Regulatory mechanism of stem cell scaffold porosity and stiffness on the processes of intramembranous ossification and bone regeneration

Abstract

Bone defects remain a significant clinical challenge due to the limited regenerative capacity. Biomaterials can offer effective therapeutic approaches to address this challenge, where their physical properties play a critical role in determining the success of regeneration. Here, we fabricated cryogelated 3D porous scaffolds with independently controlled pore sizes and stiffness. These scaffolds were then seeded separately with bone marrow stromal cells (BMSCs) and MLO-A5 cells to create osteogenic cell-scaffold composites. To determine the optimal physical properties for promoting bone regeneration, we evaluated the scaffolds using both in vitro osteogenic cultures and a rat critical-sized calvarial defect model. Transcriptomic analysis was performed concurrently to explore the molecular link between physical scaffold cues and osteogenic differentiation. The results indicated that the small-pore high-stiffness scaffold exhibited the optimal osteogenic performance in vitro. In rats, the small pore scaffold groups also showed superior bone regeneration. Furthermore, transcriptomic analysis revealed that scaffold pore size significantly changed the expression of Apolipoprotein E (ApoE) and Osteoglycin (OGN). Our findings suggest a feasible strategy to enhance the efficiency of bone regeneration by leveraging the properties of tailored scaffolds, enabling personalized repair approaches.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s13287-026-05170-7.

Introduction

Bone defects often result from trauma or diseases such as osteoporosis, bone tumors, and metabolic disorders [1–3]. Clinical studies report that about 15 to 26% of the patients had traumatic bone defects following open long bone fractures [4, 5]. A global epidemiological data estimate more than 178 million new fractures each year, which means tens of millions of people suffer from pain and functional impairment caused by bone defects [6, 7]. Although bone has an intrinsic capacity for self-repair, the spontaneous healing is often insufficient for critical-size defects—the smallest intraosseous wound that fails to heal spontaneously [8]. This clinical challenge has therefore driven increasingly exploring bone tissue regeneration as a promising therapeutic pathway.

Bone regeneration relies on a precisely coordinated sequence of cellular events. Mesenchymal stem cells (MSCs) possess a potent capacity for osteogenic lineage differentiation among these, as the core cellular population initiating and sustaining regenerative process. In response to osteoinductive signals such as bone morphogenetic proteins (BMPs), MSCs proliferate and commit to osteoprogenitor cells. These progenitors then mature into osteoblasts which synthesize and secrete bone matrix. Eventually, a group of osteoblasts embedded within the new matrix and terminally differentiate into osteocytes. This osteogenic process is regulated through a series of signaling events. For instance, Runx2 and Osterix form a core transcriptional network that promotes osteogenesis. They jointly drive the expression of genes associated with production of bone matrix—COL1A1, ALP, and OCN, dynamically reflecting the progression of MSC osteogenic differentiation [9]. While MSCs are commonly used to study osteogenesis and for therapeutic purposes, MLO-A5 cells are often chosen to model the mineralization stage of osteogenic differentiation. As a postosteoblast-preosteocyte cell line, MLO-A5 can secrete abundant bone matrix proteins in vitro and mineralize rapidly [10]. Surprisingly, recent findings also indicate that MLO-A5 is not fully terminally differentiated, exhibiting significant mesenchymal lineage plasticity. It is capable of differentiating into both adipocytes and chondrocytes [11]. This lineage plasticity, a feature MLO-A5 shares with MSCs, is what makes it a particularly suitable counterpart for modeling in vivo osteogenesis. MSCs and MLO-A5 can respectively simulate the early and late stages of bone formation, while also capturing the inherent multi-lineage complexity of osteogenic process. However, successful bone regeneration not only depends on cellular differentiation but also on the intricate cellular microenvironment where these processes take place.

Bone tissue regeneration technology is fundamentally about recreating a biomimetic microenvironment with functional cells and a biocompatible physicochemical framework. This approach simulates the natural bone healing process, ultimately leading to the regenerative repair of bone tissue. To create such an environment, three-dimensional (3D) porous scaffolds are highly effective because they provide a physical structure that closely mimics the native extracellular matrix (ECM). Compared with conventional hydrogel systems [12, 13], this interconnected porous network structure more effectively supports biological activities including cell migration, communication, and interaction [14]. Natural organic biopolymers such as collagen and chitosan are frequently used to construct scaffolds due to their excellent biocompatibility and biological activity. However, these materials generally offer inferior mechanical properties. More critically, their macroscopic physical properties—such as pore size and stiffness—are strongly coupled, making it difficult to regulate them independently [15]. While, electrospun nanofiber scaffolds offer highly controllable composition and function [16, 17], achieving precise control over their microscale pore architecture—specifically its uniformity and stability—remains a significant challenge [18].

Scaffold pore size [19] and stiffness [20] are recognized as two vital parameters regulating osteogenic differentiation of MSCs. Stiffness is one of the most direct mechanical signals perceived by MSCs through mechanotransduction [21]. MSCs use their internal actin cytoskeleton to exert contractile forces on their surroundings to sense scaffold stiffness. For example higher stiffness significantly promotes the osteogenic differentiation of MSCs [22–24]. The influence of pore size is more complex. It is critical in deciding whether MSCs can survive and organize within the scaffold [25]. Excessively small pores provide a large adhesion surface area, but they can impede cell migration, nutrient transport, and metabolite removal. More importantly, they restrict vascularization, leading to cell death due to hypoxia and nutrient deficiency, ultimately causing regeneration to fail [26]. Conversely, oversize pores reduce the scaffold mechanical properties and cell adhesion efficiency. Regenerating cells, lacking structural sufficient support, struggle to form continuous tissue, resulting in slow matrix deposition and delayed tissue healing [27]. Therefore, pore size requires finding an optimal balance: providing enough space for cell growth and vascularization while maximizing the available adhesion surface area for cells adhesion [28]. Although the effects of stiffness or pore size have been studied individually, there is a lack of understanding of how the interplay between these two key physical parameters affects the MSCs osteogenic efficiency. present a compelling solution. This technique uses cryoprotectants as porogens to control the growth of ice crystals, which then template an interconnected porous network. Crucially, this method enables the separate control of scaffold pore size and stiffness at a low cost. This ability to independently control physical properties makes the scaffold an ideal platform for systematically investigating how cells respond to specific physical cues in a 3D environment.

