Dynamic hydrogels orchestrate the differentiation fate of mesenchymal stem cells through epigenetic regulation of SETD7 to accelerate bone defect repair

Abstract

Dynamic mechanical signaling of the extracellular matrix is a key determinant of mesenchymal stem cell (MSC) fate, closely regulating their proliferation, differentiation and migration. Previously, we developed a highly cell-adaptive dynamic hydrogel (HA-ADA) that modulates MSC fate through unknown mechanisms. Here, using human bone marrow-derived mesenchymal stem cells (hMSCs), we found that sustained mechanical stimulation provided by HA-ADA hydrogel induced rapid spreading and significantly enhanced their osteogenic differentiation while inhibiting adipogenesis. Mechanistically, miRNA sequencing revealed that this process was mediated by the downregulation of miR-376a-3p and miR-127-5p, thereby relieving their inhibitory effect on the methyltransferase SETD7. Elevated SETD7 expression catalyzed methylation of β-catenin and accelerated its nuclear translocation. In the nucleus, β-catenin further formed a transcriptional complex with YAP to synergistically amplify downstream signals and potently activate the expression of Runx2, a key transcription factor for osteogenesis, which ultimately drove osteogenic differentiation and inhibited adipogenesis. The present study elucidated a novel mechanism by which cell-adaptive hydrogels regulate the β-catenin/YAP signaling loop through the miR-376a-3p/miR-127-5p-SETD7 axis, thereby determining the osteogenic/adipogenic differentiation of stem cells, which not only deepens our understanding of mechanotransduction but also provides new targets and material design strategies for bone regeneration.

Graphical abstract

Schematic illustration of the HA-ADA hydrogel system for enhanced bone regeneration through dynamic matrix-mediated mechanotransduction. (a) Experimental workflow. (b) Chemical composition of hydrogels showing HA-ADA or HA-CA polymers decorated with RGD motifs, which crosslink with Ac-β-CD under blue light to form distinct network architectures. (c) Comparative illustration of hydrogel performance in bone regeneration. HA-ADA hydrogels exhibit high network adaptability that enables cell spreading through dynamic “gate opening” mechanisms involving actin polymerization and F-actin bundle formation, leading to enhanced bone regeneration. In contrast, HA-CA hydrogels display low network adaptability with restricted “gate closed” states that limit cell spreading and result in reduced bone regeneration capacity. (d) Molecular mechanism underlying HA-ADA hydrogel-promoted osteogenesis. The dynamic matrix downregulates miR-376a-3p and miR-127-5p expression, thereby relieving their repression of SETD7. Elevated SETD7 subsequently methylates β-catenin, facilitating YAP/β-catenin signaling cascade activation to drive osteogenic differentiation.

Highlights

• •The dynamic hydrogel (HA-ADA) directs MSC lineage commitment toward osteogenesis via mechanical cues.
• •Mechanical stimulation downregulates miR-376a-3p and miR-127-5p to activate the SETD7/β-catenin signaling pathway
• •The HA-ADA hydrogel-MSC construct accelerates bone repair in vivo.

1. Introduction

Bone defects, resulting from trauma, disease, or congenital anomalies, present substantial challenges in regenerative medicine, necessitating innovative strategies that effectively guide tissue regeneration. The extracellular matrix (ECM) is not merely a static scaffold but a dynamic entity that delivers biomechanical and biochemical signals instructing cellular behavior [[1], [2], [3]]. Among the cell types involved, mesenchymal stem cells (MSCs) are key players in bone repair due to their multipotency and regenerative potential [4,5]. Mechanical cues from the ECM are well-established regulators of MSC fate, governing processes such as proliferation, migration, and differentiation [6]. In our previous work, we developed a dynamic hydrogel (HA-ADA) that features dynamic crosslinking to effectively recapitulate the ECM and promote osteogenic lineage commitment of MSCs [7]. However, the underlying mechanism remains elusive.

Osteoblastic lineage commitment, proliferation, and differentiation are orchestrated by defined genetic programs [8]. MicroRNAs (miRNAs), short non-coding RNAs of 20–22 nucleotides, silence gene expression by guiding Argonaute (AGO) proteins to complementary sequences within the 3′-untranslated regions (3′UTRs) of target mRNAs, thereby destabilizing the transcripts or repressing their translation [9,10]. These miRNAs are initially transcribed as primary transcripts and subsequently processed into their mature forms through sequential cleavage by Drosha and Dicer [11]. Functionally, miRNAs modulate a multitude of biological processes, including the complex cascades of osteogenic differentiation and osteoblast-mediated bone formation, through the targeted regulation of specific genes [[12], [13], [14]]. Mounting evidence indicates that miRNAs are integral to mechanotransduction pathways that regulate cellular behaviors [15,16], including the fate decisions of MSCs [17]. We therefore hypothesized that the mechanical stresses imparted by the dynamic hydrogel matrix may direct the differentiation of MSCs by modulating their intracellular miRNA levels.

Epigenetic modifications represent a critical mechanism through which mechanotransduction regulates cell fate [18,19]. Mechanical signals remodel chromatin architecture and govern gene expression patterns through epigenetic modifications, including DNA methylation and histone modifications [20]. Among epigenetic regulators, the lysine methyltransferase SETD7 methylates both histones and non-histone substrates such as β-catenin, playing a pivotal role in cell differentiation [21]. Although mechanical stimuli are known to modulate the expression and activity of epigenetic enzymes, how dynamic hydrogels orchestrate mesenchymal stem cell fate through epigenetic mechanisms remains elusive.

In this study, we fabricated two types of physically cross-linked supramolecular hydrogels. Hyaluronic acid (HA) was functionalized with monoacryloyl-β-cyclodextrin (Ac-β-CD), followed by pre-complexation via CD–ADA or CD–CA host–guest interactions prior to photopolymerization. The resulting hydrogels—denoted HA–ADA and HA–CA—were stabilized exclusively by CD–ADA or CD–CA host–guest complexes, respectively [7]. Compared with HA–CA gels, the HA–ADA hydrogel, with its highly dynamic and adaptive network, significantly promoted the spreading and osteogenic differentiation of human bone marrow-derived mesenchymal stem cells (hMSCs). Further mechanistic investigation revealed that the HA–ADA hydrogel downregulated miR-376a-3p and miR-127-5p in hMSCs, thereby relieving their repression of the methyltransferase SET domain containing protein 7 (SETD7). The subsequent upregulation of SETD7 promoted β-catenin methylation and nuclear translocation. Within the nucleus, β-catenin cooperated with YAP to initiate a transcriptional program driving osteogenic lineage commitment. Our work elucidates a mechanotransduction pathway through which the cell-adaptive HA–ADA hydrogel directs encapsulated cell fate, providing valuable insights for the rational design of bioactive materials to accelerate bone regeneration.

