously within the joint cavity and act in synergy, such that AF-mediated ROS scavenging continuously provides a low-oxidative-stress niche that supports TGF-β-driven chondrogenic differentiation throughout the repair process [79,80].

Although the role of biochemical factors in cartilage regeneration has been widely corroborated, the underlying mechanisms by which physical properties of biomaterials synergize with biochemical signals to reshape cell fate remain to be fully elucidated. Previous studies have demonstrated that AKT phosphorylation is an indispensable prerequisite for TGF-β3-induced chondrogenesis, and inhibiting AKT activity completely abrogates the pro-differentiative effects of TGF-β3 [80]. Crucially, we reveal for the first time the core bottleneck behind the failure of cartilage regeneration in OA at the molecular level: excessive ROS within the pathological OA microenvironment potently suppress the integrin–PI3K–AKT signaling axis in MSCs, thereby disrupting the mitochondrial homeostasis of MSCs and impeding their chondrogenic differentiation [66]. Addressing this challenge, the Chonichspheres system developed herein demonstrates exceptional regulatory capabilities.

Specifically, the platform mediates its therapeutic effects through a parallel integration of biochemical and physical cues. Concurrently with the early action of TGF-β3, the rapid ROS-scavenging activity of the sustainedly released AF continuously relieves the oxidative suppression imposed by the pathological microenvironment. Critically, by rescuing mitochondrial homeostasis and restoring the intracellular ATP supply, AF thoroughly resolves the metabolic energy crisis within MSCs. Empowered by this bioenergetic recovery, the cells regain the capacity to actively engage with the GelMA/AHAMA matrix scaffold. Functioning as a pivotal mechanotransduction hub, this matrix then initiates the highly energy-consuming process of cytoskeletal reorganization by promoting integrin clustering and focal adhesion formation on the cell membrane. These physical cues, dictated by the surface topology and stiffness of the material, are precisely translated into biochemical directives, triggering downstream phosphorylation and fully activating the PI3K–AKT–mTOR axis. The activation of this core pathway exerts a unique dual effect. On the one hand, it directly ameliorates the mitochondrial function of MSCs within the OA microenvironment and promotes endogenous ATP production, fulfilling the bioenergetic demands for chondrogenesis and anabolism. On the other hand, this metabolic restoration converges with the early-initiated TGF-β3/Smad signaling pathway to cooperatively enhance anabolism by upregulating the expression of COL2A1 and SOX9, while simultaneously attenuating catabolism by downregulating MMP13 expression. Consequently, the integrin–PI3K–AKT–mTOR axis mediated by Chonichspheres constitutes a critical mechanism that synergistically and potently reverses the degenerative progression of OA at the transcriptional, translational, and post-translational modification levels. This discovery of precisely manipulating intracellular signal transduction via a combination of “biochemical clearance” and “physical property reprogramming” establishes a novel theoretical paradigm for the design of next-generation, multidimensional bioactive materials with intrinsic disease-modulating functions.

A critical methodological advancement in this study is the establishment of a high-fidelity MPS, designed to circumvent the poor predictive value associated with traditional static culture models and the clinical translation limitations of small animal models [81,82]. Articular cartilage regeneration is a dynamic process involving complex fluid dynamics and multi-tissue communication (cartilage-synovium-SF). Conventional 2D or static 3D models fail to recapitulate this physiological complexity [83]. Furthermore, animal models are inherently limited by interspecies differences and challenges in dynamic monitoring; consequently, relying solely on them often results in a significant “translational gap” between preclinical data and clinical outcomes. To bridge this gap, we developed a dynamic MPS model incorporating primary human cells. Our MPS system innovatively incorporates a dynamic flow module that recapitulates the continuous perfusion of the local defect microenvironment, validating the structural stability of Chonichspheres under a dynamic setting. In terms of physiological relevance, in vivo fluid dynamics within the knee joint exhibit extreme spatial heterogeneity, ranging from high-shear sliding friction at the superficial articular surface to ultra-low interstitial fluid flow within the deep cartilage ECM or defect niches. As demonstrated by our computational analysis, the microenvironment within our MPS accurately recapitulates the ultra-low shear conditions characteristic of this deep ECM niche. Replicating the high-shear environment of the superficial joint would require exceptionally high flow velocities. Such extreme flow rates would drastically reduce the local residence time of active molecules—including ROS and critical cytokines—around the Chonichspheres, leading to a rapid washout that precludes the establishment of a stable OA pathological microenvironment. Therefore, our MPS design deliberately decouples mechanical stimulation from the experimental system. By maintaining an ultra-low shear environment, we effectively strip away fluid-induced mechanical stress and prevent anoikis. This strategic design strictly isolates the biochemical variables, ensuring that the simulated dynamic system functions exclusively to facilitate the convective transport and interaction of pathological molecules. Consequently, the observed cellular responses and the cytoprotective efficacy of the Chonichspheres can be attributed entirely to the modeled inflammatory and oxidative microenvironment, fulfilling our primary experimental objective without the confounding effects of physiological-grade shear stimulation. Crucially, the system simulates the complex paracrine communication between cells under OA pathological conditions. This elevates in vitro evaluation in regenerative medicine from simple “cellular observation” to “dynamic biomimetic emulation,” establishing a standardized, universal paradigm for screening anti-OA drugs and optimizing tissue engineering parameters.

