Adaptive hydrogel loaded with pre-coordinated stem cells for enhanced osteoarthritis therapy

Abstract

Osteoarthritis (OA) is a prevalent chronic joint disease with no currently available cure. Despite the promise of mesenchymal stromal cells (MSCs) in promoting OA management, direct intra-articular administration of MSCs faces several critical challenges, including rapid cell clearance from the joint cavity, limited survival in the hostile inflammatory environment, and insufficient control over their differentiation. In this study, we present a strategy that enhances the functionality of MSCs via pre-coordinated with Mg2+ and hypoxia-mimicking agent dimethyloxalylglycine (DMOG) integrated within an adaptive hydrogel for OA treatment. Mg2+ regulates macrophage polarization toward an anti-inflammatory phenotype, inhibits osteoclast activation, and preserves subchondral bone integrity by activating the PI3K-Akt signaling pathway. Concurrently, DMOG, activates the HIF-1α pathway, mimicking hypoxic microenvironment that support chondrocyte repair and stimulate cartilage matrix synthesis. MSCs pre-coordinated with Mg2+ and DMOG exhibit enhanced chondrogenic differentiation and immunomodulatory capacity, thus improving their regenerative potential in OA. To facilitate localized and sustained delivery, a self-healing tissue-adhesive hydrogel composed of phenylboronic acid and methacrylate-modified hyaluronic acid (HAMA-PBA) is synthesized to encapsulate the pre-coordinated MSCs. This hydrogel ensures cellular retention and functionality at the injury site. In vivo, the system significantly reduces joint inflammation, enhances cartilage regeneration, and improves joint function. Overall, these findings demonstrate a synergistic and effective stem cell-based therapeutic strategy for OA treatment through biochemical pre-conditioning and biomaterial-enabled delivery.

Headnote

ABSTRACT

Osteoarthritis (OA) is a prevalent chronic joint disease with no currently available cure. Despite the promise of mesenchymal stromal cells (MSCs) in promoting OA management, direct intra-articular administration of MSCs faces several critical challenges, including rapid cell clearance from the joint cavity, limited survival in the hostile inflammatory environment, and insufficient control over their differentiation. In this study, we present a strategy that enhances the functionality of MSCs via pre-coordinated with Mg2+ and hypoxia-mimicking agent dimethyloxalylglycine (DMOG) integrated within an adaptive hydrogel for OA treatment. Mg2+ regulates macrophage polarization toward an anti-inflammatory phenotype, inhibits osteoclast activation, and preserves subchondral bone integrity by activating the PI3K-Akt signaling pathway. Concurrently, DMOG, activates the HIF-1α pathway, mimicking hypoxic microenvironment that support chondrocyte repair and stimulate cartilage matrix synthesis. MSCs pre-coordinated with Mg2+ and DMOG exhibit enhanced chondrogenic differentiation and immunomodulatory capacity, thus improving their regenerative potential in OA. To facilitate localized and sustained delivery, a self-healing tissue-adhesive hydrogel composed of phenylboronic acid and methacrylate-modified hyaluronic acid (HAMA-PBA) is synthesized to encapsulate the pre-coordinated MSCs. This hydrogel ensures cellular retention and functionality at the injury site. In vivo, the system significantly reduces joint inflammation, enhances cartilage regeneration, and improves joint function. Overall, these findings demonstrate a synergistic and effective stem cell-based therapeutic strategy for OA treatment through biochemical pre-conditioning and biomaterial-enabled delivery.

ARTICLE INFO

Keywords:

Osteoarthritis

Cell therapy

Hypoxia

Chondrogenic

Injectable hydrogel

1. Introduction

Cartilage repair remains a formidable clinical challenge, largely due to the tissue's limited intrinsic regenerative capacity and the complex pathological milieu associated with inflammation [1-3]. Osteoarthritis (OA), a degenerative joint disorder, is characterized by progressive degradation and loss of articular cartilage, remodeling of the subchondral bone, osteophyte formation [4], and inflammatory responses within synovial tissue [5-7]. These pathological changes exacerbate joint deterioration and impede effective cartilage regeneration. Among current clinical treatments, matrix-induced autologous chondrocyte implantation (MACI) employs tissue engineering strategies but is constrained by the limited availability and poor differentiation capacity of harvested chondrocytes [8,9]. Alternatively, microfracture techniques leverage mesenchymal stem cells (MSCs) for cartilage repair through multilineage differentiation. However, this method is often compromised by fibroblast-mediated fibrous tissue formation [10,11], and the joint's inflammatory microenvironment further diminishes regenerative efficacy [12].

Scaffold-based approaches incorporating biofactors to recruit endogenous MSCs or chondrocytes offer a promising solution [13-15], yet significant hurdles persist in the selection, immobilization, and controlled release of these biofactors [16], as well as in achieving efficient cellular recruitment [17]. Moreover, inflammation may impair the lineage-specific differentiation of recruited cells [18,19]. Although MSC-based therapies have demonstrated potential for OA treatment, challenges such as cell leakage and high cell death following intra-articular injection continue to limit therapeutic outcomes [20-22]. A promising alternative involves pre-coordinated MSCs and delivering them via biomaterial matrices, enhancing their in vitro functionality and improving in vivo reparative efficacy.

MSCs are well-established for their regenerative capacity and multipotency, including their ability to differentiate into chondrocytes and osteoblasts [23]. They have been widely explored as cell sources for cartilage, bone, and osteochondral repair [24,25]. In OA, synovial MSCs have been found to migrate into the synovial fluid, with their abundance correlating with disease severity. Intra-articular injection of synovial-derived MSCs in rat models has been shown to inhibit cartilage degeneration in a frequency-dependent manner [26]. Similarly, multiple administrations of allogeneic bone marrow MSCs have attenuated joint damage in anterior cruciate ligament (ACL) - deficient OA models [27]. Notably, the need for repeated injections and the limited ability to prevent subchondral bone deterioration remain significant limitations. Furthermore, the inflammatory joint environment often impairs the regenerative capacity of transplanted MSCs.

To address these limitation, genetic and biochemical modification of BMSCs has been explored to enhance the expression of antiinflammatory cytokines or growth factors [28-30]. Among biochemical strategies, biofunctional metallic ions such as magnesium (Mg2+) have emerged as potent enhancers of stem cell activity and tissue regeneration [31,32]. Mg2+, a bioavailable and low-cytotoxic element, plays critical roles in orthopedic applications [33,34]. In bone repair, Mg2+ supports cell migration, angiogenesis, and osteogenesis, and has been shown to stimulate osteoblast proliferation and differentiation [35, 36]. Adequate Mg2+ levels are essential for maintaining bone integrity and preventing loss. In cartilage repair,Mg2+ promotes extracellular matrix (ECM) synthesis, modulates chondrocyte metabolism, and inhibits cartilage degradation [37,38]. It also exhibits immunomodulatory properties by influencing macrophage polarization and reducing oxidative stress through regulation of NF-kB and AKT signaling pathways [39-41]. Clinically, intra-articular MgCl, injections have been investigated for OA management, showing potential in improving subchondral bone quality and alleviating cartilage degeneration [42-44]. Co-administration with vitamin C has further demonstrated symptom alleviation in OA models [45]. Compared to other regulatory factors,Mg2+ is cost-effective, biocompatible, and easily incorporated into cell culture systems to enhance the chondrogenic differentiation of BMSCs.

In addition to biochemical cues, hypoxic stimulation plays a critical role in promoting cartilage regeneration, as native articular cartilage exists in a hypoxic microenvironment (pO; - 1-2 %) [46]. Hypoxia is essential for chondrogenesis [47,48], and studies have shown that low oxygen tension enhances chondrocyte density, matrix production [49, 50], and promotes the chondrogenic differentiation of MSCs [51,52]. However, implementing gas-controlled hypoxic systems can be technically demanding, prompting the use of hypoxia mimetics as practical alternatives.

Dimethyloxalylglycine (DMOG), a prolyl hydroxylase inhibitor, stabilizes hypoxia-inducible factor-1œ (HIF-1a) by inhibiting its degradation under normoxic conditions, thereby mimicking hypoxia [53-55]. HIF-10 is a key regulator of cellular adaptation to low oxygen, exerting both anabolic and anti-catabolic effects in chondrocytes and making it a promising therapeutic target in OA [56,57]. Compared with other hypoxia mimics such as Co2+ and deferoxamine (DFO), DMOG offers distinct advantages, including chemical stability, potent bioactivity, high safety, and ease of use. Notably, DMOG can inhibit both PHD2 and FIH, thereby enhancing HIF signaling and promoting the chondrogenic differentiation. Furthermore, studies have shown that DMOG induced a more chondrogenic transcriptional profile than the other such HIF stabilizing compounds [58]. Therefore, DMOG represents a promising agent for improving the functionality of BMSCs in OA stem cell therapy.

Building on these insights, this study aims to pre-coordinate BMSCs withMg2+ and DMOG to enhance their anti-inflammatory properties and chondrogenic differentiation. These pre-coordinated cells are then encapsulated within an injectable hydrogel for intra-articular delivery to address OA-related cartilage degeneration (Scheme 1). The hydrogel is synthesized from phenylboronic acid and methacrylate-modified hyaluronic acid (HAMA-PBA), forming a dual-crosslinked network consisting of covalent crosslink (ie., photo crosslinked methacrylate) and dynamic crosslink (ie., reversible interaction between phenylboronic and polyhydroxy groups). These features impart hydrogel with selfhealing and tissue-adhesive properties, facilitating the retention of BMSCs at the defect site and minimizing clearance by adhering to cartilage tissue. In vivo, at early stage, a mixture of Mg2+/DMOG is coinjected during the early post-operative phase to mitigate inflammatory macrophage activity, thereby preserving the functionality of the pre-coordinated BMSCs. Collectively, this strategy presents an effective approach for stem cell-mediated cartilage repair, with the potential to reduce cartilage degradation and osteophyte formation while promoting chondrogenesis in OA-affected joints.

2. Materials and methods

2.1. Effects of Mg", DMOG, Co2+, and DFO on chondrocyte behaviors

2.1.1. Proliferative activity

Rat-derived chondrocytes were used in this study. The chondrocyte culture medium was replenished with DMEM/F12 (Solarbio, China), 10 % FBS (Gibco, Invitrogen, USA), and 1 % penicillin-streptomycin (HyClone, Thermo, USA). The conditioned medium was prepared based on chondrocyte culture medium with different concentrations of Mg2+ (MgCl,, Aladdin, China), DMOG (APE x BIO, USA), Co?" (C°Cl,, Aladdin, China), and DFO (Sigma, USA), respectively. The chondrocytes were cultured to P2 at 5000 cells, in which the tissue culture plates (TCPs) control group was cultured using chondrocyte culture medium, and the experimental group was cultured with the conditioned medium, respectively. The incubation atmosphere contained 95 % air and 5 % CO, at 37 °C. The proliferation of chondrocytes in different culture mediums was investigated using CCK-8 test. The proliferation curves obtained were normalized to the TCPs group to obtain the heatmap of cellular viability.