2. Results

2.1. Development and Characterization of dynamic hydrogels

Evidence from previous studies has established a clear link between the mechanical properties of culture substrates and stem cell behavior [22,23]. To mimic the dynamic extracellular matrix environment in vivo, we developed a hydrogel with a highly dynamic matrix based on adamantane-modified hyaluronic acid (HA-ADA hydrogel). First, we confirmed the successful synthesis of Ac-β-CD, cholic acid-modified hyaluronic acid (HA-CA), and HA-ADA by ¹H nuclear magnetic resonance (¹H NMR) spectroscopy, with both HA-CA and HA-ADA exhibiting identical degrees of substitution (DS = 0.30) (Fig. 1a–c). Initial characterization by scanning electron microscopy (SEM) revealed that HA-CA hydrogels displayed small, densely packed pores, whereas HA-ADA hydrogels possessed larger, more open pores (Fig. 1d). Rheological measurements demonstrated that HA-ADA hydrogels exhibited a lower storage modulus and a reduced G'/G″ ratio compared to HA-CA hydrogels. These results indicated that HA-ADA hydrogels were softer and more dynamic, potentially providing a more permissive environment for cell migration and proliferation, while HA-CA hydrogels were stiffer and may restrict cellular activities (Fig. 1e). Further mechanical characterization through compressive testing revealed distinct stress-strain behaviors between the two hydrogel systems (Fig. S1), with HA-ADA hydrogels showing a significantly lower compressive modulus compared to HA-CA hydrogels (Fig. S2), confirming their enhanced mechanical compliance.

Fig. 1.

Development and Characterization of HA-CA and HA-ADA hydrogels. (a) ¹H NMR spectrum of Ac-β-CD. (b) ¹H NMR spectrum of HA-CA. (c) ¹H NMR spectrum of HA-ADA. (d) SEM images of HA-CA and HA-ADA hydrogels. (e) Rheological properties of HA-CA and HA-ADA hydrogels. (f) Stress relaxation profiles of HA-CA and HA-ADA hydrogels. (g) Schematic illustration of the procedure for encapsulating hMSCs within HA-CA and HA-ADA hydrogels. (h) SEM images of HA-CA and HA-ADA hydrogels after hMSCs encapsulation. (i) Representative live/dead staining images of hMSCs encapsulated in HA-CA and HA-ADA hydrogels after 24 h or 72 h of culture. (j) Quantitative analysis of cell viability in (i). n = 3. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Stress relaxation experiments revealed that HA-CA hydrogels exhibited a τ₁/₂ of 208 s, whereas HA-ADA hydrogels showed a markedly shorter τ₁/₂ of 13 s, suggesting that HA-ADA hydrogels possessed superior dynamic adaptability and provided cells with a more mechanically responsive microenvironment, while HA-CA hydrogels remained relatively static (Fig. 1f). Additionally, degradation studies demonstrated differential degradation kinetics between the two hydrogel systems, with HA-ADA hydrogels exhibiting more rapid degradation than HA-CA hydrogels (Fig. S3), further supporting their dynamic nature. To evaluate hydrogel performance under cell-laden conditions, we encapsulated hMSCs within the hydrogels, followed by crosslinking via 3-min blue light irradiation (Fig. 1g). SEM analysis further confirmed that HA-ADA hydrogels exhibited a more uniform and porous architecture compared to HA-CA hydrogels (Fig. 1h). After 24 h of culture, live/dead staining revealed minimal red fluorescence in HA-CA hydrogels, while no apparent red fluorescence was observed in HA-ADA hydrogels. This difference became more pronounced at 72 h, with HA-ADA hydrogels supporting significantly enhanced cell viability and proliferation, as evidenced by abundant green fluorescence and minimal red fluorescence (Fig. 1i and j). CCK-8 assays further corroborated the superior cell viability in HA-ADA hydrogels (Fig. S4). Taken together, the results demonstrated that we successfully engineered a highly dynamic hydrogel system with exceptional biocompatibility in vitro.

2.2. The HA-ADA hydrogel governs the differentiation fate of hMSCs

We next investigated the influence of HA-CA and HA-ADA hydrogels on the differentiation fate of hMSCs. First, hMSCs were encapsulated within the hydrogels, and their morphological changes were monitored. As shown in Fig. 2a, confocal 3D imaging revealed that hMSCs spread extensively and established cell-cell contacts within HA-ADA hydrogels, whereas cells in HA-CA hydrogels remained spherical with limited extension. To visualize the cytoskeletal architecture in detail, we performed 3D reconstruction of F-actin staining. The results demonstrated that hMSCs in HA-ADA hydrogels developed elaborate three-dimensional cytoskeletal networks with extensive filopodia and lamellipodia (Supplementary Video 1), while cells in HA-CA hydrogels maintained a spherical actin organization (Supplementary Video 2). Subsequently, 2D imaging of hMSCs further confirmed that cells cultured in HA-CA hydrogels remained predominantly rounded at both 24 and 72 h, whereas cells within HA-ADA hydrogels exhibited extensive spreading (Fig. 2b).

Fig. 2.

The HA-ADA hydrogel governs the differentiation fate of hMSCs. (a) 3D rendered images of F-actin cytoskeleton (phalloidin staining) in cells encapsulated within HA-CA and HA-ADA hydrogels. (b) Cytoskeleton staining images of the encapsulation of hMSCs within the HA-CA and HA-ADA hydrogel networks. (c) Representative images of ALP, ARS, and ORO staining after osteogenic or adipogenic induction for specified durations. (d) Western blot analysis of protein levels of osteogenic markers (Runx2, OCN) and adipogenic marker (PPARγ) in hMSCs harvested from HA-CA and HA-ADA hydrogels following induction for designated periods. (e) Quantitative analysis of Western blot results. n=3. (f) mRNA expression levels of RUNX2, ALP and PPARG determined by qRT-PCR. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Supplementary video related to this article can be found at https://doi.org/10.1016/j.bioactmat.2026.01.019

The following is/are the supplementary data related to this article.

To assess differentiation, hydrogel-encapsulated cells were cultured in either osteogenic or adipogenic induction medium for specified periods, followed by staining with alkaline phosphatase (ALP) activity, calcium deposits (Alizarin Red S, ARS), and lipid droplets (Oil Red O, ORO). The results demonstrated that hMSCs in HA-ADA hydrogels exhibited significantly enhanced ALP activity and increased ARS-positive mineralization nodules, but markedly reduced ORO-positive lipid accumulation compared to those in HA-CA hydrogels, indicating that HA-ADA hydrogels promoted osteogenic differentiation while suppressing adipogenic commitment of hMSCs (Fig. 2c).

Western blot analysis revealed that hMSCs in HA-CA hydrogels expressed lower levels of the osteogenic markers Runx2 and osteocalcin (OCN), but higher levels of the adipogenic marker PPARγ, compared to those in HA-ADA hydrogels (Fig. 2d and e). Consistent with these findings, qRT-PCR analysis confirmed that hMSCs in HA-ADA hydrogels expressed significantly higher mRNA levels of RUNX2 and ALP, and lower levels of PPARG (Fig. 2f). Collectively, these data demonstrated that the highly dynamic nature of HA-ADA hydrogels facilitates extensive cell spreading and directs the differentiation fate of hMSCs by promoting osteogenesis while inhibiting adipogenesis.