To investigate how the interaction between these two key physical parameters influences the osteogenic efficiency of mesenchymal stem cells (MSCs), our research team, building upon previous studies [29], successfully developed a novel bone organoid platform—termed the cell-scaffold composite—by innovatively employing cryoprotectants to regulate ice crystal growth and combining them with crosslinkers to independently modulate pore architecture and mechanical stiffness. The composite was constructed by seeding bone marrow stromal cells (BMSCs) onto a 3D porous scaffold fabricated via DMSO-regulated cryogelation. We then evaluated its efficacy in promoting osteogenesis and enhancing bone regeneration, while identifying the physical properties of materials that could yield superior therapeutic outcomes. We also performed transcriptomic analysis on the cell-scaffold composite during osteogenic differentiation to investigate how its physical properties influence therapeutic molecular mechanisms. Our research ultimately utilized this composite to construct a controllable bone regeneration model for rat calvarial osteogenesis. This approach not only provides a viable solution for enhancing the efficiency and quality of bone regeneration but also offers new insights for developing personalized treatment.

Results

Study design

Figure 1 illustrates the overarching schema of our investigation. By varying the concentrations of the cross-linker glutaraldehyde (GA) and the cryoprotectant dimethyl sulfoxide (DMSO), we achieved separate control over the scaffold’s stiffness and pore size. This strategy enabled us to fabricate four distinct groups of cryogelated 3D porous scaffolds with independently tuned physical properties. To compare the responses of early and late-stage bone cells, we cultured rat BMSCs and MLO-A5 cells separately on each of the four distinct scaffold groups. This process yielded our target constructs: cell-scaffold composites poised for osteogenic differentiation. The composites were then cultured in an osteogenic induction medium for 7 to 21 days. Finally, their regenerative capabilities were rigorously assessed using a dual-validation approach that combined in vitro assays with in vivo functional tests. For the in vitro analysis, we examined the cell-scaffold composites via molecular biology assays and histological staining to validate their osteogenic differentiation at the gene expression and protein histological morphology. And for the in vivo functional validation, the four groups of cell-scaffold composites pre-cultured with BMSCs for 14 days, were implanted into a rat critical-size calvarial defect model to assess their therapeutic efficacy.

Fig. 1.

Schematic of the experimental outline. The left section describes the four groups of the cryogelated 3D porous scaffold with varying in pore size and stiffness. Notation: large-pore/low-stiffness (L/l); large-pore/high-stiffness (L/h); small-pore/low-stiffness (S/l); small-pore/high-stiffness (S/h). The right section details in vivo and in vitro experiments validating the bone regeneration capacity of the cell-scaffold composite. Created with BioRender.com

Fabrication and characterizations of the cryogelated 3D porous scaffold

The cryogelated 3D porous scaffolds were fabricated using a cryogelation method that enables the separate control of pore size and stiffness. Briefly, a precursor solution was prepared by mixing gelatin with the cross-linker GA and varying concentrations of the cryoprotectant DMSO. The solution then underwent cryogelation at -20℃ for 16 h to allow for polymer network crosslinking. Finally, the resulting cryogels were thawed, washed to remove residual reagents, and freeze-dried to yield the porous scaffolds (Fig. 2A).

Fig. 2.

Construction and Characterization of the Cryogelated 3D porous scaffold. A Fabrication and physical characterization of cryogelated 3D porous scaffolds. Created with BioRender.com B SEM images of the cryogelated 3D porous scaffolds, demonstrating the correlation between pore size and DMSO concentration. Scale bar: 100 μm. C Column Scatter Plot of Pore size from SEM image analysis revealed a difference in scaffold pore size resulting from the varied DMSO concentration. D Column Scatter Plot of results from AFM test revealed difference in local scaffold stiffness from the varied GA concentration. (center line) (n ≥ 11, ANOVA). E Quantification of BMSC viability and proliferation on the four scaffold groups at days 1, 3, and 7 using a CCK-8 assay. F Live/Dead fluorescence microscopy of BMSCs cultured on the scaffolds for 3 days. Live cells are stained green (Calcein-AM), and dead cells are stained red (Ethidium Homodimer-1). Scale bar: 100 μm

The principle of pore size control is based on the function of DMSO as a cryoprotectant. As the precursor solution freezes, the growth of ice crystals leads to the enrichment of DMSO in the remaining unfrozen liquid phase. This continues until the DMSO concentration reaches a critical threshold of ~ 40%, at which point the freezing point of the surrounding solution drops to the ambient − 20℃ and further ice crystal growth is halted. The final pores are formed in situ as templates of these ice crystals. Therefore, by varying the initial DMSO concentration, we could precisely control the extent of ice crystal growth and thus the final scaffold pore size. Scaffold stiffness, meanwhile, was independently regulated by the concentration of the GA cross-linker.

SEM analysis of the four scaffold groups visually revealed the relationship between DMSO concentration and pore size (Fig. 2B). Quantitative analysis confirmed that pore size was dictated by the DMSO concentration. Scaffolds fabricated with high concentration DMSO (the S/l and S/h groups) exhibited significantly smaller pores than those made with low concentration DMSO (the L/l and L/h groups). Conversely, pore size was unaffected by the concentration of GA at any DMSO level. This result confirms the successful separate control of two properties (Fig. 2C).

To characterize the stiffness that cells experience within the scaffold microenvironment, we measured the local stiffness of fragmented cryogelated scaffolds via atomic force microscopy (AFM). The local Young’s modulus was calculated from the force-displacement curves to quantify this property.The AFM measurements revealed that scaffold stiffness was governed by the GA concentration. The high GA groups (L/h and S/h) were significantly stiffer than their low GA counterparts (L/l and S/l). And stiffness was not significantly affected by the initial DMSO concentration (Fig. 2D).

Having established the physical properties of the scaffolds, we next evaluated their biocompatibility. To ensure the scaffolds provided a suitable microenvironment for cell growth, we first assessed their biocompatibility. We quantified BMSCs proliferation on the four scaffold groups at days 1, 3, and 7 using a CCK-8 assay. Absorbance values (OD at 450 nm) increased significantly over time across all four groups, demonstrating robust proliferative activity on all materials. Importantly, no significant differences in absorbance were observed between the four groups at any given time point. This suggests that the base material is inherently biocompatible and that the tested variations in stiffness and pore size do not induce cytotoxicity (Fig. 2E).

We performed a Live/Dead assay on day 3 of culture to further visualize cell viability. Fluorescence microscopy confirmed the high viability of cells across all scaffold groups. The images showed that the vast majority of cells were alive, stained green by Calcein-AM, while only a few dead cells were visible, stained red by Ethidium Homodimer-1 (Fig. 2F). These quantitative and qualitative results demonstrate the excellent biocompatibility of all four scaffold groups.

In summary, the combination of tunable physical properties and excellent biocompatibility makes our cryogelated 3D porous scaffolds a reliable platform for the subsequent investigations.