In summary, the Chonichspheres strategy offers a promising clinical solution for OA treatment through a multifaceted mechanism comprising structural support, oxidative protection, homing/integration, and guided differentiation. The modular design of Chonichspheres, by integrating cartilage-targeting adhesion (AHAMA) and controlled release of chondrogenic factors (AF/TGF-β3), effectively overcomes the limitations of free-drug delivery in an OA microenvironment [84,85]. Beyond achieving functional repair of damaged cartilage via activation of the integrin–PI3K–AKT–mTOR axis, the MPS-based evaluation system established herein provides a versatile methodological framework for regenerative research in other load-bearing tissues, such as bone. Despite these encouraging results, we acknowledge certain limitations. For instance, the GelMA/AHAMA photocrosslinkable microsphere system adopted in this study still faces inherent bottlenecks in long-term clinical translation. Residual methacrylated moieties carry potential pro-inflammatory risks, while the cytotoxicity of photoinitiators and cellular damage induced by ultraviolet crosslinking cannot be ignored. Moreover, systematic verification regarding the long-term biocompatibility and in vivo metabolic patterns of material degradation products remains insufficient. In addition, the standardized fabrication of multi-component formulations and complicated medical device approval procedures further hinder its clinical translation progress. In follow-up studies, the substitution degree of methacrylate groups can be regulated and dynamic covalent bonds can be introduced to precisely modulate material degradation behavior. Visible light crosslinking or material self-crosslinking strategies can be adopted to avoid phototoxicity-related issues. Incorporation of natural biological components is conducive to improving biocompatibility. Systematic biosafety evaluation and standardized optimization of preparation protocols should be performed to further enhance its potential for clinical translation. Similarly, although AF exhibits excellent ROS-scavenging capabilities, its widespread clinical application is currently limited by the lack of clear regulatory frameworks for carbon-based nanomaterials. While the localized working concentration of AF utilized in this study (400 ng/mL) falls substantially below reported cytotoxicity and genotoxicity thresholds [86], its long-term in vivo biodistribution, tissue accumulation, and ultimate clearance pathways still require rigorous longitudinal evaluation.

Furthermore, we must acknowledge the limitations of our current MPS model. OA is fundamentally an immune-mediated inflammatory disease [87]. However, our system lacks circulating and resident immune cells, especially macrophages. As a result, it cannot fully replicate complex biomaterial-immune interactions or in vivo macrophage polarization [88]. We must therefore be cautious when extrapolating these in vitro results to in vivo pathological conditions. Currently, our MPS primarily validates the robust antioxidative and mechanoprotective capabilities of Chonichspheres against specific stressors. To bridge this gap, future studies will develop a next-generation “Joint-on-a-chip”. To further comprehensively verify in vivo MSC survival, retention and host immune responses, future work will combine multiple analytical approaches. By integrating an immune cell module, this model will enable a more comprehensive pre-clinical evaluation. Additionally, further research must align the long-term degradation rate of the microspheres with in vivo cartilage matrix deposition. While the mouse DMM model provides a robust platform for evaluating the initial biological efficacy of Chonichspheres, the inherent interspecies differences in joint mechanical loading between rodents and humans cannot be overlooked. For the successful clinical translation of this Chonichsphere-based therapeutic strategy, further validation in large weight-bearing animal models that closely recapitulate human joint physiology is essential. Such large-animal studies not only help confirm the long-term structural integrity of microspheres and functional ECM maturation under complex biomechanical microenvironments, but also better mimic human physiological characteristics, thereby greatly accelerating the clinical translation of this repair strategy.

4. Materials and methods

4.1. Materials

Hyaluronic acid (HA, MW: 100–200 kDa), sodium periodate (NaIO₄), ethylene glycol (C₂H₆O₂), sodium hydroxide (NaOH), methacrylic anhydride (C₈H₁₀O₃, 94%, containing 0.2% Topanol stabilizer), liquid paraffin (light, density: 0.84–0.86 g/mL), Coumarin-6, and L-ascorbic acid were purchased from Macklin Biochemical Co., Ltd. (Shanghai, China). Span 80 was obtained from Solarbio Science & Technology Co., Ltd. (Beijing, China). Gelatin Methacryloyl (GelMA) was sourced from Yongqinquan Intelligent Equipment Co., Ltd. (Suzhou, China). Recombinant Human TGF-β3 was purchased from Novoprotein (Suzhou, China). Fullerenes (C₆₀) were obtained from Yitai Technology Co., Ltd. (Shanghai, China). Cell culture reagents including Phosphate-buffered saline (PBS), DMEM, and DMEM/F12 were purchased from Gibco (USA). Fetal bovine serum (FBS) was supplied by VivaCell (Shanghai, China). Other biochemical reagents and kits, including Collagenase Type II, Hyaluronidase, Toluidine Blue O, Rhodamine-Phalloidin, and the Cell Counting Kit-8 (CCK-8), were acquired from Solarbio. Hydrogen peroxide (H₂O₂) was purchased from Sigma–Aldrich (USA). Additional reagents such as Trypsin-EDTA, Penicillin-Streptomycin (100×), Trizol, RIPA lysis buffer, PMSF, SDS-PAGE loading buffer, DMSO, and assay kits for ROS, BCA protein, Calcein/PI, JC-1, Mito Deep Red, DCFH-DA, MitoSOX, and DAPI-containing antifade mounting medium were obtained from Beyotime Biotechnology (Shanghai, China). The TGF-β3 ELISA kit was procured from a supplier in Jiangsu, China. All primary and secondary antibodies were purchased from Proteintech (China) or Abcam (USA). RNA extraction and cDNA synthesis kits were obtained from Yeasen Biotechnology (Hunan, China) and MedChemExpress (MCE, Shanghai, China), respectively.