2.3. miR-376a-3p and miR-127-5p were downregulated in hMSCs encapsulated in HA-ADA hydrogels and converged on SETD7 as a common target

To elucidate the mechanism by which HA-ADA hydrogels regulate hMSC fate, we performed high-throughput miRNA sequencing on hMSCs cultured in HA-CA and HA-ADA hydrogels, given that miRNAs are key regulators of gene expression in biological systems through binding to target mRNAs, leading to translational repression or mRNA degradation [24,25]. Differentially expressed miRNAs were defined based on the criteria of |log2FC|≥1 and adjusted P <0.05. As shown in Fig. 3a, compared to HA-CA hydrogels, HA-ADA hydrogels upregulated 19 miRNAs and downregulated 13 miRNAs in hMSCs. Among these, miR-376a-3p and miR-127-5p exhibited the most pronounced downregulation. Specifically, miR-376a-3p expression decreased more than two-fold from an average of 130,328.3 to 54,178, while miR-127-5p also decreased more than two-fold from 5291 to 2003 (Fig. 3b). These results were subsequently validated by qRT-PCR in hMSCs within the HA-CA and HA-ADA hydrogels (Fig. 3c).

Fig. 3.

miR-376a-3p and miR-127-5p were downregulated in hMSCs encapsulated in HA-ADA hydrogels and converged on SETD7 as a common target. (a) Volcano plot depicting differentially expressed miRNAs from sequencing data of hMSCs encapsulated in HA-CA versus HA-ADA hydrogels. (b) Heatmap of miRNA expression profiles derived from sequencing of hMSCs in HA-CA and HA-ADA hydrogels. (c) Expression levels of miR-376a-3p and miR-127-5p in hMSCs cultured in HA-CA and HA-ADA hydrogels. n = 3. (d) Venn diagram showing overlapping target genes of miR-376a-3p and miR-127-5p predicted by TargetScan and miRWalk. (e, f) Dual-luciferase reporter assays for miR-376a-3p and miR-127-5p target validation. n = 3. (g) Predicted binding sites of miR-376a-3p and miR-127-5p on SETD7 3′UTRs, and schematic of the constructed SETD7 mutant variant. (h, i) Dual-luciferase reporter assays evaluating the interaction of miR-376a-3p (h) and miR-127-5p (i) with wild-type and mutant SETD7 sequences. n = 3. (j) Western blot analysis of SETD7, Runx2, and OCN protein levels in hMSCs transfected with miR-376a-3p or miR-127-5p mimics or inhibitors, followed by 4 days of osteogenic induction. (k–m) ALP, ARS, and ORO staining of hMSCs transfected with miR-376a-3p or miR-127-5p mimics or inhibitors after osteogenic or adipogenic induction for specified durations. (n–p) Quantitative analysis of ALP activity (OD values) (n), ARS-positive area (o), and ORO-positive area (p). n = 3. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Bioinformatics analysis using TargetScanHuman 8.0 and miRWalk predicted three common target genes for miR-376a-3p and miR-127-5p: SETD7, MBD5, and ABI2 (Fig. 3d). Subsequently, dual-luciferase assays confirmed that miR-376a-3p and miR-127-5p reduced luciferase activity of the SETD7 construct but not the MBD5 and ABI2 constructs (Fig. 3e and f). We then performed dual-luciferase reporter assays using wild-type and mutant SETD7 mRNA constructs (Fig. 3g–i). The results demonstrated that co-transfection with miR-376a-3p significantly reduced luciferase activity of the wild-type SETD7 construct but not the mutant. Similarly, miR-127-5p decreased luciferase activity of wild-type SETD7, with no significant effect on the mutant (Fig. 3i), indicating direct binding of both miRNAs to SETD7 3′UTR.

For further in vitro validation, we synthesized mimics and inhibitors for miR-376a-3p and miR-127-5p. Efficient cellular uptake of these mimics and inhibitors by hMSCs was confirmed (Fig. S5), providing a basis for subsequent experiments. Western blot analysis revealed that, compared to control, transfection with miR-376a-3p and miR-127-5p mimics significantly reduced SETD7 protein levels, with the effect of miR-376a-3p mimic being more pronounced. Conversely, inhibitors of miR-376a-3p and miR-127-5p markedly increased SETD7 protein expression, an effect that was most significant when both inhibitors were used in combination (Fig. 3j). Additionally, the mimics suppressed expression of the osteogenic markers Runx2 and OCN, while the inhibitors promoted their expression (Fig. 3j). We next evaluated osteogenic and adipogenic differentiation using ALP, ARS, and ORO staining. The results indicated that miR-376a-3p and miR-127-5p mimics significantly reduced ALP activity and ARS-positive area, whereas the inhibitors enhanced these parameters. The combination of both inhibitors produced the most robust promotion of osteogenesis. In contrast, the mimics increased the ORO-positive area, indicative of enhanced adipogenic differentiation, while the inhibitors decreased it (Fig. 3k–p). Collectively, these findings demonstrate that HA-ADA hydrogels upregulate SETD7 expression through downregulating miR-376a-3p and miR-127-5p.

2.4. Upregulated SETD7 in hMSCs promotes osteogenic differentiation

Previous research has identified SETD7 as a key promoter of osteogenic differentiation in hMSCs [26,27]; however, its potential role in mediating the HA-ADA hydrogel-induced regulation of hMSC lineage commitment remains unclear. hMSCs cultured in HA-CA and HA-ADA hydrogels were harvested and analyzed by Western blotting to determine SETD7 protein levels. The results revealed that hMSCs cultured in HA-ADA hydrogels expressed significantly higher levels of SETD7 compared to those in HA-CA hydrogels (Fig. 4a). To further investigate the functional role of SETD7 in osteogenic differentiation, we first overexpressed SETD7 in hMSCs and monitored their differentiation. Western blot analysis confirmed successful SETD7 overexpression (Fig. 4b). Subsequent evaluation demonstrated that SETD7-overexpressing hMSCs exhibited enhanced ALP activity, increased ARS staining, and reduced ORO staining, indicating that SETD7 promoted hMSCs toward osteogenic differentiation (Fig. 4c and d). Conversely, SETD7 was knocked down in hMSCs, with knockdown efficiency confirmed by Western blotting (Fig. 4e). SETD7-silenced hMSCs displayed reduced ALP activity, decreased ARS-positive areas, and markedly increased ORO-positive areas, indicating a shift toward adipogenic differentiation (Fig. 4f and g). Furthermore, we employed (R)-PFI-2, a potent and selective inhibitor of SETD7 methyltransferase activity [28,29], to assess its impact on hMSC fate. (R)-PFI-2 treatment significantly suppressed SETD7 methyltransferase activity (Fig. 4h). Consistent with genetic knockdown, inhibitor-treated cells exhibited diminished ALP activity, reduced ARS staining, and enhanced ORO staining compared to controls, confirming that pharmacological inhibition of SETD7 promotes adipogenic differentiation (Fig. 4i and j). Collectively, these results demonstrated that compared to the static HA-CA hydrogel, HA-ADA hydrogels enhanced SETD7 protein expression in hMSCs, thereby directing osteogenic differentiation.