Regulation of ossification by material properties in vitro

To investigate how different scaffold physical properties regulate osteogenic potential, we performed a series of in vitro assays. We used rat BMSCs and MLO-A5 cells, which were seeded separately into the four distinct scaffold groups to form our cell-scaffold composites. These composites were then maintained under osteogenic induction conditions for 7 to 21 days, and their subsequent differentiation was assessed by molecular and histological analyses (Fig. 3A).

Fig. 3.

In vitro experimental schematic and results. A Schematic of the in vitro osteogenic differentiation of the cell-scaffold composite. Created with BioRender.com. B, C mRNA expression levels of osteogenic markers in BMSCs at B day 7 and C day 14 of osteogenesis. D mRNA expression levels of late-stage osteogenic markers in BMSCs following 14 days of osteogenic induction. E ALP, OCN, and VonKossa staining of BMSCs after 14 days of osteogenic induction. F mRNA expression levels of osteogenic markers in MLO-A5 at day 21 of osteogenesis. G Representative images of ALP, OCN, and VonKossa staining in MLO-A5 following 21 days of osteogenic induction. Scale bar: 100 μm

We focused on BMSCs as our seed cells to model the early stages of osteogenesis, because they are upstream progenitors in this lineage. We first assessed early osteogenic differentiation on day 7. We then used qPCR to measure the gene expression of key osteogenic markers, Runx2 and Osterix (Osx), and the early marker ALP (Fig. 3B). At this early time point, the expression of all three genes, Runx2, OSX and ALP, was significantly higher in the small-pore composite groups compared to the large-pore groups. These results indicate that while all composites had initiated an osteogenic gene program by day 7, the small-pore scaffolds demonstrated superior early osteogenic potential.

Recognizing that osteogenic differentiation is a dynamic process, we extended our analysis to day 14. At this mid-stage of differentiation, we again measured the expression of Runx2, Osx, and ALP, and additionally included the matrix synthesis marker COL1A1. The expression levels of all four markers in the small-pore composite groups were substantially higher than the large-pore groups at mid-stage. Furthermore, the expression difference for Runx2, OSX, and ALP was more pronounced on day 14 than on day 7 (Fig. 3C). This trend extended to the late stage of osteogenesis. We observed a similar significant upregulation of the mature osteoblast marker, Osteocalcin (OCN), in the small-pore scaffolds (Fig. 3D). Collectively, these findings demonstrate that the small-pore BMSC-scaffold composites exhibit superior osteogenic potential throughout all tested stages of differentiation. Overall, these findings show that small-pore BMSCs-scaffold composites have superior osteogenic potential at all stages of tested differentiation.

Synthesizing the results from Fig. 3B–D, it indicates that the S/h group consistently exhibited the strongest expression of osteogenesis-related genes and significantly surpassed the other three groups. The only exception was observed for Runx2 expression on day 7, where no single group displayed a clear advantage.

To complement the molecular analysis, we next performed a histological evaluation of the osteogenic capacity of the BMSC-scaffold composites. On day 14 of osteogenic culture, we performed Alkaline Phosphatase (ALP) staining to determine if the composites had entered an early osteogenic state. We also conducted Osteocalcin (OCN) and Von Kossa staining to evaluate late-stage osteoblast activity and matrix mineralization.

The qualitative staining results were consistent with the gene expression data. In the ALP staining, the S/h group formed the most widespread and dark blue deposits. In contrast, the other three groups, especially the large pore groups formed far fewer blue-purple nodules. For the late-stage markers, OCN staining showed clear brown deposits in the S/h group, Von Kossa staining also showed more black mineralized plaques in this group. The comparison of both stains indicated that the S/h group was more positive than the S/l group and significantly more positive than the L/h and L/l groups (Fig. 3E). These histological findings corroborate our molecular data. They can confirm that the S/h group possesses the highest osteogenic potential throughout the entire process of bone regeneration.

Generality of the optimal physical properties

To investigate if the optimal physical properties of our scaffold were broadly applicable, we selected a different cell type, the MLO-A5 cell line. This approach was designed to test whether the S/h group would maintain its osteogenic advantage when modeling bone regeneration initiated from the late-stage osteogenic cell. We quantified the gene expression of Runx2, ALP, COL1A1, and OCN in the MLO-A5-scaffold composites after 21 days of osteogenic culture. Consistent with the BMSC results, the S/h group exhibited the highest expression of these osteogenic genes compared to the other three groups (Fig. 3F).

We also performed histological analysis on the MLO-A5-scaffold composites after 21 days of culture, including staining for Collagen Type I (Col-1), Masson’s Trichrome, and Alizarin Red. Col-1 staining showed that the S/h group formed evenly distributed and dense brown positive regions, confirming active bone matrix synthesis. Similarly, the S/h group displayed the largest area of positive blue (Masson) and red (Alizarin) staining (Fig. 3G).

These results confirm the generality of the small-pore, high-stiffness cryogelated 3D porous scaffold as it maintains its superior osteogenic performance regardless of the initial cell type used.

Mechanistic insights into material property-driven osteoinduction

To systematically investigate the molecular mechanisms by which pore size and stiffness independently regulate osteogenic differentiation, we first performed transcriptome sequencing across all experimental conditions, using two-dimensional (2D) cultured cells as the control group for comparative analysis. Pearson correlation analysis (Fig. 4A) revealed that, compared to the 2D control, the other four groups (comprising distinct combinations of pore size and stiffness) exhibited distinct yet well-clustered transcriptional profiles, suggesting that variations in material physicochemical properties (specifically pore size and stiffness) drive significant transcriptional divergence.

Fig. 4.

Molecular mechanisms of osteogenesis on scaffolds with varying properties. A Pearson correlation of the groups. Analysis of Kyoto Encyclopedia of Genes and Genomes (KEGG) of B L/l, C L/h, D S/l, E S/h VS 2D BMSC (Ctrl). Volcano plot of differentially expressed genes (DEGs) of F S/l vs. L/l, G S/h vs. L/h, H L/h vs. L/l and I S/h Vs S/l. J Venn diagram analysis of differentially expressed genes (DEGs) among A (S/l vs. L/l), B (S/h vs. L/h), C (S/l vs. L/l) with D (S/h vs. L/l). K A heatmap illustrating the specific DEGs (In J) between the four groups. mRNA expression levels of Apoe in BMSC following 7 days (L) and 14 days (M) of osteogenic induction. mRNA expression levels of OGN in BMSC following 7 days (N) and 14 days (O) of osteogenic induction

To validate the osteoinductive capacity of the scaffold materials, we conducted Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis comparing the four experimental groups with the 2D control. The results demonstrated that: under large-pore conditions, both the low-stiffness (L/l group, Fig. 4B) and high-stiffness (L/h group, Fig. 4C) scaffolds significantly enriched the “ECM-receptor interaction” and “Cytokine-cytokine receptor interaction” pathways—key candidates for mediating scaffold-regulated osteogenic differentiation. In contrast, small-pore conditions exhibited broader pathway enrichment: small-pore low-stiffness (S/l group, Fig. 4D) and small-pore high-stiffness (S/h group, Fig. 4E) scaffolds not only commonly enriched canonical osteogenesis-related pathways (“PI3K-Akt signaling pathway” and “Focal adhesion”) but also uniquely highlighted the “Calcium signaling pathway” in the S/l group, which is directly linked to mineralization and calcification. These findings not only confirm the robust osteoinductive activity of our scaffold designs but also reveal that the small-pore high-stiffness group (S/h) demonstrates superior osteogenic potential among the tested conditions.

To dissect the individual contributions of pore size and stiffness, we performed targeted comparative analyses of the four groups based on these two parameters (Fig. 4G and I). Specifically, Fig. 4G presents the volcano plot of differentially expressed genes (DEGs) from the small-pore low-stiffness group (S/l) versus the large-pore low-stiffness group (L/l), clearly depicting the top 10 upregulated DEGs (blue), top 10 downregulated DEGs (yellow), and non-significant genes (white).

We then integrated the top 10 upregulated and downregulated DEGs from four key comparisons: Fig. 4G (S/l vs. L/l), Fig. 4F (S/h vs. L/h), Fig. 4G (S/l vs. L/l), and Fig. 4F (S/h vs. S/l). Venn diagram analysis (Fig. 4J) of these DEGs identified potential core regulatory factors under the influence of single parameters. The results showed that: in the pore size comparison groups (A: S/l vs. L/l; B: S/h vs. L/h), ApoE and Lyz2 were consistently enriched; in the stiffness comparison groups (C: L/h vs. L/l; D: S/h vs. S/l), LOC100910944 (Romo1) and NEWGENE_1308171 (OGN) were commonly enriched. These four genes are likely the primary regulators driving the observed differences in osteogenic differentiation efficacy across the four material groups.

To further characterize their regulatory roles, we performed hierarchical clustering analysis (heatmap, Fig. 4K) on all top 10 upregulated and downregulated DEGs from groups A, B, C, and D. Integrating literature evidence and heatmap patterns, we focused on the two pore-size-associated DEGs (ApoE and Lyz2): compared to Lyz2, ApoE has been more extensively implicated in osteogenic processes, suggesting its functional relevance. qPCR validation confirmed ApoE as a critical regulator of pore-size-dependent osteogenic differentiation. As shown in Fig. 4L, at day 7 of induction, ApoE expression was significantly higher in small-pore groups (S/l and S/h) than in large-pore groups (L/l and L/h). Intriguingly, by day 14, ApoE levels in large-pore groups exceeded those in small-pore groups (Fig. 4M). This dynamic temporal pattern may reflect stage-specific roles of ApoE during osteogenic differentiation (e.g., early vs. late phases), thereby elucidating how pore size modulates osteogenesis through ApoE regulation.

qPCR analysis of the OGN gene (Fig. 4K) was performed to validate the underlying mechanism of matrix stiffness (Fig. 4N and O). In the large-pore-size group, we observed that OGN expression was upregulated with increasing stiffness at day 7. However, this difference disappeared by day 14, at which point the expression levels among the groups became comparable. Conversely, in the small-pore-size group, OGN expression showed no significant difference across stiffness levels at day 7. Interestingly, a distinct phenomenon emerged at day 14: under small-pore-size conditions, the OGN expression level in the low-stiffness group was significantly higher than that in the high-stiffness group.

Repair of rat critical-sized calvarial defects

We cultured four groups of BMSC-scaffold composites with distinct physical properties in vitro for 14 days, then implanted them into rat calvarial defect models to evaluate their bone regeneration capacity (Fig. 5A).

Fig. 5.

The rat critical-size calvarial defect model. A Schematic of bone regeneration in the calvarial defect model. Created with BioRender.com. B 3D micro-CT reconstructions at 12 weeks. C Quantification of bone volume fraction (BV/TV) and bone mineral density (BMD). D H&E and Masson’s trichrome staining of calvarial defect sections at 12 weeks, scale bar, 500 μm

Micro-CT 3D reconstructions visually demonstrated defect repair in each group at 12 weeks post-surgery. We observed that the small pore size groups were more clearly filled with new bone formation and integrated to the host bone edge (Fig. 5B). We then quantified the newly formed bone mineral density (BMD) and bone volume fraction (BV/TV). The results showed the same trend as the 3D micro-CT reconstruction images. The BMD and BV/TV of the BMSC-Scaffold Composite in the small pore groups were significantly higher than those in the large pore groups (Fig. 5C).

Histological analysis of H&E and Masson’s trichrome-stained sections further showed the composites’ in vivo bone regeneration capabilities (Fig. 5D). Under H&E staining, large blocks of new bone tissue formation were observed in the small pore groups, which exhibit good osseointegration with the native bone at the defect margins. Masson staining similarly revealed that the small-pore groups had more widespread, intensely stained, and continuous blue areas, indicating richer collagen deposition and stable osteogenic regions. Collectively, these findings demonstrate that the small pore BMSC-scaffolds significantly promoted calvarial bone regeneration and mineralization in this model.

Discussion

This study demonstrated that a combination of small pore size and high stiffness most effectively promotes the osteogenic differentiation of BMSCs in vitro. Finally, small pore cell-scaffold composites correspondingly achieved superior bone regeneration in a rat critical-size calvarial defect model.

In our in vitro experiments, we found that the small pore composite groups consistently showed higher expression of all tested osteogenic markers than the large pore groups, including Runx2, Osx, ALP, COL1A1, and OCN. Within the small pore groups, the high-stiffness group (S/h) also outperformed the low-stiffness group (S/l) in most assays, except for Runx2 during the early stages of differentiation. These findings are consistence with recent studies, emphasizing the significant influence of pore size and stiffness on osteogenic differentiation.

For instance, studies show that hMSC expression of Runx2 and COL-1 significantly alters with changes in scaffold stiffness and porosity, accompanied by varying degrees of ALP activity upregulation [30]. Similarly, another study showed that a combination of 3D porous surface treatment combined with stiffness features significantly enhanced the expression of early osteogenic markers [31]. While our results are similar to these published findings, there are differences due to factors such as absolute pore size values, material types, and stiffness scales. Nevertheless, the overall trend strongly supports the conclusion that physical properties are potent regulators of osteogenesis, and that a combination of small pore size and high stiffness is particularly favorable for this process.

Under the premise of consistent macroscopic stiffness, the superior osteogenic activity exhibited by small-pore scaffolds primarily originates from their unique micro-topological cues. First, small pore size means a smaller radius of curvature at the microscopic scale, and a study has shown that curvature is an osteogenic driving force independent of the material [32]. The curvature of the scaffold converts physical geometric constraints into intracellular mechanical stress through the Contact Guidance mechanism. Specifically, curvature induces the directional alignment of BMSCs and changes the cytoskeleton tension, thereby activating the nuclear translocation of YAP/TAZ [33]. After entering the nucleus, YAP/TAZ initiates the osteogenic differentiation program, leading to the up-regulation of key markers such as Runx2 and Alpl at the transcriptional level. This completely aligns with the observed expression trends of osteogenic factors. In addition to the influence of curvature, this study employs a DMSO-mediated preparation process, and cells generate stronger mechanical sensing within the confined space provided by small pores [29]. Meanwhile, the DMSO preparation endows the scaffold with a natural pore structure characterized by high topological randomness and continuous curvature changes, which differs from 3D printing.

Some studies suggest that complex pore features can provide more mechanical anchoring points and strengthen osteogenic signal transduction by inducing greater mechanical strain on the cell nucleus. This structural induction effect synergistically promotes osteo-angiogenesis coupling, further ensuring the effect of bone regeneration [34]. Furthermore, the high specific surface area possessed by small-pore scaffolds is also a key factor in promoting osteogenesis. Smaller pore sizes can significantly increase the surface area available for cell adhesion, thereby promoting early cell attachment. At the same time, the total amount of adsorbed serum proteins is increased. These highly dense adhesion sites provide a sufficient physical basis for the rapid clustering of integrins, thereby strengthening the initial mechanical anchoring signals and triggering the downstream osteogenic differentiation program [35].

Admittedly, the optimal choice for the influence of scaffold pore size on osteogenesis still has certain controversies in the academic community. For example, a study on 3D-printed ceramic scaffolds found that in a long bone defect model, larger channels are more conducive to the deep ingrowth of bone tissue, which is primarily attributed to the superior mass transport capacity of nutrients brought by large pore sizes [36]. However, according to the transcriptomic KEGG enrichment results, we did not observe typical transcriptional features dominated by hypoxia or glycolysis in our cranial bone defects. Instead, the differences are primarily concentrated in mechanosensation, cell adhesion, and related signal transduction pathways.

This indicates that under our experimental conditions, nutrient diffusion is not the core factor limiting osteogenesis.This phenomenon may stem from the differences in the requirements for the bone regeneration microenvironment at different anatomical sites. Long bones, due to their high blood supply requirements, rely more on the mass transport provided by large pore sizes; whereas the cranium, as a typical flat bone, has a physiological structure closer to dense cortical bone. A study indicated that the small-pore structure simulating cortical bone is superior to the cancellous-like large-pore structure in providing early physical confinement and enhancing cellular mechanotransduction [37]. Therefore, we believe that by simulating the micro-topology of cortical bone, the positive physical induction generated by small-pore scaffolds compensates for potential metabolic loss, achieving a superior pro-osteogenic effect in this specific physiological environment of the cranium.

To further explore the influence of the scaffold’s physical properties, we used transcriptomic analysis to identify two key genes: OGN and ApoE. OGN has previously been recognised for its role in promoting bone formation, serving as an osteogenic marker [38–40]. However, the precise underlying mechanisms remain unclear. When we examined the dynamics of OGN expression specifically within the small-pore composites between days 7 and 14, we observed a unusual expression trend. On day 14 in particular, OGN expression levels were significantly lower in the S/h group than in the S/l group. This finding contrast sharply with the trends observed for other osteogenesis related genes, where the S/h group generally exhibited higher expression. A recent review revealed that the effects of OGN on bone turnover are conflicting based on the current evidence. While OGN can increase the expression of ALP and OCN, thereby promoting mineralisation, it also appears to decrease Runx2 and Osx expression, which could reduce cell recruitment. OGN also appears to suppress the rate of COL1 fibrillogenesis [41]. The review concludes that OGN likely acts predominantly as a negative regulator, but does not discount its function as an established osteogenic marker. Moreover, another study reported that OGN deficient mice exhibit increased bone mass [42]. Considering the substantially elevated expression of other osteogenesis related genes in the S/h group, we propose that the downregulation of OGN on day 14 in this highly osteogenic group might paradoxically reflect its negative regulatory capacity advancing the differentiation process. This interpretation is consistent with previous reports suggesting that OGN has complex and conflicting roles in osteogenic differentiation. Our experimental results offer a theoretical basis for further exploration of the still unclear mechanisms of OGN. Moreover, our results are the first to report, to our knowledge, OGN downregulation during scaffold-mediated bone defect repair, filling a gap in the literature.

ApoE is generally accepted as a factor that promotes adipogenic differentiation. However, a recent study suggests that ApoE acts as a regulator maintaining the osteogenesis-lipogenesis balance, specifically within the context of material-mediated bone regeneration [43]. We supplemented our culture medium with dexamethasone, which is known to promote ApoE secretion by cells. On day 7 of osteogenic culture, we observed that at the same stiffness level, ApoE expression was higher in the small-pore groups compared to the large-pore groups. Other recent research has indicated that ApoE promotes osteogenic differentiation via the JAK-STAT pathway. Following ApoE treatment, activated STAT proteins further regulate the levels of four key osteogenic proteins—ALP, Runx2, COL1A1, and Osx, leading to their significant upregulation [44]. This is consistent with our own observations. Interestingly, this trend reversed by day 14. At this later time point, the large-pore groups exhibited higher levels of ApoE expression. We note that ApoE has been described as a distinctive gene in the early stage of implant osseointegration [45]. Given its position upstream in the MSC osteogenic differentiation pathway, the peak expression of ApoE is theoretically expected to precede the rise of other osteogenic markers. On day 14, we observed that the small-pore groups displayed strikingly higher expression of other osteogenic markers. We hypothesize that the substantial early secretion of ApoE upregulates downstream osteogenic genes, but ApoE expression in the small-pore groups may have already peaked before this time point. In contrast, ApoE secretion in the large-pore groups had probably not yet peaked, resulting in lower expression of these downstream genes. Given the lack of studies linking the role of APOE in osteogenic differentiation directly to the Wnt pathway, it is possible that Wnt signaling becomes the dominant driver of osteogenesis after day 7, effectively circumventing ApoE’s influence on the differentiation process.

In the future, we aim to determine a more detailed temporal profile of ApoE expression by adding more measurement time points and extending the culture duration. This will enable us to validate our hypothesis by establishing whether and when an ApoE peak occurs and whether this peak is delayed in the large-pore groups. We also plan to monitor MSC adipogenesis to ascertain whether ApoE plays a definitive positive role in scaffold mediated osteogenesis. Further exploration of Wnt and STAT pathway activation could also help elucidate the downstream mechanisms of ApoE. The precise impact of ApoE on differentiation is currently a subject of considerable debate. Given the significant dynamic changes observed in our system, we propose that ApoE appears to be a key Regulator of osteogenic differentiation. Our ability to modulate ApoE expression by controlling scaffold pore size establishes our experimental model as a valuable platform for further dissecting mechanisms of ApoE.

We further explored the reversal in the expression trends of OGN and APOE observed at day 14. A study has shown that ApoE expression can significantly reduce the expression of the adipogenic factor PPARγ, thereby inhibiting lipid accumulation and decreasing the differentiation of MSCs into adipocytes [43]. OGN knockout in mice was also shown to induce hyperlipidemia and consequently result in bone nonunion. Another study found that MSCs overexpressing OGN via lentivirus also exhibited inhibition of PPARγ expression [46]. We propose that high APOE and OGN expression in the early stages prevents PPARγ production, thereby inhibiting BMSC differentiation towards adipocytes. Once osteogenic differentiation is established, however, the suppression of adipogenesis via PPARγ inhibition may no longer be required, potentially leading to the observed downregulation of APOE and OGN. Additionally, research indicates that OGN production is vitamin D dependent, and adipose tissue contains significant amounts of vitamin D [47]. Since adipogenesis was suppressed by OGN and ApoE in the early stages, this could lead to reduced local adipose tissue formation. As adipose tissue is a source of vitamin D, we speculate that the subsequent decrease in local vitamin D availability could also contribute to the downregulation of OGN.

Furthermore, the downstream markers Osx and ALP showed a much larger expression difference between the small and large pore groups, whereas the upstream regulator Runx2 exhibited only a modest change. This finding may be attributed to Runx2 activation triggering a cascade amplification effect at the transcriptional level, which further enhances the differential transcriptional activity of downstream factors Osx and the osteogenic marker ALP. This differential expression between upstream and downstream molecules was even more evident in the BMSC group at day 14. However, the specific mechanisms underlying this phenomenon require further investigation.

Interestingly, while the S/h group showed absolute superiority in promoting the in vitro osteogenic differentiation of both BMSCs and MLO-A5 cells, this was not the case in our in vivo rat calvarial defect model. In the in vivo experiments, although the small-pore scaffolds still significantly outperformed their large-pore counterparts, the high stiffness (S/h) group showed no clear osteogenic advantage over the low stiffness (S/l) group. In fact, S/h group’s regenerative capacity was observed to be slightly weaker, although this difference was not statistically significant.

This result suggests that in vitro differentiation of a single osteogenic lineage may not fully represent the complex process of bone regeneration in vivo. Some research indicates that while a high stiffness (> 20 kPa) is favorable for the osteogenic differentiation of stem cells, a moderate stiffness (about 5–20 kPa) is more conducive to myogenic and chondrogenic differentiation [20, 48, 49]. Because bone tissue regeneration in vivo is a multicellular cooperative process, the stiffness of scaffold materials for repairing defects must balance support for diverse cell types with biocompatibility.

Recent studies have further illuminated the complex role of mechanical signals in the bone microenvironment. For instance, a peri-arteriolar niche has been identified in the bone marrow that is highly sensitive to mechanical stimuli through the mechanosensitive channel PIEZO1. And the niche is crucial for maintaining the osteogenic potential of osteoprogenitor cells. When mechanical stimulation is excessively strong or weak, the homeostasis of this niche is disrupted, leading to reduced osteogenic and immunogenic cell production [50]. Viewed in light of our own findings, this suggests that mechanical signals during in vivo bone repair are not a matter of “stronger is better.” Instead, they must exist within a dynamic equilibrium to maintain the functional stability of the bone marrow microenvironment and the osteo-immune coupling. This may explain why a higher scaffold stiffness promoted osteogenic differentiation in vitro, whereas in vivo, an excessive mechanical stiffness may disrupt microenvironmental homeostasis, thereby impairing the regenerative efficacy.

While this study highlights the pivotal role of physical scaffold cues in governing the osteogenic differentiation of BMSCs, several limitations merit consideration. Bone regeneration remains an intricate physiological process involving a diverse milieu of cell types. Beyond the osteogenic effects emphasized in our work, the extent of vascularization—marked by CD31/Emcn expression—and shifts in the immune microenvironment are also vital determinants of the ultimate repair outcome. Given that our initial experimental design was tailored to investigate the direct inductive effects of physical signals, early-stage angiogenesis and immune cell infiltration post-implantation were not fully characterized. This focus, while intentional, precludes a comprehensive mapping of the scaffold’s integrated regenerative mechanisms.

The use of a calvarial defect model primarily addresses intramembranous ossification, underscoring the specific benefits of small pore sizes in flat bone repair. It remains to be validated whether this preference for smaller pores holds true in the more complex mechanical landscapes of endochondral ossification, such as fracture healing. Our future efforts will therefore aim to establish cross-model verification systems to dissect the synergistic interplay between mechanical cues, immunomodulation, and angiogenesis. Additionally, while our transcriptomic data suggest the involvement of ApoE and OGN in scaffold-mediated osteogenesis, a direct causal link has yet to be definitively established. Subsequent studies will focus on functional validation and the elucidation of underlying signaling pathways, potentially offering more refined guidance for the rational design of biomaterials in bone tissue engineering.

The findings of this study are consistent with previous reports on the regulation of osteogenic differentiation by pore size and stiffness. Furthermore, our work reveals a critical concept: the pore size-stiffness interplay. This principle demonstrates that a scaffold’s optimal osteogenic performance arises not from a single dominant physical factor, but from the synergistic regulation between multiple parameters. This study employs the cryogelated 3D porous scaffolds as a platform to achieve separate control over physical properties, filling a research gap in understanding how multi-scale physical characteristics of scaffolds influence bone tissue regeneration.

Conclusion

In this study, we achieved the separate control of stiffness and pore size in cryogelated 3D porous scaffolds by regulating DMSO concentration. This allowed us to identify the optimal physical features for BMSC osteogenesis and thereby construct a cell-scaffold composite with high osteogenic potential. Furthermore, we demonstrated that the superiority of these physical properties has abroad generality, effectively promoting the osteogenic differentiation of different cells in the osteogenic lineage. Our in vitro experiments verified that a combination of small pore size and high stiffness was most favorable for osteogenic differentiation. Subsequently, in our in vivo rat critical-size calvarial defect model, we found that the small-pore scaffolds produced superior bone tissue repair compared to the large-pore scaffolds. RNA-seq analysis revealed that the cryogelated 3D porous scaffolds exhibited significant osteoinductive properties. Further research demonstrated that ApoE and OGN genes independently regulated the osteogenic differentiation process of cells within the scaffold microenvironment, and this regulatory mechanism was consistent with the osteogenic outcomes observed in both in vitro and in vivo experiments. The prospects of this tunable scaffold system offer a novel paradigm for the personalized clinical treatment of bone defects.

Materials and methods

Materials and reagents

Gelatin was obtained from Sigma-Aldrich (cold water fish skin, Cas. 9000-70-8, MW ~ 60 kDa), Cryoprotectant DMSO (SolarBio) and glutaraldehyde (Macklin). Other reagents included ascorbic acid (Sigma-Aldrich), β-glycerophosphate (Sigma-Aldrich), and dexamethasone (Sigma-Aldrich). Cell culture reagents included αMEM (Gibco, Carlsbad, CA, USA), Penicillin/Streptomycin (P/S) and fetal bovine serum (FBS) (Wisent).

Preparation of cryogelated 3D scaffold

In this study, porous gelatin scaffolds with tunable pore size and mechanical properties were fabricated using a cryogelation method. Briefly, a 4% (w/v) cold fish skin-derived gelatin solution was dissolved at 65 °C and then cooled on ice. The cooled gelatin solution was mixed with different final concentrations of the cryoprotectant dimethyl sulfoxide (DMSO; 0.5% or 4.5% v/v, to regulate pore size) and the crosslinker glutaraldehyde (GA; 0.175% or 0.4% w/v, to regulate mechanical hardness). The resulting mixture was injected into pre-cooled poly (methyl methacrylate (PMMA) molds held at 4 °C, followed by cryogelation at − 20 °C for 20 h to form a porous structure via ice-templating. After gelation, unreacted GA was neutralized using a 1% (w/v) sodium borohydride (NaBH₄) solution (pH ≈ 9.7) to reduce cytotoxicity, and the samples were thoroughly rinsed with deionized water. The scaffolds were then freeze-dried, sterilized under ultraviolet (UV) light, and stored under vacuum.

By varying the concentrations of DMSO and GA, four types of scaffolds were prepared: large pore size/low stiffness (L/s), large pore size/high stiffness (L/h), small pore size/low stiffness (S/l), and small pore size/high stiffness (S/h).

Physical characterization of cryogelated 3D scaffold

Scanning electron microscope (SEM)

The four groups of freeze-dried scaffold samples were mounted onto conductive adhesive tape, sputter-coated with gold for 90 s, and then observed using scanning electron microscopy (SEM) to examine the microstructure, pore morphology, and take photographs. ImageJ software was used to process the SEM images. The long and short axes of each pore in the gelatin scaffolds were measured, and the average of the two values was taken as the final diameter of the pore.

Atomic force microscope (AFM)

Local stiffness of the porous scaffolds was measured by atomic force microscopy (AFM) using a previously reported method with minor modifications. Briefly, scaffolds were gently fragmented into small debris to expose individual pore-wall regions, and the debris were immobilized on glass coverslips for testing. A spherical probe was used for indentation, and force–displacement curves were recorded during the loading–unloading process. The local Young’s modulus was calculated from the obtained curves based on the Hertz contact model. For each scaffold group, multiple debris were analyzed and repeated measurements were performed to ensure reliability.

Cell culture

Rat bone marrow mesenchymal stem cells (rBMSCs) were isolated from the femurs and tibias of 3–4-week-old Sprague–Dawley (SD) rats. The bone marrow cavity was flushed to collect the cells, which were then seeded in α-MEM medium supplemented with 10% FBS and 1% PS, and cultured at 37 °C in a 5% CO₂ atmosphere. When the cell confluence reached 80–90%, the cells were passaged using 0.25% trypsin-EDTA (GIBCO; 25200072; USA), and P3–5 were used for subsequent experiments. MLO-A5 was cultured under the same conditions using the same medium.

Biocompatibility evaluation of cryogelated 3D scaffold

Live and dead cell staining

The viability of cells on the scaffolds was evaluated using a Calcein/PI Cell Viability/Cytotoxicity Assay Kit (Beyotime, C2015M). Rat bone marrow mesenchymal stem cells (rBMSCs) were seeded onto the scaffolds at a density of 1.0 × 10⁶ cells/uL. After 3 days of culture, the samples were washed with phosphate-buffered saline (PBS), and then incubated with a mixture of Calcein-AM and propidium iodide (PI) at 37 °C in the dark for 30 min. Finally, the samples were observed under a confocal laser scanning microscope (CLSM), where live cells exhibited green fluorescence and dead cells exhibited red fluorescence.

CCK-8 assay

rBMSCs were seeded onto four groups of scaffolds placed in a 96-well plate at a density of 2.0 × 10⁵ cells/µL. At days 1, 3, and 7 of culture, 100 µL of culture medium containing 10% CCK-8 reagent was added to each well. The plates were incubated at 37 °C for 2 h. The supernatant was then transferred to a new 96-well plate, and the absorbance (OD) at 450 nm was measured using a microplate reader (Bio-DL).

Osteogenic induction

rBMSCs / MLO-A5 were digested and resuspended at a concentration of 8.0 × 10⁵ cells/µL. A volume of 8 µL of the cell suspension was slowly pipetted onto each of the four groups of scaffolds. The scaffolds were incubated in the cell culture incubator for 2 h to allow sufficient cell infiltration and attachment. Subsequently, the osteogenic induction medium (containing 10% FBS, 1% P/S, 500 µM ascorbic acid, 10 mM β-glycerophosphate, and 100 nM dexamethasone) was added for culture. Samples were collected at days 7, 14, and 21 for subsequent analysis.

RNA extraction and RT-qPCR

At days 7, 14, and 21 of induction, total RNA was extracted from the cell-scaffold composite using Trizol reagent. The RNA concentration was measured by ultraviolet spectroscopy (Microdrop; Bio-DL Corporation). One microgram (1 µg) of RNA was reverse-transcribed into cDNA. The primers were synthesized by Sangon Biotech. Quantitative reverse transcription PCR (qRT-PCR) was performed using the Bio-Rad CFX Connect detection system (Bio-Rad, Hercules, USA).The PCR cycling conditions were as follows: initial denaturation at 95 °C for 3 min, followed by 40 cycles of 94 °C for 5 s, 58 °C for 15 s, and 72 °C for 15 s, and a final extension at 72 °C for 30 s. The osteogenesis-related genes examined included RUNX2, ALP, Osx, OCN, and COL1A1. GAPDH was used as the internal reference gene, and the relative expression levels of the target genes were calculated using the 2^−ΔΔCt method.

The primer we used as followed:

	Forward sequence	Reverse sequence
R-Gapdh	GGTGATGCTGGTGCTGAGTATGTC	GGCGGAGATGATGACCCTTTTGG
R-Runx2	GAACCAAGAAGGCACAGACAGAAG	GGCGGGACACCTACTCTCATAC
R-ALP	GCGACAGCAAGCCCAAGAGAC	TCACTGTGGAGACGCCCATACC
R-COL1A1	GTGCGATGGCGTGCTATGC	TCCTATGACTTCTGCGTCTGGTG
R-Osx	CCCTGGGAAAAGGAGGCACAAAG	GAGGAGTCCATTGGTGCTTGAGAAG
R-OCN	GGTGCAAAGCCCAGCGACTC	GGCTCCAAGTCCATTGTTGAGGTAG
R-ApoE	GGCGGCTGGAGGAAGTGG	AGGCGTATCTGCTGGGTCTG

Establishment of critical-size rat calvarial defect and cell-scaffold composites implantation

All animal experimentation protocols reported in this study were approved by the Animal Ethics Review Committee of Peking University People’s Hospital (Approval No.: 2023PHE099). The reporting of this work adheres to the ARRIVE guidelines 2.0 (Supplementary Material 1). A total of forty (n = 40) Sprague-Dawley (SD) rats were used, aged 8 weeks, and weighing 220–240 g. All rats were housed under Specific-Pathogen-Free (SPF) conditions, with ad libitum access to food and water, and maintained under a controlled 12-hour light/dark cycle. For animals within the same experimental group, four rats were randomly placed in the same cage, and the cages were kept in the same location within the animal facility.

Following a one-week acclimatization period, the n = 40 SD rats were randomly divided into five experimental groups, with n = 8 animals per group. The group assignments were completed by HL and QQC using a random method. The experimental groups were: the Ctrl group (defect untreated), and the L/l, L/h, S/l, and S/h groups (defects treated with the corresponding scaffolds). Blinding measures were implemented at different stages of the experiment to reduce bias and enhance the reliability of the study. Specifically, the experimenters responsible for outcome assessment and data analysis were blinded to the group assignment to avoid potential bias.

Rats were anesthetized with 4–5% isoflurane for induction and 1–2% isoflurane for maintenance in oxygen during the surgical procedure. A longitudinal incision was made on the scalp to expose the cranium. A 6 mm diameter calvarial defect was carefully created using a trephine bur, with continuous saline irrigation to minimize thermal injury. The dura mater was meticulously preserved. The appropriate gelatin scaffolds for each group were then implanted into the defects, and the surgical site was closed with sutures. Post-operative care included the administration of 100 mg/kg cefazolin for two days to prevent infection. Rats were monitored daily. All rats with infected incisions were excluded from the data processing; rats without infection were included as data processing. At the experimental endpoint, rats were euthanized by gradual-fill CO₂ inhalation at a flow rate equivalent to 30–70% of the chamber volume per minute until respiratory arrest, followed by continued exposure for at least 1 min to ensure death.

Micro-CT scanning and analysis

Twelve weeks post-operation, rats were euthanized with carbon dioxide. The excised calvarial specimens, encompassing the defect area, were fixed in 4% paraformaldehyde for 24 h. Subsequently, they were assessed using micro-CT (SCANCO µCT-100, Switzerland). Bone volume/tissue volume (BV/TV) and bone mineral density (BMD) were quantified using the Evaluation V6.5 software (SCANCO, Switzerland).

Histology and immunohistochemistry

Histological and immunohistochemical (IHC) analyses were performed on both the in vitro osteogenic induction model and in vivo calvarial samples. The calvarial specimens were decalcified using 10% EDTA solution over a period of four weeks. Following a sequential dehydration process in graded ethanol series, the samples were embedded in paraffin at a section thickness of 7 mm. Staining was carried out with hematoxylin and eosin (H&E), Masson’s trichrome, Alizarin Red, and Vonkossa staining. For immunohistochemical analysis, antibodies targeting OCN (Servicebio, GB11233, 1:200), ALP (Servicebio, GB111364, 1:500), and Col-X (Servicebio, GB11022, 1:200) were employed.

RNA sequencing

After rBMSCs were cultured on the four types of scaffolds under osteogenic induction for 14 days, total RNA was extracted according to the method described in Sect. "RNA extraction and RT-qPCR". RNA-seq was conducted by Anoroad Gene Technology (Beijing, China). Differential expression genes were identified by the DESeq2 package, and functional enrichment for GO and KEGG was performed with the GOstats package.

Statistical analysis

All data were derived from at least three independent experiments and are presented as the mean ± SEM. Differences between two groups were analyzed using the unpaired Student’s t-test, and multiple group comparisons were performed using one-way ANOVA followed by Tukey’s post hoc test. All analyses were conducted using GraphPad Prism (version 10.1.2; La Jolla, CA, USA). A p-value < 0.05 was considered statistically significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Qianqian Chen and Shichun Xu contribute equally to this work. Qianqian Chen: Writing—review & editing, Writing—original draft, Methodology, Investigation, Formal analysis. Shichun Xu: Writing—review & editing, Funding acquisition, Formal analysis, Conceptualization. Yiqi Su: Methodology. Senyang Xiao: Methodology. Qingchen Meng: Methodology. Wenjing Li: Methodology. Yunwei Wang: Methodology. Dan Xing: Supervision, Project administration, Data curation, Conceptualization. Hui Li: Writing—review & editing, Funding acquisition, Formal analysis, Conceptualization.

Funding

This work was funded by grants from National Key Research and Development Program of China (2024YFA1108600), Natural Science Foundation of Beijing Municipality (L222087, L232094, and L244019), the National Natural Science Foundation of China (82302776) and Peking University People’s Hospital Scientific Research Development Funds (RDGS2023-04, RDX2023-12, andRS2024-04). Qianqian Chen and Shichun Xu equally to this work.

Data availability

The RNA sequencing data generated in this study have been made publicly available in the NCBI SRA database under accession number PRJNA1371997. All other data and materials presented in this article will be made available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The animal experiments were conducted under the approval of the PKUPH Animal Use Protocol & Ethic Review Committee (2023PHE099). (1) Name of the institutional approval committee or unit: PKUPH Animal Use Protocol & Ethic Review Committee; (2) Approval number: 2023PHE099; (3) Date of approval: November 27, 2023. The study was reported following Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.The original source of the MSCs used in this study has confirmed that the isolation of these cells was approved by the PKUPH Animal Use Protocol & Ethic Review Committee.

Competing interests

The authors declare no competing interests.