4.2. Methods

4.2.1. scRNA-seq analysis

We retrieved scRNA-seq datasets of cartilage (GSE169454) and synovial tissues (GSE216651) from healthy individuals and OA patients via the NCBI GEO database. The sample descriptions and their corresponding accession numbers detail the specific clinical characteristics of the whole knee joints. The provided GEO Series (GSE) identifiers denote the specific repositories where the raw sequencing data are publicly accessible.

All downstream scRNA-seq data processing and the generation of Seurat objects were executed utilizing the “Seurat” R package. Following appropriate data normalization and scaling, the “Harmony” algorithm was employed to integrate the datasets and mitigate batch effects. Unsupervised cellular clustering was subsequently performed, and the resulting distinct cell populations were visualized using Uniform Manifold Approximation and Projection (UMAP).

To identify marker genes for each specific cluster, the Wilcoxon rank-sum test was applied via the FindAllMarkers function. The threshold criteria for marker gene selection were established as follows: 1) a log-fold change (logFC) > 0.25; 2) a p-value <0.05; and 3) a minimum fractional expression (min.pct) > 0.1. Furthermore, to quantitatively evaluate the activity of specific signaling pathways, we calculated pathway module scores utilizing the AddModuleScore function, and divided the cells into two groups (Normal, OA).

4.2.2. Synthesis of AF

AF was synthesized via a nucleophilic substitution reaction. Briefly, 50 mg of pristine fullerene (C₆₀) was dissolved in toluene, followed by the addition of 10 mL ethylenediamine (EDA) under nitrogen protection. The mixture was magnetically stirred at 60 °C for 3 days until the solution transitioned from purple to reddish-brown.

The solvent was then removed by rotary evaporation at 100 °C to yield crude AF. To purify the product, the crude AF was redissolved in 1 mM HCl and dialyzed against ultrapure water for 4 days using a dialysis membrane. The hydrodynamic diameter and Zeta potential of the purified AF were characterized using a Zetasizer Nano ZS90 (Malvern, UK), and the morphology was examined via transmission electron microscopy (TEM, FEI Tecnai F20).

4.2.3. Synthesis of AHAMA

Synthesis of Aldehyde-HA (AHA): First, 1 g of HA was dissolved in 100 mL of deionized water (DI water) under mechanical stirring (350 rpm). A 5 mL aqueous solution of 0.5 M NaIO₄ was added dropwise, and the reaction proceeded in the dark for 2 h (250 rpm) to oxidize the vicinal diols. The reaction was quenched by adding 1 mL of ethylene glycol with stirring for an additional 1 h. The resulting solution was dialyzed (MWCO: 8–14 kDa) against DI water for 72 h and lyophilized to obtain AHA.

Synthesis of AHAMA: Subsequently, 1 g of AHA was dissolved in 50 mL of DI water (350 rpm). Methacrylic anhydride (MA, 20 mL) was added dropwise under dark conditions. The reaction mixture was maintained in an ice-water bath, and the pH was maintained by the addition of 5 M NaOH (20 mL) via a syringe pump over 2 h. After incubation at 4 °C overnight, the mixture was centrifuged at 7000 × g for 15 min to remove insoluble byproducts. The supernatant was dialyzed (MWCO: 8–14 kDa) for 72 h in the dark and lyophilized to yield AHAMA. The chemical structure was confirmed by ¹H NMR spectroscopy (Bruker Avance NEO 600 MHz, Germany).

4.2.4. Fabrication of GAM@AF microspheres

Composite HMs were fabricated using a microfluidic electrospray system combined with cryogenic treatment. The aqueous phase consisted of 1% (w/v) GelMA and 2% (w/v) AHAMA containing 400 ng/mL AF and 0.5% (w/v) photoinitiator [86]. The oil phase comprised liquid paraffin supplemented with 5% (v/v) Span 80. Both phases were injected using precision syringe pumps (Lead Fluid, China).

