Type A porcine skin gelatin (Sigma, St. Louis, USA) was fully dissolved in phosphate buffer saline (PBS) solution at 50°C (10% w/v). The methacrylic anhydride (MAA, 8% v/v) was then added and stirred at 200 rpm for 24 h. Dulbecco’s phosphate-buffered saline (DPBS) was preheated to 50°C under the same volume with PBS. The preheated DPBS was then mixed with the above mixed solution at 50°C for 10 min. The final solution was dialyzed against distilled water at 40°C for 5 days, using a dialysis membrane with 12–14 kDa cutoff (Fisher Scientific, Waltham, USA), then lyophilized for 5 days to yield the purified gelatin methacrylate (GelMA).

The 3D-printed porous Ti2448 scaffolds were prepared by an Arcam A1 EBM system (Arcam, Gothenburg, Sweden), following the same method as previously reported (Xu et al., 2022). In brief, the scaffold models (9.50 mm × 9.50 mm × 2.50 mm) with a nominal 70% porosity were created. The Ti2448 powders (particle size ∼45–106 µm) were preheated to 773 K before melting. The samples were produced with a scan speed of 130 mm/s, a voltage of 60 kV, and a vacuum of 2 × 10⁻³ mbar.

The lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) photoinitiator (0.1% w/v) was added to the GelMA solution in PBS (5% w/v), and then filtered through 0.22 µm membrane filter. 100 μL GelMA solution was added to the Ti2448 scaffolds, which was exposed to the UV light (365 nm) both top and bottom sides for 2 min. The samples were then lyophilized for 24 h, denoted as Ti2448-GelMA. The DFO powder was added to the above 5% w/v GelMA solution, with the final DFO concentration of 0.5 wt%, and LAP was then added. Following the same procedure as Ti2448-GelMA preparation, Ti2448-GelMA scaffolds loaded with DFO are acquired and denoted as Ti2448-GelMA/DFO.

The chemical structure of gelatin and GelMA was characterized by ¹H nuclear magnetic resonance spectrometer (¹H-NMR) with a Bruker Avance Neo 400 spectrometer (Bruker, Bremen, Germany) using D₂O as the solvent.

Fourier transform infrared spectrometry (FTIR, Agilent Technologies, California, USA) was conducted in the range of 400–4,000 cm⁻¹. The crystal information was measured by X-ray diffraction (Rigaku, Tokyo, Japan) using the radiation of Cu Kα radiation (λ = 1.5418 Å). The surface morphology and elemental analysis was investigated by a scanning electron microscope (SEM, Carl Zeiss AG, Jena, Germany) equipped with energy dispersive spectroscopy (EDS, Oxford Instruments, Abingdon, UK). The mean pore size of GelMA network was evaluated by the ImageJ software based on SEM images.

MC3T3-E1 cells were purchased from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China), and incubated in the α-MEM medium (Hyclone, GE Healthcare, Chicago, USA) supplemented with 10% FBS (Lonsera, Uruguay) and 1% antibiotic/antimycotic (Gen-View Scientific Inc., Calimesa, USA). MC3T3-E1 cells (3 × 10⁴ cells/scaffolds) were seeded on both Ti2448 and Ti2448-GelMA scaffolds in a 48-well plate for 1, 3, and 5 days. The cytotoxicity was evaluated by the CCK-8 reagent (Gen-View Scientific Inc., Calimesa, USA). Briefly, the 100 μL mixture (10 μL CCK-8 and 90 μL culture medium) was added to each well for 3 h and the OD value was measured with a microplate reader (Multiskan Go, Thermo Fisher Scientific, Waltham, USA) under 450 nm.

MC3T3-E1 cells (3 × 10⁴ cells/scaffolds) were seeded on Ti2448, Ti2448-GelMA and Ti2448-GelMA/DFO scaffolds. After culturing for 24 h, cells on the scaffold were gently rinsed by PBS, and immobilized in 4% paraformaldehyde for 2 h, following by dehydration in a graded series of ethanol solutions (50%, 70%, 90%, 95% and 100%) for 10 min each and dried with hexamethyldisilane for 20 min. The morphology was observed by SEM.

Alkaline phosphatase (ALP) activities of MC3T3-E1 cells cultured on both Ti2448 and Ti2448-GelMA scaffolds were evaluated. MC3T3-E1 cells were seeded on the scaffolds (3 × 10⁴ cells/scaffold) for 7 and 14 days, which were lysed in 100 μL of Triton X-100 (1% v/v) for 5 min. The ALP activity was determined by ALP quantification kit (Beyotime, Shanghai, China) and the OD value was measured at 520 nm.

