CFO nanoparticles, 3-(trimethoxysilyl)-propyl methacrylate (TMSPMA) and 2-hydroxy-4-(2-hydroxyethoxy)-2-methylpropiophenone (Irgacure-2959, I2959) were purchased from Macklin (Shanghai, China). GelMA was purchased from Engineering For Life (Suzhou, China). All the reagents for cell cultures were purchased from VivaCell (Shanghai, China). Other reagents for cell experiments were purchased from Beyotime Biotechnology (Shanghai, China) unless otherwise mentioned.

MeCFO was synthesized from a slightly modified procedure reported previously (Cha et al., 2014). CFO nanoparticles were conjugated with methacrylate functional groups by reacting with TMSPMA. Briefly, CFO nanoparticles were evenly dispersed in anhydrous ethanol by ultrasound for 20 min. The TMSPMA solution was slowly added to CFO suspension with a ratio of 50 μL per mg CFO. The reaction mixture was stirred mechanically for 12 h at 50°C. And it was dialyzed in anhydrous ethanol for 3 days. Finally, MeCFO was obtained by vacuum drying and kept in a dark place.

The hydrogel was prepared by crosslinking under ultra-violet (UV) (405 nm) light with I2959 as the photoinitiator. It mainly consisted of 5% GelMA doped with different concentrations of MeCFO (Table 1). Briefly, GelMA was dissolved in phosphate buffered saline (PBS) at 60°C. The photoinitiator I2959 was completely dissolved in PBS at 40°C in dark. MeCFO was sonicated and dispersed in PBS for 2 h. Then GelMA was added to the MeCFO dispersion and thoroughly mixed. At last, the I2959 solution was mixed into the above system with a final concentration of 0.1%. The solution was then transferred to a polymethyl methacrylate mold and formed by UV light for 10 min.

To increase the GelMA viscosity and make GelMA undergo reversible physical cross-linking, the prepared MeCFO/GelMA solution was placed at 4°C for 20 min before printing. Then the printer cartridge was loaded onto the biological 3D printer with the low-temperature mode. The platform temperature was controlled at 4°C. The temperature of the print nozzle was 24°C. The pressure of the printing extrusion was controlled to be about 200 kPa. The moving speed of the dispensing unit was set to 200 mm/s, and the printing height and spacing were respectively 0.23 mm and 1 mm. After setting the parameters, we adjusted the size and generation path of the 3D model as required. The prepared MeCFO/GelMA solution was easily extruded from the needle due to shear thinning. When it dropped onto the printing platform (4°C), it quickly restored viscosity. To solidify the printed scaffolds, UV light curing was carried out synchronously in the printing process. Finally, the printed pre-scaffold was irradiated with UV light for 10 min.

All experimental data were analyzed by Origin 2021 software (OriginLab Corporation, United States). Their results were presented as a mean ± standard deviation. Statistical analysis was performed using the two-tailed Student’s t-test between different groups. One-way analysis of Variance (ANOVA) was used to compare and analyze multiple data sets. The level of significant statistical differences was set as a p-value less than 0.05.

The surface of CFO nanoparticles in this study was chemically modified to reduce their agglomeration (Atila Dinçer et al., 2014; Sundar et al., 2014). For this purpose, methacrylate groups were conjugated on the surface of CFO by reaction with TMSPMA to obtain MeCFO (Figure 1A). TEM images illustrated the morphology and size of MeCFO. As shown in Figure 2A, CFO and MeCFO were both polygonal in shape with an average diameter of around 20–100 nm. Compared with CFO, there was a slight difference in the morphology of the magnetic nanoparticles after surface modification. The modified CFO nanoparticles were thicker as they were coated with a thin TMSPMA layer. Moreover, CFO nanoparticles were aggregated and overlapped while MeCFO nanoparticles were dispersed with little overlap. This indicates that TMSPMA enhanced the dispersion of magnetic nanoparticles, which was conducive to the dispersion of nanoparticles in the substrate biomaterial (Atila Dinçer et al., 2014; Tanasa et al., 2019).