The generated droplets were immediately frozen at −40 °C and photocrosslinked under 405 nm UV light for 5 min. The crosslinked microspheres were collected by centrifugation (5000 rpm, 5 min), washed three times with diethyl ether to remove residual oil, and subsequently washed with DI water. Finally, the microspheres were lyophilized to obtain GAM@AF microspheres. Sterilization was performed via UV irradiation, and samples were stored at 4 °C. To determine AF loading efficiency, FITC-labeled AF (Xi'an Ruixi Biological Technology) was used, and absorbance was measured before and after encapsulation.

4.2.5. Preparation of Chonichspheres and TGF-β3 release profile

To fabricate the final bioactive system (Chonichspheres), 1 mg of lyophilized GAM@AF microspheres was incubated with a TGF-β3 solution (300 ng/mL, pH 7.4) overnight at 4 °C. The loading efficiency was calculated by measuring the concentration of unbound TGF-β3 in the supernatant after centrifugation.

For release kinetics, the TGF-β3-loaded microspheres were resuspended in 1 mL of PBS containing 0.1% BSA (pH 7.4) and incubated at 37 °C with horizontal shaking (80 rpm). At predetermined time points, the supernatant was harvested to monitor the sustained release kinetics.

The concentration of TGF-β3 was quantified using an enzyme-linked immunosorbent assay (ELISA) kit according to the manufacturer's instructions. Briefly, blank wells, standard wells (50 μL of standards at varying concentrations), and sample wells (10 μL of sample + 40 μL of diluent) were prepared. Except for the blank wells, 100 μL of HRP-conjugate reagent was added to each well. After sealing and incubating at 37 °C for 60 min, the plate was washed thoroughly and blotted dry. Chromogen solutions A (50 μL) and B (50 μL) were added sequentially and incubated in the dark at 37 °C for 15 min. The reaction was terminated with stop solution, and the absorbance was measured at 450 nm using a microplate reader. The loading efficiency was then calculated based on the concentration of TGF-β3 remaining in the initial supernatant.

4.2.6. In vitro degradation assay

To evaluate biodegradability under pathological conditions, freshly prepared Chonichspheres were immersed in a degradation medium containing Hyaluronidase (100 U/mL) and Collagenase Type II (2 U/mL) at pH 6.5. The samples were incubated at 37 °C with shaking (80 rpm). The degradation medium was replenished every 3 days. Morphological changes were monitored at scheduled time intervals.

4.2.7. Physical characterization

The particle size and general morphology of the microspheres were analyzed using bright-field microscopy (Nikon, Japan). The surface topography and porous structure of the lyophilized microspheres were visualized via SEM (S-400, Japan). Elemental distribution was qualitatively assessed using energy-dispersive X-ray spectroscopy (EDS).

4.2.8. Cell culture

4.2.8.1. Culture of cell lines and stem cells

Human bone marrow mesenchymal stem cells (hBMSCs) (YB-H3277, Yubo Biotechnology, Shanghai, China) were cultured in α-MEM supplemented with 20% fetal bovine serum (FBS; Gibco, Australia), 100 U/mL penicillin, and 100 μg/mL streptomycin. The human chondrocyte cell line C28/I2 (IM-H846, Yimo Biotechnology, China) was maintained in DMEM containing 10% FBS and antibiotics. Human Synovial Fibroblasts (CP-H241, Procell, China) were cultured in DMEM/F12 supplemented with 15% FBS and antibiotics. All cells were maintained in a humidified incubator at 37 °C with 5% CO₂. For chondrogenic differentiation, a specialized chondrogenic induction medium (Procell, China) was used.

4.2.8.2. Isolation and culture of primary cells and spheroid formation

Human tissue samples were obtained from patients undergoing total knee arthroplasty (TKA), intramedullary nailing for tibial fractures, or knee arthroscopy. This study was conducted in strict accordance with the Declaration of Helsinki, and ethical approval was obtained from the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (Approval No. 2026-0350-01). Informed consent was obtained from all donors.

4.2.8.3. Isolation and culture of primary chondrocytes

Worn and unworn hyaline cartilage tissues were harvested from the femoral condyles and tibial plateaus during TKA to represent OA and normal cartilage, respectively. Tissues were washed three times with PBS containing 1% penicillin-streptomycin, minced into 1 mm³ fragments, and digested with 0.2% Collagenase Type II at 37 °C for 12–16 h with constant agitation.

The digest was filtered through a 70 μm cell strainer and centrifuged at 1200 rpm for 5 min. Cells were resuspended in DMEM/F12 containing 10% FBS and seeded into culture flasks. For spheroid formation, primary chondrocytes at the logarithmic growth phase were harvested and resuspended at a density of 6 × 10⁵ cells/mL. A volume of 200 μL (1.2 × 10⁵ cells) was dispensed into each well of a 96-well U-bottom ultra-low attachment plate. Half or complete medium changes were performed every 2–3 days.

4.2.8.4. Isolation and culture of primary synoviocytes

Synovial tissues collected during TKA or arthroscopy were classified as OA or normal synovium. After removing adipose and fibrous connective tissues, the samples were minced and allowed to adhere to the bottom of culture dishes for 4–6 h before the addition of DMEM/F12 containing 20% FBS. Cells were passaged upon reaching 80% confluence. Spheroid construction followed the same protocol as chondrocytes, with 1.2 × 10⁵ cells per well in U-bottom ultra-low attachment plates.

