Injectable Hydrogel Systems with Multiple Biophysical and Biochemical Cues for Bone Regeneration

Abstract

Bone regeneration is a complex process in which angiogenesis and osteogenesis are crucial. Introducing multiple angiogenic and osteogenic cues simultaneously in a single system and tuning these cues to optimize the niche remains a challenge for bone tissue engineering. Here, based on our injectable biomimetic hydrogels composed of silk nanofibers (SNF) and hydroxyapatite nanoparticles (HA), deferoxamine (DFO) and bone morphogenetic protein-2 (BMP-2) were loaded on SNF and HA to introduce more angiogenic and osteogenic cues. The angiogenesis and osteogenesis capacity of the injectable hydrogels could be regulated by tuning the delivery of DFO and BMP-2 independently, resulting in vascularization and bone regeneration in cranial defects. The angiogenesis and osteogenesis outcomes accelerated the regeneration of vascularized bones toward similar composition and structure with natural bones. Therefore, the multiple biophysical and chemical cues provided by the nanofibrous structures, organic-inorganic compositions, and chemical and biochemical angiogenic and osteogenic inducing cues suggests the potential for clinical applicability of these hydrogels in bone tissue engineering.

Table of Contents Entry

Tunable multiple angiogenic and osteogenic cues were introduced to hydrogel systems simultaneous to optimize the niches for bone regeneration.

1. Introduction

Large bone defects caused by trauma, tumor, osteoporosis, arthritis and infection occur with a high incidence worldwide.^1,2 Autogenic bone grafts are considered the gold standard to achieve functional recovery in such large bone defects, but are limited due to the lack of donors or donor site morbidity.^3,4 As an emerging option, bone tissue engineering is becoming a promising alternative to treat bone defects.^5–7 Although many natural and synthetic biomaterial grafts have been developed to accelerate bone regeneration,^8–10 the regenerated bones are usually inferior to natural bones in terms of structure and mechanics. Natural bone regeneration is a complex and dynamic process in which tunable multiple biochemical and biophysical factors regulate cell behaviors and bone remodeling through different stages of repair and regeneration.^11,12 Simulating the multiple chemical and physical cues simultaneously is a useful strategy to optimize the functional restoration of bones, but remains as a significant challenge.

Different biomaterials have been fabricated to simulate specific physical niches and functions of native bones, such as organic-inorganic compositions, nanostructures and mechanical cues.^13–15 Growth factors and active macromolecules have also been loaded into bone grafts to endow biochemical regulation at defect sites.^16–18 Recently, several chemical/physical collaboration strategies were developed to design bone biomaterials with both vascularization and osteogenic capacity.^19–21 Although these studies suggested the possibility of optimizing bone regeneration through the smart simulation of the microenvironments around bone tissues, a significant gap exists between these types of bone biomaterials and the more accepted autogenic grafts. The booming demand for better bone grafts are stimulating innovation of bioactive grafts with multiple better chemical and physical cues, including the need for the presentation of multiple regulators from the same bone biomaterial system.

Silk fibroin (SF) has been widely used in bone tissue engineering due to its biocompatibility, controlled biodegradability and ease of manufacturing.^22–24 Tough SF biomaterials were developed through tuning the beta-sheet (crystalline) content and protein concentration to provide mechanical cues and osteoinduction.^25–27 Water dispersible HA nanoparticles assembled with SF as a template were also blended with SF solutions to prepare bone scaffolds with organic-inorganic compositions mimicking natural bones.²⁸ Different bioactive molecules such as BMP-2 were loaded into SF biomaterials and released slowly to accelerate bone regeneration.²² Therefore, SF biomaterials are suitable platforms to involve different physical and chemical cues for the optimization of bone regeneration. Multiple physical and chemical regulators have also been introduced into SF-based scaffolds and resulted in improved osteoinduction and bone repair.^29,30 Recently, beta-sheet rich silk fibroin nanofibers (SNFs) were reported and assembled.^31,32 As a versatile building unit, these nanofibers could be used to fabricate biomimetic bone scaffolds and hydrogels, and also to load macromolecules and small molecules, to provide enhanced outcomes when compared to traditional SF materials.^22,33 Injectable SNF bone materials were fabricated since the hydrophobic-hydrophilic properties of the SNFs endowed the nanofibers with shear-thinning performance.^28,33 Unlike traditional silk solutions, different chemical and physical cues could be introduced and optimized on the nanofiber units, respectively, and then assembled to form scaffolds/hydrogels. It is feasible to fabricate SNF-based bone materials with multiple optimized physical and chemical cues, overcoming the unwanted interactions of the multiple cues in SF bone materials as reported previously.^34–36

Here, as a proof of concept, SNFs and HA nanoparticles were blended to form injectable hydrogels with an optimal ratio of SNFs and HA, providing improved organic/inorganic compositions. The small molecule, desferrioxamine (DFO), was loaded onto the SNFs t to induce vascularization based on our recent findings.³³ Compared to growth factors and other small molecules, DFO are more stable molecular and could induce the secretion of multiple angiogenesis-related factors and cytokines simultaneously, to facilitate more stable vascularization in bone regeneration.³³ Bone morphogenetic protein-2 (BMP-2) was also loaded onto the SNFs and HA nanoparticles to optimize osteoinduction. The DFO and BMP-2 loaded SNFs and HA particles were assembled to prepare the injectable hydrogels with multiple osteoinduction cues including organic-inorganic compositions, and angiogenic/osteoinductive factors. Both in vitro and in vivo results revealed that the hydrogels created suitable niches to accelerate the regeneration of vascularized bone when compared with other silk-based hydrogels recently reports.^22,28

2. Materials and methods

2.1. Preparation of silk nanofiber hydrogels

Beta-sheet rich silk nanofiber solutions were prepared as described previously.^22,28,37 Briefly, raw silk was boiled in 0.02 M Na₂CO₃ for 20 min and washed with distilled water to degum sericin proteins. The degummed silk was dissolved in 9.3M LiBr solution at 60°C for 4 h and dialyzed against distilled water for above 72 h. After centrifugation at 9,000 rpm for 20 min at 4°C twice, fresh aqueous solution with concentration of about 6 wt% was obtained and concentrated to above 20% slowly over 24 h at 60°C. The concentrated solution was diluted to 2 wt% with distilled water and cultured at 60°C until hydrogel formation. Following the gelation process, silk fibroin transformed into nanofibers and these were then stored at 4°C until use.

2.2. Preparation of SF-coated HA nanoparticles

Water dispersible HA nanoparticles were prepared with silk fibroin as template and stabilizer. Fresh silk fibroin solution (6 wt%) was incubated at 60°C for 24 h to form silk nanoparticles. The nanoparticle solution (20 mL) was blended with 20 mL H₃PO₄ (0.06 M), following by the addition of Ca(OH)₂ solution (100 mL, 0.02M) at speed of 90mL h⁻¹ at 70°C. The solution became milky, indicating the HA nanoparticle formation. The nanoparticles were collected after the centrifugation at 9,000 rpm for 20 min.

2.3. Preparation of BMP-2 and DFO loaded composite hydrogels

According to our previous studies, SNF hydrogels with concentrations of 2 wt% were used. The ratio of SNF and HA was fixed to 4:6 wt/wt, to provide an organic-inorganic composition similar to natural bone. According to our recent study,³³ DFO (Sigma-Aldrich, St Louis, USA) was blended with SNF hydrogels directly and tuned the concentration to 120 μmol/L, which could induce good vascularization. Simply, to prepare the composite hydrogels, DFO was added into 2 wt% silk nanofiber hydrogel with magnetic stirring for 2h, forming DFO-loaded hydrogels with concentration of 120 μmol/L and termed SNF-D. Then, 300 ng/ml BMP-2 (Ruibang Company, Shanghai, China) was added into SNF-D hydrogel directly and stirred for 2h to further load BMP-2 on the nanofibers. The hydrogels were termed SNF-D-B. HA particles were blended with SNF, SNF-D and SNF-D-B hydrogels and termed SNF/HA, SNF-D/HA, and SNF-D-B/HA, respectively. Different BMP-2 loading processes were used to tune the release behavior of BMP-2. BMP-2 was loaded on HA particles according to the method reported and then blended with the SNF-D hydrogel.²² The hydrogel that formed was termed SNF-D/HA-B. Half the amount of BMP-2 (150 ng/mL) was also loaded on SNF-D and HA particles, respectively, and then blended to form composite hydrogels. The hydrogels were termed SNF-D-B/HA-B. All the formed composite hydrogels were listed in Table 1.

Table 1.

Abbreviation of the composite hydrogels and their compositions

SNF/HA	HA nanoparticles mixed with silk nanofiber
SNF-D/HA	DFO-loaded on silk nanofiber and then mixed with HA nanoparticles
SNF-D-B/HA	DFO-loaded on silk nanofiber, further load BMP-2 on the nanofibers, and then mixed with HA nanoparticles
SNF-D-B/HA-B	DFO-loaded on silk nanofiber termed SNF-DFO, half amount of BMP-2 loaded on SNF-DFO and HA particles, respectively, and then blended to form composite hydrogel
SNF-D/HA-B	DFO-loaded on silk nanofiber termed SNF-DFO, BMP-2 loaded on HA particles termed HA-BMP-2, and then mixed

2.4. Characterization

The microstructure of the composite hydrogels was observed with scanning electron microscopy (SEM, S-4800, Hitachi, Tokyo, Japan) at 3.0 KV. Before observation, the samples were sputter-coated with gold. The crystal structure of SNF and HA was analyzed with X-ray diffraction (XRD, Nano ZS90, Malvern instruments, Malvern, U.K.). The mechanical properties were measured by rheometry (AR2000, TA Instruments, New Castle, DE) under a 20 mm cone plate (Ti, 20/1°). One milliliter of hydrogel was dispensed on the bottom plate. All samples were stabilized for 20 min before the measurement. Frequency sweeps were collected continuously over a wide frequency range from 100 to 1 rad s⁻¹ at 25 °C.

2.5. The release of BMP-2 and DFO from the hydrogels

According to our previous method,^22,33 different hydrogels (2 mL) were placed into a dialysis tube (50,000-MWCO) and then immersed in phosphate-buffered saline solution (PBS, 10 mL). The samples were incubated in an oscillating bath at 37°C for 42 days. BMP-2 release was monitored by removing and replacing 1 mL of the PBS solution at the indicated time points and measured by a BMP-2 ELISA kit (Biovision, San Francisco, CA). Similar to the BMP-2 release measurement process, the DFO-containing hydrogels were transferred to a dialysis tube and incubated in PBS (10 mL) at 37°C in an oscillating water bath for 40 days. Then 1 mL of solution was collected and refilled with fresh PBS at the indicated time points. The released DFO was determined by combing ferric chloride with the collected PBS and detected using multiscan spectra at 485 nm (Biotek, USA).^17,38

2.6. In vitro cytocompatibility of the composite hydrogels

All the composite hydrogels were sterilized with ⁶⁰Co γ-irradiation at the dose of 25kGY and used to culture bone mesenchymal stem cells (BMSCs). BMSCs were extracted from male Sprague-Dawley (SD) rats (40g). All animal procedures were performed in accordance with the guidelines for care and use of laboratory animals of Soochow University and approved by the animal ethics committee of Soochow University. After passaging to the third generation, flow cytometric analysis was performed to identify the stemness of the cells. BMSCs were labeled with fluorescein isothiocyanate (FITC)- conjugated antibodies against CD90, CD44, CD45 and CD34 (Cyagen Biosciences, Guangzhou, China). FITC-conjugated mouse IgG1 (Cyagen Biosciences, Guangzhou, China) was used as negative control. Expression distribution was assessed by single-channel flow cytometry (BD Biosciences, USA) (Fig. S1). BMSCs were cultured on the surface of different hydrogels at a cell density of 5×10⁴ in plates, and supplied with basal culture medium, dulbecco’s modified Eagle medium (DMEM, low glucose) with 10% fetal bovine serum and penicillin-streptomycin (100 U/mL, Gibco, Grand Island, USA) at 37°C and 5% CO₂. At day 1, 3 and 7, BMSCs were washed with PBS (PH 7.4) three times and fixed with 4% paraformaldehyde solution (Sigma-Aldrich, St. Louis, MO, USA) for 15 min at room temperature. After permeabilization with 0.1% (v/v) Triton X-100 (Sigma, St. Louis, USA) for 10 min, the cell actin and nuclei were stained with FITC-phalloidin (Thermo Fisher, Waltham, MA, USA) and DAPI. Fluorescence images of BMSCs were collected by a confocal microscope (Olympus FV10 inverted microscope, Nagano, Japan). Cell proliferation was evaluated with DNA content assay when the cells cultured for 1, 3, and 7d. All samples were digested with proteinase K overnight at 56°C.³² Then PicoGreen DNA assay (Invitrogen, Carlsbad, CA, USA) was used to analysis the DNA content according to the method published.²²

2.7. Tube formation assay

Tube formation assay in vitro was used to analyze the influence of the released DFO on angiogenesis. Matrigel (100μL/well, BD Biosciences, Beford, MA, USA) was covered on the gels.³³ Human umbilical vein endothelial cells (HUVECs, from the Cell Bank of the Chinese Academy of Sciences, Shanghai, China) were cultured at a density of 6×10⁴ cells/each well. After culturing for 3 and 6 h, the cells were fixed with 4% paraformaldehyde solution for 15min at room temperature, and stained with FITC-phalloidin (cell actin) and DAPI (nuclei), respectively. The fluorescence images of HUVE tubes were observed with a laser scanning confocal microscope (LSCM, Olympus FV10 inverted microscope, Nagano, Japan) with excitation/emission at 358/462 nm and 494/518 nm. The tube length was calculated from three random fields via ImageJ software.

2.8. Cell osteodifferentiation in vitro

Bone related gene expression and immune-fluorescence staining were performed to analyze the osteogenic differentiation of BMSCs. The cells were seeded on the hydrogel surface in 24-well plates with normal medium for 24 h, and then transferred to osteogenic differentiation medium (low glucose-DMEM, 10% FBS, 100 U/mL penicillin-streptomycin, 10 nM dexamethasone, 10 mM sodium-β-glycerophosphate, and 0.05 mM ascorbic acid-2-phosphate) for 7, 14, 21 days, respectively. Two early markers of osteogenic maturation (alkaline phosphatase (ALP) and runt-related transcription factor 2 (Runx2)) and two later markers of osteogenic differentiation and mineralization (osteocalcin (OCN) and osteopontin (OPN)) were analyzed with quantitative real time polymerase chain reaction (PCR).

The immunofluorescence expression of ALP, OCN and OPN were also visualized with LSCM. After culture for 7, 14, 21 days, ALP, OCN and OPN were labeled with the following primary antibody: ALP (BOSTER, BM4284), OCN (Abcam, ab13420), OPN (Abcam, ab8448). After incubation with primary antibody overnight, the cells were washed and conjugated with secondary antibodies including goat anti-mouse IgG (Servicebio, GB21303) and goat anti-rabbit IgG (Servicebio, GB21301) conjugated with cyanine 3 (Cy3). Nuclei were stained with DAPI. LSCM images were analyzed to evaluate the osteodifferentiation of the cells.

2.9. In vivo study

A rat calvarium defect model was applied to evaluate the osteoinduction of the composite hydrogels in vivo.^28,39,40 Ninety 8-week-old male SD rats (Animal Resource Center, Soochow University) were prepared and divided into 6 groups: Blank, SNF/HA, SNF-D/HA, SNF-D-B/HA, SNF-D-B/HA-B, SNF-D/HA-B. All animal procedures were performed in accordance with the guidelines for care and use of laboratory animals of Soochow University and approved by the animal ethics committee of Soochow University. The operation was performed according to our previous procedures.²⁸ Critical-sized full-thickness calvarium bone defects (5 mm diameter) were created at two sides of parietal bone. Each defect was filled with the composite hydrogels. After surgery, the rats were kept in pathogen free (SPF) laboratory animal room and raised for 3 months. At 1, 2, 4, 8 and 12 weeks, the rats were sacrificed using chloral hydrate to obtain specimens.

The collected samples were fixed with 10% buffered neutralized formalin for 24 h. Micro-computed tomography (μCT, SkyScan 1176, SkyScan, Aartselaar, Belgium) was used to measure the healing process of the bone defects. The scanning parameters were as follows: voltage 65 kV, current 385 μA and resolution 18 μm. The images were analyzed and reconstructed with the system software.

At 1, 2, 4, 8 and 12 weeks, the specimens were collected and treated for histological analysis. The specimens were decalcified with 10% EDTA at room temperature according to previous procedures.^41,42 Then, the decalcified specimens were dehydrated with gradient alcohol solutions (75%, 85%, 90%, 95%, 100%) and embedded in paraffin. After sectioned, the specimen sections were stained with hematoxylin and eosin (HE) and Masson’s trichrome, immunohistochemistry targeting CD31 (Abcam, ab64543), ALP (BOSTER, BM4284), OCN (Abcam, ab13420), OPN (Abcam, ab8448), respectively to reveal the tissue ingrowth, vascularization and new bone formation at defect sites.^22,28

2.10. Statistical analysis

All quantitative results were presented as the mean ± standard deviations and evaluated by one-way or two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison test. GraphPad Prism Software (GraphPad Software Inc.) were also used to compare the significant difference between the groups. P<0.05 was considered statistically significant.

3. Results and discussion

3.1. Design and Characterization of injectable hydrogels with multiple cues

Unlike previous silk fibroin solutions, SNFs can be utilized to load both small molecules and macromolecules effectively and achieve sustained release.^22,33 DFO, a small molecule drug, was loaded into SNF hydrogels and used to induce angiogenesis in wound healing.^43,44 Similar to our recent study, here DFO was loaded on the SNF while also avoiding aggregation of the nanofibers (Fig. S2). The amount of DFO loaded was 120 μmol/L, selected as this level induced angiogenesis in vivo in our prior work.³³ BMP-2 was also loaded on the SNFs and HA nanoparticles, respectively, achieving different sustained release behavior. The SNFs, HA, DFO-loaded SNFs, BMP-2-loaded SNFs and BMP-2-loaded HA were then used as building units (Fig. S2) to assemble injectable hydrogels with the multiple cues (Scheme 1). According to our previous study,²⁸ the concentration of SNFs remained at 2 wt% while the ratio of SNFs and HA was kept at 40:60 wt/wt, suitable organic-inorganic compositions for osteoinduction. SEM images indicated that the SNFs retained nanofibrous structures in the hydrogels after the assembly process. HA nanoparticles were dispersed homogeneously in the hydrogels without aggregation at micron scale (Fig. 1a). All of the drug-loaded hydrogels had similar nano-micro hierarchical structures as found with the drug-free control, suggesting that the introduction of DFO and BMP-2 had no negative influence on hydrogel morphology (Fig. 1a). Further, even without the use of chemical modifications or harsh physical process, both the SNFs and HA nanoparticles remained unchanged in term of their crystal structures after hydrogel formation (Fig. 1b). All of the hydrogels, with and without DFO and BMP-2, showed similar mechanical properties, with moduli around 21 kPa, sufficient to facilitate osteodifferentiation of stem cells (Fig. 1c).^45,46 The modulus of the hydrogels remained constant in the temperature range of 22-40°C, implying stability at body temperature (Fig. 1d). Similar to reported SNF hydrogels,³³ all of the drug-loaded hydrogels exhibited typical shear-thinning property (Fig. 1e). The hydrogels could be injected from syringes with needles (27G) by hand and remained in the solid hydrogel state after the removal of the applied force (Fig. 1f). Therefore, besides the physical cues (composition and microstructure), the DFO and BMP-2 loaded hydrogels were also injectable (Fig. 1g).

Scheme 1.

Schematic of injectable hydrogels with multiple cues and its functional repair of bone regeneration in bone defect model.

Fig.1.

Characterization of the composite hydrogel: (a) SEM images of different hydrogels. The images show similar microporous structures in SNF/HA, SNF-D/HA, SNF-D-B/HA, SNF-D-B/HA-B and SNF-D/HA-B hydrogels, respectively and homogeneous distribution of HA nanoparticles inside the SNF matrix. Scale bars: 200 μm (first line) and 1 μm (second line). (b) XRD curves of the SNF, HA and different composite hydrogels. (c-e) Rheological characterization of composite hydrogels: (c) frequency sweep, (d) modulus (G*) changes following the increase of temperature, and (e) viscosity property of different hydrogels. (f) Inversion test of the different hydrogels at 0 h and 24 h. (g) Injectability of the hydrogels. All the hydrogels showed similar injectability and the SNF-D/HA hydrogel was used as a model. (h) DFO release behaviors from the hydrogels. (i) BMP-2 release behavior from the composite hydrogels.

3.2. In vitro BMP-2 and DFO release

DFO loaded on the SNF hydrogels was slowly released for >40 days to provide angiogenetic stimulation.³³ After the DFO-loaded SNFs were blended with the HA nanoparticles, the composite hydrogels exhibited similar DFO sustained release behavior (Fig. 1h). Further loading of BMP-2 on the SNFs and HA nanoparticles had no influence on the delivery of DFO. Considering the similar release of DFO from these composite hydrogels, the amount of DFO (120μmol/L) optimized previously was used in the present study.³³

Different release behaviors of BMP-2 could be used to tune the osteoinductive capacity of the hydrogel systems.^36,47 BMP-2 has been loaded on silk fibroin scaffolds and HA nanoparticles, respectively, to change the release behaviors for osteoinduction.¹⁵ Here, BMP-2 was loaded on the SNFs and HA nanoparticles, respectively, and assembled into the composite hydrogels with tunable release behavior (Fig. 1i). The sustained release for above 40 days was found with all the composite hydrogels, and slower release was achieved when BMP-2 was loaded on the HA nanoparticles.

3.3. In vitro biocompatibility of the composite hydrogels

BMSCs were used to evaluate the cytocompatibility of the different composite hydrogels. SNF/HA, an injectable bone hydrogel with biocompatibility, was used as a control. BMSCs were cultured on different hydrogels for 1 week. The cells adhered and spread well on all of the hydrogels at day 1 and then proliferated to form continuous monolayers up to 7 days (Fig. 2a). DNA content confirmed the increase of cell numbers (Fig. 2b). The cell proliferation on all the hydrogels was similar, suggesting that the addition of DFO and BMP-2 had no negative influence on cytocompatibility. Similar to DFO-loaded SNF hydrogels,²³ the toxicity of DFO was significantly attenuated after loading on the SNFs in the composite hydrogels.

Fig. 2.

In vitro biocompatibility and angiogenesis assay of the composite hydrogels: (a) Confocal microscopy images of BMSCs cultured on the different hydrogels for 1, 3, and 7 days. Scale bar, 50 μm. (b) The proliferation behavior of BMSCs on the different hydrogels at day 1, 3, and 7. (c) Endothelial network formation of HUVECS when the cells cultured on the hydrogels for 3 and 6 h. Scale bar: 100 μm. (d) Number of meshes (HPF, high power field) calculated from different hydrogel groups. (e) Total meshes area (HPF) calculated from different hydrogel groups. Statistically significant *P≤0.05, **P≤0.01, and ***P≤0.001.

3.4. In vitro angiogenesis of HUVECs

Previous studies revealed that both DFO and BMP-2 could facilitate angiogenesis in vitro and in vivo.^48–52 To study the effect of the loaded DFO and BMP-2 on vascularization of the hydrogels, HUVEC tube formation was investigated (Fig. 2c). After culture for 3 and 6 h, a few fragile tubes appeared on the SNF-HA hydrogels, suggesting inferior angiogenic capacity when compared to the introduction of DFO. More thicker tubes assembled on the SNF-D/HA hydrogels. The number of complete tubes increased with the presence of both DFO and BMP-2, indicating a synergistic effect on angiogenesis. When BMP-2 was loaded on the HA nanoparticles or SNFs to form the composite hydrogels, similarly thicker tubes assembled on the different hydrogels. Significantly higher total length of complete tubes with larger areas was also achieved for the hydrogels containing both DFO and BMP-2 when compared to other groups, confirming the vascularization optimization through double loading of DFO and BMP-2 (Fig. 2d, e). Both confocal images and quantitative tube length results revealed that similar tubes assembled on the different DFO-BMP-2 loaded hydrogels without significant difference. Different BMP-2 loading processes resulted in various BMP-2 release behaviors and then tuned the osteoinductive capacity as shown in the following osteogenic differentiation study. The angiogenesis results suggest that angiogenesis and osteoinduction of the hydrogels could be tuned separately by changing the amount of DFO and BMP-2, and then adjusting the BMP-2 loading method.

3.5. In vitro osteogenic differentiation of BMSCs on the hydrogels

We have previously introduced BMP-2 to silk-HA composite scaffolds and improved osteoinductive capacity through tuning the BMP-2 release behavior.²² Similarly, the osteogenetic capacity of the injectable SNF-HA based hydrogels was also regulated by controlling the release behavior of BMP-2. After the optimization of angiogenic capacity, the osteogenic differentiation of BMSCs on the DFO-BMP-2 loaded hydrogels with different BMP-2 release rates was investigated to optimize the osteogenesis of the composite hydrogels. The sustained release behavior with gradually slower rates was achieved when more BMP-2 was loaded on HA nanoparticles in the composite hydrogels. Different osteogenic genes including ALP, Runx2, OCN and OPN were measured when the BMSCs were cultured for 1 month (Fig. 3a–d). Similar to previous studies,^17,53 the loaded DFO stimulated angiogenesis and facilitated osteodifferentiation, resulting in the upregulation of the different osteogenic genes. The introduction of BMP-2 further promoted osteogenesis in the composite hydrogels. Compared to BMP-2 free hydrogels, BMP-2 loaded hydrogels exhibited significantly higher expression of RUNX2 and OCN, which confirmed the synergistic action of BMP-2 and DFO (Fig. S3). Tuning BMP-2 release behavior further improved the osteogenic capacity of the hydrogels. Similar to BMP-2 loaded SNF-HA composite scaffolds, slower BMP-2 release from the hydrogels resulted in improved stem cells osteodifferentiation. The highest osteogenic gene expression was achieved when all the BMP-2 was loaded on the HA nanoparticles. Immunofluorescence staining for ALP, OCN and OPN also revealed the different osteogenic differentiation behavior of the BMSCs on the hydrogels. The cells cultured on the hydrogels containing DFO and BMP-2 exhibited stronger immunofluorescent expression of the osteogenic markers and achieved the most significant osteodifferentiation when cultured on the SNF-D/HA-B hydrogels where all the BMP-2 was loaded on the HA nanoparticles (Fig. 3e).

Fig. 3.

Osteo-differentiation of BMSCs cultured on the different composite hydrogels: (a-d) ALP activity and mRNA levels of RUNX2, OCN and OPN detected by real time PCR, respectively. (e) Immunofluorescence assay for ALP expression on day 7, OCN expression on day 14 and OPN expression on day 21. Scale bar, 200 μm.*P≤0.05, **P≤0.01, and ***P≤0.001.

3.6. In vivo bone regeneration

To evaluate the influence of angiogenic and osteogenic capacity of the hydrogels on bone regeneration in vivo, a rat cranial defect model was utilized and filled with the different hydrogels. After implantation of the hydrogels for 12 weeks, no animal death or infection occurred for all the rats. Although SNF based hydrogels were physically crosslinked, high beta sheet content of SNF endowed the hydrogels with suitable stability for tissue engineering.^28,33 Better stability was then achieved after the introduction of HA. The SNF/HA based hydrogels degraded slowly in vivo, accelerating bone regeneration in rat body.²⁸ Considering that DFO has no negative influence on the stability of SNF based hydrogels,³³ slow degradation remained for DFO-loaded SNF/HA hydrogels, strengthening its application in bone regeneration. Micro-CT images revealed dynamic bone healing processes in the defects (Fig. 4a,b,c). Similar to our previous study,²⁸ the SNF/HA hydrogel-filled defects showed progressive tissue regeneration in the peripheral area of the defect while the unfilled defects remained void after 12 weeks of implantation. The results revealed that the hydrogels with bone-similar organic-inorganic compositions stimulated the bone repair. The regeneration of osteoid tissues further improved after the introduction of DFO. Compared to that of SNF/HA samples, the vacant area decreased from 14% to 41% after 12 weeks. Better osteogenic stimulation was achieved through the released BMP-2. After 12 weeks, about 90% of the defect was filled with new tissue for rats treated with SNF-D-B/HA hydrogels, while the defects healed completely when filled with SNF-D/HA-B and SNF-D-B/HA-B hydrogels. Even at 8 weeks after surgery, the defects treated with SNF-D/HA-B hydrogels were closed by >95%, which was significantly better than the defects treated with the other hydrogels. The micro-architectural analysis of bone formation at the defect sites further supported the different tissue regeneration outcomes (Fig. 4d,e,f,g). At week 4, new bone volumes (BV) increased from 8.41±0.66 to 10.85±1.22, 13.68±0.24, 13.7±1.02 and 15.86±0.49 mm³ when the defects were treated with SNF/HA, SNF-D/HA, SNF-D-B/HA, SNF-D-B/HA-B, SNF-D/HA-B, respectively. The values further increased from 14.12±0.73 to 18.56±1.01, 22.02±1.27, 24.58±1.33 and 26.03±1.31mm³ at 12 weeks after implantation. The trabecular thickness (Tb. Th) and bone tissue volume per total tissue volume (BV/TV) exhibited similar trends for different hydrogels where best values were achieved for the defects treated with SNF-D/HA-B hydrogels. The trabecular separation (Tb. Sp) values further confirmed the regeneration of the densest new bone tissues when the defects were filled with SNF-D/HA-B hydrogels. The 2D images of the defects also revealed different bone healing. After 12 weeks, the defects treated with SNF/HA remained nonunion while the new bone formed at defects had connected at the defects treated with other composite hydrogels. The thickest bone formed at the defects filled with SNF-D/HA-B hydrogels, confirming the best osteogenic capacity for the samples studied. The results revealed faster tissue healing with improved new bone quality with both osteogenic and angiogenic features of the injectable composite hydrogels.

Fig. 4.

Micro-CT images and micro-architectural analysis of new bones at defect sites treated with different composite hydrogels: (a) 3D reconstructed images of defect sites at 4, 8 and 12 weeks after implantation. Scale bar, 2 mm. (b) 2D images of coronary sections at 4, 8 and 12 weeks after implantation. Scale bar, 2 mm. (c) 2D images of sagittal sections at 4, 8 and 12 weeks after implantation. Scale bar, 2 mm. (d) Bone volume/total volume ratio (BV/TV). (e) Bone volume. (f) Trabecular separation (Tb. Sp). (g) Trabecular thickness (Tb. Th). *P≤0.05, **P≤0.01, and ***P≤0.001.

To evaluate the influence of angiogenesis on bone regeneration, the new bone tissues at the defects were collected after 1, 2, 4, 8 and 12 weeks for H&E staining (Fig. 5). H&E staining indicated better early vascularization inside the hydrogels following the introduction of DFO and then BMP-2. Compared to the DFO-free samples (SNF/HA), significantly more blood vessels formed in the defects treated with SNF-D/HA, SNF-D-B/HA, SNF-D-B/HA-B, SNF-D/HA-B hydrogels at weeks 1, 2, 4 and 8. The improved angiogenesis stimulated ingrowth of tissues, resulting in faster defect healing. The released BMP-2 further accelerated new bone formation, with more bone regenerated in the samples loaded with BMP-2 and DFO. At week 12, vacant areas remained at the defects treated with SNF/HA, SNF-D/HA and SNF-D-B/HA, while complete healing appeared for the SNF-D-B/HA-B samples. When the defects were filled with the SNF-D/HA-B hydrogels, complete bone healing was achieved at week 8 and then new bone tissues with mature bone structure were regenerated after 12 weeks (Fig. 6). Therefore, similar to the micro-CT results, the best bone regeneration was achieved for the SNF-D/HA-B hydrogels in our present system.

Fig. 5.

H&E staining of the defects treated with blank group and different composite hydrogels at weeks 1, 2, 4, 8 and 12 after surgery. black arrows point to blood vessels, IM indicates implant material, HB means host bone and NB shows new formed bone. Scale bar, 20X: 1 mm, 200X: 100 μm.

Fig. 6.

The vascularization and the quality of the regenerated bones at the defects: (a) CD31 expression of the tissues grown inside the different composite hydrogels at peripheral region and central region at 1 and 2 weeks after surgery. Scale bar, 100 μm. (b) Quantification of vessel area calculated from the images (a). (c) Quantification of ALP-positive cells from the tissues grown inside the different composite hydrogels at 1 and 2 weeks after surgery. (d) ALP expression from the tissues grown inside the different composite hydrogels at 1 and 2 weeks after surgery. Scale bar, 100 μm. (e) Masson’s trichrome staining, OCN and OPN expressions of the defects treated with different composite hydrogels at 12 weeks after surgery. The surrounding normal bone tissues were also stained as control. Scale bars, 20X: 1 mm, 200X: 100 μm, OCN and OPN: 100 μm. Black arrows point to blood vessels, HB shows host bone and NB shows new bone formation. *P≤0.05, **P≤0.01, and ***P≤0.001.

The sections collected at 1 and 2 weeks after surgery were stained with CD 31 to visualize the early angiogenesis in the bone healing processes. Unlike the DFO-free groups where little neovascularization happened at 1 and 2 weeks, many neo-vessels formed around the peripheral and central regions of the defects treated with the DFO-loaded hydrogels (Fig. 6a). Quantitative analysis of the vessel numbers suggested similar angiogenesis for the different DFO-loaded hydrogel groups (Fig. 6b). No significant vascularization difference appeared for the DFO-loaded hydrogels when BMP-2 loaded on SNF and HA, respectively, suggesting that the introduction of DFO did not influence the angiogenesis of the hydrogel systems. Previous study confirmed that sustained release could stimulate the rapid in-growth of vessels into wounds.³³ Similar stimulation was achieved in our injectable bone systems where the sustained release of DFO also promoted the vascularization of the defects, which provided suitable vascularized microenvironments for improved bone tissue regeneration. Although further study is necessary to find optimal long-term vascularized niches for bone regeneration, the present in vivo results revealed that the sustained release of DFO facilitated the bone regeneration, significantly superior to DFO-free systems. The immunohistochemical (IHC) staining of ALP was carried out to evaluate the osteogenic environments at early stage (week 1 and 2) (Fig. 6c,d). Similar ALP expression was observed in the two BMP-2 free groups (SNF/HA and SNF-D/HA), which should be due to the high content of HA particles. The ALP expression was upregulated in the BMP-2-loaded groups (SNF-D-B/HA, SNF-D-B/HA-B and SNF-D/HA-B), confirming that the BMP-2 provided stronger osteogenic cues. Although the BMP-2 loaded groups contained the same amount of BMP-2 and exhibited similar angiogenic capacity in vivo, the different release behaviors resulted in different osteogenic capacity and the most extensive osteogenic environment was found with the SNF-D/HA-B group. Masson’s trichrome staining results indicated gradually quicker and better bone regeneration at the defects treated with SNF/HA, SNF-D/HA, SNF-D-B/HA, SNF-D-B/HA-B and SNF-D/HA-B hydrogels, respectively (Fig. S4). Besides Masson’s trichrome staining, the quality of the regenerated bone after 12 weeks was studied through immunohistochemical staining of OCN and OPN, typical markers of mature bones. As shown in Fig. 6e, more mature bone tissue occupied the defects treated with the BMP-2 and DFO-loaded hydrogels. The regenerated bone tissue with richer collagen deposition and better osteoid matrix was achieved with the SNF-D/HA-B hydrogel group, which exhibited similar composition and structure to the surrounding natural bones (Fig. 6e). The results confirmed that tuning multiple osteogenic and angiogenic cues simultaneously in the injectable hydrogel systems created a suitable niche for bone regeneration, superior to previous injectable silk-based bone systems.²⁸

4. Conclusion

Silk nanofibers and HA nanoparticles were used as nanocarriers to load DFO and BMP-2, forming injectable composite hydrogels with multiple angiogenic and osteogenic cues. The angiogenic and osteogenic capacity could be regulated independently through tuning the release behaviors of DFO and BMP-2, achieving microenvironments for bone tissue regeneration. Both in vitro and in vivo results revealed the adjustability of angiogenic and osteogenic cues in the composite hydrogel systems which stimulated the bone regeneration. Compared to previous injectable silk bone systems, these new silk nanofiber-based hydrogels introduced multiple physical/chemical osteogenic and angiogenic cues for improved bone regeneration (quality and quantity).