Fig. 4.

Upregulated SETD7 in hMSCs promotes osteogenic differentiation. (a) Western blot analysis of SETD7 expression in hMSCs encapsulated within HA-CA and HA-ADA hydrogels. (b) Protein levels of SETD7 detected by Western blot after SETD7 overexpression in hMSCs. (c) ALP, ARS, and ORO staining of hMSCs overexpressing SETD7 following osteogenic or adipogenic induction for specified periods. (d) Quantitative analysis of ALP activity (OD values), ARS-positive area, and ORO-positive area from (c). n = 3. (e) SETD7 protein expression measured by Western blot after SETD7 knockdown in hMSCs. (f) ALP, ARS, and ORO staining of SETD7-knockdown hMSCs after osteogenic or adipogenic induction for designated durations. (g) Quantification of ALP activity (OD values), ARS-positive area, and ORO-positive area from (f). n = 3. (h) Measurement of SETD7 methyltransferase activity. (i) ALP, ARS, and ORO staining of (R)-PFI-2-treated hMSCs subjected to osteogenic or adipogenic induction for indicated time periods. (j) Quantitative analysis of ALP activity (OD values), ARS-positive area, and ORO-positive area from (i). n = 3. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

2.5. SETD7-mediated β-catenin methylation enhances its nuclear accumulation and requires binding to nuclear YAP for osteogenic activity

SETD7 is a lysine methyltransferase with a broad spectrum of substrates [21]. Studies have shown that SETD7 can methylate a range of non-histone proteins, including ERα [30], YAP [31], TAF-10 [32], and STAT3 [33], thereby modulating diverse cellular pathways and functions. However, the regulatory role of SETD7 in determining the differentiation fate of hMSCs remains unclear. Based on the miRNA sequencing results described above, Gene Ontology (GO) enrichment analysis identified the Wnt signaling pathway as significantly enriched among the predicted target genes (Fig. 5a). Wnt signaling is crucial for directing the osteogenic differentiation of MSCs [34,35]. Notably, β-catenin serves as the central signaling molecule in the canonical Wnt pathway [36]. We therefore hypothesized that SETD7 might promote the osteogenic differentiation of hMSCs by binding to and methylating β-catenin, enhancing its nuclear translocation, and consequently activating Wnt/β-catenin signaling.

Fig. 5.

SETD7-mediated β-catenin methylation enhances its nuclear accumulation and requires binding to nuclear YAP for osteogenic activity. (a) GO analysis of predicted miRNA targets (top 20 terms). (b) Co-IP of cytoplasmic β-catenin complexes from hMSCs in HA-CA and HA-ADA hydrogels, immunoblotted for SETD7 and methyl-lysine. (c) Nuclear β-catenin and H3 levels. (d, e) Quantification of (b) and (c). (f) Co-IP analysis following SETD7 overexpression (MOI = 20, 12 h transduction, 48 h culture). (g) Nuclear β-catenin levels after SETD7 overexpression. (h, i) Quantification of (f) and (g). (j) Co-IP analysis following SETD7 knockdown (MOI = 10, 24 h transduction, 48 h culture). (k) Nuclear β-catenin levels after SETD7 silencing. (l, m) Quantification of (j) and (k). (n, o) Nuclear YAP and H3 expression with quantification. (p, q) Co-IP analysis of nuclear YAP-β-catenin interaction in hMSCs cultured in HA-CA and HA-ADA hydrogels. (r) β-catenin/YAP co-localization by immunofluorescence. (s) β-catenin and YAP staining after gene silencing. (t, u) ARS staining and quantification of osteogenic differentiation following knockdown. n = 3. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

To test this hypothesis, we first performed co-immunoprecipitation (Co-IP) assays using hMSCs cultured in HA-CA and HA-ADA hydrogels. Analysis of β-catenin immunocomplexes revealed enhanced interaction between β-catenin and SETD7, along with elevated levels of β-catenin methylation, in hMSCs from HA-ADA hydrogels compared to those from HA-CA hydrogels (Fig. 5b–d). Consistently, nuclear β-catenin levels were significantly higher in HA-ADA-cultured hMSCs (Fig. 5c–e). To further validate the functional relationship between SETD7 and β-catenin, we overexpressed SETD7 in hMSCs and assessed changes in β-catenin methylation and nuclear localization. As shown in Fig. 5f and h, SETD7 overexpression markedly enhanced the association between SETD7 and β-catenin and increased β-catenin methylation. Correspondingly, nuclear β-catenin levels were significantly elevated upon SETD7 overexpression (Fig. 5g–i). Conversely, silencing SETD7 resulted in reduced SETD7-β-catenin interaction, decreased β-catenin methylation, and concomitant reduction in nuclear β-catenin levels (Fig. 5j–m).

YAP signaling plays a crucial role in mechanotransduction [37]. Given the importance of YAP in mechanotransduction, we next investigated whether YAP is involved in the SETD7-β-catenin signaling axis. First, we found that hMSCs in HA-ADA hydrogels exhibited higher levels of nuclear YAP compared to those in HA-CA hydrogels (Fig. 5n and o). Concurrently, the nuclear interaction between YAP and β-catenin was increased in cells from HA-ADA hydrogels (Fig. 5p and q). Immunofluorescence co-localization of YAP and β-catenin in hMSCs cultured in HA-CA and HA-ADA hydrogels further confirmed this observation (Fig. 5r). Subsequently, to explore whether nuclear β-catenin-YAP interaction is essential for downstream biological functions, we silenced YAP in hMSCs. Compared to controls, YAP silencing markedly reduced nuclear β-catenin-YAP binding and significantly attenuated osteogenic effects, and these effects could not be reversed by SETD7 overexpression (Fig. 5s–u). These findings indicate that nuclear β-catenin-YAP interaction is a critical step in the miRNA-SETD7 signaling axis. Collectively, these data demonstrate that SETD7 promotes nuclear translocation of β-catenin by mediating its methylation, where it forms a complex with YAP in the nucleus to drive osteogenic differentiation of hMSCs.

2.6. Dynamic hydrogel loaded with rMSCs promotes bone repair

Implantation of autologous MSCs represents a promising strategy for bone defect treatment. Due to the imbalanced regenerative microenvironment at the defect site, transplanted MSCs often fail to differentiate properly, resulting in poor therapeutic efficacy. In addition, maintaining the regulated differentiation capacity of MSCs during delivery remains a significant challenge. Based on our preceding findings, we demonstrated that dynamic hydrogels can significantly promote osteogenic differentiation of MSCs while inhibiting adipogenic differentiation. Mechanistically, dynamic hydrogels modulate matrix dynamics-sensitive miRNAs, which in turn regulate signaling pathways governing MSC fate determination. Accordingly, we proposed a therapeutic strategy that integrates MSC delivery with fate regulation for bone defect repair.