4.2.8.5. Cryopreservation and thawing

Cells were cryopreserved using serum-free freezing medium (Procell, China) at a density of 1 × 10⁶ cells/mL. Vials were placed in a Mr. Frosty™ freezing container at −80 °C overnight and subsequently transferred to liquid nitrogen (−196 °C). Thawing was performed rapidly in a 37 °C water bath, followed by immediate washing with pre-warmed medium to remove cryoprotectants.

4.2.9. Extraction and storage of SF

SF was aspirated from the knee joints of patients undergoing TKA or arthroscopy using an 18G needle after partial capsulotomy. The collected SF was centrifuged at 3000 × g for 15 min at 4 °C to remove cellular debris and erythrocytes. The supernatant was aliquoted (500 μL/tube) and stored at −80 °C.

4.2.10. Preparation of bio-hybrid microspheres

To fabricate cell-loaded microspheres, GAM microspheres were co-cultured with a BMSC suspension (1 × 10⁶ cells/mL) in 96-well U-bottom ultra-low attachment plates at 37 °C with 5% CO₂. After 48 h of 3D co-culture, uniformly loaded GAM@MSCs were obtained. The GAM@MSCs@TGF-β3 and Chonichspheres@MSCs groups were prepared using the same cell loading protocol on their respective functionalized microspheres.

4.2.11. Biocompatibility assessment

To evaluate the biocompatibility of the Chonichspheres, a Calcein-AM/PI Live/Dead Assay Kit (Beyotime, China) was employed, where live cells were labeled with green fluorescence and dead cells with red fluorescence. Furthermore, the morphological characteristics and cytoskeletal organization of MSCs on the Chonichspheres were assessed on days 1, 7, and 14.

The samples were stained with Rhodamine-Phalloidin (Solarbio, China) to visualize F-actin filaments and DAPI (Beyotime, China) to counterstain the nuclei. After a 15-min incubation, fluorescence images were acquired using a laser scanning confocal microscope (Olympus, Japan). To comprehensively visualize the distribution and morphology of MSCs on the 3D microsphere surfaces, Z-stack maximum intensity projections were generated. Quantitative analysis of the relative fluorescence intensity was performed using ImageJ software.

4.2.12. In vivo longitudinal tracking of MSC retention

To evaluate the retention capability of Chonichspheres in vivo, we employed a longitudinal bioluminescence imaging (BLI) study using Luciferase-labeled MSCs. First, MSCs were transduced with lentiviral vectors stably expressing firefly luciferase (Luc) and selected with puromycin to establish a stable Luc-expressing cell line. The Luc-MSCs were suspended in serum-free DMEM at a density of 1.0 × 10⁸ cells/mL. As described previously, the Luc-MSCs were co-incubated with sterilized Chonichspheres to construct cell-laden HMs, denoted as Chonichspheres@Luc-MSCs. Subsequently, 10 μL of Chonichspheres@Luc-MSCs or an equivalent dose of free Luc-MSCs (1.0 × 10⁶ cells) was intra-articularly injected into the right knee joint of the C57BL/6 mice using a 31-gauge Hamilton syringe. Following injection, the needle was held in place for 60 s to prevent fluid backflow. At days 0, 7, 14, and 21 post-injection, the mice were intraperitoneally injected with D-luciferin potassium salt (150 mg/kg) for BLI. Ten minutes later, mice were anesthetized with isoflurane and positioned on a 37 °C heating pad. Bioluminescence signals in the knee joint region were captured using an in vivo imaging system (IVIS, BLT, USA). Data were quantified as total photon flux within a fixed region of interest (ROI) and normalized to the initial signal intensity recorded on day 0 to assess the clearance kinetics of the transplanted cells. All animal experiments were approved by the Institutional Animal Care and Use Committee of Chongqing Medical University.

4.2.13. Design and fabrication of the tri-Chamber MPS

The MPS was fabricated using a combination of 3D printing and soft lithography. Briefly, molds were designed and produced via 3D printing, into which a mixture of polydimethylsiloxane (PDMS) base and curing agent (10:1 w/w) was cast. Following degassing and thermal curing, the PDMS structures were demolded.

To establish a stable microenvironment, the chip was engineered with multiple independent compartments separated by internal partitions. Notably, the height of each chamber was designed to be 100–200 μm higher than the partition height to allow controlled fluidic communication while preventing solid-phase overflow. Within each compartment, microwells with diameters of 0.46–0.50 mm were precisely machined to accommodate cellular components.

After oxygen plasma surface treatment, the PDMS layers were aligned and irreversibly bonded to a glass substrate to yield the finalized MPS. This optimized architecture addresses common limitations in conventional microfluidic devices, such as cell detachment, disordered microsphere movement, and cross-contamination between chambers. The design ensures the stable immobilization of microspheres and prevents their migration across compartments, thereby significantly enhancing experimental stability and reproducibility.