2.1.2. Characterization of chondrocyte morphology

The growth status of chondrocytes was further visualized by immunofluorescence staining with Rhodamine - Phalloidine (Solarbio, China) and DAPI (Solarbio, China). The main operation process is briefly described as follows. The chondrocytes were fixed with 4 % paraformaldehyde (Aladdin, China) for 1 h. After fixation, the samples were rinsed three times with PBS (Hyclone, USA) and permeabilized using 0.1 % Triton-X 100 (Sigma, USA) for 10 min. Then, cells were blocked with 1 % BSA (Solarbio, China) for 1 h. The cytoskeleton was visualized by staining the cellular samples with staining solutions, and images were captured with confocal laser scanning microscope (CLSM, TCS SP5-X, Leica, Germany).

2.1.3. Cell scratch test

Chondrocytes were seeded at 5 x 10° cells/mL in 6-well plates. When cells reached to 90 % confluence, a straight line was drawn with a 200 pL pipet tip in the well center. Meanwhile, the medium in the well was replaced with FBS-free medium for starvation treatment. After 12 h, the cells were treated with conditioned medium with the particular concentration of Mg2+, DMOG, Co··, and DFO for another 12 or 24 h, then the nuclei were stained with DAPI and imaged with CLSM. Based on the fluorescent images, cell migrations in different cases were quantitatively compared using ImageJ software.

2.1.4. Chondrogenic differentiation

Rat-derived BMSCs were used in this study. Chondrogenic differentiation was induced in BMSCs using a chondrogenic induction medium supplemented with Mg2+, DMOG, Co··, and DFO. The induction medium comprised DMEM/F12 containing 10 % FBS, 50 pm ascorbic acid (Sigma, USA), 1 % ITS (Sigma, USA), 100 nM Dex (Sigma, USA), and 10 ng/mL transforming growth factor 83 (ТСЕ-ВЗ) (Bio-Techne, USA). BMSCs were seeded at a density of 2000 cells/well in a 24-well culture plate and cultured for 14 d. Cells cultured in chondrogenic induction medium without supplementation of Mg2+, DMOG, Co?·, or DFO served as control.

To evaluate the efficacy of the specific concentration of Mg2+, DMOG, Co?+, and DFO, Alcian blue solution (Solarbio, China) was used to stain the glycosaminoglycan (GAG) at day 14. Additionally, immunofluorescence staining results of HIF-1a/SOX9/COL-II were used to observe the chondrogenic differentiation of BMSCs after 14 d.

At day 3, 7, and 14, the medium in each well was discarded and the cells were lysed with lysis buffer (containing 1 % Triton X-100, 200 mM Tris, and 150 mM NaCl), followed by three cycles of freezing and thawing. Then, the protein expression level of HIF-1a and COL-II were quantified using ELISA kits (Bluegene, China). Meanwhile, samples at day 3,7 and 14 were collected for gene expression detection by qRT-PCR (TaKaRa, USA), including the gene expression of SOX9 and ACAN for checking the differentiation of BMSCs in Mg2+/DMOG/Co··/DFOconditioned medium, as well as BCL-2 and AGO for checking the apoptosis of BMSCs. The main steps were as follows: 500 pL RNA-Quick Purification kit (Esunbio, China) was added into wells to extract the total RNA for quantitative reverse transcription polymerase chain reaction (qRT-PCR) assay in ABI PRISM 7500 Real-Time PCR system (Applied Biosystems, USA). The genes (SOX9, ACAN, BCL-2, and AGO) were evaluated by 2-44 method using a housekeeping gene GAPDH as reference, and their specific primers were designed as listed in Table S1.

2.2. Evaluation of combination effects

The combination effects of Mg2+ and three other hypoxia mimics (DMOG, Co··, and DFO) at the particular concentration on cytocompatibility and chondrogenesis were evaluated. Cytocompatibility was assessed using a lactate dehydrogenase (LDH) assay (Beyotime, China). Briefly, BMSCs were seeded into 96-well culture plates at a density of 2 x 10° cells/well and cultured for 24 h to ensure cell attachment and spreading. Subsequently, the culture medium was replaced with a conditioned medium (200 pL/well) supplemented with specific concentrations of Mg2+ and three other hypoxia mimics (DMOG, Co?·, and DFO), followed by incubation for an additional 24 h. The conditioned media was then collected and added with the LDH working solution. After incubation for 30 min, a stop solution was added, and optical density (OD) values were measured at 490 nm using a microplate reader. Dead cell ratios were calculated according to the manufacturer's protocol.

To evaluate chondrogenesis, BMSCs were seeded at a density of 1 x 10 cells/well into 24-well plates and cultured for 2 days until cell proliferation exceeded 80 %. The medium was subsequently replaced with a two-component conditioned medium containingMg2+ and one hypoxia mimic (DMOG, Co··, or DFO) to induce chondrogenesis. Cells were cultured under these inductive conditions for 14 days, with the medium refreshed every two days. Comparative chondrogenic induction was assessed via immunofluorescence staining for HIF-1x and SOX9 (Santa Cruz, USA). At predetermined intervals, cells were lysed with 200 pL of cell lysis buffer, followed by three cycles of freezing and thawing. The expression levels of HIF-1a and SOX9 were quantitatively determined using ELISA kit, using BSA as the standard. For quantitative reverse transcription-polymerase chain reaction (qRT-PCR) analysis, total RNA was extracted at predetermined time points (day 3, 7, and 14) using the RNA-Quick Purification kit (Esunbio, China). Four genes (HIF10, ACAN, OCT4, and TNF-a) were evaluated, with specific primers sequences detailed in Table S2.

2.3. Investigation for the mechanism of Mg2+ synergizing DMOG in inducing chondrogenic differentiation through RNA-seq analysis

The BMSCs were cultured in the chondrogenic induction medium containing 200 ppmMg2+ and 50 ppm DMOG for 14 days. In addition, a chondrogenic differentiation medium containing only Mg2+ or DMOG was used as a control. The total RNA was isolated with Trizol regent for labeling reactions. The next-generation sequencing was performed by the Beijing Genomics Institute (Shenzhen, China) as per the protocol of the manufacture. The sequencing data was filtered with SOAP-nuke (v1.5.2). The clean reads were mapped to the reference genome using HISAT?2 (v2.0.4). Bowitie 2 (v2.2.5) was applied to align the clean reads to the reference coding gene set, then the gene expression levels were calculated by RSEM (v1.2.12). Essentially, different expression analysis was performed using the DESeq2 (v1.4.5) with |log,FC| > 0 and Q-value <0.05. The cluster profiler function package in R language was used for enrichment analysis of Gene Ontology (GO analysis). Significant enrichment of GO pathway was screened by p-value <0.05. Gene Set Enrichment Analysis was used by plotting with the Dr. Tom platform (htt ps://biosys.bgi.com) provided from BGI.

2.4. Investigation for the anti-inflammatory activity of macrophages upregulated by Ме?" and hypoxia mimics

RAW264.7 were seeded at a density of 1 x 10· cells/well in a 24-well plate and then supplemented with M1 inflammatory phenotype induction medium (High glucose DMEM, 10 % FBS, 1 % penicillinstreptomycin and 1 pg/mL lipopolysaccharide (LPS, Solarbio, China)). This was employed as both a negative control group and a positive control. The negative group was RAW264.7 cultivated with macrophage cell culture media that did not include LPS. RAW264.7 incubated with M1 inflammatory induction medium was used as a positive control. Furthermore, 200 ppmMg2+ and 50 ppm DMOG were added to the M1 inflammatory phenotype induction medium as the experimental group. After 3 days of growth, the samples from each group were harvested and processed for nuclear and cytoplasmic staining, as well as immunofluorescence staining using iNOS (Affinity, USA) and Arg-1 (Affinity, USA) antibodies. Subsequently, these samples underwent flow cytometry testing, and qRT-PCR testing to evaluate the expression of M1 phenotype-related genes (IL-1ß and IL-6) and M2 phenotype-related genes (IL-4 and IL-10). The specific primers for these genes were meticulously designed and are listed in Table S3.

2.5. Isolation and culture of chondrocytes from clinical specimens

Human OA cartilage samples were obtained from the knee joint of patients undergoing total knee arthroplasty. All protocols were approved by the Ethics Committee of Peking University Third Hospital (Number: (2022) 581-02). Informed consent was obtained from all donors prior to tissue collection. The human-derived chondrocytes (hCHO) were isolated from the knee joints of donors of patients by digestion with 0.25 % trypsin (Sigma, USA) for 20 min, followed by 0.4 % type II collagenase (BioFroxx, Germany) for 5 h. Then, the isolated chondrocytes were cultured to approximately 90 % confluency, followed by the replacement of control medium or conditioned medium. The control medium consisted of DMEM/F12 (Solarbio, China) supplemented with 10 % FBS and 1 % penicillin-streptomycin.

Upon reaching confluency, cells were washed with PBS and then cultured in either control or experimental conditioned media. The conditioned medium was formulated by supplementing the control medium with 200 ppmMg2+ and 50 ppm DMOG. For comparative analysis, parallel groups treated with either 200 ppm Mg"? alone or 50 ppm DMOG alone were included. Media were refreshed every two days. After 7 days of culture, OA chondrocytes were qualitatively tested by cytoplasmic staining and immunocytochemical staining of HIF-1a and MMP13, and quantitatively tested by qRT-PCR for specific genes related to chondrocyte phenotype, COL-II, SOX9, ACAN and HIF-10, and for cartilage matrix degradation indicators, MMP13, and inflammatory phenotype-related factors, IL-18, CCL2 and TNF-a. The specific primers for these genes were meticulously designed and are listed in Table S4.

2.6. Synthesis of methacrylic acid modified-hyaluronic acid (HAMA) and phenylboronic acid-modified HAMA (HAMA-PBA)

Methacrylated hyaluronic acid (HAMA) was synthesized via esterification of hyaluronic acid (HA) with methacrylic anhydride (MA). Briefly, 1 g of HA (MW = 100,000, Heowns, China) was dissolved in 100 mL of deionized water and cooled in ice water bath. The pH of the mixed solution was adjusted to 8 using NaOH. Under continuous agitation, methacrylic anhydride was added dropwise. After 24 h of reaction, the product was subjected to dialysis (molecular weight cut-off: 1000 Da) against deionized water for 3 days. The dialyzed product was subsequently lyophilized to yield a white, sponge-like HAMA powder.

To prepare phenylboronic acid-modified HAMA (HAMA-PBA), 0.5 g of HAMA was dissolved in 150 mL of MES buffer. Subsequently, 0.7 g (0.25 mmol) of 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholinium chloride (ОМТММ; Aladdin, China) was added to activate the carboxyl groups of HAMA, and the solution was stirred for 30 min. Then, 0.057 g (0.416 mmol) of 3-aminophenylboronic acid (PBA; Aladdin, China) was added, and the reaction was allowed to proceed under continuous stirring for 24 h. The resulting product was dialyzed extensively against deionized water for 3 days, followed by lyophilization to obtain the HAMA-PBA conjugate. The final product was characterized by 'H NMR (Brucker AV400, Germany).