Figure 2B showed the main elemental composition of CFO and MeCFO by EDX spectrum. The presence of the elements, C, Fe, Co, and O, were appeared in the MeCFO. But Si in MeCFO was not found by EDX spectrum analysis due to the low content of Si. FTIR spectra (Figure 2C) confirm the successful synthesis of MeCFO by identifying the functional group of its organic molecules. The FTIR spectrum CFO and MeCFO were found characteristic peaks at 580–590 cm^-1 due to the stretching vibration of the Fe-O bond (Bohara et al., 2014). The absorption bands around 3,400 cm^-1 and 1,631 cm^-1 are attributed to the bending and stretching vibrations of water molecules (Penchal Reddy et al., 2015). The absorption bands near 1,382 cm^-1 and 1,013 cm^-1 respectively corresponded to the vibration of the Si-C bond and vibration of the Si-O bond from TMSPMA (Cha et al., 2014; Koohestani and Sadjadi, 2021). The hydroxyl functional group of CFO nanoparticles was transformed into the methacrylate group by silanization. Because of the existence of C=O bonds in TMSPMA, the absorption peak of MeCFO near 1,630 cm^-1 was significantly stronger than that of CFO.

For further surface composition analysis of CFO and MeCFO, XPS detection was performed (Figures 2D–J). XPS spectrum can identify the Si in the MeCFO sample compared to CFO samples (Figure 2D). The XPS data of CFO nanoparticles could not be identified as silicon by Thermo Avantage v5.948 software. The peak of MeCFO nanoparticles’ binding energy at 285.08 eV was attributed to C1s, and the weak peak at 102.08 eV is related to the presence of Si2p, which further proved that TMSPMA was coated on the surface of CFO nanoparticles (Tan et al., 2013). The high-resolution Fe2p, O1s, and C1s XPS spectra were also investigated for an in-depth analysis of the MeCFO composition. In the Fe2p spectrum of CFO and MeCFO (Figures 2E, F), two obvious peaks were observed at 724.3 and 710.5 eV, ascribed to Fe2p _1/2 and Fe2p _3/2 of CFO, respectively (Rodewald et al., 2019). Figure 2G highlighted that the O1s peak of CFO nanoparticles had two deconvolution peaks. The first peak was concentrated at 529.5 eV, which could be attributed to Fe-O (Márquez et al., 2012). The second peak was concentrated at 530.8 eV, possibly corresponding to hydroxyl bonding within the magnetic nanoparticles (Tanasa et al., 2019). Furthermore, as shown in Figure 2H, four different kinds of O1s peaks were observed in the XPS spectra of MeCFO at 529.5, 529.9, 531, and 532.1 eV. They are assigned to Fe-O, Si-O, C=O, and C-O, respectively (Xiong et al., 2016). In the C1s spectrum of MeCFO (Figure 2I) four types of C1s peaks correspond to the following functional groups: C-C (284.5 eV), C-O (284.9 eV), C-N (285.5 eV) and O=C-O (288.1 eV), respectively (Shi et al., 2016). As seen in Figure 2J, the Si2p curve simulates two characteristic signals of 101.9 eV (C-Si-O) and 102.5 eV (O-Si-O) (Yan et al., 2022). These results further prove there are methacrylate groups on the surface of the CFO.

The thermal behavior of CFO nanoparticles before and after functional modification is shown in Figures 3A, B. Figure 3A displayed that the weight of CFO and MeCFO decreased with the increase in temperature. There was a significant weight loss within 360°C. It could be the evaporation of water and pyrolysis of the unstable hydroxyl group from the magnetic nanoparticles (Zhao et al., 2012; Alam et al., 2017). The weight loss rate of MeCFO was higher than that of CFO, which is related to the continuous loss of methacrylate groups (Duong et al., 2017). Their decline rates tended to be the same above 360°C. When the temperature rose to 800°C, the weight loss ratios of CFO and MeCFO were respectively 2.337% and 2.839% (Table 3). Thermogravimetric results also demonstrated that TMSPMA was successfully grafted on the surface of CFO nanoparticles. The derivative thermogravimetric (DTG) curve of CFO and MeCFO were shown in Figure 3B. The maximum degradation temperature of CFO and MeCFO respectively were 323.57°C and 345.92°C, which is consistent with the literature description (Nguyen et al., 2020). The intensity and area of the CFO degradation peak were greater than those of the MeCFO. This also indicated that the methacrylate groups on MeCFO were decomposed by heat. Figure 3C showed the hysteresis loops of CFO and MeCFO. The saturation magnetization (Ms) values for the CFO and MeCFO samples were 58.52 and 59.56 emu/g, respectively. The coating did not significantly change the magnetization. In addition, the coercivity (Hc) of CFO and MeCFO samples were 2.17 and 2.19 KOe. According to the literature, an increase in the Hc value of MeCFO may be attributed to its increased diameter as a result of surface functionalization (Almasi Kashi et al., 2020). The coating effect of TMSPMA on CFO is close to the results reported in the references (Atila Dinçer et al., 2014; Tanasa et al., 2019; Agnihotri et al., 2022). VSM measurements show that surface modification does not cause significant changes in magnetic properties. To simulate the magnetothermal effect of MeCFO, magnetic fluid was simulated by finite element method. The temperature distribution diagrams of magnetic fluid were obtained by COMSOL Multiphysics 5.6 software (Figures 3D–F). As shown in Figure 3D, we first preset our geometric model using the software. With the condition of magnetic field, the temperature of blank group could be calculated and analyzed to be concentrated on the coil surface. At the same time, the temperature did not rise significantly (Figure 3E). Different from the blank group, the temperature distribution of MeCFO group was diffused outward by MeCFO. Moreover, the temperature of magnetic fluid reached 40°C–45°C, which could help MeCFO nanoparticles kill tumor cells (Hervault et al., 2016). The temperature of magnetic nanoparticles can be controlled by changing the concentration, intensity, and time of the magnetic field (Darwish et al., 2019). This will be conducive to the promotion of MeCFO antitumor therapy. The ion concentrations released by CFO and MeCFO nanoparticles at 1 and 4 days are shown in Table 4. It could be found that the release of Co²⁺ from the MeCFO is significantly less than that of the CFO. The Fe²⁺ of the MeCFO is significantly less than that of the CFO. Because the released Si⁴⁺ is too little, no change can be detected by ICP-OES within 4 days when compared with the soaking solution with Si-containing impurity (not shown). In conclusion, MeCFO nanoparticles are expected to be promising in magnetic hyperthermia.

The synthetic route of MeCFO/GelMA hydrogel is shown in Figure 1B. CFO nanoparticles were incorporated into GelMA solution and formed by UV light photocrosslinking. FTIR spectrogram (Figure 4A) of the composite hydrogels showed the characteristic peaks of GelMA significantly observed at 1,635, 1,532, and 1,191 cm⁻¹ were respectively related to the −C=O (amide Ⅰ), N-H (amide Ⅱ), and C-O-C bonds (Rahimi Mamaghani et al., 2018; Bansal et al., 2022). An absorption band around 557 cm⁻¹ appeared in the spectrum of MeCFO/GelMA hydrogel was ascribed to the MeCFO nanoparticles (Supriya et al., 2019). The cross-section microstructure of MeCFO/GelMA hydrogel was studied by SEM. The microstructure of MeCFO/GelMA hydrogel with different proportions of MeCFO is indicated in Figure 4B. All MeCFO/GelMA hydrogels exhibited a porous microstructure due to the process of freeze drying. With the increase in MeCFO content, the pore size of MeCFO/GelMA hydrogel gradually decreased. This is because the presence of MeCFO fills the porous microstructure and the number of holes in the hydrogel (Chen et al., 2022b).

Osteosarcoma resection often results in bone defects of varying sizes. This means that MeCFO/GelMA hydrogels need some mechanical support. Hydrogels are flexible and can resist bending and squeezing to a certain extent. The images of the mechanical experiment were shown in Figure 5A. The mechanical properties of MeCFO/GelMA hydrogels were changed with the addition of MeCFO (Figure 5B). With the increase of MeCFO content, the mechanical properties of MeCFO/GelMA hydrogel were also gradually enhanced, which revealed that magnetic nanoparticles improved the mechanical properties of hydrogels. It is due to the presence of nanoparticles that can arrange the orientation of polymer chains, leading to the formation of more regular structures and electrostatic interactions between nanoparticles and polymer functional groups. This resulted in increased cross-linking density and thus improved mechanical properties (Farzaneh et al., 2021). Figure 5C showed that MeCFO/GelMA hydrogels exhibited tunable compressive modulus in the range of 2.7–6.5 kPa. For hydrogels containing MeCFO, this upward trend was clear and statistically significant. The maximum value was also recorded in the composite containing 0.1% MeCFO, which was significantly higher than that of the blank group. The stress of 0%, 0.01%, 0.05%, and 0.1% MeCFO/GelMA hydrogels under 50% strain is 3.85 ± 0.07, 4.58 ± 0.48, 5.14 ± 0.33 and 8.18 ± 1.59 kPa, respectively (Figure 5D). Therefore, 0.1% MeCFO/GelMA hydrogel demonstrated better resistance to fracture at 50% strain, which possessed considerable hydrogel stiffness.

As a temporary substrate for bone repair, the shear force of the hydrogel is also important in addition to the compression force, while new bone is formed. In Figure 6A, the viscosity of hydrogels varies with the shear rate. The relationship between shear stress and shear rate is nonlinear, which is manifested as shear thinning. The viscosity of MeCFO/GelMA gels at the low shear rate increased with the increase in the MeCFO ratio, which means that shear thinning performance was enhanced. This simulated the shear process of the hydrogel extruded from the printing cylinder (Meng et al., 2020). So MeCFO/GelMA hydrogel system can be extruded for 3D printing, which is beneficial for designing scaffolds for bone defects with different shapes. For the amplitude scanning (Figure 6B) and frequency scanning (Figure 6C), the energy storage modulus (G′) of hydrogels remained stable with the increase of shear strain or frequency, but their loss modulus (G″) increased when the shear strain was higher than 2% and the frequency was higher than 20 rad/s. With the increase of MeCFO content, the G′ and G″ was increased. It is because the nanoparticles act as reinforcers, improving the viscoelasticity of the hydrogel (Kuang et al., 2019). This result is consistent with the compression performance mentioned above. The results of time scanning measurements are shown in Figure 6D. The variation curve of the G′ of hydrogel with time is roughly stable, which indicates that hydrogel has sufficient mechanical properties and strong stability. With the increase of MeCFO content, the G′ of hydrogel was also enhanced. This stability over time is one of the desirable properties of extrudable materials (injectable materials, 3D printable materials, and so on) in filling defect sites (Arun Kumar et al., 2016). Rheological results show that MeCFO can significantly enhance G′ and G''. In particular, 0.1% MeCFO/GelMA hydrogel showed a wide linear viscoelastic region and good shear resistance, indicating satisfactory toughness.

One of the most important properties of hydrogel materials is their ability to absorb water. The swelling properties of scaffolds used in tissue engineering are essential for cell attachment. For this purpose, the swelling behavior of magnetic hydrogels with different content of MeCFO was studied, which also indirectly indicated porosity (Ganji et al., 2010). The swelling state of all samples can be divided into two stages. The first stage occurs within 2 h and increases rapidly, while the second stage is a high plateau after 24 h, as shown in Figure 7A. All GelMA hydrogels swelled fastest within 2 h, then it starts to stabilize. After 24 h, the equilibrium state of swelling is reached. The equilibrium swelling rate (Figure 7B ) of 0%, 0.01%, 0.05%, and 0.1% MeCFO/GelMA hydrogels are 751.7% ± 102.4%, 733.7% ± 8.2%, 680.9% ± 24.4%, and 679.6% ± 9.6%, respectively. In particular, GelMA hydrogel has the fastest swelling speed and the highest equilibrium swelling rate. As the number of nanoparticles in the hydrogel increased, the number of pores increased, but the size of the pores and their interconnections decreased (Bonhome-Espinosa et al., 2017). Therefore, the water absorption ability of the hydrogel decreased with the increase of nanoparticles. In addition to water absorption capacity and 3D-connected porous morphology, bone defect repair scaffolds should also have effective porosity, which should match the porosity range of natural bone (40%–65%) (Saravanan et al., 2017). The porosity of MeCFO/GelMA hydrogel was analyzed by the ethanol replacement method (Zhang et al., 2013). Figure 7C shows that the porosity of MeCFO/GelMA hydrogels is more than 60%. There was no statistically significant difference between the groups. It meets the porosity requirement of bone tissue.

The prepared MeCFO/GelMA hydrogel can be extruded into self-supporting uniform filaments by the virtue of the rheological properties (Figure 1C and Figure 8A). Figure 8A also showed the macro and micro morphology of 3D-printed MeCFO/GelMA hydrogel. The results showed that 3D printed hydrogels had a complete shape without distortion, deformation, and collapse. The pore size of the 3D-printed hydrogel was uniform. In addition, the hydrogel grid structure was visible and the diameter of the fibers looks alike. There is some distance between the layers, which conformed to the imported 3D scaffold model. This can provide a good three-dimensional culture environment for the cells (Yuan et al., 2018). Meanwhile, the performance of MeCFO/GelMA hydrogel can meet the requirements of 3D printing.

In this study, we first conducted cytotoxicity experiments with different concentrations of MeCFO extracts to screen out the gradient of better cytocompatibility. Figure 8B showed no cytotoxicity at MeCFO concentrations less than 50 μg/mL. The concentration was consistent with 0.1% MeCFO content in GelMA hydrogel. In recent years, a large number of cell experiments have been conducted to evaluate the cytotoxicity of CFO nanoparticles in vitro (Abudayyak et al., 2017; Ahmad and Zhou, 2017; Panda et al., 2022). It has been reported that CFO nanoparticles with surface modification have lower cytotoxicity (Mushtaq et al., 2017). The cytocompatibility of MeCFO hydrogel on BMSCs was studied by the CCK-8 method. Compared with 50 μg/mL MeCFO, CFO at the same concentration showed significant cytotoxicity (Figure 8C). This indicates that methacrylate groups can improve the cytocompatibility of CFO. Saos-2 cells were selected to verify the antitumor effect of MeCFO. The CCK-8 results showed that 50 μg/mL MeCFO significantly reduced the activity of tumor cells (Figure 8D). In addition to anti-tumor through magnetothermal effects, MeCFO also had direct cytotoxicity to tumor cells. This may be related to tumor cells uptaking more nanoparticles, which was conducive to killing tumor cells targetable (Prijic et al., 2010; Wang and Thanou, 2010). Hence, 50 μg/mL MeCFO nanoparticles have good cytocompatibility with normal cells but have obvious cytotoxicity to tumor cells. Moreover, Fe³⁺ and Co²⁺ have a certain osteogenic effect (Glenske et al., 2018). It has previously been shown that the differentiation of BMSCs affected bone formation (Ghosh et al., 2019). RT-PCR results in this study showed that the expression of the osteogenic differentiation gene of BMP-2 in the 50 μg/mL MeCFO group was significantly increased compared with the control group (Figure 8E). Although there was no statistical difference in VEGF gene expression, there was also an upward trend in the MeCFO group when compared with CFO and the control group. In addition, osteoblasts produce extracellular calcium-deposited minerals to form bone tissue. Alizarin red staining was used to detect the amount of mineralization caused by the formation of bone nodules. Alizarin red staining results showed that the MeCFO group had the deepest staining color among the three experimental groups (Figure 8F). This suggests that 50 μg/mL MeCFO is beneficial to bone regeneration through bone mineralization.

For MeCFO/GelMA hydrogels, the cytotoxicity assay at day 4 in Figure 8G showed that all hydrogels had good cytocompatibility when cultured with BMSCs and there was no significant statistical difference between groups. It has been reported in the literature that the cytotoxicity of CFO is displayed in a dose- and size-dependent manner (Ansari et al., 2018). The higher concentration and smaller the size of CFO, the more cytotoxic it is. In this study, the final concentration of MeCFO was less than 0.1% and the particle size was large in the range of 20–100 nm. So the hydrogel had good cytocompatibility, which is similar to the conclusion of related literature (Ansari et al., 2016). The staining results of living cells and dead cells were consistent with the results of the cytotoxicity assay. Live and dead cell experiments showed that the cells grew well within 4 days after incubation with MeCFO/GelMA hydrogels. Most of the cells showed living cells with green fluorescence, while dead cells with red fluorescence were almost absent (Figure 8H). And the cells could be seen on the hydrogel pseudopodia and spread out well. These results suggest that MeCFO/GelMA hydrogels function as effective matrix supporting cell adhesion and growth. The introduction of MeCFO nanoparticles in small quantities does not affect the cytocompatibility of hydrogels. Considering the osteogenic differentiation ability of low content of MeCFO, 0.1% MeCFO-incorporated GelMA hydrogels will be promising for the bone regeneration.