The MPS comprises three functional zones: the OA Cartilage Chamber, the OA Synovial Chamber, and the MSC Chamber, which house chondrocyte spheroids, synoviocyte spheroids, and MSC-laden Chonichspheres, respectively. Chondrogenic medium (Procell, China) supplemented with SF was introduced into the system, flowing sequentially from the OA cartilage and synovial chambers through to the MSC chamber to establish a biomimetic pathological crosstalk.

4.2.14. Construction and experimental grouping of the tri-Chamber MPS

Primary human chondrocytes and synoviocytes were utilized to generate multicellular spheroids. Briefly, cell suspensions of synoviocytes and chondrocytes (6 × 10⁵ cells/mL) were cultured in 96-well ultra-low attachment (ULA) U-bottom plates at 37 °C with 5% CO₂. After 72 h of incubation, the resulting synoviocyte spheroids and chondrocyte spheroids were harvested and strategically placed into the synovial and cartilage chambers of the MPS, respectively.

Simultaneously, the MSC-laden microspheres (GAM@MSCs, GAM@MSCs@TGF-β3, or Chonichspheres@MSCs) were inoculated into the MSC chamber. A peristaltic pump was integrated into the MSC chamber to drive the dynamic recirculation of chondrogenic medium and SF. The system was maintained at a constant flow rate of 10 μL/min, recapitulating the physiological fluid shear stress and nutrient exchange [[89], [90], [91]].

The experimental groups within the MPS were defined as follows: Control group: Nor-synoviocyte spheroids (synovial chamber), Nor-chondrocyte spheroids (cartilage chamber), and GAM@MSCs (MSC chamber), with recirculation of pure chondrogenic medium. Blank group (OA Model): OA-synoviocyte spheroids, OA-chondrocyte spheroids, and GAM@MSCs, with recirculation of a mixture containing 90% (v/v) chondrogenic medium and 10% (v/v) SF. GAM@MSCs@TGF-β3 group: OA-synoviocyte spheroids, OA-chondrocyte spheroids, and GAM@MSCs@TGF-β3, with recirculation of 90% chondrogenic medium and 10% SF. Chonichspheres@MSCs group: OA-synoviocyte spheroids, OA-chondrocyte spheroids, and Chonichspheres@MSCs, with recirculation of 90% chondrogenic medium and 10% SF [92].

To evaluate the therapeutic efficacy of Chonichspheres within the pathological microenvironment, a multi-stage analytical timeline was established. Early-stage antioxidant response was assessed by measuring ROS levels after 24 h of dynamic culture. The activation of intracellular signaling pathways and global transcriptomic changes were investigated via Western blot and RNA-seq analysis, respectively, at the 72 h mark. Long-term chondrogenic differentiation and tissue integration were characterized following 14 days of continuous induction, utilizing RT-qPCR, immunofluorescence, and histological staining. This strategy allowed for a comprehensive understanding of the transition from immediate cytoprotection to definitive tissue regeneration.

4.2.15. ROS detection

Intracellular ROS generation was assessed using a Reactive Oxygen Species Assay Kit (Beyotime, China). Following 24 h of dynamic culture in the MPS, the bonded PDMS layers were detached, and the microsphere-cell complexes were harvested. The samples were incubated with 10 μM DCFH-DA at 37 °C for 20 min to detect ROS production. Nuclei were counterstained with Hoechst 33342 Live Cell Stain (Beyotime, China).

Fluorescence images were acquired using a laser scanning confocal microscope (Olympus, Japan). To visualize ROS levels within MSCs on the Chonichspheres, Z-stack maximum intensity projections were generated. For quantitative analysis, relative fluorescence intensity was measured using ImageJ software, with individual microspheres serving as the statistical unit. In each of the 3 independent biological replicates, 3 microspheres per group were randomly selected and analyzed. Each data point in the bar graphs represents the mean fluorescence intensity of one biological replicate. The results are presented as mean ± standard deviation (SD). Statistical differences among groups were evaluated by one-way analysis of variance (ANOVA), followed by Tukey's post hoc test.

4.2.16. Immunofluorescence staining

Following chondrogenic induction, the protein expression levels of COL2A1, SOX9, p-FAK, p-AKT, and p-mTOR were evaluated via 3D immunofluorescence staining. The microsphere-cell complexes were washed three times with PBS, fixed with 4% paraformaldehyde for 2 h, permeabilized with 0.2% Triton X-100 for 5 min, and blocked with 2% bovine serum albumin (BSA) for 30 min.

Samples were then incubated overnight at 4 °C with primary antibodies against COL2A1 (Proteintech, China), SOX9 (Proteintech, China), p-FAK (Proteintech, China), p-AKT (Abcam, USA), and p-mTOR (Abcam, USA). After washing, the samples were incubated with CoraLite488-conjugated goat anti-rabbit or anti-mouse IgG secondary antibodies (Proteintech, China) for 1 h at room temperature.