2.7. Characterization of HAMA and HAMA-PBA hydrogels

Hydrogels were lyophilized for morphological and swellingdegradation test examination. The two groups of hydrogels (HP-PBS and HP-DMEM) with different degrees of cross-linking were obtained by dissolving the HAMA-PBA in PBS buffer (HP-PBS) and DMEM highglucose medium (HP-DMEM), respectively, at a concentration of 3 % in a 37°C-water bath. HAMA hydrogel was obtained as the control group at the same concentration and preparation as the HP hydrogel group. The hydrogel abbreviations were shown in Table S5. The interior morphology in hydrogels were observed by SEM, and pore size was determined by using ImageJ software. The Mass gain ratio of hydrogels was determined from the weight changes after soaking in PBS for 40 days at 37 °C.

Compressive modulus was tested on water-swollen hydrogels with a cylindrical shape using an Instron universal testing machine (Instron 9500, USA). The samples were compressed at a speed of 1 mm/min and the uniaxial compression was evaluated from О % strain to 40 % strain.

The rheological properties of the HP-PBS and HP-DMEM hydrogels were evaluated using a rheometer (AR-2, TA, USA). The dynamic frequency sweep was conducted from 0.1 to 100 rad/s at 1 % strain, and the dynamic strain sweep was performed from 0.1 to 100 % strain at the speed of 10 rad/s. The storage modulus (G') and the loss modulus (G") were plotted against strain or angular frequency.

2.8. Hemolysis test

To conduct the hemolysis test, mouse blood was centrifuged at 1000 rpm for 10 min to obtain red blood cells. The red blood cells were then washed three times with PBS. The purified red blood cell concentration was diluted to 20 % (v/v). To create the hydrogel scaffold, 200 pL of HAMA, HP-PBS, or HP-DMEM precursor solution was added into the Teflon mold (ф8 mm x 4 mm) and cross-linked with UV light. After allowing HAMA /HP-PBS/HP-DMEM hydrogels at 37 °C, 1 mL of the red blood cell suspension was added to a 1.5 mL test tube and gently mixed by pipetting. The sample was then incubated at 37 °C for 1.0 h and subsequently released. The clear supernatant was measured using a microplate reader at 540 nm. Triton X-100 (0.1 %) and PBS were used as positive and negative controls, respectively. Each sample was copied three times during the test. The percentage of hemolysis was calculated using the following equation.

...

An, Ap, and A; represent the fractions of absorbance values for the supernatant samples, negative control (PBS), and positive control (Triton X-100), respectively.

2.9. 3D cell culture in hydrogel

BMSCs were collected from SD rats and cultured to the third passage for subsequent use. Hydrogel precursor solutions were made by dissolving the materials in DMEM (Hyclone, USA), sterilized by filtration via a 0.22 pm membrane, and then the cells (1 million/mL) were dispersed into the solutions, followed by being transferred into plate wells and cross-linking with UV light. In all of the cell cultures, the medium was refreshed every 2 days.

The condition of the cells growing inside the hydrogel was assessed using the live/dead staining. Briefly, at day 1 and 3, hydrogels were taken out to be stained with AM and PI (Aladdin, China) and imaged with CLSM. Living cells were represented by green fluorescence (AM, 500-550 nm), dead cells by red fluorescence (PI, 570-620 nm). In addition, a red hydrogel dye that reacts with olefins was added to the precursor solutions. After co-culturing with the cells for 3 days, the cells were stained with AM and PI, then inspected and photographed using CLSM. Additionally, to evaluate the protective efficacy of hydrogels against shear stress-induced damage to stem cells, we conducted live/ dead staining on samples before and after syringe injection of BMSCsloaded HP-DEME hydrogels. The stained samples were subsequently analyzed by 3D photography using CLSM.

2.10. Pre-coordinated BMSCs in hydrogels

The M + D conditioned medium was created by combining 200 ppmMg2+ and 50 ppm DMOG into the chondrogenic differentiation medium. BMSCs were stimulated for 14 days before being digested and loaded into HAMA hydrogel (H + Sti-C) and HP-DMEM hydrogel (HP + Sti-C), respectively. Cells cultured in general culture medium that had not been stimulated were imposed in HAMA hydrogel (H + C) and HP-DMEM hydrogel (HP + C) as control. After 4 weeks of culture in hydrogels, cells were stained with Alcian blue (Solarbio, China) and immunostained with antibodies against COL-II and HIF-1a. 3D confocal images were acquired with a CLSM. Table 56 listed the cell processing methods and abbreviations. Additionally, Table S7 provided specific grouping of hydrogel-cell complexes for further clarification.

2.11. Subcutaneous implantation

All animal experiments were performed at the Sinoresearch (Beijing) Biotechnology Co., Ltd., which were approved by the Animal Care and Use Committee of Peking University Third Hospital and followed the international standards on animal welfare (Approval number: ZYZC2022-1012). The BMSCs embedded hydrogel discs (p8 mm x 4 mm, 1 million cells/mL) were prepared as aforementioned. On the back of the rats, four separated sites (left-up, left-bottom, right-up, rightbottom) were ready for sample implantation after hair removal and sterilized with iodophor. The rats were randomly divided into 4 groups with different samples being implanted: (i) H + C, (ii) HP + C, (iii) H + Sti-C, and (iv) HP + Sti-C. At 4 weeks postoperation, 4 rats in each group were sacrificed to collect the tissues containing implants for evaluation. The retrieved tissues were fixed in 10 % neutral formaldehyde buffer for 24 h, embedded with OCT and sliced into 5 mm thickness slices for histological and immunofluorescence staining for HIF-1la and COL-II expressions, and fluorescence images were taken with CLSM. The histological staining, including Hematoxylin and Eosin (H&E, Beyotime, China) and Alcian blue staining were performed following the manufacturer's instructions. Images were recorded on a digital slice scanning device (Nanozoomer, Japan).

2.12. OA model in rats

To induce rat OA model, 40 male SD rats weighting 250-300 g were subjected to Anterior Cruciate Ligament Transection (ACLT) and Destabilization of the Medial Meniscus (DMM). Briefly, after routine anesthetic (isoflurane 2.5 % maintenance) and surgical preparation, the two hindlimbs of each rat were draped in a sterile manner, and a medial parapatellar incision was created to expose the knee joint, followed by a longitudinal incision to open the joint capsule. A lateral dislocation of the patella was performed, and the fat pad was removed to further expand the joint space. When the ACL and medial meniscus were exposed, the Sham group was complete. For animals subjected to ACLT and DMM, C and Sti-C cell suspensions were encapsulated within HPDMEM hydrogel and subsequently injected into the joint cavity. After the hydrogels had self-adapted within the joint cavity, they were crosslinked using a handheld UV curing apparatus. The patella was gently repositioned, and the joint capsule and skin incisions were sutured. Starting on the day of surgery, sterile Mg2++DMOG (M + D) injection was administered into the joint cavity once weekly for a consecutive period of four weeks, and another group with saline injections were simultaneously conducted as comparisons to the corresponding M + D injection groups.

All rats were treated with prophylactic antibiotics for 7 days, and allowed unrestricted movement in their individual cages. Rats were randomly divided into seven groups: (i) placebo surgery (Sham, positive Control), Saline treatment: (ii) ACLT + DMM (OA modeled, negative Control), (iii) OA with HP + C (Saline, HP + C), (iv) OA with HP + Sti-C (Saline, HP + Sti-C); M + D treatment: (v) OA (M + D, Control), (vi) OA with HP + C (M + D, HP + C), (vii) OA with HP + Sti-C (M + D, HP + StiC). 4 and 8 weeks postoperatively respectively, all rats were euthanized, and the knee joints were harvested. The specific experimental grouping and corresponding methodological details for animal experiment are fully elucidated in the Supplementary Information, and outlined in Table S8.

2.13. MRI and Micro-CT analysis in vivo

Owing to the stringent freshness requirements for magnetic resonance imaging (MRI) analysis, one animal specimen was randomly selected from each cohort of five parallel biological replicates at both week 4 and week 8 for MRI-based assessment (Siemens, Erlangen, Germany). Each knee joint was harvested from the remaining rats, fixed in 10 % neutral buffered formaldehyde (NOVON, China) for 24h at R.T., and subjected to Micro-CT scanner (Bruker Skyscan 1176, Belgium). Micro-CT scanning and reconstruction are used to assess the knee joint conditions and condition of osteophyte formation. Then, bone structure parameters including the new bone volume to total volume (BV/TV), bone mineral density (BMD), trabecular thickness (Tb. Th), and trabecular number (Tb. N) was analyzed using CTAn software (Version 1.14).

2.14. Histological and immunofluorescence staining in vivo

The harvested knee joints were fixed in 10 % neutral buffered formaldehyde for 72 h at R.T., and decalcified in 10 % formic acid solution with 10 % neutral buffered formaldehyde at R.T. The decalcified joints were trimmed, dehydrated in a graded ethanol series, and embedded in paraffin. Serial sections (8 pm thick) through the center of the repair site were cut and characterized by histological analysis and immunohistochemistry staining. The sections were subjected to H&E, Safranin O/fast green (Saf O/fast green, Beyotime, China), and immunohistochemical staining for COL-II and MMP13. Pictures of medical imaging modalities and immunohistochemical stains were blindly and independently scored by four evaluators, according to the OARSI criteria and Mankin Score system (Tables 59 and S10). TRAP staining was used to characterize the presence of osteoclasts in subchondral bone plates. The procedure for staining is as described in the TRAP staining instructions (Solarbio, China). Picrosirius Red was utilized to stain collagen and directionality plugin from ImageJ was used to determine the orientation distribution of the collagen fibers observed in the superficial and deeper zones of articular cartilage.

Immunofluorescence staining of iNOS and Arg-1 was performed on the joint sample sections to evaluate the polarization behavior of macrophages in OA rat. Additionally, quantitative analysis of immunofluorescence staining was performed using Image Pro Plus software.

2.15. Analysis of osteoclast-chondrocyte interaction in vitro

The bone marrow mononuclear cells (BMMCs) were collected from C57BL/6J mice and cultivated to a sufficient concentration in M-CSF (MedChemExpress, MCE, USA) factor-containing medium before use. BMMCs were seeded at a density of 1 x 10° cells/well in 24-well plate and then supplemented with macrophage medium-based osteoclast differentiation medium containing 100 ng/mL RANKL (MedChemExpress, MCE, USA) as the control group. The macrophage medium, in turn, was formulated using High glucose DMEM, enriched with 10 % FBS and 1 % penicillin-streptomycin. In addition, the М + D conditioned medium was created by combining 200 ppm Mg2+ and 50 ppm DMOG into the osteoclast differentiation medium. After 5 days of growth, the cells were assessed qualitatively and quantitatively using skeleton staining, cytotoxicity, the osteoclast marker TRAP, and qRT-PCR for osteoclast-associated TRAP, MMP9, and Cathepsin K.