To evaluate the therapeutic potential of HA-ADA hydrogels loaded with rat bone marrow-derived mesenchymal stem cells (rMSCs) for bone defect repair in vivo, we established a rat model of lateral femoral condyle bone defect (diameter: 3.0 mm). The defects were filled with HA-ADA hydrogel alone (Control), ADA + MSCs, ADA + MSCs@mimics, or ADA + MSCs@inhibitors. Rats were euthanized at 4 and 8 weeks post-implantation, and femurs were harvested for micro-computed tomography (micro-CT) and histological analysis (Fig. 6a). First, H&E staining of the heart, liver, spleen, lungs, and kidneys at 8 weeks post-surgery revealed intact structures with no apparent organ toxicity (Fig. S6). As shown in Fig. 6b–d, micro-CT analysis at 4 weeks revealed that the ADA + MSCs group exhibited enhanced bone regeneration compared to the control group. This reparative effect was significantly attenuated in the ADA + MSCs@mimics group, whereas it was markedly potentiated in the ADA + MSCs@inhibitors group. These differences in repair efficacy became more pronounced at 8 weeks. H&E staining of the bone defect sites further confirmed these reparative effects (Fig. 6e). Subsequent immunohistochemical staining demonstrated that Runx2 and SETD7 protein levels were highest in the ADA + MSCs@inhibitors group, followed by the ADA + MSCs and ADA + MSCs@mimics groups, with the Control showing the lowest levels. Expression of both markers was further elevated at 8 weeks (Fig. 6f). Additionally, we collected callus tissue from the bone defect sites at 4 and 8 weeks for Western blot analysis, which revealed that nuclear β-catenin levels were elevated in the ADA + MSCs@inhibitors group compared to controls, followed by the ADA + MSCs and ADA + MSCs@mimics groups, consistent with the bone repair efficacy (Fig. S7). Collectively, these in vivo data demonstrate that the HA-ADA hydrogel promotes osteogenic commitment of MSCs and accelerates bone repair, potentially through a mechanism involving the downregulation of miR-376a-3p and miR-127-5p, thereby relieving their inhibitory effect on SETD7 expression.

Fig. 6.

Dynamic hydrogel loaded with MSCs promotes bone repair. (a) Schematic diagram of the animal experimental design. (b) micro-CT evaluation of femoral condyles collected from Control, ADA + MSCs, ADA + MSCs@mimics, and ADA + MSCs@inhibitors groups at 4 and 8 weeks post-implantation. (c) Quantitative analysis of bone mineral density (BMD) in the defect region at 4 and 8 weeks. n = 3. (d) Quantitative analysis of bone volume fraction (BV/TV) in the defect region at 4 and 8 weeks. n = 3. (e) Histological analysis of femoral defects by H&E staining. (f) Immunohistochemical staining of Runx2 and SETD7 in the femoral defect area at 4 and 8 weeks. Data are presented as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

3. Discussion

This study developed a dynamic HA-ADA hydrogel based on reversible host-guest crosslinking, which downregulates miR-376a-3p and miR-127-5p through mechanical cues, thereby derepressing SETD7. The upregulated SETD7 directly methylates β-catenin, promoting its nuclear translocation and association with YAP to drive osteogenic differentiation of stem cells. In vivo experiments confirmed that this hydrogel significantly enhanced bone defect repair, revealing a novel pathway whereby biomaterial mechanics regulate cell fate through epigenetic mechanisms (Fig. 7).

Fig. 7.

Schematic illustration of the HA-ADA hydrogel system for enhanced bone regeneration through dynamic matrix-mediated mechanotransduction. (a) Experimental workflow. (b) Chemical composition of hydrogels showing HA-ADA or HA-CA polymers decorated with RGD motifs, which crosslink with Ac-β-CD under blue light to form distinct network architectures. (c) Comparative illustration of hydrogel performance in bone regeneration. HA-ADA hydrogels exhibit high network adaptability that enables cell spreading through dynamic “gate opening” mechanisms involving actin polymerization and F-actin bundle formation, leading to enhanced bone regeneration. In contrast, HA-CA hydrogels display low network adaptability with restricted “gate closed” states that limit cell spreading and result in reduced bone regeneration capacity. (d) Molecular mechanism underlying HA-ADA hydrogel-promoted osteogenesis. The dynamic matrix downregulates miR-376a-3p and miR-127-5p expression, thereby relieving their repression of SETD7. Elevated SETD7 subsequently methylates β-catenin, facilitating YAP/β-catenin signaling cascade activation to drive osteogenic differentiation.

The core advantage of the designed HA-ADA hydrogel lies in the inherent “dynamic nature” of its network, which originates from its cross-linking mechanism based on host-guest interactions between CD and ADA. Unlike conventional static covalent cross-linking, the CD-ADA interaction is non-covalent and reversible, enabling continuous dynamics among the cross-linking units within the hydrogel network [38,39]. This physically dynamic property not only facilitates nutrient diffusion but, more importantly, provides appropriate mechanical cues to support cell spreading and osteogenic differentiation. However, while our study demonstrates that SETD7 upregulation promotes osteogenesis in the HA-ADA system, whether all dynamic hydrogels elicit similar effects remains uncertain. The specific range of dynamic mechanical properties likely plays a determining role. Different stress relaxation timescales, viscoelastic moduli, and network reorganization kinetics have been shown to differentially regulate cell behavior and fate decisions [40,41]. Our HA-ADA hydrogel may provide an optimal dynamic window for activating the miRNA-SETD7 axis, suggesting that systematic mapping of dynamic parameters and their correlation with SETD7 expression is essential for rational biomaterial design.

Through enrichment analysis of miRNA target genes, we found significant enrichment in the Wnt signaling pathway, providing a bioinformatics basis for exploring the interaction between SETD7 and the core Wnt protein β-catenin. Given SETD7's broad substrate spectrum, whether SETD7 can also interact with other mechanosensitive signaling molecules remains to be elucidated. For example, YAP, as a core transducer of mechanical signals, convert mechanical cues such as ECM stiffness and cell morphology into transcriptional responses [42,43], and their potential interaction with SETD7 may coordinately regulate stem cell fate. Additionally, the mechanosensitive ion channel Piezo1 plays important roles in various pathological processes, including neurological diseases, cardiovascular diseases, and cancer [44,45], and mediates atherosclerotic inflammation in endothelial cells through regulating YAP activation [46]. Future studies need to systematically investigate the interaction networks between SETD7 and these mechanosensitive molecules to comprehensively understand how dynamic mechanical signals coordinately regulate cell fate through multiple pathways.

The key translational significance of this work lies in the successful integration of “cell delivery” and “fate regulation” into a synergistic therapeutic strategy. Our in vivo experiments (Fig. 6) convincingly demonstrate the superiority of the “HA-ADA hydrogel + MSCs” approach. By implanting MSCs pre-treated with miRNA mimics or inhibitors into a rat bone defect model, we observed that inhibition of miR-376a-3p/127-5p further enhanced regeneration, while their overexpression compromised the therapeutic effect. This provides direct causal evidence that the miR-SETD7-β-catenin axis plays a central regulatory role in bone regeneration in vivo, signifying that the hydrogel functions as an intelligent platform that actively “instructs” MSCs.

