Allograft microspheres that could induce autologous bone regeneration have been widely applied in clinical trials (Krauss, 1999). In this research, allograft microspheres were immersed in mSBF for surface mineralization to form nanostructured mineral coatings (Scheme 1). After mixing PEG with CS, the CS/PEG gel was formed by the cross-linking effect. Compared with the CS/PEG gel, CS/PEG-MCM had a rougher surface due to the presence of MCM. As shown in Figure 1A, MCMs were uniformly covered by a layer of coating with a typical plate-like structure after immersion in mSBF, and the surface of the MCMs was determined as having significant amounts of calcium and phosphorus by energy-dispersive spectroscopy (EDS) (Figure 1B), which was helpful for the loading of protein and promoting osteogenesis. CS has been widely used in biomedical fields because of its good biodegradability and excellent cell affinity. After mixing the CS solution with the PEG solution, the hydrogel (CS/PEG gel) was formed by the cross-linking effect. A uniform network structure was observed using a scanning electron microscope (SEM). Then, MCMs were dispersed with the PEG solution, followed by cross-linking with CS for the generation of the CS/PEG-MCM gel. As shown in Figure 1A, there is a big difference between the CS/PEG gel and the CS/PEG-MCM gel. The surface of the CS/PEG gel was relatively smooth, while that of the latter was covered with many microspheres. Furthermore, the results of EDS further demonstrated that MCM was successfully embedded in the hydrogel because the intensity of calcium and phosphorus in the CS/PEG gel (Figure 1C) was much lower than that in the CS/PEG-MCM gel (Figure 1D).

Then, in order to further demonstrate the successful preparation of the CS/PEG gel and the loading of MCM in the hydrogel, the gelation process of the hydrogel was observed (Figure 2A). Before gelation, both the mixtures of CS/PEG and CS/PEG-MCM had excellent flowability. After approximately 4 min of the reaction, the solution transformed into solid hydrogel, which did not flow as the bottle was turned upside down. In addition, the surface of the CS/PEG-MCM gel was much rougher than that of the CS/PEG gel due to the presence of MCM. Next, we further examined the Fourier transform infrared (FT-IR) spectrum of the various samples. The absorption peak of CS at 2,970 cm⁻¹ had disappeared in the CS/PEG gel, indicating the chemical reaction between CS and PEG, which may be the mechanism of the cross-linking of the hydrogel. Compared with the CS/PEG gel, the absorption peak of the CS/PEG-MCM gel around 1,110 cm⁻¹ had significantly decreased, indicating that MCM was successfully introduced to the hydrogel (Figure 2B).

Subsequently, to investigate the degradation of the hydrogel, the CS/PEG gel and CS/PEG-MCM gel were soaked in PBS for different time periods. As shown in Figure 2C, both degraded quickly within 6 h, and then the degradation tended to be slow. They had only lost about 20% after 1 week, which indicates their superb retention ability in defects. Then, in order to determine the sustained-release ability of the composite hydrogel, the release behavior of BMP-2 in MCM@BMP-2 and CS/PEG-MCM@BMP-2 was observed. As shown in Figure 2D, the cumulative release of BMP-2 from the MCM@BMP-2 gel was 86.9% ± 0.7% on day 16 and continued increasing to 98.0% ± 3.7% on day 75. As for the CS/PEG-MCM@BMP-2 gel, the cumulative release of BMP-2 was only 30.5% ± 3.6% on day 75. Thus, it is obvious that the sustained-release effect of the composite hydrogel was much better than that of MCM, which exhibited a promising strategy in tissue engineering.

Live/dead staining (Sang et al., 2022) was used to determine the biocompatibility of the hydrogels. Calcein-AM can pass through the cell membrane and produce calcein that can emit green fluorescence under the action of intracellular enzymes, while PI as a nuclear dye cannot pass through the living cell membrane—it can only pass through the cell membrane of dead cells to the nucleus, react with DNA, and emit red fluorescence (Luo et al., 2021a; Luo et al., 2021b). In order to measure the biocompatibility of the hydrogel scaffolds, HUVECs were treated with the supernatant from the hydrogels for 24, 48, and 72 h. As shown in Figure 3A, all groups displayed a good morphology at 24 h, while they still maintained a favorable proliferative ability at 48 and 72 h. In addition, there was no significant difference among the three groups, and they all exhibited excellent proliferative activity and showed green fluorescence, which indicates the good biocompatibility of the hydrogel. Interestingly, HUVECs treated with these hydrogel extracts seemed to undergo tube formation after 48 h of incubation. Then, we employed ImageJ to analyze the pictures (Figure 3B). To our surprise, all of them possessed the ability to promote tube formation, and there was no significant difference in meshes, total length, and junctions (Figures 3C–E). Then, we further performed an MTT assay in HUVECs to quantitatively detect the cell proliferation ability. In addition, the optical density (OD) values were also similar, and the statistics had no difference. In addition, the OD values were increased over time, which further demonstrated the superior biocompatibility of the hydrogels. Thus, both results showed good compatibility of the hydrogels.

In order to verify the osteogenic differentiation ability of the hydrogel, we performed an immunofluorescence experiment to study osteogenic-related protein expression in BMSCs. RUNX2 is a specific transcription factor for osteoblasts and plays an important role in bone formation and reconstruction. It determines the differentiation of pluripotent stem cells to osteoblasts and promotes chondrocyte maturation and cartilage vascularization (Komori, 2019). In addition, osteopontin is an important bone matrix protein, which is closely related to bone formation and development (Si et al., 2020). As for osteocalcin (Komori, 2020), it is mainly synthesized by osteoblasts and dentin cells, which play an important role in regulating bone calcium metabolism. Therefore, these osteogenic proteins play an important role in osteogenic differentiation. In order to have a more accurate image, F-actin (red) was used to observe the cytoskeleton of the cells, and DAPI (blue) was used as a dye to observe the nucleus. As shown in Figures 4A, B, the cells treated with both the CS/PEG gel and CS/PEG-MCM@BMP-2 gel had a good morphology, which further demonstrated the good biocompatibility of the hydrogels. However, the intensities of the green fluorescence of RUNX2, osteopontin, and osteocalcin in the CS/PEG-MCM@BMP-2 gel were all stronger than that in the CS/PEG gel. In addition, according to the results of ImageJ analysis, the expressions of RUNX2 (Figure 4C), osteopontin (Figure 4D), and osteocalcin (Figure 4E) in the cells treated with CS/PEG-MCM@BMP-2 gel were all higher than that in the CS/PEG gel, which indicated that this composite hydrogel greatly supported osteogenic differentiation.

To investigate the ability of the composite hydrogel to reconstruct bone defects in vivo, a rat cranium defect model was constructed by electric drill after anesthesia, and then the various hydrogels were implanted into the defects, followed by the suture. After rearing for 4 and 8 weeks, the skulls were extracted from the rats and immersed in paraformaldehyde solution for further determination. In order to intuitively observe the regeneration of the skull in rats, micro-computed tomography (micro-CT) was performed. From the 3D reconstruction of the cranium defect by micro-CT (Figure 5A), the new bone coverage areas of the same samples at 8 weeks were all higher than those at 4 weeks, which indicated that these hydrogels all had some bone regeneration ability, but there was a difference in efficacy, which indicated the sustained release of the hydrogel and a little self-healing ability. The CS/PEG gel had the weakest efficacy, whether at week 4 or 8, the new bone coverage area was the smallest. The CS/PEG@BMP-2 gel and CS/PEG-MCM gel had a better ability in bone regeneration, which may be contributed to the presence of BMP-2 or MCM in the hydrogel, respectively. Most importantly, the CS/PEG-MCM@BMP-2 gel was clearly higher than other hydrogels at 4 weeks (Figure 5B). However, at week 8, the fracture of new bone in defects of CS/PEG-MCM@BMP-2 had no significant difference from that of CS/PEG-MCM, which may be due to the presence of MCM or the sustained release of BMP-2 (Figure 5C), which requires further exploration.

In order to further explore the osteogenic effect of the hydrogels, the cranium tissues of various samples at 4 and 8 weeks were harvested from the rats for hematoxylin and eosin (HE) staining to evaluate the formation of new bone (Miao et al., 2019). The pink part in the middle of HE staining image indicated the formation of the new bone. As illustrated in Figure 6A, the new bone coverage area of the CS/PEG-MCM@BMP-2 gel was 35.7% ± 0.9% (Figure 6C) at 4 weeks and continued increasing to 56.5% ± 1.5% (Figure 6D) at 8 weeks, both of which had a significant difference from other groups. Then, in order to further determine the ability of bone regeneration, Masson staining was also employed (Cheng W. et al., 2020). As reported earlier, the newborn collagen fibers can react with aniline blue to appear blue, and tendon fibers can react with acid fuchsin to appear red. In addition, the new bone is rich in collagen, which indicates that the larger the blue area, the more bone is produced. As shown in Figure 6B, the CS/PEG-MCM@BMP-2 gel possessed the most area of new bone. After quantitative analysis by ImageJ, the percentage of the new bone in CS/PEG-MCM@BMP-2 both at 4 weeks (Figure 6E) and 8 weeks (Figure 6F) had a statistical difference with that of other hydrogels, which further demonstrated the better osteogenic ability of the CS/PEG-MCM@BMP-2 gel.

Subsequently, we further performed immunofluorescent staining to explore the expression of collage to clarify the osteogenic ability of the hydrogels. Col-Ⅰ is a significant marker of the formation of bone mechanical strength. With time, the expression of Col-Ⅰ was higher at 8 weeks than at 4 weeks (Figure 7A). Moreover, the tissues treated with the CS/PEG-MCM@BMP-2 gel showed the highest intensity of Col-Ⅰ at 8 weeks (Figures 7B, C). Furthermore, we also explored the expression of TGF-β by immunofluorescent staining (Figure 8A). TGF-β is a protein that plays an important part in bone formation. As shown in Figures 8B, C, the expression of TGF-β was also higher in the group of the CS/PEG-MCM@BMP-2 gel, which showed the same results as the expression of Col-Ⅰ. Thus, this composite hydrogel showed a good effect on bone defect repairing in vivo.

Allograft bone microsphere was purchased from Yapeng Biological. Briefly, 1 mg of allograft bone microspheres was immersed in 0.5 mL of SBF for a week for mineralization, and then the fresh SBF was added to replace the original solution every day. Finally, the MCM was collected after lyophilization.

For the preparation of the CS/PEG gel, we first configured the solution of the pre-hydrogel. A total of 0.15 g of chitosan (CS) and 0.35 g of polyethylene glycol (PEG) were dissolved in 2 mL of PBS under ultrasonication. Then, the CS/PEG gel was prepared after 4 min of mixing the CS solution with the aforementioned PEG solution.

For the preparation of the CS/PEG-MCM gel, microspheres (allograft, Yapeng Biological) were immersed in mSBF for the generation of MCM. Then, 10 mg of MCM was dispersed in 50 μL PEG solution, followed by the mixture with 50 μL CS.

As for the CS/PEG-MCM@BMP-2 gel, 10 μg BMP-2 and 10 mg MCM were added together to 1 mL PBS at 37°C for 4 h incubation to allow the binding of BMP-2. Then, the aforementioned MCM@BMP-2 was dispersed in the PEG solution, followed by cross-linking with the CS solution.

The morphology and energy-dispersive X-ray spectroscopy (EDS) of the freeze-dried hydrogel and MCM were observed using a scanning electron microscope (SEM, HITACHI S-4800). The infrared spectrogram of the samples was determined by Fourier transform infrared spectroscopy.

The degradation behavior of the hydrogel was determined in the PBS solution (pH 7.4) (Cheng et al., 2020a). First, the primary mass of the freeze-dried hydrogel was recorded. Then, the freeze-dried hydrogel was immersed in 2 mL PBS in an oscillator shaker with constant temperature (37°C). At each time point, the hydrogel was taken from the tube, followed by the recording the mass after being freeze-dried.

The drug release behavior of BMP-2 from MCM@BMP-2 or CS/PEG-MCM@BMP-2 was evaluated in the PBS solution (pH 7.4). The supernatants were taken from the solutions in an oscillator shaker with constant temperature (37°C) at various time points by centrifugation, and fresh PBS was added to the tubes for further immersion. Finally, the concentration of BMP-2 was determined using an ELISA kit.

Human umbilical vein endothelial cells (HUVECs) were cultured in DMEM containing 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin (PS) at 37°C under 5% CO₂. For the biocompatibility test, live–dead staining was performed using a living/dead cell double staining kit (Luo et al., 2021b). Calcein-AM is able to easily penetrate into living cell membranes, followed by green fluorescence, which is only suitable for living cells. In addition, pyridine iodide (PI) is only used for dead cells for red fluorescence. HUVECs were seeded on a 24-well plate overnight, and the leaching solutions of the CS/PEG gel and CS/PEG-MCM gel were added to the plate. After 24-, 48-, and 72-h incubation, the mixture of 2 μM calcein-AM and 8 μM PI was added to the plate for 30 min of incubation. Then, the cells were washed and observed using a fluorescent microscope. In addition, the tube-forming behavior was analyzed by ImageJ software.

We further performed an MTT assay to determine the cell viability of the hydrogel. HUVECs were seeded on a 96-well plate overnight, and the leaching solutions of the CS/PEG gel and CS/PEG-MCM gel were added to the plate for further incubation. After 24-, 48-, and 72-h incubation, the cells were washed, and the MTT solution was added for 4 h of incubation. Finally, the cell viability was measured by the addition of DMSO, followed by the measurement using a microplate reader.

Bone marrow mesenchymal stem cells (BMSCs) wer cultured in α-MEM containing 10% FBS and 1% PS at 37°C under 5% CO₂. First, BMSCs were seeded on a glass in a 6-well plate in a cell incubator overnight, and the leaching solutions of CS/PEG gel and CS/PEG-MCM@BMP-2 gel were added into the plate for 24 h of incubation. In addition, the cells were washed and incubated with 4% paraformaldehyde for 20 min at 37°C. Next, the cells were washed with PBS three times, and 0.5% Triton X-100 was added to the plate for 20 min of incubation at room temperature. After washing with PBS three times again, the cells were incubated with 5% BSA for 30 min for closure. Then, osteogenic antibodies including RUNX-2, osteopontin, and osteocalcin were added to the plates, followed by further incubation at 4°C overnight. The next day, the glasses were washed with PBST three times, followed by 1 h of incubation with goat anti-rabbit IgG-H&L at room temperature. Then, DAPI was added to the plate for 5 min of incubation. Finally, the glasses were washed with PBST and observed using a fluorescent microscope (Miao et al., 2019). The results were analyzed by ImageJ.

A rat skull defect model was constructed for the research model (Cheng et al., 2020b). The CS/PEG gel and CS/PEG-MCM@BMP-2 gel were implanted into the skull defect for a certain period of time. Then, the area of the skull was detected using an X-ray image analysis system and extracted from the rat for further immunohistochemical analysis.

The skull extracted from the rat was fixed in 4% polyformaldehyde for 24 h, followed by decalcification with 10% EDTA for 1 week at 37°C. Then, the samples were embedded in paraffin and sliced into 5-μm-thick sections. Finally, the slides were reacted with hematoxylin and eosin, Masson, Col-Ⅰ, and TGF-β (Qiao et al., 2020).

All data were reported as mean ± standard deviation (SD), and the data were analyzed by the GraphPad software (ns, non-significance, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001).