Cytoskeletal F-actin was stained with Rhodamine-Phalloidin (Solarbio, China) for 30 min, and nuclei were counterstained with DAPI (Boster, China) for 10 min. Fluorescence images were acquired using a laser scanning confocal microscope (Olympus, Japan). Z-stack maximum intensity projections were generated to visualize the protein expression in MSCs on the Chonichspheres. Quantitative analysis of relative fluorescence intensity was performed using ImageJ software. For quantitative analysis, fluorescence intensity was measured using ImageJ software, with individual microspheres as the statistical unit. In each of the 3 independent biological replicates, 3 microspheres per group were randomly selected and analyzed. Each data point in the bar graphs represents the mean fluorescence intensity of one biological replicate. The results are presented as mean ± SD. Statistical differences among groups were evaluated by one-way analysis of variance (ANOVA), followed by Tukey's post hoc test.

4.2.17. Transcriptome sequencing

Total RNA was extracted using Trizol reagent (Beyotime, China) and purified using an RNA extraction kit (Yeasen, China) according to the manufacturer's protocol. RNA concentration and purity were quantified using a spectrophotometer (A260/A280 ratio: 1.8–2.1), and RNA integrity was verified using an Agilent 2100 Bioanalyzer, ensuring an RNA Integrity Number (RIN) > 7.0.

Differential expression analysis was performed using the edgeR package in R. This tool employs a statistical method based on the negative binomial distribution to estimate biological variation and detect DEGs via exact tests. Pairwise comparisons were conducted to identify DEGs between the OA and Control groups, as well as between the OA and Treatment groups.

Subsequently, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed using the clusterProfiler R package to elucidate biological functions associated with the identified DEGs. Heatmaps were generated using the pheatmap package to visualize hierarchical clustering and expression patterns, specifically focusing on genes related to the integrin-PI3K-AKT-mTOR axis. This analysis provided a visual representation of gene expression profiles across the Control, OA, and Treatment groups, highlighting key disease-related genes and therapeutic response mechanisms.

4.2.18. RT-qPCR

Total RNA was extracted using an RNA extraction kit (Accurate Biology, China) following the manufacturer's protocol. Reverse transcription was performed using the RT Master Mix (MCE, China).

Quantitative real-time PCR (qPCR) was conducted using SYBR Green qPCR Master Mix (MCE, China) on a CFX96 Real-Time PCR Detection System (Bio-Rad, USA). The thermocycling conditions were as follows: initial denaturation at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 5 s and annealing/extension at 60 °C for 30 s. Relative gene expression levels were normalized to the housekeeping gene GAPDH using the 2^−ΔΔCt method. All experiments were performed in triplicate.

4.2.19. Western blotting

Total protein was extracted using RIPA lysis buffer supplemented with phenylmethanesulfonyl fluoride (PMSF) (Beyotime, China) on ice. Protein concentration was determined using a BCA Protein Assay Kit (Beyotime, China).

Equal amounts of protein (20 μg) were resolved by 12.5% SDS-PAGE at a constant voltage of 120 V for 1.5 h and subsequently transferred onto Polyvinylidene Fluoride (PVDF) membranes (MilliporeSigma, USA) at 100 V for 1.5 h. Membranes were blocked with 5% non-fat milk in Tris-buffered saline with Tween-20 (TBST) for 1 h and washed three times.

The membranes were then incubated with primary antibodies against PI3K, p-PI3K, AKT, p-AKT, mTOR, p-mTOR, FAK, p-FAK, and GAPDH (diluted 1:1000) overnight at 4 °C. Following three washes with TBST, the membranes were incubated with HRP-conjugated secondary antibodies (Abcam, UK) for 1 h at room temperature. Protein bands were visualized using an Enhanced Chemiluminescence (ECL) detection system (Xi'an Mishi Biotechnology, China), and band intensity was quantified using ImageJ software (NIH, Bethesda, MD, USA).

4.2.20. Mouse model of OA

All animal experiments were conducted in strict accordance with the guidelines approved by the Institutional Animal Care and Use Committee (IACUC) of Chongqing Medical University (Approval No. IACUC-CQMU-2023-0283). A knee OA (KOA) model was established in C57 mice via destabilization of the medial meniscus (DMM). Twenty-five 8-week-old male C57 mice were randomly assigned to the OA group (n = 20) and the Sham group (n = 5). Sample size per group was determined according to the standard sample size adopted in published studies of OA and cartilage repair. Considering biological individual variation among experimental animals and statistical test power, n = 5 per subgroup was sufficient for reliable intergroup statistical analysis and result evaluation.

Anesthesia was induced with 2–3% isoflurane and maintained at 1.5–2.5% via inhalation, with the depth of anesthesia continuously monitored to ensure adequate sedation. Upon achieving anesthesia, the mice were secured on a sterile surgical board, and the right knee area was shaved and disinfected with povidone-iodine and 70% ethanol. A 3–5 mm longitudinal incision was made medial to the patellar ligament.

The joint capsule was exposed via blunt dissection and opened to visualize the medial meniscotibial ligament (MMTL), which was subsequently transected to destabilize the meniscus. For the Sham group, a similar arthrotomy was performed without transecting the MMTL. The incision was sutured in layers. To prevent infection, a single dose of penicillin (50,000 U/kg) was administered subcutaneously. Analgesia was provided via buprenorphine (0.05–0.1 mg/kg) injections every 8–12 h for 24–48 h.