Alizarin Red S (ARS) staining was used to evaluate the biomineralization level. After cultured for 7 and 14 days, cells were fixed with 4% paraformaldehyde for 15 min. The fixed samples were stained with ARS solution (Beijing Solarbio Science and Technology Co., Ltd., Beijing, China) for 30 min and cleaned with PBS. Finally, the cells are observed utilizing a confocal microscope (Olympus Corporation, Tokyo, Japan). For quantification analysis, 10% hexadecyl pyridinium chloride monohydrate (Sigma-Aldrich, St. Louis, USA) was used to dissolve the mineralized nodules and the OD value was measured at 562 nm.

In vitro angiogenesis of Ti2448-GelMA and Ti2448-GelMA/DFO scaffolds was investigated by the tube formation assay. The pipette tips, 96-well plates and Matrigel matrix (ABW, Shanghai, China) were first placed at 4°C overnight. 50 μL Matrigel matrix were then added to the plates for 30 min Ti2448-GelMA and Ti2448-GelMA/DFO scaffolds were first immersed into 2 mL serum-free α-MEM at 100 rpm for 24 h. The serum-free α-MEM was incorporated into the extracts to generate a range of diluted extracts (1/2, 1/4 and 1/8 concentrations). The serially diluted extraction medium was prepared by supplemention with 10% FBS and 1% antibiotic/antimycotic, hereafter referred to as 1, 1/2, and 1/4 dilution group. Cells (3 × 10⁴ cells/scaffold) with different diluted extraction were seeded on the Matrigel coated plate for 3, 6 and 9 h. The morphology of HUVEC cells were observed by an Eclipse Ti inverted fluorescent microscope (Nikon Instruments Inc., Tokyo, Japan). Quantitative analysis for the number of vascular meshes were calculated by ImageJ software.

The release of DFO from the Ti2448-GelMA/DFO scaffolds was evaluated by a mould incubator (Shanghai Longyue Instrument Equipment Co., Ltd, Shanghai, China). The Ti2448-GelMA/DFO scaffolds were first incubated in the centrifuge tube at 37°C under 100% humidity. After 1, 3, 7 and 14 days, the scaffolds were removed and rinsed with deionized water. Then more deionized water was collected and supplied in the centrifuge tube until the final volume is 2 mL. The obtained solution was then chelated with FeCl₃ (10 mg/mL) at a 9:1 volumetric ratio for 5 min. The OD value was measured at 485 nm. The amount of DFO released was calculated according to the standard DFO calibration curve. The drug release rate is calculated using the amount of released drug divided by the total drug content.

All data were expressed as means ± standard deviation. One-way analysis of variance (ANOVA) was conducted for multiple samples with varying time points. Statistical significance was determined by p < 0.05.

GelMA was first fabricated by the reaction between the MAA and the functional groups of lysine and hydroxyl lysine in gelatin (Supplementary Figure S1A). GelMA was further photo crosslinked to form three-dimensional network structures. The ¹H spectra of gelatin and GelMA were conducted to confirm the synthesis of GelMA (Supplementary Figure S1B). Two new peaks were found at 5.45 ppm and 5.68 ppm, corresponding to vinyl from the methacrylate group. However, both peaks were not present in the gelatin sample. The peak at 2.81 ppm represented the methylene proton (2H) of lysine, which showed high intensity peak of the unmodified gelatin and low intensity peak of the GelMA, consistent with previous findings (Cao et al., 2021; He et al., 2023). Supplementary Figure S2 showed that GelMA as prepared was amorphous. FTIR results (Supplementary Figure S1C) revealed the characteristic peaks of gelatin and GelMA. The peak at 1,543 cm^-1 from the amide bond was related to N-H bending. Furthermore, the peak at 1,641 cm^-1 for gelatin was associated with the C=O stretching of amide bond, while that for GelMA associated with both C=O and C=C stretching exhibited higher intensity, indicating the presence of methacrylate groups in the GelMA sample. The presence of these characteristic peaks again confirmed the successful synthesis of GelMA.

Previous study reported that Ti2448 have good biocompatibility and osteoconductivity (Nune et al., 2017b; Liu et al., 2019). Ti2448 scaffolds showed the porous structure with the pore size of ∼600 μm (Figure 1A). GelMA, mimicking the extracellular matrix, offered a three-dimensional structure and support the osteogenic differentiation of bone marrow mesenchymal stem cells, leading to further bone regeneration (Piao et al., 2021). For Ti2448-GelMA scaffolds, GelMA was incorporated into the porous interconnected structure of Ti2448 scaffolds (Figure 1B) and formed a porous network with the mean pore size of 120.2 ± 25.1 μm (Figures 1C,D). The interconnectivity and proper pore size (>100 μm) were crucially important for the transportation of oxygen/nutrients, cell attachment as well as the vascular growth, which were beneficial for the bone regeneration (Cimenoglu et al., 2011; Wu et al., 2014; Safaei-Yaraziz et al., 2021; Uiiah et al., 2021; Yao et al., 2021). Taking the advantages of 3D printing technology, the porous Ti2448-GelMA scaffolds provides the good interconnectivity with proper pore size (120.2 ± 25.1 μm), which may enhance the vessel growth and further facilitate the bone regeneration.