To investigate the effect of chondrocytes induced with M + Dconditioned medium on the inhibition of BMMCs differentiation into osteoclasts. BMMCs at a density of 1 x 10° cells/well were seeded into the upper chamber of Transwell plates, and the chamber was supplied with osteoclast differentiation medium. The lower chamber was seeded with chondrocytes at a density of 1 x 10· cells/well, and the M + Dconditioned medium was supplied to each of the sample groups, respectively. The chondrocyte medium with no modifications was used as a control. The medium was changed every other day, and osteoclasts in the upper chamber of the Transwell were examined qualitatively and quantitatively for TRAP on the day 7 of culture. Chondrocytes in the lower chamber were stained with COL-II, ALP, and COL-X. Fluorescence distribution statistics of cells in COL-II and COL-X immunofluorescence labeling was analyzed using ImageJ software.

2.16. Statistical analysis

All the experimental analyses in this study were performed at least three independent times. The experimental data are expressed as the mean + standard deviation (п > 4). The data provided in this study are representative. Statistical analyses were performed using unpaired T - tests for comparison of two groups of samples and ANOVA for multiple groups. Differences were considered significant at ·p < 0.05, ··p < 0.01, ···D < 0.001, ····p < 0.0001. All statistical analysis was performed using IBM SPSS Statistics 25.

3. Results and discussion

3.1. Regulation effects of Mg2+ and hypoxia mimics on chondrocytes

In this study,Mg2+ and DMOG were selected to pre-coordinate BMSCs for joint cavity injection. To determine appropriate working concentrations, preliminary experiments were conducted to evaluate the effects ofMg2+ and DMOG on chondrocyte behavior, with a focus on optimizing efficacy while minimizing cytotoxicity. For comparative analysis, other hypoxia mimetics - cobalt ions (Co··) and deferoxamine (DFO) - were also tested to underscore the advantages of DMOG. Given the inherent sensitivity of chondrocytes to oxygen tension, they were employed as a representative model for assessing the bioactivity of these compounds. Chondrocyte proliferation displayed a clear dosedependent response (Fig. S1).Mg2+ exhibited minimal cytotoxicity even at concentrations up to 500 ppm, followed by DMOG up to 400 ppm. In contrast, Co?· and DFO significantly reduced cell viability at concentrations of 0.5 ppm and 1 ppm, respectively. Based on these findings, Mg2+ (200 ppm), DMOG (50 ppm), Co"? (0.1 ppm), and DFO (0.5 ppm) were selected for subsequent cell culture studies, ensuring cell survival remained above 80 % in all the cases (Fig. 52).

Chondrocytes were cultured in medium supplemented with Mg2+, DMOG, Co··, or DFO to conduct morphological observation and chondrogenic-related markers quantification (Fig. 1A-J). Under standard culture conditions, chondrocytes displayed an elongated fibroblastlike morphology at low density, progressing to a characteristic "paving brick" morphology upon reaching confluence, with individual chondrocyte adopting a rounded shape (Fig. 1B). Treatment with Mg2+, DMOG, or Co?" did not significantly alter the morphology at confluence. However, chondrocytes exposed to DFO exhibited notable morphological dedifferentiation by day 7. Quantification of cell aspect ratios indicated that Mg2+-treated cells retained a morphology closer to native chondrocytes, while DFO-treated cells deviated from 1 (Fig. 1D). Cell migration assessed via a scratch assay revealed thatMg2+ most strongly promoted chondrocyte migration, followed by the DMOG group, then the DFO and Co·· groups (Fig. 53). Glycosaminoglycan (GAG) production, measured by Alcian blue staining on day 7, was highest in the Mg"? group, followed by DMOG and Co··, while DFO-treated cells showed the least GAG production, reflecting a dedifferentiated phenotype (Fig. 1B). Immunofluorescence analysis of HIF-1a, SOX9, and COL-II (Fig. 1C) coupled with corresponding ELISA quantification (Fig. 1E and F) reveal that Mg"? strongly enhanced expressions of all three markers, with the order of Mg? > DMOG > DFO > Co··. qRT-PCR analysis on chondrogenic-related genes (SOX9, ACAN), as well as, on apoptosisrelated genes (BCL-2, AGO) further confirmed the superior performance ofMg2+ and DMOG in maintaining chondrocyte phenotype (Fig. 1G-J). BCL-2, an anti-apoptotic factor, was upregulated by both Mg2+ and DMOG (Fig. 11) [59], while AGO, a pro-apoptotic gene, was downregulated in these groups compared to controls (Fig. 1J) [60]. In summary, Mg"? can enhance chondrocyte-related phenotypic matrix synthesis and secretion by regulating HIF-1a, and has the characteristics of maintaining cell viability and inhibiting apoptosis. Different from Co?" and DFO, the hypoxia mimic DMOG has low cytotoxicity, and obvious enhancement effects on biological responses of chondrocytes.

To explore potential synergistic effects,Mg2+ was co-administered with DMOG, Co··, or DFO to assess their combined impact on BMSC activity. LDH assay reveals elevated cytotoxicity (-20 %) only in theMg2+ +DFO group, while the Mg?·+DMOG and Mg2++Co·· combinations maintained higher viability (Fig. 54). Co-treatment with Mg2+ and hypoxia mimetics enhanced the expression of HIF-la and SOX9 in BMSCs, with the Mg2++DMOG group (M + D) showing the strongest immunofluorescence signal (Fig. 1L), corroborated by ELISA results (Fig. S5). qRT-PCR further confirmed that the Mg2++DFO group exhibited the highest expression of HIF-1a, ACAN, and OCT4 (a stemness marker), along with the lowest expression of tumor necrosis factoralpha (TNF-a), an inflammation-associated gene (Fig. 1M). These results suggest thatMg2+ and DMOG synergistically enhance both chondrogenic differentiation and stemness retention in BMSCs while mitigating pro-inflammatory responses. Consequently, the M + D group also demonstrated superior performance in promoting BMSC chondrogenesis. (Fig. 1N). Based on these findings,Mg2+ (200 ppm) and DMOG (50 ppm) were selected for the pre-coordination of BMSCs in subsequent mechanistic investigations and for in vivo intra-articular injection studies targeting OA-related cartilage repair.

3.2. Evaluation ofMg2+ +DMOG on BMSCs chondrogenic differentiation

Transcriptome analysis was conducted to investigate differential gene expression in BMSCs treated with the M + D combination, using Mg2+ or DMOG treated cells as respective controls (Fig. 2). Principal component analysis of biological replicates demonstrated a good dispersion between each entry, suggesting the dependability of experimental data (Fig. S6). Comparative analysis revealed that both Mg2+ and M + D treatments induced a greater number of cartilage-related differentially expressed genes (DEGs) than DMOG alone (Fig. S7). Notably,Mg2+ exhibited a stronger regulatory effect than DMOG on genes associated with ion regulation, cartilage matrix synthesis, and antiinflammatory responses (Fig. 2B). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis showed significant upregulation of the ECM receptor-related pathway in theMg2+ -treated groups, consistent with enhanced cell adhesion and matrix interactions. Enrichment of the HIF-la signaling pathway was also observed, particularly in theMg2+ and M + D groups (Fig. 2C). Additionally, the PI3K-Akt signaling pathway - often implicated in inflammatory processes - was significantly downregulated in response to Mg2+.

Gene ontology (GO) enrichment analysis (Fig. 2D) further supported these findings: the Mg"? group exhibited upregulation of genes involved in cell migration, adhesion, and ECM production, particularly in response to hypoxic stimuli. The DMOG group predominantly activated genes linked to hypoxia signaling, promoting both motility and ECM assembly. Importantly, the M + D group demonstrated a synergistic enhancement across these processes, particularly in augmenting ECM synthesis and hypoxia responsiveness. Moreover, the M + D treatment more effectively maintained BMSC stemness-related gene expression compared to either treatment alone (Fig. 2E). Gene Set Enrichment Analysis (GSEA) revealed stronger enrichment scores and greater pathway deviation in the M + D group within the HIF-1a and TGF-B signaling pathways compared to control (Fig. 2F and G), reinforcing the synergistic regulatory effects of the combined treatment on chondrogenic differentiation. Further analysis focused on genes involved in three core biological functions: inflammatory response (PI3K-Akt, Rap1 signaling), immune modulation (MAPK, Ras signaling), and cartilage development (proteoglycan biosynthesis, focal adhesion). The results showed functional overlap among these pathways (Fig. 2H). Heatmap visualization of representative genes involved in cartilage differentiation and hypoxia signaling (Fig. 21) showed that Mg"? primarily upregulated cartilage-associated genes, whereas DMOG predominantly enhanced hypoxia-related gene expression. Strikingly, the М + D combination induced simultaneous upregulation of both gene categories, thereby promoting BMSC chondrogenic differentiation through activation of the hypoxia signaling axis.

3.3. Evaluation ofMg2+ +DMOG in suppressing inflammation and enhancing cartilage-related matrix secretion

The crosstalk between macrophages and chondrocytes is of paramount importance in the progression of OA. To evaluate the immunomodulatory potential ofMg2+ and DMOG, RAW264.7 macrophages were polarized toward a pro-inflammatory M1 phenotype via lipopolysaccharide (LPS) stimulation, followed by treatment with conditioned media containing Mg", DMOG, or their combination (M + D). The LPSonly group served as the inflammatory control (Control), while non- stimulated RAW264.7 cells maintained in growth medium constituted the MO baseline (Medium group). As depicted in Fig. 3A, LPS-treated macrophages exhibited hallmark M1 features, including strong inducible nitric oxide synthase (iNOS) fluorescence, weak expression of the M2-associated marker arginase-1 (Arg-1), altered cell morphology, and diminished adhesion. Upon treatment with Mg2+, DMOG, or M + D, a reduction in iNOS expression and a concurrent increase in Arg-1 fluorescence were observed (Fig. 3B), indicating a shift toward the antiinflammatory M2 phenotype. Notably, the M + D group demonstrated the most pronounced Arg-1 expression and near-complete suppression of iNOS, closely resembling non-inflamed macrophage behavior. Flow cytometry analysis further validated these findings (Fig. 3C and D). In the M + D group, the proportion of CCR7· M1 macrophages decreased from 8.01 % to 4.67 %, while CD206· M2 macrophages increased from 23.35 % to 31.72 %. While bothMg2+ and DMOG individually influenced macrophage polarization, the combined treatment elicited a more substantial shift. qRT-PCR results (Fig. ЗЕ-Н) confirmed a marked reduction in the expression of pro-inflammatory cytokines IL-1p and IL6, alongside upregulation of anti-inflammatory cytokines IL-4 and IL-10 in the M + D group. These outcomes suggest that activation of hypoxia signaling contributes to the enhanced anti-inflammatory effects of the M + D combination.