Animals were housed individually in warmed cages until full recovery from anesthesia and were provided with soft food and water. Over the subsequent 4 weeks, all C57 mice underwent treadmill training (5 sessions per week) using an SA101D small animal treadmill equipped with an air-puff stimulus (SA Instruments, Stony Brook, NY, USA) to accelerate OA progression.

Concurrently during the training period, the OA group (n = 20) was randomized into four subgroups to receive intra-articular injections of PBS, GAM@MSCs, GAM@MSCs@TGF-β3, or Chonichspheres@MSCs, respectively. The injection dosage was optimized based on preliminary dose-response studies to ensure safety and therapeutic efficacy. Based on previous literature utilizing similar nanoparticle delivery systems, the total injection volume was set at 20 μL per knee. Specifically, each injection contained approximately 1.2 × 10⁵ microsphere-adherent MSCs.

4.2.21. Radiographic evaluation

Eight weeks following DMM surgery, the C57 mice were euthanized, and their knee joints were harvested for radiographic assessment. High-resolution micro-CT imaging was performed using a μCT100 system (Scanco Medical, Brüttisellen, Switzerland) to evaluate joint space width, osteophyte volume, and subchondral bone density. The knee joint samples were scanned at a voltage of 55 kV and a current of 200 μA. Three-dimensional (3D) reconstruction of the skeletal structures was conducted using μCT Ray V4.2 software, while quantitative data analysis was performed using the μCT Evaluation Program V6.6.

4.2.22. Histological and immunofluorescence analysis

Following radiographic evaluation, the knee joint samples were fixed in 4% paraformaldehyde, decalcified in EDTA solution, and embedded in paraffin. Sagittal sections (5 μm thick) were prepared and stained with Hematoxylin and Eosin (H&E) and Safranin O-Fast Green to assess cartilage histopathology. The severity of OA was quantified using the Mankin scoring system.

For immunofluorescence analysis, tissue sections were incubated overnight at 4 °C with rabbit monoclonal primary antibodies against COL2A1 and Aggrecan (Abcam, USA). On the following day, the sections were washed and incubated with fluorophore-conjugated secondary antibodies for 1 h at room temperature. Cytoskeletal actin filaments were stained with Rhodamine-Phalloidin, and nuclei were counterstained with DAPI. Images were acquired using an upright fluorescence microscope (Olympus DP74, Tokyo, Japan). Quantitative analysis of fluorescence intensity and histological scoring were performed using ImageJ software (NIH, Bethesda, MD, USA).

4.2.23. Statistical analysis

The sample size for animal experiments was determined using G∗Power 3.1 software based on an a priori power analysis. With an effect size (d) of 0.8, a statistical power of 0.80, and a significance level (α) of 0.05, it was determined that five joint samples per group were sufficient to detect meaningful differences in repair outcomes. This sample size is consistent with standard practices in studies of OA and cartilage defect repair, ensuring robust statistical justification for our experimental design [93]. All experiments were performed independently at least three times (n ≥ 3). Data are presented as mean ± SD. Data collection and management were performed using Microsoft Excel (Microsoft Corporation, Redmond, WA, USA). Statistical analyses and graphing were conducted using GraphPad Prism version 8.0 (GraphPad Software, San Diego, CA, USA). Comparisons between two groups were analyzed using two-tailed unpaired Student's t-tests. For comparisons among three or more groups, one-way analysis of variance (ANOVA) was performed followed by the Tukey's post hoc test. For longitudinal data evaluated across multiple time points, statistical significance was determined using two-way repeated-measures ANOVA followed by appropriate post hoc tests. Statistical significance is indicated as follows: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001; “ns” indicates no significant difference.

Ethics approval and consent to participate

All human-related experiments were conducted in strict accordance with the Declaration of Helsinki, and ethical approval was obtained from the Ethics Committee of The First Affiliated Hospital of Chongqing Medical University (Approval No. 2026-0350-01). Informed consent was obtained from all donors.

All animal experiments were conducted in strict accordance with the guidelines approved by the Institutional Animal Care and Use Committee (IACUC) of Chongqing Medical University (Approval No. IACUC-CQMU-2023-0283).

CRediT authorship contribution statement

Jinping Chen: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Pengcheng Xiao: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing – original draft. Xingkuan Wang: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Jianye Tan: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft. Chengcheng Du: Data curation, Investigation, Visualization. Zhuolin Chen: Data curation, Investigation, Visualization. Bochen Tang: Data curation, Investigation, Validation. Pengrui Zhang: Data curation, Investigation, Visualization. Jiacheng Liu: Data curation, Investigation, Visualization. Yinsong Sun: Data curation, Investigation, Visualization. Shengwen Cheng: Data curation, Investigation, Visualization. Yichi Zhang: Data curation, Investigation, Visualization. Zhong Alan Li: Resources, Supervision, Writing – review & editing. Junyi Liao: Conceptualization, Resources, Supervision, Writing – review & editing. Wei Huang: Conceptualization, Funding acquisition, Resources, Supervision, Writing – review & editing. Yiting Lei: Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following are the Supplementary data to this article: