Microstructure-tailored β-tricalcium phosphate scaffolds utilize degradation-guided osteoclast suppression to accelerate bone regeneration

Abstract

β-Tricalcium phosphate (β-TCP) has been a valuable artificial bone material in clinical applications owing to its bioactivity, osteoconductivity, and degradability. However, the inappropriate degradation rate of β-TCP implants is one of the most important factors affecting the bone regeneration process, and it remains a challenge to effectively regulate β-TCP degradation for adapting to bone regeneration. To address this, we fabricated β-TCP scaffolds with strut sizes ranging from 300 to 900 µm by using digital light processing (DLP) printing. The results in vitro showed that a strut size-dependent degradation gradient, and the scaffolds with a 300 µm strut size had the fastest degradation rate. Furthermore, these scaffolds with smaller strut sizes (particularly 300 µm) enhanced the adhesion and proliferation of mBMSCs, increased alkaline phosphatase activity, and promoted the osteogenic-related gene expression. Notably, these scaffolds with smaller strut sizes, especially 300 µm, inhibited osteoclast differentiation due to the suppressive effect of Ca²⁺ and PO₄³⁻ from β-TCP degradation on osteoclastogenesis. Finally, the scaffold with a 300 µm strut size significantly inhibited the formation of multinucleated osteoclasts, accelerating bone regeneration and promoting the maturation of new bone during a 24-week tibial defect repair. This work suggests that microstructure-driven degradation tuning of β-TCP scaffolds can optimize bone repair by synchronizing material resorption with osteogenesis and inhibiting osteoclast differentiation.

---

1. Introduction

β-Tricalcium phosphate (β-TCP) is well known as a representative biodegradable bioactive ceramic. Its biodegradability plays a crucial role in bone regeneneration,¹ since the degradation of β-TCP can enhance the local concentration of calcium and phosphate ions, promoting the formation of bone minerals.² However, clinical trials have shown that certain β-TCP implants can persist in the body for several years when used in craniofacial applications, and the slow degradation delays the growth of new bone into the defect.^3–5 Regulating the degradation adaptability of β-TCP is an important challenge for its development and clinical applications. The ideal implants for bone defect repair should possess biodegradability that matches tissue regeneration.⁶ β-TCP implants with favorable biodegradability not only provide structural support for new bone growth, but also accelerate the repair process by creating an ionic microenvironment conducive to bone defect healing.⁷ Rapid degradation may result in premature collapse of the implant structure, leading to incomplete healing,⁸ while slow degradation can hinder new bone growth and increase the risk of inflammation, infection, and rejection.⁹

The degradation of calcium phosphate in vivo is influenced by three primary mechanisms: physical fracture, chemical dissolution, and cell-mediated biodegradation.¹⁰ Among these mechanisms, chemical dissolution serves as the predominant mode of degradation, which occurs when bodily fluids interact with the material, penetrating its porous structure and leading to dissolution.⁶ This process is closely related to the compositional properties, crystallinity, and structural parameters of calcium phosphate, including pore size, porosity, and specific surface area. But altering the phase composition and crystallinity can introduce problems related to compositional complexity and stability control. And, for the structural parameters, the increased pore size, porosity, and specific surface area enhance the permeability of the implant for accelerating chemical dissolution. Nevertheless, traditional manufacturing techniques for calcium phosphate, such as gas foaming, freeze-drying, and gel casting, cannot accurately control pore morphology, pore size, porosity and interconnectivity.¹¹ To address these limitations, three-dimensional (3D) printing technology, especially light-curing 3D printing technology, has been employed to design and fabricate implants with controllable structural parameters.^12–15 Modifying the pore structural parameters of the β-TCP implant is a promising strategy for regulating the degradation rate.¹⁶

Degradation of calcium phosphate ceramics plays an important role in osteogenesis and osseointegration by providing essential Ca²⁺ and PO₄³⁻ ions during bone regeneration to create a proper ionic microenvironment.¹⁷ In this process, calcium ions facilitate bone formation and maturation through the process of calcification.^2,7 And, the calcium ion influences bone regeneration via cellular signaling pathways, including the activation of ERK1/2 and PI3K/Akt pathways, to stimulate osteoblast adhesion, proliferation, and differentiation, and support osteoblast function. Additionally, the calcium ion has been shown to regulate the formation and resorptive activity of osteoclasts.^2,18 Similarly, the phosphate ion significantly modulates mineralization, and it promotes osteoblast differentiation and proliferation activating pathways such as IGF-1 and ERK1/2 as well. Moreover, it also upregulates the expression of BMP-2.^19,20 Notably, phosphate ions can inhibit osteoclast differentiation and bone resorption by regulating the signaling of the RANK ligand (RANKL) and its receptor in vivo.²¹ In summary, calcium and phosphate ions collectively promote osteoblast proliferation and differentiation while inhibiting osteoclast differentiation to some extent. Thus, modulating the degradation of β-TCP implants can control the concentration of calcium and phosphate ions in the local microenvironment, ultimately regulating the behaviors of osteoblasts and osteoclasts during bone repair. Improving the degradation adaptability of bioceramic implants should be a key consideration for effective bone regenerative repair.

The diamond truss structure is inspired by natural diamond configurations and is widely utilized in the fabrication of 3D-printed implants for bone defect repair. This architecture closely resembles the trabecular structure of natural bone and exhibits excellent osteoconductivity.²² And, our previous work indicated that β-TCP scaffolds with diamond truss exhibit excellent osteogenic performance in vivo.²³ In this study, we evolved the structures with different strut structure sizes (300, 500, 700, and 900 µm) based on diamond truss structures with consistent pore size. These scaffolds were designated as D300, D500, D700, and D900, respectively. Then, four different sizes of β-tricalcium phosphate (β-TCP) bioceramic scaffolds were precisely prepared by digital light processing (DLP) 3D printing. As shown in Scheme 1, we systematically investigated the effects of these scaffolds on the viability, proliferation, and differentiation of mouse bone marrow mesenchymal stem cells (mBMSCs). Furthermore, we examined the influence of these scaffolds on osteoclast differentiation and the paracrine factors released by osteoclasts on the behaviors of mBMSCs and endothelial cells. Finally, a tibial defect model was used to assess the osteogenic properties and long-term degradation of the scaffold and to study the osteoclast behaviors in vivo. Thus, this study meticulously investigated the effects of strut structural parameters on the degradation rate of β-TCP scaffolds and the influence of multiple osteo-related cell behaviors (including osteoblasts, osteoclasts, and endothelial cells) for bone regeneration. This work provides an important reference for the development of tricalcium phosphate scaffolds with controlled degradation and better adaptability for bone regeneration and clinical applications.

Scheme 1 (A) Design and fabrication of microstructure-tailored β-TCP scaffolds and (B) the impact of their microstructure on ion release, in vitro and in vivo cellular responses, and bone tissue regeneration.

2. Experimental

2.1. Design and preparation of porous scaffolds

2.1.1. Design of porous scaffolds. First, we established the frameworks with cylindrical struts of four aspect ratios (1.93, 1.45, 1.23, and 1.12) arranged in diamond configurations (Scheme 1A) by using SOLIDWORKS. Then, the diamond structural units were used for filling into the cylindrical space by precisely adjusting unit cell sizes (1360, 1680, 2000, and 2320 µm) in Materialise Magics. Finally, we designed the models of four scaffolds, which have a consistent pore size (450 µm) and strut diameter gradients (300 µm, 500 µm, 700 µm and 900 µm).

2.1.2. Preparation of β-TCP scaffolds with different strut sizes. The β-TCP paste for photocurable ceramic 3D printing powder developed by our research group was used to prepare the designed scaffolds using a DLP printer (Beijing Ten Dimensions Technology Co., Ltd, China).²³ After 3D printing, a high-temperature process was used to sinter remove the scaffold and retain only pure β-TCP bioceramics for cellular experiments and in vivo implantation experiments.

2.2. Characterization of porous β-TCP scaffolds

2.2.1. Porosity of scaffolds. The scaffold volume (V) was calculated by measuring the diameter and height of these scaffolds, and the scaffold mass (m) was measured as well. The porosity was calculated according to the following formula (1):

Porosity (%) = m/ρV × 100% (1)

where ρ is the density and for β-TCP, ρ = 3.14 g cm⁻³.

2.2.2. Protein adsorption. The scaffolds were placed in a 24-well plate. Then, 1 mL BSA solution (1 mg mL⁻¹) was added to each well. The plate was incubated in a constant temperature shaker at 37 °C and 90 rpm for 24 hours. Afterward, the scaffolds were removed with tweezers, and the total protein concentration was determined by using the Protein Assay Kit (Beyotime, China).

2.2.3. Microscopic morphology observation. The surface morphology characteristics of the prepared scaffold were observed and analyzed using a field emission scanning electron microscope (Sigma 300, Zeiss, Germany).

2.2.4. In vitro studies of the degradability. In vitro degradation experiments were conducted on all scaffolds using Tris–HCl buffer. Four scaffolds were selected from each group, and their masses were measured using an electronic balance. The scaffolds were immersed in Tris–HCl buffer (0.1 mM, pH = 7.4) according to a ratio of 1 g/50 mL respectively. The scaffolds were degraded in a thermostatic shaking table (37 °C, 90 rpm). The scaffolds were removed from the solution every week, rinsed with deionized water and anhydrous ethanol, and then dried in an oven at 60 °C. Finally, the mass was measured and recorded. The concentrations of released Ca²⁺ and PO₄³⁻ ions were detected using an ICP-OES instrument (iCAP 7200 Duo, Thermo Fisher Scientific, USA). Fresh Tris–HCl was refilled into centrifuge tubes based on the mass measurements. The tubes were then placed in a holder and continued to be shaken for degradation.

The surface morphology of the scaffolds after degradation in Tris–HCl buffer for 30 days was observed and analyzed using a field emission scanning electron microscope (Sigma 300, Zeiss, Germany).

2.3. Bone marrow mesenchymal stem cell viability, proliferation and differentiation

2.3.1. Cell culture. To evaluate the effects of different strut sizes on cellular behavior, TCP scaffolds (ϕ 9 mm × 2 mm) with varying strut sizes (300, 500, 700, and 900 µm) were fabricated for in vitro multi-cellular experiments, with triplicate samples used for each experimental group. The sterilized β-TCP scaffolds were pre-wetted with complete medium for 1 h and carefully drained before cell seeding. The concentrated cell suspension (1 × 10⁶ cells per mL) was then spot-seeded onto the scaffolds and allowed to adhere for 30 min before the addition of complete medium.

Mouse bone mesenchymal stem cells (mBMSCs, P3, Procell, China) were cultured in α-MEM complete medium, which contained 10% fetal bovine serum (FBS). The medium was changed every other day until the degree of mBMSC fusion reached 80–90%. The scaffolds were sterilized and added into 48-well plates. Then, 2 × 10⁴ cells were added to each scaffold and cultured at 37 °C with 5% CO₂.

2.3.2. Cell viability of mBMSCs on scaffolds with different strut sizes. mBMSCs (mBMSCs, P3, Procell, China) were seeded onto the scaffolds and cultured for 7 days. The cell viability was assessed using a calcein/PI assay kit (Beyotime, China). The staining procedure distinguished live and dead cells in scaffold co-cultures. Fluorescence imaging was performed using an Axio Observer 7 microscope (Zeiss, Germany). All experiments included triplicate samples per test group.

2.3.3. Proliferation of mBMSCs on scaffolds with different strut sizes. mBMSCs were seeded into the scaffolds. After 1, 3, 5, and 7 days, the scaffolds were washed with PBS. Then, 300 µL of 10% cell counting kit-8 (CCK8) working solution (GLPBio, China) was added to each well and incubated for 1 hour. Then, the OD value of the reaction solution was measured at 450 nm using a microplate reader based on the manufacturer's instructions.

2.3.4. Alkaline phosphatase (ALP) assay. mBMSCs were seeded on scaffolds for 7 and 14 days. For ALP staining, scaffolds were first fixed in 4% paraformaldehyde for 30 minutes. After PBS washing, mBMSCs on scaffolds were stained with BCIP/NBT alkaline phosphatase color development kit (Beyotime, China) for 30 minutes. Finally, the stained constructs were observed under a CKX31 biomicroscope (Olympus, Japan). For ALP activity assay, the ALP activity of mBMSCs was detected using an alkaline phosphatase activity kit (Beyotime, China) according to the manufacturer's instructions.

2.3.5. Expression of osteogenic genes. The expression levels of osteogenic gene collagen I (COL I), osteopontin (OPN) runt-related transcription factor 2 (Runx2), and osteocalcin (OCN) in mBMSCs cultured on scaffolds for 7 and 14 days were measured by RT-qPCR. The total RNA was extracted with the Trizol reagent (Invitrogen, USA). Next, the extracted RNA was reverse transcribed to obtain cDNA using a prime-script RT reagent (Takara, Japan). A reaction system (qPCR Kit, Genecopoeia, USA) was mixed for the PCR. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as the housekeeping gene in this study. The primer sequences used for osteogenesis-related genes in this study are shown in Table S1. The relative osteogenic expression level of the target genes in cells is analysed using the 2^−ΔΔCT calculation method.

2.4. Osteoclast differentiation on scaffolds with different strut sizes

2.4.1. Cell culture. Raw264.7 cells (P3, Procell, China) were selected as osteoclast precursor cells. These cells were maintained in DMEM containing 10% FBS and 50 ng mL⁻¹ RANKL (R&D Systems, USA) for 6 days. During this period, the medium was changed on day 3.

2.4.2. TRAP staining. Raw264.7 cells were seeded into 24-well plates at a density of 1 × 10⁵ each well, and the cells were incubated in a cell culture incubator for 1 hour to make the cells fully adhere to the wall. Then, the scaffolds were soaked in PBS beforehand, and the group without scaffolds was used as the positive control group (OC group). Raw264.7 cells cultured in medium without RANKL were used as the negative control group (Raw264.7 group). After osteoclastic differentiation of Raw264.7 cells was induced with RANKL for 6 days, osteoclasts were stained with the TRAP/ALP double staining kit (Wako, Japan) and then photographed by optical microscopy. The osteoclast area was counted using ImageJ software.

2.4.3. TRAP activity. Raw264.7 cells were cultured as described in Section 2.4.2 above. After 6 days of induction with RANKL, the TRAP activity of osteoclasts was detected using a tartrate resistant acid phosphatase activity kit (Beyotime, China) according to the manufacturer's instructions.

2.4.4. PDGF-BB ELISA assay. Raw264.7 cells were cultured as described in Section 2.4.2 above. After 6 days of induction with RANKL, the concentration of PDGF-BB secreted by osteoclasts co-cultured with scaffolds was detected using the PDGF-BB ELISA kit (Meimian, China) according to the manufacturer's instructions.

2.4.5. Osteoclast-related gene expression. The expression levels of the osteoclast-related genes TRAP, recombinant cathepsin K (CTSK), and recombinant nuclear factor of activated T-Cells, and cytoplasmic 1 (NFATc1) in osteoclasts cultured with different scaffolds for 6 days were measured using the RT-qPCR. The total RNA was extracted with the Trizol reagent. Next, the extracted RNA was reverse transcribed as described in Section 2.3.5 above. Similarly, a PCR reaction system was prepared. GAPDH was used as the housekeeping gene. The primer sequences used for osteoclast-related genes in this study are shown in Table S2. The relative expression is calculated using the 2^−ΔΔCT method.

2.5. The effects of scaffold-incubated osteoclast medium on mBMSCs and endothelial cells

2.5.1. Preparation of osteoclast conditioned medium. Raw264.7 cells were seeded onto the scaffolds in 24-well plates at a density of 5 × 10⁴ per scaffold. The medium was collected after 3 days of incubation using DMEM containing 50 ng mL⁻¹ RANKL (R&D Systems, USA) and 10% FBS. Finally, it was mixed with fresh medium (mBMSCs) corresponded to α-MEM containing 10% FBS, and MS1 corresponded to DMEM containing 5% FBS at a ratio of 1 : 3 to form conditioned medium (OC-CM). The medium with osteoblasts directly induced on the well plates without co-culture with the scaffolds was used as the control group (OC group).

2.5.2. The proliferation of mBMSCs with OC-CM. mBMSCs and mouse islet endothelial cells (MS1, P3, Procell, China) were seeded into 24-well plates at a density of 2 × 10⁴ per well and cultured with OC-CM to form different scaffolds, respectively. After 1, 3, and 5 days, the medium was carefully removed from the well plate and washed with PBS. Then, CCK-8 working solution (GLPBio, China) was added and incubated for 1 hour. Then, the OD value of the reaction solution was measured at 450 nm using a microplate reader based on the manufacturer's instructions.

2.5.3. The migration of mBMSCs with OC-CM. Raw264.7 cells were seeded on the scaffolds in 24-well plates at a density of 5 × 10⁴ per scaffold and cultured in ’Dulbecco's modified Eagle's medium containing 50 ng mL⁻¹ RANKL (R&D Systems, USA), 10% FBS and 1% penicillin/streptomycin for 6 days. The mBMSCs were cultured on transwell chambers adapted to the 24-well plates at a density of 1.5 × 10⁴ cells per chamber and incubated for 30 minutes to allow the cells to adhere on the chamber. The chambers were transferred to the above 24-well plates and further incubated for 24 hours, after which the cells were stained with crystal violet staining solution (Beyotime, China) for 30 minutes. And then the mBMSCs that had not migrated to the top of the transwell membranes were scraped off with a clean swab, and those that had migrated to the bottom of the membranes were observed under an inverted biomicroscope (CKX31, Olympus, Japan). Finally, the number of migrated cells was counted using ImageJ software.

2.5.4. The ALP assay of mBMSCs with OC-CM. mBMSCs were seeded into 24-well plates and cultured for 7 days with OC-CM. As described in Section 2.3.4 above, alkaline phosphatase (ALP) assays were performed on the samples.

2.5.5. The wound healing of MS1 with OC-CM. As described in Section 2.5.1, we collected conditioned medium (CM) from RANKL-induced RAW264.7 cells cultured on various scaffold groups. The CM was then sterilized by filtration through a 0.22-µm membrane and centrifuged at 1200 × g for 5 min. The resulting supernatant was then mixed with high-glucose DMEM medium (containing 5% fetal bovine serum and 1% penicillin/streptomycin) at a 1 : 3 ratio to prepare the CM for MS1 cell culture. MS1 cells were seeded into 6-well plates at a density of 1.5 × 10⁶ per well and cultured in an incubator until the cells were in a monolayer spread well plate state. After scoring and positioning the well plates, a straight-line wound was made at the bottom of the plate with a clean 1 mL pipette tip and the dead cells were rinsed 3 times with PBS. Cells were cultured with OC-CM and photographs of the scratches were taken with an inverted biomicroscope (CKX31, Olympus, Japan) after the addition of culture medium, which was recorded as 0 h, and the same area was photographed again after 24 h. The 0-hour and 24-hour photographs were processed using ImageJ software, and the percentage of cell migration or wound healing at 24 hours was calculated.

2.6. Animals and surgical procedures for the rabbit's tibial defect

Twenty-seven New Zealand white rabbits (2.5 ± 0.1 kg, male) were purchased from Longgui Xingke Animal Farm, China. After intramuscular infusion of neoadjuvant anesthesia at 0.1 mL kg⁻¹, 3% pentobarbital sodium was injected intravenously at a dose of 0.1 mL kg⁻¹ for adequate anesthesia. The surgical site was prepared with an electric shaver and then modelled on the operating table. A sterile surgical drape was placed to expose the surgical site, and a 3-cm surgical incision was made 1 cm below the bilateral joints of the New Zealand White rabbit, and the muscle and periosteum were separated with hemostatic forceps to reveal the position of the tibial plateau. The tibia was rinsed with saline, and the drilled tibia was removed from the marrow cavity to complete the creation of the defect, and D300, D500, and D700 sterilized scaffolds with a diameter of 9 mm and a height of 2.7 mm, were implanted into the defect site, respectively. And, after the position was fixed, the periosteum and muscle were sutured. Finally, the epidermis was sutured. The animals were asphyxiated with CO₂ at the planned time points, and bilateral tibiae were taken as samples. All protocols were approved by the Animal Care and Experiment Committee of South China University of Technology, Guangzhou, China And all methods were performed in accordance with relevant guidelines and regulations (AEC: CV2023001).

2.7. Micro-CT assay

At 8, 16, and 24 weeks after the operation, rabbits were euthanized, and the femoral condyles were explanted and fixed in 4% neutral paraformaldehyde for 48 h. The femoral condyles were scanned using a Micro-CT scan (Zhongkekaisheng, China) at 30-µm resolution. As there are density differences between β-TCP and new bone, the Mimics software used in this study can distinguish one from the other. Then, the osteogenesis in the scaffolds and the degradation of the scaffolds were observed and the degradation rate of the scaffolds was calculated by calculating the volume of the remaining scaffold using the following formula (2):

Degradation rate (%) = (1 − V_R/V_O) × 100% (2)

where V_R is the volume of the scaffolds remaining after degradation and V_O is the original volume of the scaffolds, respectively.

2.8. Histopathological analysis

After 8, 16, and 24 weeks of implantation, samples were dehydrated through a graded alcohol series and embedded in polymethylmethacrylate (PMMA). After hardening, the sagittal sections of the central segment were cut into slices using a microtome (AiMco, China) and polished. Samples of hard tissue sections after 8, 16, and 24 weeks of implantation were stained with hematoxylin–eosin staining solution (Zhongkekaisheng, China) for histological observation and biodegradation analysis. In addition, other samples of hard tissue sections after 16 weeks of implantation were stained with TRAP staining solution (Zhongkekaisheng, China) for osteoclast observation and differentiation of osteoclast analysis.

2.9. Statistical analysis

All data results are presented as mean ± standard deviation and each experiment was repeated independently at least three times. Data between different groups were analyzed using one-way ANOVA for testing. The * in the experimental data plots indicates significant differences, *P < 0.05, **P < 0.01, and ***P < 0.001.

3. Results and discussion

3.1. Physicochemical properties of β-TCP scaffolds with different strut sizes

We successfully designed and prepared β-TCP scaffolds with different strut sizes and different specific surface areas while the side pore size was kept at 450 µm by adjusting the size of the diamond truss model.²⁴ Based on the preliminary studies and the currently accepted optimal strut parameters for bone regeneration,^25–27 the strut sizes of these scaffolds were 300, 500, 700 and 900 µm, which were named D300, D500, D700, and D900 accordingly (Table 1). As shown in Fig. 1A, the pore size distribution matched the designed model, with an average pore diameter of 450 µm and a deviation of ±1.08% across all groups. And, the strut size and pore size of the 3D printed scaffolds were highly consistent with the designed parameters of the scaffolds, minimizing its potential confounding effects, while the pillar diameters follow a gradient variation. Additionally, the surfaces of all these scaffolds had uniformly distributed submicron pores, and there is no significant difference in their micro–nanomorphology. Unavoidably, the porosity of the scaffold showed a slight gradient reduction (Fig. 2B). In short, all of the above results indicated that the strut size in the diamond truss structure of scaffolds were precisely controlled by DLP 3D printing.

Table 1 Pore characterization of the 3D printed scaffold model and real scaffold

Name	Strut size (µm)		Pore size (µm)		Specific surface area (mm^−1)
	Model	Real scaffold	Model	Real scaffold
D300	300	300.81 ± 12.23	450	441.67 ± 14.10	14.49
D500	500	490.33 ± 13.66	450	437.67 ± 11.06	8.98
D700	700	684.74 ± 11.17	450	439.13 ± 12.15	7.02
D900	900	860.95 ± 11.68	450	448.44 ± 11.78	5.23

---

Fig. 1 The characteristics of β-TCP scaffolds with different strut sizes. (A) STL model of the scaffolds and optical photographs of the scaffolds and SEM micrographs. (B) Porosity of the scaffolds. (C) Protein adsorption of the scaffolds. (D) SEM micrographs of the scaffolds after 30 days of immersion in Tris–HCl buffer. (E) and (F, G) Weight loss curves and the accumulated release of Ca²⁺ and PO₄³⁻ ions of the scaffolds after Tris–HCl buffer immersion (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and ***p < 0.001.).

Fig. 2 Cell behaviors of mBMSCs on β-TCP scaffolds with different strut sizes. (A) Calcein AM/PI staining images of mBMSCs on different scaffolds after culturing for 7 days. (B) Proliferation ability of mBMSCs on scaffolds detected using CCK-8. (C) ALP staining images of mBMSCs on different scaffolds after culturing for 7 and 14 days. (D) ALP activity in mBMSCs on scaffolds after culturing for 7 and 14 days. (E) Osteogenic-related gene expression of Col I, OPN, Runx2, and OCN in mBMSCs cultured on different scaffolds on day 7 and day 14 (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and***p < 0.001.).

The proteins could enhance cell adhesion and promote mineralization and osseointegration of bioceramics; thus, protein adsorption influences bioceramic degradation in many ways.^28,29 To investigate the effect of modifying the strut size on the protein adsorption of the scaffolds, the scaffolds were immersed in BSA solution for 24 hours and their protein adsorption was measured. As shown in Fig. 1C, the scaffolds with smaller strut sizes had higher protein adsorption. This suggests that a significant increase in the protein adsorption of the scaffold can be achieved by reducing the strut size, and the D300 scaffold has approximately 4 times the protein adsorption of the D900 scaffold, which is attributed to the increased specific surface area.

A prolonged immersion treatment of the scaffolds in Tris–HCl buffer for 30 days was used to simulate their degradation process in vitro. The surface morphology of β-TCP scaffolds with different strut sizes before and after being immersed is shown in Fig. 1A and D. The β-TCP surface became rougher, showing the crystal dissolution. And more cracks appeared between the struts, indicating the degradation of β-TCP scaffolds. Notably, the diamond structured β-TCP scaffolds with smaller strut sizes exhibited increased degradation rates, because the increased specific surface area led to an increased contact area between the scaffold surface and the buffer solution.

As shown in Fig. 1E, β-TCP scaffolds with different strut sizes exhibit varying degradation rates in Tris–HCl buffer. The D300 scaffold showed a significantly higher degradation rate at 8 and 12 weeks with values of 4.6 ± 0.3% and 8.3 ± 0.6% compared to the D500 scaffold (3.9 ± 0.2% and 6.5 ± 0.8%), the D700 scaffold (3.5 ± 0.3% and 5.0 ± 0.3%), and the D900 scaffold (2.0 ± 0.4% and 3.3 ± 0.3%). Apparently, the mass loss curve showed that the scaffolds with smaller strut sizes have a faster degradation rate. After calculation, the degradation simulation curves and functions are shown in Fig. S1. This curve indicated that the degradation rate of these scaffolds was stable, and the D300 scaffold had the fastest degradation rate of 0.0972 wt% day⁻¹ over three months. Subsequently, the degradation rates of the D500, D700 and D900 groups were 0.0816 wt% day⁻¹, 0.0671 wt% day⁻¹, and 0.0419 wt% day⁻¹. Obviously, the scaffolds degraded slowly in simulated body fluids, and the results showed that the degradation rate of the four scaffolds was distributed in a gradient as expected. As shown in Fig. 1F and G, the β-TCP scaffolds continuously released Ca²⁺ and PO₄³⁻ ions within 12 weeks, with minimal variations in ion release rates over time. Indeed, strut size modulation resulted in a gradient distribution of scaffold degradation rates, which would also directly affect the local microenvironmental concentration of Ca²⁺ and PO₄³⁻ ions and cellular behaviors for bone regeneration.

3.2. The mBMSC behaviors on β-TCP scaffolds with different strut sizes

To investigate the effect of strut sizes on cell adhesion, calcien AM/PI staining and the CCK-8 testing were used to assess mBMSC activity and proliferation. As shown in Fig. 2A, the few red fluorescence-labeled dead cells in all groups indicated that the β-TCP scaffolds had good biocompatibility. Importantly, the scaffolds with smaller struts exhibited higher cell densities. The D300 scaffold is almost completely covered by mBMSCs, while the D900 scaffold has significantly fewer mBMSCs. This is because the scaffolds with smaller strut sizes can adsorb more proteins, which facilitates cell adhesion. Furthermore, the CCK-8 results showed that the D300 scaffold promoted the proliferation of mBMSCs (Fig. 2B). The reason was that the higher permeability of the D300 scaffold facilitates nutrient exchange within the scaffold, and the large specific surface area of the scaffolds provides more space and attachment points for cell adhesion.

Furthermore, to investigate the role of strut modulation on osteogenic properties, we examined the osteogenic differentiation of mBMSCs on these scaffolds. The ALP staining and activity (Fig. 2C and D) showed that the D300 group exhibited the highest ALP expression level at day 7. And there was no significant difference between the D700 and D900 groups, but these two groups exhibited significantly lower ALP expression levels than the D500 group. At day 14, ALP expression levels of the D300 group remained significantly higher than those of other groups, showing that D300 promotes osteogenic differentiation better. Similarly, the RT-PCR (Fig. 2E) results demonstrated that the expression level of the osteogenesis-related genes OPN, RUNX2 and OCN in the D300 and D500 groups was significantly higher than those in the D700 and D900 groups. Obviously, the scaffolds with smaller strut sizes exhibited a greater capacity to upregulate the expression of osteogenesis-related genes. Consistently, the expression levels of COL1, RUNX2 and OCN genes in scaffold D300 remained significantly higher than the other groups at day 14, indicating that BMSCs in the D300 group were in an active state of osteogenic differentiation. These results confirmed that the D300 scaffold with the fastest degradation rate induces the best osteogenic differentiation of mBMSCs among these scaffolds. This may be related to the promotion of osteogenic differentiation of cells by calcium and phosphate ions released from β-TCP scaffolds.^30–32 Notably, strut morphology also exerts significant regulatory effects on both BMSC alignment and osteogenic differentiation. Substantial evidence from prior studies confirms that cylindrical struts with smaller diameters (typically 200–300 µm) induce more effective axial cell alignment and greater osteogenic differentiation capacity in BMSCs relative to larger strut dimensions (e.g., 500 µm).³³ Our experimental data consistently support this established relationship, with quantitative analyses revealing statistically significant enhancement (p < 0.05) of osteogenic differentiation on 300 µm struts compared to other diameter groups. The underlying mechanism involves preferential cell elongation and oriented attachment along narrower strut surfaces, where the resulting cytoskeletal tension generates pro-osteogenic intracellular mechanical stimuli.^23,34

3.3. The degradation-controlled β-TCP scaffolds with different strut sizes suppress osteoclastic differentiation

The release of Ca²⁺ and PO₄³⁻ ions during the degradation of calcium phosphate-based biomaterials has been known to stimulate osteoblasts,^35,36 and these Ca²⁺ and PO₄³⁻ ions can inhibit the proliferation, differentiation and resorption functions of osteoclasts as well.^37–39 To investigate the effect of β-TCP scaffolds with different strut sizes on osteoclast differentiation, we selected Raw264.7 macrophages as progenitors for osteoclastogenesis³⁹ and assayed osteoclastic differentiation of Raw264.7 co-cultured with each group of scaffolds for 6 days under RANKL-induced (Fig. 3A and Fig. S2). After exposure to RANKL for 6 days, in contrast to untreated Raw264.7, the majority of Raw264.7 cells induced by RANKL were stained red to show tartrate resistant acid phosphatase (TRAP) positive. TRAP is considered a key cytochemical marker of bone resorption and osteoclast activity. Interestingly, though osteoclasts in β-TCP scaffold groups showed TRAP-positive, they were mostly mononuclear or osteoclast precursor cells (OPCs). Especially, scaffolds with small strut sizes showed fewer and smaller multinucleated osteoclasts in co-culture. The D300 scaffold showed almost no multinucleated osteoclasts, while there were more mature and large multinucleated osteoclasts (OC) with ≥3 nucleus in the control without β-TCP scaffolds (Fig. 3A and B). Quantitative analyses further demonstrated a decrease in TRAP activity with accelerated degradation of the β-TCP scaffold (Fig. 3C). Since calcium and phosphorus ions from the β-TCP scaffold inhibit osteoclast maturation,^2,40 the D300 scaffold could more strongly inhibit osteoclast differentiation due to the release of more calcium and phosphorus ions in a short time.

Fig. 3 Differentiation of osteoclast co-cultured with scaffolds with different strut sizes after 6 days of cell culture. (A) Optical microscopy images of TRAP staining of Raw264.7 co-cultured with different scaffolds cultured in Dulbecco's modified Eagle's medium containing 50 ng mL⁻¹ RANKL (black arrows indicate multinucleated osteoclasts.). (B) Osteoclast area. (C) TRAP activity in osteoclast co-cultured with different scaffolds. (D) Secretion of PDGF-BB using the ELISA. (E) Osteoclast-related gene expression of TRAP, CTSK, and NFATc 1 in osteoclast co-cultured with different scaffolds on day 6 (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and ***p < 0.001.).

Furthermore, the OPC also promotes the migration of endothelial progenitor cells (EPCs) and induces H-capillary formation through the specific secretion of platelet-derived growth factor-BB (PDGF-BB), whereas mature osteoclasts cannot secrete PDGF-BB.^41,42 An ELISA was used to detect the concentration of PDGF-BB in the culture medium supernatant after osteoclast differentiation co-cultured with the scaffolds (Fig. 3D). The concentration of PDGF-BB was higher in the scaffold groups than the control group without scaffolds. The result illustrated that the increased rate of scaffold degradation facilitates the inhibition of OPC maturation and promotes the secretion of PDGF-BB. Thus, the faster degradation rate of the D300 scaffolds can inhibit osteoclast maturation and also enhance the release of PDGF-BB from OPCs for vascularization.^43,44

Additionally, the RT-qPCR was used to detect the expression of the osteoclast phenotype genes such as TRAP, osteoclast bone resorption function-related genes like cathepsin K (CTSK), and the mouse nuclear factor of activated T cells 1 (NFATc1) (Fig. 3E). The results showed that TRAP expression in the D300 group was lower than those of other groups, consistent with the results of TRAP activity. Apparently, as the scaffold strut size decreases, the gene expression of TRAP was sequentially down-regulated in the corresponding groups. Similarly, the gene expression of CTSK in the D900 group was up-regulated compared to other groups. CTSK is an important functional protein for osteoclasts during bone resorption, and the results suggested that osteoclasts in the D900 scaffold had stronger bone resorption capability. Furthermore, the gene expression of NFATc1 in the D300 and D500 groups was significantly lower than that in the D700 and D900 groups. NFATc1 is known as a major regulatory factor for osteoclastogenesis and its upregulation is crucial for osteoclasts obviously showed an inhibitory effect on osteoclastic differentiation and resorption due to the role of Ca²⁺ and PO₄³⁻ ions released during their degradation.^45–47 Ca²⁺ and PO₄³⁻ can inhibit osteoclast formation via Ca²⁺-calcineurin-NFATc1 and RANKL-induced NF-κB signaling pathways.^48,49

3.4. The degradation-controlled scaffolds promote osteogenesis and angiogenesis by regulating paracrine secretion of osteoclasts

The coupling mechanism between osteoclasts and osteoblasts during bone remodeling is important.^50,51 To investigate the effects of osteoclast-derived paracrine factors on the behaviors of mBMSCs and mouse islet endothelial cells (MS1), we collected OC-CM from different scaffolds to culture mBMSCs (Fig. 4A). The control group OC is the conditioned medium of osteoclasts collected without co-culture conditions with the scaffolds. The CCK-8 results showed that there was no significant difference in cell proliferation between all the groups, so OC-CM had no significant impact on the proliferation of mBMSCs (Fig. 4B). Subsequently, ALP staining and quantitative analysis were used to assess the effect of scaffolding on the osteogenic differentiation of mBMSCs. As shown in Fig. 4C and D, the D300 group showed the highest ALP activity, while the OC group showed the lowest. The result suggested that the D300 scaffolds with faster degradation rates promoted osteogenic differentiation of mBMSCs. In addition, paracrine factors of the osteoclast precursor can also recruit more BMSCs. We examined mBMSCs crossing over transwells in OC incubations in different groups. As shown in Fig. 4E and F, the results showed that OC-CM collected in the scaffold group significantly promoted the migration of mBMSCs compared to the OC group without scaffolds. Indeed, different scaffold degradation rates modulated the paracrine behavior of osteoclasts. With increasing scaffold degradation rates, the corresponding collected OC-CMs better promoted osteogenic differentiation and migration of mBMSCs.⁵² The D300 scaffold may enhance cross-talk between osteoclasts and mBMSCs, promoting osteogenic differentiation of mBMSCs.⁵³

Fig. 4 Analysis of scaffolds with different degradation rates to regulate osteoclast-promoting osteogenesis and angiogenesis. (A) Schematics of the preparation of osteoclast conditioned medium and cultivation of mBMSCs and MS1 cells. (B) Proliferation ability of mBMSCs cultured with OC-CM detected using CCK-8. (C) ALP activity in mBMSCs cultured with OC-CM for 7 days. (D) Optical microscopy images of ALP staining of mBMSCs cultured with OC-CM for 7 days. (E) and (F) Optical microscopy images and statistics of mBMSCs co-cultured with osteoclasts across the transwell. (G) Proliferation ability of MS1 cultured with OC-CM detected using CCK-8. (H) Optical microscopy images of 24-h wound healing of MS1 cultured with OC-CM. (I) 24-h wound healing rate of MS1 cultured with OC-CM (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and ***p < 0.001.).

Additionally, we cultured MS1 with OC-CM to investigate the effects of osteoclast-derived paracrine factors on MS1 behaviors. Differing from mBMSCs, the CCK-8 results showed that OC-CM in D300 and D500 scaffolds with fast degradation rates better promoted MSI proliferation compared to the D700 and D900 groups (Fig. 4G). Furthermore, the scratch test was used to investigate the effect of OC-CM on MS1 wound healing (Fig. 4H and I). The results showed that OC-CM collected from the D300 group was more effectively promoted wound healing of MS1 compared to the control. Moreover, the result of MS1 tube formation (Fig. S3) also indicated that the OC-CM collected from these scaffolds group could promote tube formation of MS1, while the control group barely formed tubes. Furthermore, the OC-CM from the D300 scaffold with the fastest scaffold degradation rate showed the strongest ability to promote MS1 tube formation. Thus, β-TCP scaffolds with controllable faster degradation can promote paracrine secretion of factors such as PDGF-BB from osteoclasts to improve migration and tubule formation of endothelial cells,⁵⁴ and these scaffolds have the better potential to enhance angiogenesis.

These results indicated that regulating the degradation rate of β-TCP scaffolds can affect the formation of osteoclasts and the paracrine effects of osteoclasts. Higher concentrations of Ca²⁺ and PO₄³⁻ ions inhibited the differentiation of macrophages into osteoclasts in vitro, and the formed mononuclear osteoclasts released platelet-derived growth factor PDGF-BB through paracrine effects. The paracrine effects could not only enhance the migration and osteogenic differentiation of mBMSCs, but also promote the proliferation, migration, and tubule formation of MS1. In summary, within an appropriate range, modulation of β-TCP scaffolds with small stent sizes can accelerate their degradation rate, which is contributing to the improvement of the scaffold's osteogenic properties.

3.5. Small-strut modulation of β-TCP scaffolds achieves controlled degradation for bone regeneration in vivo

Furthermore, a tibial defect model was used to assess the degradation properties of these scaffolds in vivo and their effect in accelerating bone regeneration (Fig. 5A), and the surgical procedure is detailed in Fig. S4. Given the slow degradation rate of the D900 scaffold and the principle of reducing the number of animals used, we chose D300, D500 and D700 scaffolds as the experimental groups in the animal experiments. The appearance of β-TCP scaffolds with different degradation rates after implantation into rabbit tibial defects for 8, 16, and 24 weeks is shown in Fig. 5B. In the 24-week long study, there were no abnormalities such as scaffold displacement or collapse, suggesting that these scaffolds could always provide stable support for new bone growth at the defect site. After implantation for 8 weeks, all scaffolds were tightly integrated with the host bone, with new bone growth from the periphery toward the center. There was more new bone on the surface of the D300 scaffold than that of D500 and D700 groups. After 16 weeks, since the new bone almost completely covered D300 scaffolds, the outline of these scaffolds became blurred, while the scaffold contours of the D700 scaffold were still clearly visible. Until 24 weeks, the contours of D700 and D500 scaffolds could be observed. In contrast, the new bone had completely covered the defect site in the D300 group and fully integrated with the surrounding bone. The apparent images of these defect sites preliminarily illustrated that the D300 group with small strut sizes had a faster rate of bone growth than that of D500 and D700 groups.

Fig. 5 (A) Model of the rabbit tibial defect. (B) Photographs of the tibial plateau and surrounding tissues 8, 16 and 24 weeks after scaffold implantation. (C) Degradation rates of different scaffolds at 8, 16, and 24 weeks after surgery. (D) Micro-CT reconstruction of bone regeneration with the scaffolds at 8, 16, and 24 weeks after surgery (scale: 3 mm) (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and ***p < 0.001).

To determine the overall new bone growth in the scaffold, the micro-computed tomography (Micro-CT) was used to scan all samples. The micro-CT results (Fig. 5C and D) showed that all scaffolds were well integrated with the surrounding host bone, and D300 scaffolds exhibited the fastest rate of new bone ingrowth. At 24 weeks, the defect area in the D300 scaffold was almost completely filled with new bone, although a significant amount of new bone also grew into the scaffolds of D500 and D700 scaffolds. The D300 scaffold with a faster degrading rate provided more space for new bone growth, accelerating the repair of bone defects. At the three time points of 8, 16, and 24 weeks, the degradation rates of the D300 scaffold (40.7 ± 4.4%, 54.1 ± 4.3%, and 85.4 ± 8.7%) were consistently higher than those of the D500 scaffold (22.9 ± 4.7%, 47.0 ± 4.6%, and 62.4 ± 2.3%) and D700 scaffold (7.3 ± 1.9%, 28.5 ± 8.1%, and 44.5 ± 10.9%). The degradation trend of the scaffolds aligned with the in vitro degradation. Specifically, scaffolds with smaller strut sizes exhibited a faster degradation rate. In summary, though new bone formed at the bone defect in all groups and tightly integrated with the scaffold due to the excellent osteoinductive properties of β-TCP ceramics, the D300 scaffold led to the fast rate of new bone formation, because faster degradation of the D300 scaffold provides more space for new bone growth.

To assess new bone formation and osteointegration at the bone defect site, we performed hard tissue sections and hematoxylin and eosin (H&E) staining. As shown in Fig. 6A, compared to the D700 scaffold, both the D300 and D500 scaffolds exhibited higher levels of new bone maturity at all time points, characterized by the presence of more osteons (OT), Volkmann's canals (VC), and more orderly arrangement of osteocytes. Notably, the new bone in the D300 group also had lamellar bone characteristics at 16 and 24 weeks, showing higher maturation of the new bone. This suggested that scaffolds with faster degradation rates are beneficial for the mineralization and maturation of new bone. Additionally, quantitative results of the new bone formation rate in histology (Fig. S5) indicated that the D300 scaffold with the fast degradation rate has the strongest capacity for bone regeneration and repair. During weeks 8 to 24, the D300 group with the fastest degradation rate had the highest amount of new bone production (48.0 ± 10.9%, 60.9 ± 10.3%, and 50.0 ± 6.6%) compared to the D500 group (39.2 ± 9.9%, 53.3 ± 9.5%, and 47.8 ± 4.4%), whereas the D700 scaffolds, with the slowest rate of degradation, had the lowest amount of new bone production (29.1 ± 3.9%, 32.6 ± 7.3%, and 42.0 ± 7.7%). The D300 scaffold achieved 85% degradation at 24 weeks along with over 50% new bone growth. At 16 weeks to 24 weeks, there was a slight decrease in the amount of new bone in the D300 group due to the remodeling and maturation of the new bone. The scaffold with smaller struts promoted the growth of new bone owing to providing more sites for the adhesion of osteogenic-related cells. Meanwhile, the faster degradation rate of the β-TCP scaffold can release more Ca²⁺ and PO₄³⁻ ions, supplying sufficient raw materials for new bone mineralization.

Fig. 6 Histological analysis after 8, 16, and 24 weeks in the tibial defect. (A) H&E staining of different scaffolds at 8, 16, and 24 weeks after surgery to evaluate osteogenesis (TCP: beta-tricalcium phosphate, OT: osteon, and VC: Volkmann's canal). (B) TRAP staining of different scaffolds at 16 weeks after surgery (blue arrows indicate where osteoclasts appear.). (C) Statistical results of the osteoclast area and number distribution (n = 4; asterisks indicate significant differences: *p < 0.05, **p < 0.01, and ***p < 0.001).

To further evaluate the effects of β-TCP scaffolds with different degradation rates on osteoclast formation in vivo, TRAP staining is used to detect osteoclast activity, as shown in Fig. 6B. At 16 weeks post-implantation, a significant number of TRAP-positive cell signals were observed in the implantation area. Statistical analysis of the number and area of osteoclasts (Fig. 6C) showed that the majority of osteoclasts within the D300 scaffolds were smaller, whereas osteoclasts in D700 scaffolds were larger osteoclasts. These results suggest that β-TCP scaffolds with different degradation rates exert different regulatory effects on osteoclast behavior in vivo, which also has been consistent with the results in vitro. D300 scaffolds with a faster degradation rate indeed inhibit intercellular fusion of osteoclast precursor cells and the formation of mature osteoclasts by pumping out large amounts of Ca²⁺ and PO₄³⁻ ions for promoting the formation and maturation of new bone.⁵⁵ Therefore, this work confirmed that the design of small strut structures can modulate the controlled and rapid degradation of β-TCP scaffolds to achieve inhibition of osteoclastogenesis and effectively stimulate bone regeneration.

4. Conclusions

This study systematically investigated the effects of precisely controlled microstructural parameters (strut sizes ranging from 300 to 900 µm) on the degradation behavior and osteogenic performance of β-TCP scaffolds in vitro and in vivo. The results demonstrated that scaffolds with a 300 µm strut architecture achieved optimal bone repair efficacy through two primary mechanisms: (1) enhanced osteoblast activity and angiogenesis mediated by the increased surface area and controlled ion release, and (2) effective suppression of osteoclast maturation through the time-dependent release of degradation products (Ca²⁺/PO₄³⁻). In a tibial defect model, these microstructure-optimized β-TCP scaffolds exhibited excellent synchronization between degradation kinetics and bone regeneration processes.

The proposed microstructure design strategy successfully optimized degradation performance without altering material composition, demonstrating the significance of structural adaptability in biomaterial development. This structure–performance relationship provides important insights into designing next-generation bone repair materials with improved biological adaptation. Our findings not only offer direct guidance for clinical optimization of β-TCP implants, but also advance fundamental understanding of the interplay between the material microstructure and biological responses.

Author contributions

Qiji Lu: writing – original draft, conceptualization, formal analysis, investigation, methodology, validation, and visualization. Fansan Zeng: writing – review and editing, formal analysis, investigation, methodology, and validation. Yudi Kuang: methodology, validation, and visualization. Jingjing Diao: conceptualization, investigation, methodology, and validation. Naru Zhao: writing – review and editing, supervision, and conceptualization. Yingjun Wang: supervision, conceptualization, formal analysis, visualization, funding acquisition, project administration, and resources.

Conflicts of interest

The authors declare no competing interests.

Data availability

All data supporting this article are included in the manuscript and the supplementary information (SI). See DOI: https://doi.org/10.1039/d5tb01393c.

Acknowledgements

We would like to thank the National Key R&D Program of China (2021YFB3800800), the National Natural Science Foundation of China (52172281 and 32201089), the Guangzhou Key R&D Program (202206040001 and 202007020002), the Guangzhou Science and Technology Planning Project (2024A04J6187), and the Natural Science Foundation of Guangdong Province (2021A1515011741) for their support.

References

1. A. Barba, A. Diez-Escudero, Y. Maazouz, K. Rappe, M. Espanol, E. B. Montufar, M. Bonany, J. M. Sadowska, J. Guillem-Marti, C. Öhman-Mägi, C. Persson, M.-C. Manzanares, J. Franch and M.-P. Ginebra, ACS Appl. Mater. Interfaces, 2017, 9, 41722–41736.
2. J. Jeong, J. H. Kim, J. H. Shim, N. S. Hwang and C. Y. Heo, Biomater. Res., 2019, 23, 4.
3. H. H. Horch, R. Sader, C. Pautke, A. Neff, H. Deppe and A. Kolk, Int. J. Oral Surg., 2006, 35, 708–713.
4. M. T. Ciprandi, B. T. Primo, H. T. Gassen, L. Q. Closs, P. A. Hernandez and A. N. Silva, Jr., J. Craniofac. Surg., 2012, 23, 145–148.
5. K. Sawada, K. Nakahara, M. Haga-Tsujimura, T. Iizuka, M. Fujioka-Kobayashi, K. Igarashi and N. Saulacic, Odontology, 2018, 106, 398–407.
6. K. Sarkar, V. Kumar, K. B. Devi, D. Ghosh, S. K. Nandi and M. Roy, ACS Biomater. Sci. Eng., 2019, 5, 5097–5106.
7. Y. Xia, P. Zhou, F. Wang, C. Qiu, P. Wang, Y. Zhang, L. Zhao and S. Xu, Int. J. Nanomed., 2016, 11, 3435–3449.
8. Y. Ding, Y. Li, J. Lin and C. Wen, J. Mater. Chem. B, 2015, 3, 3714–3729.
9. M. Kazemi, M. M. Dehghan and M. Azami, Mater. Sci. Eng., C, 2019, 105, 110071.
10. Z. Sheikh, M. N. Abdallah, A. A. Hanafi, S. Misbahuddin, H. Rashid and M. Glogauer, Materials, 2015, 8, 7913–7925.
11. A. Bigham, F. Foroughi, E. Rezvani Ghomi, M. Rafienia, R. E. Neisiany and S. Ramakrishna, Bio-Des. Manuf., 2020, 3, 281–306.
12. B. Zhang, X. Gui, P. Song, X. Xu, L. Guo, Y. Han, L. Wang, C. Zhou, Y. Fan and X. Zhang, ACS Appl. Mater. Interfaces, 2022, 14, 8804–8815.
13. J. W. Lee and D. W. Cho, Curr. Pharm. Des., 2015, 21, 1606–1617.
14. K. M. Ferlin, M. E. Prendergast, M. L. Miller, D. S. Kaplan and J. P. Fisher, Acta Biomater., 2016, 32, 161–169.
15. H. Zhang, C. Jiao, Z. Liu, Z. He, G. Mengxing, T. Zongjun, C. Wang, Z. Wei, L. Shen and H. Liang, J. Mech. Behav. Biomed. Mater., 2021, 121, 104642.
16. C. Schmidleithner, S. Malferrari, R. Palgrave, D. Bomze, M. Schwentenwein and D. M. Kalaskar, Biomed. Mater., 2019, 14, 045018.
17. Y. Xiao, Q. Yin, L. Wang and C. Bao, Ceram. Int., 2015, 41, 995–1003.
18. J. Tian, W. Qi, Y. Zhang, M. Glogauer, Y. Wang, Z. Lai and H. Jiang, J. Endod., 2015, 41, 1500–1506.
19. J. Nishino, M. Yamazaki, M. Kawai, K. Tachikawa, K. Yamamoto, K. Miyagawa, M. Kogo, K. Ozono and T. Michigami, J. Cell. Biochem., 2017, 118, 1151–1163.
20. H. Tada, E. Nemoto, B. L. Foster, M. J. Somerman and H. Shimauchi, Bone, 2011, 48, 1409–1416.
21. J. Wang, Q. Zhao, L. Fu, S. Zheng, C. Wang, L. Han, Z. Gong, Z. Wang, H. Tang and Y. Zhang, Bioact. Mater., 2022, 15, 446–455.
22. B. Zhang, X. Pei, C. Zhou, Y. Fan, Q. Jiang, A. Ronca, U. D'Amora, Y. Chen, H. Li, Y. Sun and X. Zhang, Mater. Des., 2018, 152, 30–39.
23. Q. Lu, J. Diao, Y. Wang, J. Feng, F. Zeng, Y. Yang, Y. Kuang, N. Zhao and Y. Wang, Bioact. Mater., 2023, 26, 413–424.
24. J. Diao, H. Ding, M. Huang, X. Fu, F. Zou, T. Li, N. Zhao, C. Mao and Y. Wang, Small Methods, 2019, 3, 1900237.
25. Y. Wang, Y. Liu, S. Chen, M.-F. Francis Siu, C. Liu, J. Bai and M. Wang, Bioact. Mater., 2025, 46, 21–36.
26. H. Liu, T. Liu, Z. Yin, X. Liu, Y. Tan, Y. Zhao and H. Yu, Regen. Biomater., 2024, 11, rbae023.
27. R. K. Athira, G. Gayathry, P. R. Anil Kumar, P. R. Harikrishna Varma, N. Kasoju and M. Komath, Ceram. Int., 2021, 47, 30051–30060.
28. X. Cao, K. Wu, C. Wang, Y. Guo, R. Lu, X. Wang and S. Chen, Int. J. Mol. Sci., 2022, 23, 8723.
29. M. Taale, F. Schütt, K. Zheng, Y. K. Mishra, A. R. Boccaccini, R. Adelung and C. Selhuber-Unkel, ACS Appl. Mater. Interfaces, 2018, 10, 43874–43886.
30. B. Liu and D. X. Lun, Orthop. Surg., 2012, 4, 139–144.
31. A. H. Mohammed Mohammed, K. A. Shariff, M. H. A. Bakar and H. Mohamad, J. Aust. Ceram. Soc., 2022, 58, 1715–1727.
32. F. R. Gavilán, A. Cerqueira, E. Anitua, F. Muñoz, I. G. Arnáez, M. Azkargorta, F. Elortza, M. Gurruchaga, I. Goñi, J. Suay and R. Tejero, J. Mater. Chem. B, 2024, 12, 2831–2842.
33. N. D. Bade, R. D. Kamien, R. K. Assoian and K. J. Stebe, Sci. Adv., 2017, 3(9), e1700150.
34. R. McBeath, D. M. Pirone, C. M. Nelson, K. Bhadriraju and C. S. Chen, Dev. Cell, 2004, 6, 483–495.
35. F. Viti, M. Landini, A. Mezzelani, L. Petecchia, L. Milanesi and S. Scaglione, PLoS One, 2016, 11, e0148173.
36. L. He, Y. Huang, J. Gu, X. Liu, J. Yin and X. Gao, J. Bionic Eng., 2024, 21, 1960–1974.
37. Y. Doi, H. Iwanaga, T. Shibutani, Y. Moriwaki and Y. Iwayama, J. Biomed. Mater. Res., 1999, 47, 424–433.
38. L. Mira-Pascual, A. N. Tran, G. Andersson, T. Näreoja and P. Lång, Int. J. Mol. Sci., 2020, 21(2), 538.
39. Y. Han, X. You, W. Xing, Z. Zhang and W. Zou, Bone Res., 2018, 6, 16.
40. H. J. Kim, J. Lee, G. R. Lee, N. Kim, H. I. Lee, M. Kwon, N. Y. Kim, J. H. Park, Y. H. Kang, H. J. Song, T. Kim, D. M. Shin and W. Jeong, J. Cell. Physiol., 2021, 236, 8239–8252.
41. J. Y. Song, K. M. Pineault, J. M. Dones, R. T. Raines and D. M. Wellik, Proc. Natl. Acad. Sci. U. S. A., 2020, 117, 7296–7304.
42. R. Russo, S. Mallia, F. Zito and N. Lampiasi, Cells, 2019, 8(2), 131.
43. A. N. Rindone, X. N. Liu, S. Farhat, A. Perdomo-Pantoja, T. F. Witham, D. L. Coutu, M. Wan and W. L. Grayson, Nat. Commun., 2021, 12(1), 6219.
44. C. Dou, J. M. Li, J. He, F. Luo, T. Yu, Q. J. Dai, Y. Q. Chen, J. Z. Xu, X. C. Yang and S. W. Dong, Bioact. Mater., 2021, 6, 4697–4706.
45. J. Tian, W. Qi, Y. Zhang, M. Glogauer, Y. Wang, Z. Lai and H. Jiang, J. Endod., 2015, 41, 1500–1506.
46. N. A. Sims and T. J. Martin, Annu. Rev. Physiol., 2020, 82, 507–529.
47. N. A. Sims, J. Bone Miner. Res., 2024, 39, 1377–1385.
48. B. R. Cao, X. Dai and W. B. Wang, J. Cell. Physiol., 2019, 234, 6831–6841.
49. X. S. Liu, R. R. Guo, S. C. Huo, H. Chen, Q. X. Song, G. Jiang, Y. Yu, J. L. Huang, S. W. Xie, X. L. Gao and L. J. Lu, J. Controlled Release, 2022, 343, 314–325.
50. M. Bahraminasab, M. Janmohammadi, S. Arab, A. Talebi, V. T. Nooshabadi, P. Koohsarian and M. S. Nourbakhsh, ACS Biomater. Sci. Eng., 2021, 7, 5397–5431.
51. B. Gao, R. Deng, Y. Chai, H. Chen, B. Hu, X. Wang, S. Zhu, Y. Cao, S. Ni, M. Wan, L. Yang, Z. Luo and X. Cao, J. Clin. Invest., 2019, 129, 2578–2594.
52. F. Chen, L. Tian, X. Pu, Q. Zeng, Y. Xiao, X. Chen and X. Zhang, Biomater. Sci., 2022, 10, 5925–5937.
53. X. P. Yu, H. S. Ma, Y. F. Wang, J. X. Hao, L. Chen, M. Gelinsky and C. T. Wu, Adv. Mater., 2024, 36(21), 2308126.
54. W. J. Liu, Q. Y. Wang, H. Y. Luo, B. C. Luo, F. J. Zhao, Y. Y. Kang, Y. L. Zhang and L. Q. Shao, ACS Nano, 2024, 18, 22390–22403.
55. M. K. Hsieh, C. Y. Wang, F. C. Kao, H. T. Su, M. F. Chen, T. T. Tsai and P. L. Lai, JBMR Plus, 2024, 8, ziae031.

---

Footnote

† These authors contributed to this work equally and should be regarded as co-first authors.

---