To further investigate the therapeutic potential of the M + D treatment in OA, its effects on primary human chondrocytes (hCHOs) derived from OA cartilage tissues (harvested during total knee arthroplasty) were evaluated (Fig. 31). hCHOs were cultured for 7 days in media supplemented with Mg2+, DMOG, or M + D and analyzed for morphological changes and expression of HIF-1a and matrix metalloproteinase13 (MMP13), a marker of cartilage degradation. As shown in Fig. 3J, control hCHOs displayed an elongated, fibroblast-like morphologytypical of inflammation-induced dedifferentiation-accompanied by increased F-actin staining. In contrast, cells treated with Mg", DMOG, or M + D exhibited reduced cell aspect ratios and spreading areas (Fig. ЗК and L), with theMg2+ and M + D groups showing the most effective restoration of the rounded chondrocyte phenotype. Fluorescence staining revealed that DMOG primarily enhanced HIF-1« expression, while М + D treatment resulted in the strongest HIF-1a signal overall. In the control group, MMP13 expression was markedly elevated, indicative of active-matrix degradation; this expression was significantly reduced in theMg2+ and M + D groups, in parallel with improved cell morphology and diminished inflammatory response.

These observations were further substantiated by qRT-PCR analysis of genes associated with inflammation (CCL2, IL-1p, TNF-a), cartilage degradation (MMP13), and chondrogenesis (COL-II, SOX9, ACAN, HIF10) (Fig. 3M-T). Mg"? notably upregulated the expression of cartilage matrix-related genes (COL-II, SOX9, ACAN), while DMOG predominantly enhanced HIF-1a transcription. The combined M + D treatment produced a synergistic effect, significantly elevating the expression of all chondrogenic markers. In contrast, MMP13 expression was markedly suppressed in all treated groups, with the M + D group showing the lowest expression level, consistent with immunofluorescence observations. Collectively, these findings demonstrate that M + D treatment effectively restores chondrocyte morphology, suppresses inflammatory and catabolic gene expression, and enhances chondrogenic potential. The observed synergy betweenMg2+ and DMOG highlights the therapeutic promise of M + D pre-conditioning or co-delivery strategies via intra-articular injection for the treatment of OA-related cartilage degeneration.

3.4. Fabrication of adaptive hydrogel

To enable the application of М + D pre-conditioned BMSCs for cartilage repair in osteoarthritic joints, an injectable hydrogel system is required to facilitate cell encapsulation and maintain cellular viability under physiological conditions. To this end, a HA hydrogel with a dual crosslinked network was developed, wherein covalent crosslinks confer structural stability, and dynamic crosslinks promote cell migration and proliferation (Fig. 4A). The hydrogel matrix was constructed by first synthesizing methacrylated hyaluronic acid (HAMA) via grafting of methacrylic anhydride, introducing photo-crosslinkable vinyl groups. Subsequently, phenylboronic acid (PBA) was grafted onto the HAMA backbone, yielding HAMA-PBA (HP), which is capable of forming reversible borate ester bonds with polyhydroxyl-containing moieties such as glucose in culture medium or polysaccharides present in native tissues.

The successful synthesis of HAMA and HAMA-PBA was confirmed via 1H NMR spectra, with characteristic peaks observed in Fig. 58. HAMAPBA was dissolved in PBS (HP-PBS) or high-glucose DMEM (HPDMEM), and HAMA was dissolved in PBS to prepare hydrogels. Due to the formation of dynamic borate-polyhydroxy interaction, spontaneous gelation occurred in the HP-DMEM solution, whereas no such gelation was observed in HP-PBS (Fig. S9). Upon photopolymerization, the HPDMEM hydrogel also demonstrated favorable injectability and adhesion to cartilage tissue surfaces (Fig. 4B). Rheological assessments confirmed the thixotropic properties of the HP-DMEM hydrogel, validating its capacity for shear-thinning, injectability, and self-healing (Fig. 4D and E). In contrast, the HP-PBS hydrogel, lacking dynamic crosslinking, showed mechanical behavior similar to that of the covalently crosslinked HAMA hydrogel. To further investigate the protective effects of the dynamic crosslinking network against shear stress, BMSCs were encapsulated within HP-DMEM and subjected to syringe extrusion. Live-dead staining revealed that HP-DMEM substantially mitigated cell damage during injection (Fig. S10), highlighting its suitability for cell delivery applications. Conversely, without hydrogel protection, the cells showed significant death during injection, caused by mechanical damage from shear forces inside the syringe needle.

The cross-sectional morphology and pore size distribution of HPDMEM, HP-PBS, and HAMA hydrogels were examined after freezedrying and fracturing (Fig. S11). All hydrogels exhibited macroporous structures (>100 pm), with HP-DMEM displaying the smallest pore size due to its dual-crosslinking network. The higher crosslinking density endowed HP-DMEM with enhanced mechanical strength (Fig. S12) and a reduced swelling ratio (Fig. S13A), while maintaining degradation rates comparable to those of HP-PBS and HAMA hydrogels (Fig. S13B). Hemolysis testing confirmed non-cytotoxic nature of all hydrogels (Fig. 4F), and CCK-8 assays further demonstrated their compatibility with cell proliferation (Fig. S14).

To evaluate cellular behavior within the hydrogel matrix, BMSCs were encapsulated in HP-DMEM and HAMA hydrogels. For visualization, the hydrogel matrices were labeled with red fluorescence and cells with green fluorescence (Fig. 4H). Confocal imaging revealed that cells within the HP-DMEM hydrogel exhibited significantly enhanced spreading and elongation compared to those in the HAMA hydrogel (Fig. 41). Scanning electron microscopy (SEM) corroborated these findings, showing improved cellular morphology and integration in HPDMEM (Fig. S15). Despite having a smaller pore size, the dynamic crosslinking within HP-DMEM facilitated cellular migration and morphological adaptation, effectively preserving cell viability and reducing post-injection cell loss through enhanced tissue adhesion.

3.5. Adaptive hydrogel encapsulation preserves the function of precoordinated BMSCs

As demonstrated by the preceding studies, the M + D treatment enhances chondrogenic differentiation by upregulating HIF-1a expression and attenuating inflammatory responses. To evaluate whether these pre-conditioned BMSCs retain their functionality for intra-articular therapy, we encapsulated them within hydrogels and assessed their performance in vitro and in vivo. Briefly, BMSCs were incubated in M + D-conditioned medium for 14 days to generate pre-coordinated cells (designated as Sti-C), while BMSCs cultured under standard conditions served as controls (C). These two cell populations were separately embedded in either HAMA or dual-crosslinked HP-DMEM hydrogels, resulting in four experimental groups: H + С, HP + С, H + Sti-C, and HP + Sti-C.

After 4 weeks of culture, all the cell-laden hydrogels maintain their initial shape and volume without notable deformation (Fig. S16A). Alcian blue staining reveals that the HP + Sti-C group exhibits the most abundant GAG secretion, followed by the H + Sti-C, HP + C, and H + C groups (Fig. S16B). Immunofluorescence staining for type II collagen (COL-II) revealed a consistent trend, confirming that the dynamic microenvironment of the HP-DMEM hydrogel significantly promotes chondrogenic differentiation of the M + D pre-conditioned BMSCs (Fig. 4J and S17). Moreover, the HP + C and HP + Sti-C groups exhibited higher cell densities and stronger HIF-1a expression than their HAMAbased counterparts, with the HP + Sti-C group displaying the highest HIF-1a levels (Fig. 4K), further correlating with its enhanced chondrogenic outcomes. These findings collectively demonstrate that the dualnetwork HP-DMEM hydrogel not only preserves the biological function of pre-coordinated BMSCs but also provides a supportive microenvironment that facilitates sustained chondrogenic activity, thereby laying a solid foundation for the development of effective stem cellbased therapies for cartilage repair.

To evaluate in vivo performance, the four hydrogel constructs were subcutaneously implanted into the dorsal region of rats to assess their structural stability and ectopic chondrogenic potential (Fig. 4L). As representative examples, the H + Sti-C and HP + Sti-C hydrogels were molded into pentagram-shaped constructs prior to implantation. Postgelation, the constructs retained their geometric fidelity. After four weeks of subcutaneous implantation, all constructs generally preserved their original morphology (Fig. S18); however, the H + Sti-C group showed greater volume expansion, consistent with the higher swelling ratio of HAMA hydrogels compared to HP-DMEM hydrogels (Fig. 513).

At week 4, all explanted hydrogels were analyzed for cartilageassociated markers. The HP + Sti-C group exhibited the highest fluorescence intensities for HIF-1a and COL-II (Fig. 4M), and the most pronounced GAG deposition, as revealed by Alcian blue staining (Fig. 4N), with the H + Sti-C group ranking second, followed by HP + Cand H + C. These data clearly indicate that the HP + Sti-C hydrogel offers superior potential for cartilage regeneration. Furthermore, hematoxylin and eosin (H&E) staining confirmed the biocompatibility of the hydrogel system, supporting its suitability for in vivo therapeutic applications.

3.6. Treatment of tissue damage associated with OA in vivo

OA modeled rats were created by ACL and DMM procedure approaches at knee joints (Fig. 5A). The OA modeled rats were divided into 6 groups for different treatments, with an extra Sham group as positive control. The OA joint receiving only saline injection served as negative control. As compared to the Sham group, Micro-CT detection shows the negative Control group exhibiting significant osteophytes, along with substantial bone loss in subchondral bone (Fig. 5B(b1 and b2)), MRI result demonstrates discontinuous, hyperintense signals on the cartilage, both confirming the success in establishing the OA model via the ACL and DMM treatments (Fig. 5B (b3)). HP + Sti-C or HP + C hydrogel was injected into joint cavity of these OA modeled rats, simultaneously with the injection of saline or M + D solution once a week. In particular, Sti-C denotes the BMSCs group that underwent pre-coordination using a conditioned medium enriched with Mg2+ and DMOG. Conversely, С represents the BMSCs group that did not receive any pre-treatment. Amelioration was achieved in different extents with these treatments. In the cases of no cell-laden hydrogel injection, the group treated with M + D displays reduced osteophytes volume as compared to the negative Control group, indicating the M + D injection able to alleviate the effect of local inflammation on tissue degeneration. In either saline or M + D injection, it is the HP + Sti-C group achieving the best repairing outcome, notably, the combination of HP + Sti-C hydrogel and M + D injection can restore the bone structure close to the Sham group. The HP + C groups also exhibit a tendency to inhibit the progression of OA, while their inhibitory efficiency is inferior to the HP + Sti-C groups. Quantitative data on the distance between the contralateral femoral condyles (Fig. 5C) and osteophytes volume (Fig. 5D) support the findings that the HP + Sti-C hydrogel alongside M + D injection has the strongest potential in reversing the OA process to reduce tissue abnormality.