Certain limitations warrant consideration. First, the upstream mechanosensory mechanisms responsible for the downregulation of these specific miRNAs remain to be identified. Second, the specific methylation site(s) on β-catenin targeted by SETD7 require further validation. Finally, the therapeutic potential merits evaluation in pathological models such as osteoporosis.

In conclusion, our HA-ADA hydrogels orchestrate stem cell fate by leveraging its dynamic properties to modulate a specific epigenetic program centered on SETD7, representing a new paradigm in regenerative biomaterial design.

4. Conclusion

In conclusion, we have identified a mechano-epigenetic pathway through which dynamic hydrogels regulate stem cell osteogenic differentiation. Our findings demonstrate that the dynamic HA-ADA hydrogels, characterized by reversible host-guest cross-linking, can transduce mechanical signals into epigenetic responses. Mechanistically, we observed that the dynamic matrix was associated with downregulation of miR-376a-3p and miR-127-5p expression in MSCs, leading to reduced translational repression of SETD7. Increased SETD7 expression was accompanied by enhanced β-catenin methylation and nuclear translocation, where β-catenin formed complexes with YAP to facilitate Runx2 transcriptional activation, thereby promoting osteogenic commitment. In vivo studies showed that the HA-ADA hydrogels significantly enhanced healing of rat femoral bone defects, with therapeutic outcomes that were notably influenced by miR-376a-3p and miR-127-5p levels. These results suggest that the hydrogel functions as an instructive platform that actively modulates cellular behavior, rather than serving merely as a passive scaffold. Our findings provide evidence that extracellular mechanical properties can be engineered to influence intracellular epigenetic programs. This work not only presents a promising strategy for bone regeneration but also contributes to the understanding of mechanotransduction-epigenetic interactions, potentially informing the design of biomaterials for various tissue engineering applications.

5. Experimental section

5.1. Materials and reagents

Sodium hyaluronate (NaHA, 40 kDa) was sourced from Shanghai Yuanye Bio-Technology Co., Ltd. (s12034). The following chemicals were obtained from J&K Scientific Ltd.: tetrabutylammonium hydroxide (TBAOH) (131688), di-tert-butyl dicarbonate (Boc₂O) (970158), sodium chloride (NaCl) (971431), 1-adamantaneacetic acid (ADA) (F228350), cholic acid (CA) (977474), sodium hydroxide (NaOH) (LGC0U69), and acryloyl chloride (A191395). β-Cyclodextrin (β-CD) (972167), dimethylformamide (DMF) (D639253), and triethylamine (TEA) (T431604) were purchased from Aladdin. Gelatin (G1890) and Dowex® resin (44514) were acquired from Sigma-Aldrich. Phenyl (2,4,6-trimethylbenzoyl) phosphinic acid lithium salt (LAP) was supplied by TCI (L0290).

5.2. Synthesis of acrylate β-cyclodextrin (Ac-β-CD)

β-CD (10 g) was dissolved in DMF (150 mL) and TEA (7 mL). The solution was cooled to 0 °C with stirring, and acryloyl chloride (5 mL) was added dropwise. After 6 h of reaction, the triethylamine hydrochloride salt was removed by filtration. The filtrate was concentrated to ∼20 mL by rotary evaporation and then dripped into acetone (600 mL) to precipitate the product. The precipitate was collected, thoroughly washed with acetone, and dried under vacuum for 3 days, yielding the Ac-β-CD with a DS of 1.31 as determined by ¹H NMR spectroscopy (Bruker Advance, 400 MHz, D₂O).

5.3. Synthesis of HA-ADA

The synthesis of HA-ADA was carried out based on a literature method, with slight modifications [7]. Briefly, NaHA (2.0 g) was first converted to its tetrabutylammonium salt (HA-TBA) via ion exchange using Dowex® resin, followed by neutralization with TBAOH in water to yield an aqueous solution (pH = 7.0). After lyophilization, the resulting HA-TBA became soluble in dimethyl sulfoxide (DMSO). HA-TBA (1.0 g, 1.4 mmol in disaccharide units, 1 equiv) was reacted with 1-adamantaneacetic acid (0.82 g, 4.2 mmol, 3 equiv) and 4-dimethylaminopyridine (DMAP) (0.13 g, 1.05 mmol, 0.75 equiv) in anhydrous DMSO (100 mL) under N₂. After dropwise addition of Boc₂O (3 equiv) at room temperature, the mixture was heated to 45 °C and stirred for 24 h. The reaction was quenched by extensive dialysis against DMSO, aqueous NaCl, and deionized (DI) water. Lyophilization afforded the final product, HA-ADA, as a white solid. The degree of adamantane modification ([X]%) was quantified by ¹H NMR spectroscopy (400 MHz, D₂O) via integration of the adamantane methylene protons (δ 1.50–1.85) against the HA backbone protons (δ 3.20–4.20).

5.4. Synthesis of HA-CA

HA-TBA (1.0 g, 1.4 mmol in disaccharide units, 1 equiv) was functionalized with cholic acid (1.72 g, 4.2 mmol, 3 equiv) and DMAP (0.13 g, 1.05 mmol, 0.75 equiv) in anhydrous DMSO (100 mL) under N₂. After dropwise addition of Boc₂O (3 equiv) at room temperature, the reaction proceeded at 45 °C for 24 h. The crude mixture was purified by sequential dialysis (DMSO, aqueous NaCl, DI water) and freeze-dried to afford the product as a white solid. The DS was determined by ¹H NMR spectroscopy, comparing the integral of the cholic acid methyl signal (δ 0.59) with the HA backbone protons (δ 3.20–4.20).

5.5. Preparation of hydrogels

HA-ADA/HA-CA, Ac-β-CD, and gelatin were dissolved in phosphate-buffered saline (PBS) at 37 °C to produce mixture solutions with fixed concentration of HA-ADA/HA-CA (1 % (w/v)), Ac-β-CD (5 % (w/v)), and gelatin (3 % (w/v)). Then initiator LAP was added at 0.05 % (w/v). The mixture was pipetted into polyvinyl chloride (PVC) molds at 37 °C, allowed to cool to 25 °C, and subsequently photo-crosslinked under 405 nm light (10 mW/cm², 3 min) to form supramolecular hydrogels.

5.6. SEM analysis

The hydrogels were flash-frozen in liquid nitrogen (5 min) and lyophilized for 3 days. The resulting xerogels were sectioned to expose the internal morphology for examination by SEM using a ZEISS Merlin FE-SEM instrument at an accelerating voltage of 5 kV.

5.7. Rheological analysis

All rheological tests were performed on a TA Instrument DHR 30 rheometer using a parallel-plate geometry (8 mm diameter, 1 mm gap). Time sweeps were conducted at 0.1 % strain and 1 Hz to monitor stability, while frequency sweeps employed a 1 % strain. Additionally, stress relaxation was evaluated by compressing gel discs (8 mm diameter, 2 mm thick) to 15 % strain at a rate of 0.3 mm/s.

5.8. Cell culture

hMSCs were expanded in α-MEM growth medium supplemented with 16.7 % fetal bovine serum (FBS), 1 % penicillin/streptomycin, and 1 % L-glutamine. The medium was replaced every 3 days, and cells were passaged at approximately 80 % confluence.

5.9. Osteogenic differentiation assay

Following encapsulation, all hydrogel constructs were cultured in osteogenic induction medium (1 mL per construct). The medium consisted of α-MEM supplemented with 16.7 % FBS (ExCell Bio, #FSP500), 1 % penicillin/streptomycin, 2 mM L-glutamine, 10 mM β-glycerophosphate disodium salt (Sigma-Aldrich, #517965C), 50 μg/mL L-ascorbic acid 2-phosphate (TCI, #G0097), and 0.1 μM dexamethasone (MedChemExpress, #HY-14648). The medium was replaced every other day.

5.10. miRNA sequencing and analysis

The small RNA sequencing libraries were subjected to paired-end 150 bp (PE150) sequencing. The raw sequencing data were initially assessed for quality using FastQC. Subsequently, adapter trimming, removal of terminal N-bases, and quality filtering (Q20) were performed with the fastp. The cleaned reads were then aligned to the Homo sapiens reference genome (GRCh38, Ensembl release 102) using the Bowtie. To enrich for miRNA sequences, reads aligning to ribosomal RNA (rRNA), transfer RNA (tRNA), and other non-coding RNAs annotated in the Rfam database were discarded. The remaining reads were quantified using the miRDeep2 package. Differential expression analysis was conducted with DESeq2, while the edgeR was employed for comparisons involving single replicates.

5.11. Quantitative Real-Time PCR (qRT-PCR)

Total RNA was extracted using TRIzol reagent, and quantified using a NanoDrop spectrophotometer with A260/A280 ratios of 1.9–2.1. RNA samples with concentrations >100 ng/μL were selected, and 1 μg of total RNA was reverse-transcribed into cDNA using the HiScript IV RT SuperMix kit. qRT-PCR was then performed on a QuantStudio 1 system with SYBR Green Premix and gene-specific primers (Table 1), under the following cycling conditions: 95 °C for 30 s; 40 cycles of 95 °C for 5 s and 60 °C for 30 s. Amplification specificity was verified by melt curve analysis, and relative gene expression was normalized to GAPDH and calculated via the 2^–ΔΔCt method.

Table 1.

Genes and primer sequences.

Gene	Forward primer (5′-3′)	Reverse primer (5′-3′)
PPARG	ACCAAAGTGCAATCAAAGTGGA	ATGAGGGAGTTGGAAGGCTCT
RUNX2	TTACTTACACCCCGCCAGTC	CACTCTGGCTTTGGGAAGAG
ALP	TAACATCAGGGACATTGACG	TGCTTGTATCTCGGTTTGAA
GAPDH	TCAAGGCTGAGAACGGGAA	TGGGTGGCAGTGATGGCA

5.12. miRNA transfection

Following cell seeding, transfection was performed when the cells reached approximately 70 % confluence and exhibited robust viability. Briefly, for transfection in 6-well plates, 6 × 10⁵ cells were seeded per well. Upon reaching the desired confluence, the culture medium was replaced. The transfection complexes were prepared by diluting 100 pmol of miRNA and 4 μL of LipoRNAi™ Transfection Reagent (Beyotime, C0535) in 125 μL of Opti-MEM® Medium. The mixture was vortexed and incubated at room temperature for 30 min to facilitate complex formation. The resulting complexes were then added dropwise to each well. After 8 h of incubation, the transfection efficiency was assessed using fluorescence microscopy. Subsequently, total RNA and protein were harvested for further analysis.

5.13. Dual-luciferase reporter assay

The dual-luciferase reporter assay was carried out following the manufacturer’s instructions (Dual-Luciferase Reporter Assay Kit, GenePharma). Cells were transfected at approximately 80 % confluence prior to analysis. The transfection mixture consisted of 2 μL of LipoRNAi™ Transfection Reagent (C0535, Beyotime), 0.5 μg of plasmid, and 2.5 μL of miRNA. After 24 h of transfection, the culture medium was aspirated, and the cells were gently washed with PBS. Subsequently, the cells were thoroughly lysed for subsequent analysis. Prior to the assay, the firefly and Renilla luciferase substrates (200 × concentrates) were briefly centrifuged and diluted to a 1 × working concentration with the supplied assay buffer. For the measurement, 50 μL of cell lysate was transferred to a microplate well, followed by the addition of 100 μL of the firefly Luciferase Reaction Reagent. The Firefly luciferase activity was measured immediately. Subsequently, 100 μL of the Renilla Luciferase Reaction Reagent was added to each well, and after brief mixing, the Renilla luciferase activity was quantified. The entire detection process was completed within 30 min, after which the data were analyzed.

5.14. Western blot analysis

Protein lysates were prepared in RIPA buffer supplemented with 1 % protease inhibitor, denatured in SDS loading buffer, and resolved by SDS-PAGE on 10–12.5 % gradient gels. Proteins were then electrophoretically transferred to PVDF membranes at 400 mA for 20–30 min (4 °C) using a rapid transfer buffer. After blocking with 5 % non-fat milk for 1 h at room temperature, membranes were incubated with primary antibodies including β-catenin (1:800, Santa Cruz Biotechnology, #sc-7963), YAP (1:1000, ABclonal, #A19134), SETD7 (1:1000, Proteintech, #24840-1-AP), methylated lysine (1:1200, Abcam, #ab23366), Runx2 (1:1000, ABclonal, #A2851), OCN (1:1000, ABclonal, #A20800), and PPARγ (1:1000, ABclonal, #A11183) overnight at 4 °C, followed by HRP-conjugated secondary antibodies for 1 h at room temperature. Signal detection was performed using a chemiluminescent substrate and captured on a ChemiDoc MP imaging system.

5.15. microCT scanning and analysis

The harvested femurs were fixed in 4 % paraformaldehyde for 7 days at 4 °C. Subsequently, the samples were rinsed twice with PBS, kept hydrated, and wrapped in plastic film for micro-CT analysis. Scanning was performed using a high-resolution micro-CT system (PerkinElmer, USA) with the following parameters: 90 kV, a 72 mm field of view (FOV) for both acquisition and reconstruction, and a 0.06 mm copper + 0.5 mm aluminum X-ray filter. Each scan was conducted in high-resolution mode with a 4-min acquisition time. Three-dimensional reconstruction and subsequent morphometric analysis were carried out using the manufacturer’s proprietary software.