EDS mapping of Ti, Nb, Zr, Sn, C, and O elements in the Ti2448-GelMA scaffold was conducted (Figure 2). The presence of C and O elements further proved the incorporation of GelMA into the Ti2448 scaffolds. Overall, GelMA was successfully loaded into the Ti2448 scaffold.

The morphology of MC3T3-E1 cells cultured on Ti2448-GelMA scaffolds was evaluated after 1 day culture (Figures 3A, B). Cells spread well with the pseudopodia extending outwards on both Ti2448 and GelMA surfaces. Notably, the cells exhibited unhindered adhesion and traversal across the Ti2448 scaffold and GelMA surfaces simultaneously (Figure 3B). Gelatin retains the key amino acid sequence, such as arginine glycine aspartate (RGD), which is critical to cell adhesion (Ahmady and Abu Samah, 2021). The proliferation of MC3T3-E1 cells was also evaluated (Figure 3C). As the cultivation time increased, the number of cells increased on both Ti2448 and Ti2448-GelMA scaffolds, without significant difference. For the Ti2448 scaffold group, the cell number at day 3 and day 5 was about 1.32 and 2.50 folds of that at day 1, respectively. For the Ti2448-GelMA scaffold group, the cell number at day 3 and day 5 was about 1.50 and 2.75 folds of that at day 1, respectively. There was no significant difference on cell numbers between these two groups at the same time point. Therefore, in our study, Ti2448-GelMA scaffold can support the cell adhesion and proliferation, demonstrating the excellent biocompatibility.

To evaluate the osteogenic differentiation of MC3T3-E1 cells on Ti2448-GelMA scaffolds, cells were cultured for 7 and 14 days and subsequently performed ARS staining to evaluate biomineralization levels (Figures 4A–H). The results showed a relatively low degree of biomineralization for MC3T3-E1 cells cultured on the Ti2448 scaffold for 7 days, with scarcely noticeable red-stained mineralized nodules. On the other hand, the amount of mineralized nodules in the Ti2448-GelMA group was higher than the Ti2448 group after culture for 7 days. Meanwhile, the ARS staining showed progressive biomineralization with increase of culture time. After cell culture for 14 days, the Ti2448-GelMA group showed a higher amount of mineralized nodules than the Ti2448 group. Additionally, quantitative analysis showed that the amount of calcium contents from the Ti2448-GelMA group were 1.8 times and 2.5 times of that from the Ti2448 group after cell culture for 7 and 14 days, respectively (Figure 4I). Compared with the Ti2448 group, MC3T3-E1 cells in the Ti2448-GelMA group also showed significantly higher expressions of ALP activity when cultured for 7 and 14 days (Figure 4J). These results suggest that Ti2448-GelMA scaffolds can facilitate osteogenic differentiation of MC3T3-E1 cells, thereby promoting further biomineralization. GelMA recently attracts more attention, which can be used to mimic the natural bone extracellular matrix and is beneficial to vascular growth and bone regeneration (Dong et al., 2019; Heltmann-Meyer et al., 2021; Li Y et al., 2021). For example, the GelMA scaffolds were prepared by the thermally induced phase separation technique, and the ALP activity of the adipose derived stem cells cultured on the GelMA group was approximate 2 times of the untreated group after 14 and 21 days (Fang et al., 2016). In our previous work, compared with the Ti2448 scaffolds, MC3T3-E1 cells derived ECM modified Ti2448 scaffold can effectively promote osteogenic differentiation of MC3T3-E1 cells, and enhance the bone integration after implantation in rabbit femoral bone defects for 1 month (Xu et al., 2022). Therefore, the Ti2448-GelMA scaffolds were prepared by introducing the three-dimensional GelMA as the bone extracellular matrix and can be beneficial for the osteogenic differentiation of MC3T3-E1 cells.

Vascularization plays an important role on the repair of large bone defects (Li et al., 2023). During bone healing and remodeling, vascular formation and bone tissue regeneration are indeed coupled processes (Ramasamy et al., 2014). Porous titanium alloys demonstrate the good osteoconductivity and osteointegration capabilities, yet the pursuit of effective internal vascularization remains a big challenge. Herein, DFO was incorporated into GelMA to enhance the Ti2448-GelMA vascularization (Figure 5A). SEM morphology revealed that doping of DFO did not change pore structures of Ti2448-GelMA/DFO scaffolds (Figures 5D, E) compared to the Ti2448-GelMA (Figures 5B, C) scaffolds.

The angiogenic effects of Ti2448-GelMA/DFO scaffolds were evaluated by co-culture with HUVEC cells (Figure 6; Supplementary Figure S3). After cell culture for 3 h, no tubes formed on both the Ti2448-GelMA group and control group, while a small amount of tubes formed in the Ti2448-GelMA/DFO group (Supplementary Figure S3A). With increase of the cultivation time, the number and size of vascular tubes increased for all groups. After cultivation for 6 h, large tubes (∼300 μm) were observed in the Ti2448-GelMA/DFO group, compared to the Ti2448-GelMA group and control group (Supplementary Figure S3B). After culture for 9 h, a greater number of small vascular tubes consolidated into large tubes, and there were much more uniform large tubes in the Ti2448-GelMA/DFO group than the other two groups (Figure 6A). The number of vascular meshes represented the closed areas delimited by the vascular tubes marked with the yellow dash line in Figure 6A ₁. The number of vascular meshes were calculated after culture for 6 and 9 h (Figures 6B, C). The results indicated that the number of meshes for all groups increased with increase of culturing time. Additionally, the meshes increased with the increase of DFO concentration, especially for the 1/2 dilution group (7.2/mm²) with the higher number of vascular meshes than the other two groups (Control group: 5.3/mm², Ti2448-GelMA group: 6.2/mm²). Therefore, HUVECs cultured with Ti2448-GelMA/DFO extractions exhibited more vascular network compared to the Ti2448-GelMA and control groups. As the concentration of DFO increases, the vascularization became more significant, demonstrating the improved vascularization effect in a dose-dependent manner. Hou et al. cultured HUVEC cells with a series of DFO concentrations (5, 20, 100 μM) for 24 h. Results showed that the group with the highest concentration of DFO (100 μM) exhibited the largest tube length and the greatest number of tubes, also demonstrating that DFO improved the tube formation of HUVECs in a concentration-dependent manner (Hou et al., 2013). Therefore, the release of DFO can effectively promote the vascularization of the Ti2448-GelMA/DFO scaffolds. Besides, the MC3T3-E1 cells attached well on the Ti2448-GelMA/DFO scaffolds (Supplementary Figure S4), which demonstrated that DFO loading on theTi2448-GelMA scaffold does not affect the cell behavior.

The DFO release profile was shown in Figure 7. At the first day, DFO exhibited the release rate of nearly 14.3%, followed by a constant sustained release. After 7 days, about 42.0% DFO has been released from the Ti2448-GelMA/DFO scaffold and the final cumulative release reaches to 82.3% after 28 days. As such, Ti2448-GelMA/DFO scaffolds exhibited a sustained release of DFO possibly through the hydrogen bonding. In fact, the regeneration of blood vessels runs through the entire process of bone repair from early bone formation, maturation to remodeling (Chen et al., 2020; Simunovic and Finkenzeller, 2021). The new blood vessels infiltrated into the newly formed bone are necessitate to supply of oxygen and nutrients, as well as removal of metabolic wastes (Marenzana and Arnett, 2013). Previous research also studied the DFO release profiles in vitro (Cheng et al., 2018; Ran et al., 2018). For example, DFO loaded titania nanotubes substrates were prepared and then immersed into PBS solution with gently shaking at 37°C. The burst release of DFO (∼83.7%) was found in the first 12 h (Ran et al., 2018). However, immersion of the sample in a liquid solution cannot truly simulate the real release behavior of DFO from the implants in vivo under a mild and humid environment. Once the scaffolds are implanted in the bone defects, which will be encapsuled by the soft callus under a high humidity environment instead of immersion in a liquid solution (Salhotra et al., 2020). Thus, it is necessary to build a model mimicking the real release profile in vivo. Herein, Ti2448-GelMA/DFO scaffold were placed under 100% humidity at 37°C, which can well simulate the environment after implantation. The DFO release profiles in Ti2448-GelMA/DFO scaffolds can effectively achieve the sustained drug release, leading to the better vascularization for the large bone defect repair.

Composite materials have been widely studied in order to improve osteogenic activity of the scaffolds and further vascular growth for large bone defect repair (Li J et al., 2021; Qian et al., 2021; Qian et al., 2024). Herein, the Ti2448 scaffold was chosen as the main part, while GelMA was implemented as the osteogenic component, with DFO loaded as the vascularization factor. This combination will provide good mechanical properties and essential biological functions to the implant materials for large bone defect repair.