It is known that articular OA is clinically characterized by cartilage loss and subchondral bone sclerosis. To further compare the repairing efficiency in different treatments, we quantified BV/TV, BMD, Tb. Th, and Tb. N based on micro-CT images (Fig. 5E-H), and stained tissue sections to show cartilage status. For the subchondral bone part, as compared to the negative Control group, the parameters representing the bone microstructure are obviously decreased with both cells and M + D injection, of note, the HP + Sti-C alongside M + D injection shows the smallest values, which are close to those data in the Sham group. These quantitative comparisons lead to the conclusion that the M + D injection is able to ameliorate local inflammatory responses, and the precoordinated BMSCs help to reduce subchondral bone sclerosis, collectively, their combination has effectively alleviated OA in inhibiting joint space narrowing.

The occurrence of subchondral bone sclerosis would reduce cartilage cushioning, increase loading stress, and ultimately accelerate cartilage degradation [61]. With the subchondral bone sclerosis being alleviated in the experimental groups the knee joints were collected for section slicing and histological staining to determine the extent of cartilage loss. H&E and Safranin O/Fast green staining (Saf O/Fast green) images were presented in Fig. 51. The cartilage can be seen almost fully destroyed in the OA model (the negative Control), while the cartilage tissue is preserved with the M + D and/or cell-laden hydrogels intervened. Certainly, it is the group of HP + Sti-C associated with M + D injection showing the smoothest cartilage surface without obvious hint of inflammatory infiltration. Accordingly, this group has the lowest OARSI score (Fig. 5J) and Mankin score (Fig. 5K) among all the groups, in contrast to those high values in the negative Control group, while sharing similarity to the Sham group. Further characterizations using immunohistochemistry staining show increased MMP13 and decreased COL-II expression in the negative Control group, in accordance with its significant cartilage loss caused by OA (Fig. 5I). In contrast, treatment with M + D and/or cell-laden hydrogels changes the situation that cartilage degeneration is inhibited as illustrated by reduced MMP13 and improved COL-II expressions. The group of HP + Sti-C hydrogel alongside M + D injection behaves the best in these evaluations, which provides the strongest protective capability in preserving the cartilage phenotype and native microstructure.

Collagen fiber orientation is a key parameter to represent the microstructure of articular cartilage [62]. The cartilage tissue is organized within a dynamically graded ECM, characterized by depth-dependent collagen compositional transitions and anisotropic fiber alignment that progress from the superficial articular zone to the subchondral bone interface. This intricate structural gradient ensures optimal biomechanical competence of diarthrodial joints by facilitating uniform load distribution and dissipating compressive stresses [63]. In this study, the injection of cells and M + D was conducted at the meantime of the OA model created, that the cartilage structure and collagen fiber orientation can be highly preserved if the administration is sufficient in counteracting OA progress. Therefore, the tissue sections were stained with Sirius red and observed under polarized light microscopy to judge collagen fiber orientation in superficial (ROI-1) and middle/deep layers (ROI-2). Compared to the natural cartilage in the Sham group (Fig. S19), the negative Control group shows a significant reduction in collagen content with disorganized fibers at 4- and 8-weeks post-operation (Fig. 6A). The percentages of orientated collagen fibers in the ROI-1 and ROI-2 zones are only 22.01 % and 27.31 %, respectively, in this OA modeled group without additional treatment. The injection with M + D or either cell-laden hydrogel (HP + C, or HP + Sti-C) is helpful in preserving collagen fiber orientation, due to their ability in reducing cartilage degradation. For the best experimental group, i.e., the HP + Sti-C with M + D injection, its collagen fiber microarchitecture displays 49.31 % tangential in ROI-1 and 68.61 % perpendicular in ROI-2, which are comparable to the Sham group (Fig. 6B). As depicted in Fig. 6C, a polar plot is employed to intuitively illustrate the variances in collagen fiber orientation angles across different groups. In summary, the counterparts in the saline injection and the M + D injection groups presented better cartilage microstructure in the latter case, as the M + D injection able to alleviating local inflammatory responses. Under the M + D injection, the collagen alignment ranks in the order of HP + Sti-C > HP + C > Control. Obviously, the combination of the M + D and HP + Sti-C injection shows the strongest capacity in maintaining the normal tissue morphology to minimize cartilage degeneration caused by OA.

3.7. Evaluation of reduced inflammation and subchondral bone resorption in vivo

Both in vitro and in vivo outcomes aforementioned reveal that the M + D administration and the M + D pre-coordinated BMSCs are able to achieve synergistical therapeutic effect in high efficiency for hCHO cells and OA modeled knee joint. As OA is a kind of inflammation associated disease, it necessitates the evaluations on the ability of the treatment in reducing local inflammation in vivo. Therefore, immunofluorescence staining on macrophages (F4/80) and inflammatory factors (iNOS, Arg1) was performed at week 8 post-surgery (Fig. 7A and B). The negative Control group displays high macrophage accumulation and iNOS expression, the M + D, HP + C and HP + Sti-C treatments reduce iNOS expression, with HP + Sti-C associated with M + D achieves the lowest iNOS expression (Fig. 7F). The Arg-1 staining intensity and quantification present an adverse trend to iNOS in these groups (Fig. 7G) The precoordinated BMSCs laden by HP-DMEM hydrogel maximizes its antiinflammatory effect in the joint cavity together with the M + D injection.

Macrophage enrichment is likely to trigger subchondral bone resorption, which is characterized by reduced bone volume and calcified cartilage thickening in early-stage OA, subchondral bone sclerosis, calcification, and plate thickening in late-stage OA. Preventing the early subchondral bone loss is a key issue in mitigating OA progression. At week 4 post-surgery, micro-CT analysis reveals the osteophytes formation, severe bone resorption and trabecular thinning in the subchondral bone plate in the negative Control group (Fig. 7C). Accordingly, its TRAP-positive stained cells are rich to indicate osteoclast activation, showing the highest value among all the groups (Fig. 7D and E). M + D injection and cell-laden hydrogel treatments are effective in downregulating osteoclast activation, reducing bone resorption, and maintaining the subchondral bone morphology. Particularly, almost no TRAP-positive cells are detected in the HP + Sti-C group with M + D injection, in consistent with the fact that the subchondral bone structure is well preserved in this group. Briefly, these findings in both macrophage phenotypes, activated osteoclasts and bone resorption confirms the feasibility of using M + D and HP-Sti-C to inhibit OA progression and protect articular cartilage from severe degeneration.

3.8. Investigation on the cross-talk between chondrocytes and osteoclasts regulated by Mg" and DMOG

With aforementioned approaches in using Mg2+ and DMOG for OA treatment, it is quite interested to interpret the cross-talk between osteoclasts and chondrocytes within the inflammatory environment, enhancing our understanding on their roles in inhibiting OA progression. At first, BMMCs were cultured in the Mg", DMOG, and M + D conditioned media for 3 and 5 days, and cell proliferation is not influenced with these additives (Fig. S20). Then, cell morphology and TRAP staining were evaluated to correlate osteoclast activity to the conditioned media, which had been stimulated by a RANKL-inducing solution. In the Control group, the formation of F-actin rings entrapping multinucleated osteoclasts is obvious after 5 days of incubation, while the other three groups show much less actin ring staining and multinucleated cells (Fig. SA). Examination on TRAP activity reveals the same trend (Fig. 8B and C). It is identified the M + D group presents the smallest area of stained F-actin rings and the least number of TRAPpositive cells, indicating that it is the most effective in inhibiting osteoclastic differentiation as compared to the Mg2+ or DMOG conditioned group. qRT-PCR analysis reveals that the expressions of osteoclastogenesis-related genes (TRAP, MMP9, and Cathepsin К) are all reduced in BMMCs cultured in conditioned media as compared to the Control group, in general, significantly decreased in the two Mg2+supplemented groups (Fig. SD-F). These results suggest that both Mg2+ and DMOG present inhibitory effect on osteoclast activation, and their combination (M + D) may achieve synergistic effect in the event. To visualize the difference, fresh bovine dental slices were prepared with BMMCs being seeded onto their surfaces, followed by soaking in RANKLsupplemented medium to induce bone resorption. Then, the slices were transferred into Mg2+, DMOG, and М + D conditioned media for 5 days before SEM observation. From Fig. 8G and H, the areas of bone resorption pits (indicated by red circles) in the conditioned media groups are significantly smaller than that observed in the Control group. TheMg2+ and М + D groups exhibit further reduction compared to the DMOG group. Overall, Mg2+ inhibits bone resorption via suppressing osteoclasts, and the combination of M + D can synergistically enhance this inhibitory effect (Fig. 81).

To mimic the crosstalk between osteoclasts and chondrocytes in OA modeled joint, we conducted a Transwell experiment, in which, chondrocytes were seeded in the lower chamber and BMMCs were placed in the upper chamber supplemented with a RANKL induction medium (Fig. 8J). After 5 days, the BMMCs were subjected to TRAP staining, while the chondrocytes were stained to analyze ALP, COL-X and COL-II expressions. As shown in Fig. 8K-M, the Control group shows high level of activated osteoclasts, accordingly, the chondrocytes in this group exhibit the obvious ALP and COL-X expressions with the least content of COL-II (Fig. 8N-P). These findings suggest that inflammation would trigger the activation of osteoclasts, which act on chondrocytes to cause their hypertrophic differentiation, thereby leading to dedifferentiation of chondrogenic phenotype. This partly interprets the cartilage degeneration in OA modeled joint. With conditioned media treatment, differently, the osteoclast activation is inhibited, and the markers (ALP, COL-X) related to calcified cartilage are downregulated in chondrocytes, while the chondrogenic potential is well preserved.

As all these data shown, the M + D treatment exhibits the most significant potential for enhancing chondrocyte functionality by attenuating the interactions between osteoclasts and chondrocytes within the inflammatory microenvironment, aligning closely with the observed in vivo results. Specifically, in vivo evaluations indicate that the M + D injection effectively reduces joint inflammation and preserves articular cartilage compared to saline treatment, primarily attributed to its capability to substantially maintain subchondral bone mass through the inhibition of osteoclast activity. This beneficial outcome is fundamentally due to the capacity of M + D to suppress the pro-inflammatory polarization of macrophages and their subsequent differentiation into osteoclasts. Additionally, sequencing analyses (Fig. 2) reveal that DMOG markedly enhances hypoxia signaling pathway expression, whereas Mg2+ significantly upregulates signaling pathways associated with PI3K-Akt, TGF-β1, MAPK, and ECM synthesis, collectively facilitating cartilage phenotype expression and suppression of inflammatory markers. Furthermore, the M + D combination group prominently elevates HIF-1α expression, indicating a hypoxia-mimicking environment conducive to chondrogenic differentiation. Overall, as illustrated in Fig. 9, the combined administration of Mg2+ and DMOG emerges as a promising therapeutic approach for OA by concurrently targeting inflammation and regulating bone metabolism.

4. Conclusion

In summary, Mg2+ demonstrates protective effects on cartilage and subchondral bone by modulating immune cells within the inflammatory environment of OA. Additionally, DMOG synergistically activates the hypoxia signaling pathway, resulting in increased HIF-1α expression and the upregulation of cartilage phenotype synthesis markers. To enhance the therapeutic potential of stem cells, we employed an in vitro functionalization strategy using M + D-conditioned medium. Simultaneously, a HA-based hydrogel with a dual network of covalent and dynamic bonds was developed to effectively encapsulate precoordinated BMSCs, supporting their proliferation and maintaining their functionalized phenotype. Notably, the PBA moiety in the hydrogel establishes reversible cyclic esters with polyhydroxy compounds on the cartilage surface, facilitating the targeted delivery of functionalized stem cells. This targeted approach ensures effective colonization of precoordinated BMSCs at the cartilage surface, promoting therapeutic effects and achieving significant cartilage repair in vivo. Overall, the adaptive hydrogel designed for functional BMSC encapsulation represents a promising platform for advancing stem cell therapy in the treatment of OA.

Ethics approval and consent to participate

Human OA cartilage samples were obtained from the knee joint of patients undergoing total knee arthroplasty. All protocols were approved by the Ethics Committee of Peking University Third Hospital (Number: (2022) 581-02). Informed consent was obtained from all donors prior to tissue collection.

CRediT authorship contribution statement

Chenyuan Gao: Writing - original draft, Visualization, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Wenli Dai: Methodology. Dingge Liu: Visualization. Xinyu Wang: Data curation. Tianyun Zhang: Methodology. Bingzheng Yu: Data curation. Yingjie Yu: Writing - review & editing, Supervision, Project administration, Formal analysis. Hua Tian: Writing - review & editing, Supervision. Xiaoping Yang: Supervision, Project administration. Qing Cai: Writing - review & editing, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

Qing Cai is an editorial board member for Bioactive Materials and was not involved in the editorial review or decision to publish this article. All authors declare that they have no competing interests.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (32371422, U22A20159), the Beijing Natural Science Foundation (7232100, L232018, L232092), the Peking University Third Hospital Fund for Interdisciplinary Research (BYSYJC2024024), and Central Universities (buctrc202220).

Sidebar

Peer review under the responsibility of editorial board of Bioactive Materials.

* Corresponding author.

*· Corresponding author.

*·· Corresponding author.

E-mail addresses: [email protected] (Y. Yu), [email protected] (H. Tian), [email protected] (Q. Cai).

Received 14 February 2025; Received in revised form 16 May 2025; Accepted 17 May 2025

Available online 28 May 2025

References

References

[1] Q. Yao, X. Wu, C. Tao, W. Gong, M. Chen, M. Qu, Y. Zhong, T. He, S. Chen, G. Xiao, Osteoarthritis: pathogenic signaling pathways and therapeutic targets, Signal Transduct. Tar. 8 (2023) 56.

[2] Y. Xie, A. Zinkle, L. Chen, M. Mohammadi, Fibroblast growth factor signalling in osteoarthritis and cartilage repair, Nat. Rev. Rheumatol. 16 (2020) 547-564.

[3] Y. Cho, H. Kim, G. Yook, S. Yong, S. Kim, N. Lee, Y.J. Kim, J.H. Kim, T.W. Kim, M. J. Chang, K.M. Lee, C.B. Chang, S.B. Kang, J.H. Kim, Predisposal of interferon regulatory factor 1 deficiency to accumulate DNA damage and promote osteoarthritis development in cartilage, Arthritis Rheumatol. 76 (2024) 882-893.

[4] L.A. Holzer, M. Kraiger, E. Talakic, G.A. Fritz, A. Avian, A. Hofmeister, A. Leithner, G. Holzer, Microstructural analysis of subchondral bone in knee osteoarthritis, Osteoporos. Int. 31 (2020) 2037-2045.

[5] R.F. Loeser, S.R. Goldring, C.R. Scanzello, M.B. Goldring, Osteoarthritis: a disease of the joint as an organ, Arthritis Rheum. 64 (2012) 1697-1707.

[6] D.J. Hunter, S. Bierma-Zeinstra, Osteoarthritis, Lancet 393 (2019) 1745-1759.

[7] Y. Wang, X. Qin, Y. Feng, T. Zhang, X. Wang, J. Li, P. Yin, Y. Yu, C. Liu, Dual-gradient silk-based hydrogel for spatially targeted delivery and osteochondral regeneration, Adv. Mater. 37 (2025) 2420394.

[8] A.J. Roelofs, J.P.J. Rocke, C. De Bari, Cell-based approaches to joint surface repair: a research perspective, Osteoarthr, Cartilage 21 (2013) 892-900.

[9] F. Migliorini, J. Eschweiler, C. Gotze, A. Driessen, M. Tingart, N. Maffulli, Matrix-induced autologous chondrocyte implantation (mACI) versus autologous matrix-induced chondrogenesis (AMIC) for chondral defects of the knee: a systematic review, Br. Med. Bull. 141 (2022) 47-59.

[10] J. Chen, Q. Li, H. Li, C. Lv, H. Yu, Q. Feng, H. Dong, Injectable acellular matrix microgel assembly with stem cell recruitment and chondrogenic differentiation functions promotes microfracture-based articular cartilage regeneration, Bioact. Mater. 44 (2025) 220-235.

[11] D.K. Bae, K.H. Yoon, S.J. Song, Cartilage healing after microfracture in osteoarthritic knees, Arthroscopy 22 (2006) 367-374.

[12] B. Peng, L. Zhou, Z. Luo, X. Bian, H. Li, G. Liu, J. Chen, Z. Ma, P. Zhang, Y. Wen, Construction of hydrogels with specific structures and functions for cartilage defects and traumatic osteoarthritis: a review, Chem. Eng. J. 503 (2025) 158242.

[13] T. Guo, M. Noshin, H.B. Baker, E. Taskoy, S.J. Meredith, Q. Tang, J.P. Ringel, M. J. Lerman, Y. Chen, J.D. Packer, J.P. Fisher, 3D printed biofunctionalized scaffolds for microfracture repair of cartilage defects, Biomaterials 185 (2018) 219-231.

[14] J. Li, Y. Yu, M. Kim, K. Li, J. Mikhail, L. Zhang, C. Chang, D. Gersappe, M. Simon, C. Ober, M. Rafailovich, Manipulation of cell adhesion and dynamics using RGD functionalized polymers, J. Mater. Chem. B 5 (2017), 6973-6973.

[15] Y. Wang, Z. Yuan, Y. Pang, D. Zhang, G. Li, X. Zhang, Y. Yu, X. Yang, Q. Cai, Injectable, high specific surface area cryogel microscaffolds integrated with osteoinductive bioceramic fibers for enhanced bone regeneration, ACS Appl. Mater. Interfaces 15 (2023) 20661-20676.

[16] N. Thakur, B. Singh, Designing protein-polysaccharides based bioactive copolymeric network by supra-molecular interactions for sustained drug delivery, Supramol. Mater. 2 (2023) 100048.

[17] Y. Zhou, M. Cui, S. Liao, B. Yuan, R. Shi, X. Hu, Y. Wang, An injectable macroporous hydrogel templated by gasification reaction for enhanced tissue regeneration, Supramol. Mater. 2 (2023) 100037.

[18] J. Ng, C.B. Little, S. Woods, S. Whittle, F.Y. Lee, S. Gronthos, S. Mukherjee, D. J. Hunter, D.L. Worthley, Stem cell-directed therapies for osteoarthritis: the promise and the practice, Stem Cell. 38 (2020) 477-486.

[19] Y. Wang, L. Li, Y.-E. Ji, T. Wang, Y. Fu, X. Li, G. Li, T. Zheng, L. Wu, Q. Han, Y. Zhang, Y. Wang, D.L. Kaplan, Y. Lu, Silk-protein-based gradient hydrogels with multimode reprogrammable shape changes for biointegrated devices, Proc. Natl. Acad. Sci. 120 (2023) e2305704120.

[20] X. Yan, B. Yang, Y. Chen, Y. Song, J. Ye, Y. Pan, B. Zhou, Y. Wang, F. Mao, Y. Dong, D. Liu, J. Yu, Anti-friction MSCs delivery system improves the therapy for severe osteoarthritis, Adv. Mater. 33 (2021) 2104758.

[21] Y. Yang, H. Zhang, Intra-articular injection of nanomaterials for the treatment of osteoarthritis: from lubrication function restoration to cell and gene therapy, Adv. Funct. Mater. 34 (2024) 2401547.

[22] M. Wu, L. Han, B. Yan, H. Zeng, Self-healing hydrogels based on reversible noncovalent and dynamic covalent interactions: a short review, Supramol. Mater. 2 (2023) 100045.

[23] Y. Shi, Y. Wang, Q. Li, K. Liu, J. Hou, C. Shao, Y. Wang, Immunoregulatory mechanisms of mesenchymal stem and stromal cells in inflammatory diseases, Nat. Rev. Nephrol. 14 (2018) 493-507.

[24] M. Hou, B. Tian, B. Bai, Z. Ci, Y. Liu, Y. Zhang, G. Zhou, Y. Cao, Dominant role of in situ native cartilage niche for determining the cartilage type regenerated by BMSCs, Bioact. Mater. 13 (2022) 149-160.

[25] Y. Chen, Y. Yan, R. Tian, Z. Sheng, L. Li, J. Chen, Y. Liao, Y. Wen, J. Lu, X. Liu, W. Sun, H. Wu, Y. Liao, X. Zhang, X. Chen, C. An, K. Zhao, W. Liu, J. Gao, D.C. Hay, H. Ouyang, Chemically programmed metabolism drives a superior cell fitness for cartilage regeneration, Sci. Adv. 10 (2024) eadp4408.

[26] I. Sekiya, H. Katano, N. Ozeki, Characteristics of MSCs in synovial fluid and mode of action of intra-articular injections of synovial MSCs in knee osteoarthritis, Int. J. Mol. Sci. 22 (2021) 2838.

[27] E.E. Mahmoud, N. Adachi, A.S. Mawas, M. Deie, M. Ochi, Multiple intra-articular injections of allogeneic bone marrow-derived stem cells potentially improve knee lesions resulting from surgically induced osteoarthritis, Bone Joint Lett. J. 101-B (2019) 824-831.

[28] A. Hassanzadeh, N. Vousooghi, R. Rahimnia, E. Razeghian, S. Rajaeian, I. Seyhoun, S. Sharif, J. Verdi, Recent advances in mesenchymal stem/stromal cells (MSCs)based approaches for osteoarthritis (OA) therapy, Cell Biol. Int. 47 (2023) 1033-1048.

[29] C. Zhang, D. Li, Z. Wang, D. Wang, У. Lu, К. Zhang, У. Yuan, X. Mei, S. Chen, M1 type macrophage targeted anti-inflammatory exosomes derived from BMSC for the treatment of acute and chronic inflammatory diseases, Mater. Des. 240 (2024) 112844.

[30] C.H. Lu, T.S. Yeh, C.L. Yeh, Y.-H.D. Fang, L.Y. Sung, S.Y. Lin, T.C. Yen, Y.H. Chang, Y.C. Hu, Regenerating cartilages by engineered ASCs: prolonged TGF-83/BMP-6 expression improved articular cartilage formation and restored zonal structure, Mol. Ther. 22 (2014) 186-195.

[31] Z. Chen, W. Zhang, M. Wang, L.J. Backman, J. Chen, Effects of zinc, magnesium, and iron ions on bone tissue engineering, ACS Biomater. Sci. Eng. 8 (2022) 2321-2335.

[32] Z. Yuan, Z. Wan, C. Gao, Y. Wang, J. Huang, Q. Cai, Controlled magnesium ion delivery system for in situ bone tissue engineering, J. Contr. Release 350 (2022) 360-376.

[33] S. Dutta, M. Roy, Recent developments in engineered magnesium scaffolds for bone tissue engineering, ACS Biomater. Sci. Eng. 9 (2023) 3010-3031.

[34] J. Chen, L. Yu, T. Gao, X. Dong, S. Li, Y. Liu, J. Yang, K. Xia, Y. Yu, Y. Li, S. Wang, Z. Fan, H. Deng, W. Guo, Nanofiber-induced hierarchically - porous magnesium phosphate bone cements accelerate bone regeneration by inhibiting Notch signaling, Bioact. Mater. 37 (2024) 459-476.

[35] Y. Chen, W. Sheng, J. Lin, C. Fang, J. Deng, P. Zhang, M. Zhou, P. Liu, J. Weng, F. Yu, D. Wang, B. Kang, H. Zeng, Magnesium oxide nanoparticle coordinated phosphate-functionalized chitosan injectable hydrogel for osteogenesis and angiogenesis in bone regeneration, ACS Appl. Mater. Interfaces 14 (2022) 7592-7608.

[36] J. Zhang, X. Ma, D. Lin, H. Shi, Y. Yuan, W. Tang, H. Zhou, H. Guo, J. Qian, C. Liu, Magnesium modification of a calcium phosphate cement alters bone marrow stromal cell behavior via an integrin-mediated mechanism, Biomaterials 53 (2015) 251-264.

[37] K. Chen, X. Xie, H. Tang, H. Sun, L. Qin, Y. Zheng, X. Gu, Y. Fan, In vitro and in vivo degradation behavior of Mg-2Sr-Ca and Mg-2Sr-Zn alloys, Bioact. Mater. 5 (2020) 275-285.

[38] Z. Liao, L. Fu, P. Li, J. Wu, X. Yuan, C. Ning, Z. Ding, X. Sui, S. Liu, Q. Guo, Incorporation of magnesium ions into an aptamer-functionalized ECM bioactive scaffold for articular cartilage regeneration, ACS Appl. Mater. Interfaces 15 (2023) 22944-22958.

[39] S. Shankar, A.N. Murthy, P. Rachitha, V.B. Raghavendra, N. Sunayana, A. Chinnathambi, S.A. Alharbi, N. Basavegowda, K. Brindhadevi, A. Pugazhendhi, Silk sericin conjugated magnesium oxide nanoparticles for its antioxidant, antiaging, and anti-biofilm activities, Environ. Res. 223 (2023) 115421.

[40] H. Yu, H. Luo, J. Chen, X. Hu, Y. Chen, J. Zhong, B. Liu, Y. Liu, G. Chen, C. Huang, Z. Peng, Hydrogel microspheres for stem cell recruitment and induction of directed differentiation in osteoarthritis therapy, Chem. Eng. J. 497 (2024) 154798.

[41] L. Zheng, S. Zhao, Y. Li, J. Xu, W. Yan, B. Guo, J. Xu, L. Jiang, Y. Zhang, H. Wei, Q. Jiang, Engineered MgO nanoparticles for cartilage-bone synergistic therapy, Sci. Adv. 10 (2024) eadk6084.

[42] N. Angrisani, C. von der Ahe, R. Willumeit-Romer, H. Windhagen, V. Scheper, M. Schwarze, B. Wiese, H. Helmholz, J. Reifenrath, Treatment of osteoarthritis by implantation of Mg- and WE43-cylinders - a preclinical study on bone and cartilage changes and their influence on pain sensation in rabbits, Bioact. Mater. 40 (2024) 366-377.

[43] N. Yao, J.K. Xu, N.Y. Zheng, J.L. Wang, S.W. Mok, Y.W. Lee, L. Shi, J.Y. Wang, J. Yue, S.H. Yung, P.J. Hu, Y.C. Ruan, У.Е. Zhang, K.-W. Ho, L. Qin, Intra-articular injection of magnesium chloride attenuates osteoarthritis progression in rats, Osteoarthr, Cartilage 27 (2019) 1811-1821.

[44] C.H. Evans, V.B. Kraus, L.A. Setton, Progress in intra-articular therapy, Nat. Rev. Rheumatol. 10 (2014) 11-22.

[45] H. Yao, J. Xu, J. Wang, Y. Zhang, N. Zheng, J. Yue, J. Mi, L. Zheng, B. Dai, W. Huang, S. Yung, P. Hu, Y. Ruan, Q. Xue, K. Ho, L. Qin, Combination of magnesium ions and vitamin C alleviates synovitis and osteophyte formation in osteoarthritis of mice, Bioact. Mater. 6 (2021) 1341-1352.

[46] H. Zhang, L. Wang, J. Cui, S. Wang, Y. Han, H. Shao, C. Wang, Y. Hu, X. Li, Q. Zhou, J. Guo, X. Zhuang, S. Sheng, T. Zhang, D. Zhou, J. Chen, F. Wang, Q. Gao, Y. Jing, X. Chen, J. Su, Maintaining hypoxia environment of subchondral bone alleviates osteoarthritis progression, Sci. Adv. 9 (2023) eabo7868.

[47] R. Phillips, Articular cartilage hypoxia is a potential target for OA therapy, Nat. Rev. Rheumatol. 18 (2022), 3-3.

[48] S. Chen, M. Gao, Z. Zhou, J. Liang, M. Gong, X. Dai, T. Liang, J. Ye, G. Wu, L. Zou, Y. Wang, X. Zou, Opposite regulation of chondrogenesis and angiogenesis in cartilage repair ECM materials under hypoxia, J. Mater. Sci. Technol. 32 (2016) 978-985.

[49] F. Zhang, B. Zhang, Y. Wang, R. Jiang, J. Liu, Y. Wei, X. Gao, Y. Zhu, X. Wang, M. Sun, J. Kang, Y. Liu, G. You, D. Wei, J. Xin, J. Bao, M. Wang, Y. Gu, Z. Wang, J. Ye, S. Guo, H. Huang, Q. Sun, An extra-erythrocyte role of haemoglobin body in chondrocyte hypoxia adaption, Nature 622 (2023) 834-841.

[50] C. Martinez-Armenta, C. Suarez-Ahedo, A. Olivos-Meza, M.C. Camacho-Rea, L. E. Martinez-Gomez, G.E. Jimenez-Gutierrez, G.A. Martinez-Nava, L.E. GomezQuiroz, C. Pineda, A. Lopez-Reyes, The critical role of hypoxia in the redifferentiation of human articular chondrocytes, Cells 11 (2022) 2553.

[51] F. Ravanetti, R. Saleri, P. Martelli, M. Andrani, L. Ferrari, V. Cavalli, V. Conti, A. P. Rossetti, E. De Angelis, P. Borghetti, Hypoxia and platelet lysate sustain differentiation of primary horse articular chondrocytes in xeno-free supplementation culture, Res. Vet. Sci. 152 (2022) 687-697.

[52] M.C. Simon, B. Keith, The role of oxygen availability in embryonic development and stem cell function, Nat. Rev. Mol. Cell Biol. 9 (2008) 285-296.

[53] A. Nowak-Stepniowska, P.N. Osuchowska, H. Fiedorowicz, Е.А. Trafny, У. Gan, Insight in hypoxia-mimetic agents as potential tools for mesenchymal stem cell priming in regenerative medicine, Stem Cell. Int. 2022 (2022) 1-24.

[54] Q. Dong, X. Fei, H. Zhang, X. Zhu, J. Ruan, Effect of dimethyloxalylglycine on stem cells osteogenic differentiation and bone tissue regeneration - a systematic review, Int. J. Mol. Sci. 25 (2024) 3879.

[55] R. Smaida, L. Pijnenburg, S. Irusta, E. Himawan, G. Mendoza, E. Harmouch, Y. Idoux-Gillet, S. Kuchler-Bopp, N. Benkirane-Jessel, G. Hua, Potential implantable nanofibrous biomaterials combined with stem cells for subchondral bone regeneration, Materials 13 (2020) 3087.

[56] K. Okada, D. Mori, Y. Makii, H. Nakamoto, Y. Murahashi, F. Yano, S.H. Chang, Y. Taniguchi, H. Kobayashi, H. Semba, N. Takeda, W. Piao, K. Hanaoka, T. Nagano, S. Tanaka, T. Saito, Hypoxia-inducible factor-1 alpha maintains mouse articular cartilage through suppression of NF-kB signaling, Sci. Rep. 10 (2020) 5425.

[57] S. Hu, C. Zhang, L. N, C. Huang, D. Chen, К. Shi, H. Jin, К. Zhang, Y. Li, L. Xie, М. Fang, С. Xiang, X. Wang, J. Xiao, Stabilization of HIF-10 alleviates osteoarthritis via enhancing mitophagy, Cell Death Dis. 11 (2020) 481.

[58] L. Chen, X. Huang, H. Chen, D. Bao, X. Su, L. Wei, N. Hu, W. Huang, Z. Xiang, Hypoxia-mimicking scaffolds with controlled release of DMOG and PTHrP to promote cartilage regeneration via the HIF-1a/YAP signaling pathway, Int. J. Biol. Macromol. 226 (2023) 716-729.

[59] J.T. Opferman, A. Kothari, Anti-apoptotic BCL-2 family members in development, Cell Death Differ. 25 (2017) 37-45.

[60] E. Yigit, P.J. Batista, Y. Bei, K.M. Pang, C.C.G. Chen, N.H. Tolia, L. Joshua Tor, S. Mitani, M.J. Simard, C.C. Mello, Analysis of the C. elegans argonaute family reveals that distinct argonautes act sequentially during RNAi, Cell 127 (2006) 747-757.

[61] W. Hu, Y. Chen, C. Dou, S. Dong, Microenvironment in subchondral bone: predominant regulator for the treatment of osteoarthritis, Ann. Rheum. Dis. 80 (2021) 413-422.

[62] C. Gao, W. Dai, X. Wang, L. Zhang, Y. Wang, Y. Huang, Z. Yuan, X. Zhang, Y. Yu, X. Yang, Q. Cai, Magnesium gradient-based hierarchical scaffold for dual-lineage regeneration of osteochondral defect, Adv. Funct. Mater. 33 (2023) 2304829.

[63] A. Pueyo Moliner, K. Ito, F. Zaucke, D.J. Kelly, M. de Ruijter, J. Malda, Restoring articular cartilage: insights from structure, composition and development, Nat. Rev. Rheumatol. 21 (2025) 291-308.