5.16. Paraffin-prepared sections and immunohistochemistry

Following micro-CT scanning, the bone samples were decalcified using a rapid decalcification solution (PhennoVision Bio, #PVB-3002) to facilitate subsequent sectioning. The decalcified samples were then dehydrated in an automated tissue processor (Leica, HistoCore PEARL) according to the manufacturer's standard protocol. Subsequently, the dehydrated tissues were embedded in paraffin blocks using an automated embedding system (Leica, HistoCore Arcadia H). Sections with a thickness of 5 μm were prepared from the paraffin-embedded blocks using a fully automatic rotary microtome (Leica). After baking, the sections were subjected to immunohistochemical staining. For immunostaining, nonspecific binding sites were first blocked with horse serum, followed by incubation with the respective primary antibodies.

5.17. 3D cell culture in hydrogels

The hydrogel precursor solution was composed of HA-CA (1 % w/v), HA-ADA (1 % w/v), porcine skin-derived gelatin (3 % w/v, Sigma-Aldrich, #V900863), and Ac-β-CD (5 % w/v, DS = 1.31) dissolved in PBS. The components were thoroughly mixed by centrifugation at 500×g for 3 min. hMSCs were encapsulated at a density of 1 × 10⁷ cells/mL within the hydrogel, which was subsequently cross-linked by exposure to blue light for 3 min. The cell-laden constructs were cultured in osteogenic induction medium. For each hydrogel, 1 mL of growth medium or osteogenic medium was supplemented. The osteogenic medium consisted of basal medium supplemented with 10 mM β-glycerophosphate, 50 μg/mL ascorbic acid 2-phosphate, and 100 nM dexamethasone, and was refreshed every 2 days. Samples were harvested on day 3 for assessment of osteogenic differentiation.

5.18. Live/dead staining assay

The viability of cells encapsulated in hydrogels was evaluated after 24 and 72 h using a live/dead assay kit (Calcein-AM/PI, Beyotime, #C2015S). Following the manufacturer’s protocol, hydrogels were briefly rinsed with PBS and incubated with the staining solution (300 μL) at 37 °C for 45 min in the dark. After removal of the dye and a final PBS wash, live (Calcein-AM, green) and dead (PI, red) cells were immediately imaged by fluorescence microscopy.

5.19. Isolation and culture of Rat mesenchymal stem cells (rMSCs)

rMSCs were aseptically isolated from one-week-old male SD rats. Following euthanasia and sterilization, femora were excised, and the bone marrow was flushed out using L-DMEM. The marrow flushate was centrifuged (1000 rpm, 10 min), and the pellet was washed and resuspended in complete growth medium (L-DMEM supplemented with 10 % FBS, 2 mM L-glutamine, 1 % NEAA, 100 U/mL penicillin, and 100 μg/mL streptomycin). Cells were seeded in a 25 cm² flask and cultured at 37 °C with 5 % CO₂. After 24 h, non-adherent cells were removed by medium change, and the adherent population was defined as rMSCs.

5.20. Adipogenic differentiation assay

Adipogenic differentiation of hMSCs was induced in vitro according to established protocols [47]. Cells were seeded in 6-well plates (2.5 × 10⁶ cells/well) and cultured for 14 days in adipogenic induction medium (α-MEM supplemented with 10 % FBS, 0.5 mM IBMX, 5 μg/mL insulin, and 1 μM dexamethasone), with medium changes every 48 h. Successful differentiation, marked by lipid droplet formation, was confirmed by ORO staining.

5.21. Osteogenic differentiation and mineralization staining

The osteogenic differentiation of hMSCs was induced by culturing the cells in 24-well plates at a density of 5 × 10⁵ cells per well using osteogenic induction medium, supplemented with 50 μg/mL ascorbic acid, 100 nM dexamethasone and 10 mM β-glycerophosphate for 14 days. Following induction, cell lysates were homogenized and subjected to an ALP activity assay via a colorimetric enzymatic kit (Roche), which spectrophotometrically quantifies the release of p-nitrophenol. Concurrently, the culture media were collected to determine secreted osteocalcin levels using an immunoassay kit (DiaSorin).

To evaluate mineralization, hMSCs were seeded in 6-well plates at 2.5 × 10⁶ cells per well and maintained in the same induction medium for 21 days. The resulting mineralized matrix was stained with 2 % Alizarin Red S (Sigma-Aldrich, pH 4.2) and visualized using a Nikon Diaphot Inverted Microscope equipped with a digital camera.

5.22. Critical bone defect model

All animal experiments were conducted following the protocols approved by the Institutional Animal Care and Use Committee (IACUC, approval No. 5032 [2025]) at Tongji Medical College, Huazhong University of Science and Technology. To establish the bone defect model, 8-week-old Sprague-Dawley rats were acclimatized for 1 week prior to surgical procedures. General anesthesia was induced via isoflurane inhalation. Following anesthetic onset, the left lateral femoral region was shaved, sterilized, and draped. A longitudinal incision was made along the femoral midline to expose the distal femur. A critical-sized bone defect (3 mm in diameter) was created in the lateral cortical bone using a dental drill under constant irrigation with ice-cold saline to prevent thermal necrosis. After the implantation of the hydrogel-cell constructs, the surgical site was sutured in layers. Postoperative analgesia was maintained by subcutaneous administration of meloxicam (1 mg/kg daily) for 3 consecutive days. Femoral samples were harvested at 4 and 8 weeks post-implantation following euthanasia (sodium pentobarbital, 100 mg/kg). The samples were fixed in 4 % formaldehyde for subsequent micro-CT analysis and histological examination.

5.23. Statistical analysis

Statistical analyses were performed using GraphPad Prism 10.0. Normality (Kolmogorov-Smirnov test) and homogeneity of variance (Levene's test) were assessed first. Data meeting both assumptions were analyzed by one-way ANOVA with Tukey's post hoc test; data with normal distribution but unequal variances were analyzed by ANOVA followed by the Games-Howell test. Non-normally distributed data were subjected to the Kruskal-Wallis test.

CRediT authorship contribution statement

Xudong Xie: Writing – review & editing, Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Liangcong Hu: Writing – review & editing, Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Yueman Zhang: Writing – review & editing, Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Bobin Mi: Writing – review & editing, Writing – original draft, Validation, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Xiaoyue Xu: Software, Methodology, Formal analysis, Data curation. Chong Ding: Software, Investigation, Formal analysis, Data curation. Yiming Li: Software, Methodology, Formal analysis, Data curation. Fawwaz Al-Smadi: Writing – original draft, Software, Investigation, Conceptualization. Xiangyu Chu: Software, Methodology, Data curation. Yuan Xiong: Software, Methodology, Data curation. Kunyu Zhang: Writing – review & editing, Visualization, Validation, Supervision, Project administration, Funding acquisition, Conceptualization. Liming Bian: Writing – review & editing, Visualization, Validation, Supervision, Methodology, Funding acquisition, Conceptualization. Guohui Liu: Writing – review & editing, Visualization, Validation, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Data availability statement

The data in this study may be requested from the authors.

Ethics approval and consent to participate

All animal procedures were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee of Tongji Medical College, Huazhong University of Science and Technology (IACUC, approval No. 5032 [2025]).

Declaration of competing interest

The authors declare no conflict of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article: