Enhanced osteogenic differentiation and biomineralization in mouse mesenchymal stromal cells on a β-TCP robocast scaffold modified with collagen nanofibers

Abstract

Calcium phosphate ceramics have been widely used in clinics as bone grafts. However, according to traditional techniques, it is difficult to create calcium phosphate bone grafts with tailored internal porous structures that may ensure optimal biocompatibility and sufficient mechanical properties. In this study, β-TCP bone scaffolds with well-defined inter-connective porous structures were fabricated by robocasting. With the aim of better mimicking the native extracellular matrix of bone without sacrificing the mechanical strength, type I collagen gel was coated on the filaments of the sintered β-TCP scaffolds to form an ultrafine fibrous network that was similar to the natural collagen nanofibers in bone. Both sintered β-TCP scaffolds and collagen gel coated scaffolds supported the growth of mouse mesenchymal stromal cells (mMSCs). Meanwhile, the thin layer of biomimetic collagen nanofibers on the β-TCP scaffolds significantly stimulated the osteoblastic differentiation of mMSCs by up-regulating the expression of ALP, Runx-2, collagen I, OPN, BSP and BMP-2. Additionally, a more active biogenesis of matrix vesicles (MVs) was induced in the mMSCs on the collagen gel coated scaffolds, evidenced by SEM and TEM results together with a significant increase of the gene expression of matrix vesicle components, indicating an enhanced initiation of matrix mineralization. Our study not only illustrates the potential use of the collagen gel coated β-TCP robocast scaffold for bone repair but also highlights the significance of the incorporation of collagen nanofibers in the scaffolds.

---

1. Introduction

Bone grafts are of vital importance for repairing bone defects caused by disease or trauma. Autografts, the current gold standard for bone grafts, however, are extremely limited by donor-site availability and morbidity. Therefore, synthetic alternatives have become attractive options. Calcium phosphate ceramics, such as hydroxyapatite, α-, β-tricalcium phosphate (TCP), octacalcium phosphate and dicalcium phosphate, are clinically favorable for use as bone scaffolds because of their chemical similarity to the inorganic phase of native bone and their excellent osteoconductivity.^1–3 In addition to the chemical composition, a well-defined 3D architecture of the bone scaffolds, especially appropriately interconnected porous structures, is required to allow for new bone in-growth, vascularization and controlled degradation.^4–6 Conventional techniques, such as chemical/gas foaming, porogen leaching, freeze drying and thermally-induced phase separation of polymer/ceramic slurries, have been used to introduce interconnected pores of calcium phosphate bone scaffolds.^5,7 Nevertheless, it is difficult to either control or reproduce the pore size, porosity, arrangement and interconnectivity with these approaches. Robocasting (RC), an additive manufacturing (AM) technology, offers a versatile tool that can fabricate scaffolds with customized shapes and with precisely controlled and interconnected porous structures from computer aided design (CAD) files.^7–12 Robocasting of calcium phosphate based scaffolds often requires polymers, such as polycaprolactone (PCL) and polylactic acid (PLA), that are dissolved in organic solvents to act as binders to ensure homogeneous particle distribution and consistent extrusion.^13–15 The use of organic solvents may cause problems because they can dissolve polymers used in most printheads. Moreover, it is extremely difficult to completely remove the organic solvents from the created scaffolds.^7,16 To overcome these problems, alternative powder-binder systems using aqueous binders, like aqueous polyvinyl alcohol solution and aqueous polyacrylic acid solution, have been developed.^17,18 However, in order to improve the mechanical properties, there is an urgent need for alternative material processing methods and optimal slurry formulation to meet the requirement of robocasting of calcium phosphate ceramics.

Skeletal bones are composed of not only calcium phosphate but also type I collagen, which is the major organic component of the bone extracellular matrix. Type I collagen is known for its excellent biocompatibility and ease of resorption by the body. Therefore, type I collagen has been increasingly used in the construction of synthetic bone scaffolds.^19–21 It has been reported that collagen sponge/HA composite scaffolds showed better osteoconductive properties compared to monolithic HA.²² Nevertheless, such collagen/calcium phosphate hybrid grafts suffer from the compromised mechanical strength and the instability of the 3D structure, because of the lack of a sintering process and the fast degradation of collagen.

In the present study, we fabricated bone scaffolds with well-defined inter-connective pores via robocasting of β-TCP. Recombinant type I collagen gel was coated on the surface of the β-TCP scaffolds after the sintering process to form a ultra-fine fibrous network that was similar to the natural collagen nanofibers of the bone extracellular matrix. The cellular responses of mMSCs, including cell proliferation, osteoblastic differentiation and matrix mineralization, to sintered β-TCP scaffolds and collagen gel coated scaffolds were systematically evaluated. Our results showed both sintered β-TCP scaffolds and collagen gel coated scaffolds supported cell adhesion and proliferation. The collagen gel coated scaffolds notably promoted proliferation and osteoblastic differentiation in mMSCs. Moreover, the collagen gel coated scaffolds stimulated mMSCs to produce matrix vesicles more actively, indicating an enhanced initiation of matrix mineralization.

2. Materials and methods

2.1 Scaffolds fabrication

β-TCP powder was prepared by the chemical precipitation method as described previously.²³ β-TCP slurries were prepared by mixing 45 vol% β-TCP in deionized water. To make a stable colloidal suspension of the β-TCP slurry, ammonium polyacrylate (PAA-NH₄) (1.5 wt% β-TCP) was added as a dispersant. The pH was adjusted to 9 by the addition of NH₃·H₂O (1 : 1 v/v). Then, high-speed ball milling was performed on a QM-BP planetary ball mill (Nanjing Nanda Instrument Plant, China) for 6 h to ensure that β-TCP had a small particle size. Hydroxypropylmethyl cellulose 4000 was added into the slurry as the viscosifier. The prepared β-TCP slurry was then filled into the cartridge and ready to be used. As shown in Fig. 1A, the 3D-Bioplotter™ system (Regenovo, China) was used to fabricate the scaffolds. The outer shapes of the scaffolds were designed using 3D CAD software SolidWorks (Massachusetts, USA). Either cubic (10.0 mm × 10.0 mm × 5.0 mm) or cylindrical (Φ 10.0 mm × 20.0 mm) scaffolds were fabricated in this study. The inner pore diameter of the scaffolds was set to 250 μm in the slicing software provided by 3D-Bioplotter system. A 250 μm tapered tip (Nordson EFD, USA) was used as the nozzle to produce β-TCP filaments with a diameter of 250 μm. The β-TCP filaments were then extruded as the designed arrangement under the pressure of compressed air. The β-TCP scaffolds were air dried for one week and then sintered at 1130 °C for 3 hours. All sintered β-TCP scaffolds were stored at room temperature and autoclaved before use.

Fig. 1 (a) Regenovo 3D-Bioplotter™ system. (b) Examples of β-TCP robocast scaffolds.

The sintered β-TCP scaffolds were further modified with collagen nanofiber following similar procedures as described previously.²⁴ Briefly, a collagen solution was prepared by dissolving type I collagen in 0.01 M HCl and stored at 4 °C. Before use, the prepared collagen solution was neutralized by adding 0.1 M NaOH on ice. The sintered β-TCP scaffolds were subsequently coated with collagen by flash dipping them in the neutralized collagen solution. After incubating the scaffolds for 2 h at 37 °C for complete gelation of collagen, the scaffolds were further air dried for 48 hours and stored at 4 °C until use. In order to make sure collagen can form nanofibrous structures without blocking the original pores of the sintered β-TCP scaffolds, collagen solution with three different concentrations (2.0 mg mL⁻¹, 1.0 mg mL⁻¹ and 0.5 mg mL⁻¹) were tested. While collagen solution with all three concentrations can form nanofibers, only nanofibers formed by 0.5 mg mL⁻¹ collagen did not block any pores of the sintered scaffolds (data not shown). Therefore, 0.5 mg mL⁻¹ collagen was used to modify the sintered β-TCP scaffolds in all following experiments.

2.2 Scanning electron microscopy (SEM) observation

The samples were sputter-coated with platinum and then examined with a Merlin scanning electron microscope (SEM) (Carl Zeiss, Germany). The samples seeded with mBMSC were fixed with 2.5% glutaraldehyde in PBS overnight at 4 °C and dehydrated in a graded ethanol series. Then, the samples were dried in a Critical Dryer K850 (Quorum, England) for 2 hours before they were sputter-coated with platinum and examined via SEM.

2.3 Micro-CT characterization

The inner porous structures of the sintered β-TCP scaffolds and collagen gel coated scaffolds were analyzed by micro-CT scanning as previously described.²⁵ The scaffolds were mounted on a rotary stage and scanned at a resolution of 14 μm using a XTV160H micro-CT (Nikon, Japan). The rotational step was 0.5° over an angle of 360°. Approximately 700 scan slices were taken and files were reconstructed by VGStudio Max 2.1. The reconstructed data was loaded onto Mimics (Materialise, Leuven, Belgium) and visualized at a thresholding range of 25 903 to 65 535. By measuring the volume of pores (V_p) and the sum of volume of pores and scaffold material (V_m), the porosity was calculated by the following equation:

Porosity = V_p/(V_p + V_m) × 100%

By measuring the total (V_p) and the interconnected pore volumes (V_ip), the pore interconnectivity was calculated according to:

Pore interconnectivity = V_ip/(V_p) × 100%.

2.4 Calcium ion release of the scaffolds

The sintered β-TCP scaffolds and collagen gel coated scaffolds were incubated in 10 mL of DMEM medium for up to 7 days at 37 °C in a humidified 5% CO₂ atmosphere. 1 mL supernatant of the leaching solution was collected for measuring the concentration of calcium ions (Ca²⁺) by atomic absorption spectroscopy (AAS) analysis (AA900T, Amoi, USA) after 0 h, 2 h, 8 h, 12 h, 24 h, 48 h, 72 h, 96 h, 120 h, 144 h, and 168 h of incubation.

2.5 Cell seeding and culture on scaffolds

Mouse mesenchymal stem cells (CRL-12424, ATCC, USA) were cultured in high glucose Dulbecco's modified Eagle's medium (H-DMEM, Gibco, USA) with 10% fetal bovine serum (FBS, Gibco, USA) at 37 °C in a humidified 5% CO₂ atmosphere. Medium was refreshed every 3 days. Cells were routinely subcultured at 80% confluence and used at passages 5–12.

To seed mMSCs onto the scaffolds, cells were trypsinized, centrifuged and resuspended in the fully supplemented medium at a final concentration of 2.0 × 10⁶ per mL. 50 μL cell suspension (1.0 × 10⁵ MSCs) were then seeded on the scaffolds and allowed to attach for 60 min before 1 mL fully supplemented medium was added. The regulation of both sintered β-TCP scaffolds and collagen gel coated scaffolds on cell behaviors were determined throughout a 14 day culture period.

2.6 Characterization of mechanical properties

After 4 days of incubation in the culture medium at 37 °C in a humidified 5% CO₂ atmosphere, the mechanical properties of both sintered β-TCP scaffolds and collagen gel coated scaffolds were tested under fully hydrated conditions using a 5960 dual column testing system (INSTRON, USA). Cylinder samples (n = 3) with a diameter of 10 mm and a length of 20 mm were tested under axial compression at a crosshead speed of 2 mm min⁻¹ until failure. The Young's modulus was calculated from the slope of the linear segment of the stress–strain curve. To evaluate the effects of cellular activities on the mechanical properties of the scaffolds, the mechanical properties of the scaffolds cocultured with mMSCs for 4 days were also tested.

2.7 Cell proliferation

Cell proliferation was determined by a Cell Counting Kit-8 (Dojindo, Japan) following the manufacturer's manual. Briefly, a standard curve over the cell number was first created (y = 0.0119x + 0.6143, R² = 0.99). The samples were harvested on days 1, 4 and 7. After removal of the media, the samples were incubated in a 200 μL working solution at 37 °C for 2 h. The absorbance at 450 nm was measured using a Varioskan Flash microplate reader (Thermo Scientific, USA).

2.8 Immunofluorescent staining

Cell adhesion was evaluated by immunofluorescent staining of the cytoskeleton protein F-actin. Briefly, cultured mMSCs were fixed with 4% paraformaldehyde for 30 min. The samples were permeabilized with 0.2% Triton X-100 in PBS. After washing with PBS, the cells were stained using phalloidin-FITC (AAT, USA) for 60 min at room temperature. Cell nuclei were stained with DAPI (Beyitime, China) for 5 min. The staining was examined under a laser scanning confocal microscopy (Leica TCS SP5, Germany).

2.9 Alkaline phosphate activity

The alkaline phosphatase activity of mMSCs on both sintered β-TCP scaffolds and collagen gel coated scaffolds was assessed after 7 days of culture. The sampling for alkaline phosphatase (ALP) staining was conducted with the BCIP/NBT phosphatase substrate (1-component) (KPL, USA) and imaged by a digital camera (Canon, Japan) (n = 3). To further quantify the ALP activity, mMSCs on both scaffolds were lysed in 30 μL of RIPA buffer containing PMSF (KeyGEN BioTECH, China) for 25 min on ice. The total protein was extracted and quantified using the BCA method. Then, the ALP activity was assessed using an alkaline phosphatase assay kit (A059-1, Nanjing Jiancheng Bioengineering Institute, China) according to the manufacturer's protocol. The results of the ALP activity were normalized to the total protein, which was defined as an activity unit.

2.10 Quantitative real-time polymerase chain reaction (qRT-PCR)

mMSCs on the scaffolds were cultured for 4, 7 or 14 days before the total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA). The isolated RNA was subjected to reverse transcription for first strand cDNA synthesis with a reverse transcription reagents kit (Invitrogen) according to the manufacturer's protocol. Quantitative real-time PCR was performed with the QuantStudio 6 Flex system (Life Technologies, USA) using the SYBR Green system (Invitrogen, USA). The mean cycle threshold (C_t) value of each target gene was normalized by the C_t value of the housekeeping gene GAPDH to acquire the relative expression. Then, the fold change of each gene was calculated using the ΔΔC_t method to compare the expression of the mRNA in mMSCs between the sintered β-TCP scaffolds and the collagen gel coated scaffolds. The sequences of the primers used for qRT-PCR in this study are shown in Table 1.

Table 1 Primer sequences used for qPCR gene expression analysis

Gene	5′–3′	Primers
ALP	Sense	5′-TGCCTACTTGTGTGGCGTGAA-3′
	Anti-sense	5′-TCACCCGAGTGGTAGTCACAATG-3′
Runx-2	Sense	5′-CACTGGCGGTGCAACAAGA-3′
	Anti-sense	5′-TTTCATAACAGCGGAGGCATTTC-3′
OPN	Sense	5′-TGCAAACACCGTTGTAACCAAAAGC-3′
	Anti-sense	5′-TCTTTACGTTTGCCGGTGACGT-3′
Collagen-I	Sense	5′-ATGCCGCGACCTCAAGATG-3′
	Anti-sense	5′-TGAGGCACAGACGGCTGAGTA-3′
BSP	Sense	5′-AGAACAATCCGTGCCACTCACTC-3′
	Anti-sense	5′-AGTAGCGTGGCCGGTACTTAAAGA-3′
BMP-2	Sense	5′-TGAGGATTAGCAGGTCTTTG-3′
	Anti-sense	5′-CACAACCATGTCCTGATAAT-3′
Annexin V	Sense	5′-CGCAGAGACCGTTGGGCCTCCC-3′
	Anti-sense	5′-CAGGGGCATCCCAAGCCT-3′
Calnexin	Sense	5′-GGCCAAGCATCATGCCATC-3′
	Anti-sense	5′-AAGCAGCTTCACATAGGCACCAC-3′
Calreculin	Sense	5′-AGGACATGCATGGAGACTCAGAATA-3′
	Anti-sense	5′-GATCAGCACATTCTTGCCCTTGTA-3′
Integrin α2	Sense	5′-ACAACCAGAGTTTGGAACAGGAC-3′
	Anti-sense	5′-CCCGCAATTATGCTGCCTATG-3′
Integrin β1	Sense	5′-ATCATGCAGGTTGCGGTTTG-3′
	Anti-sense	5′-GGTGACATTGTCCATCATTGGGTA-3′
GAPDH	Sense	5′-TGGATGGCCCCTCCGGGAAA-3′
	Anti-sense	5′-AGTGGGGACACGGAAGGCCA-3′

---

2.11 Transmission electron microscopy (TEM) observation

mMSCs were detached from the scaffolds with trypsin–EDTA (0.05%, Thermo Fisher Scientific, USA) and collected after 4 days of culture. The cells were then fixed in situ with 2.5% glutaraldehyde at 4 °C overnight. After fixation, the cells were incubated in a 1% osmium tetroxide phosphate buffer solution for 1 h. The cells were then dehydrated in a graded ethanol series and embedded in epoxy resin. Ultra-thin sections of approximately 70–100 nm were prepared, mounted on copper grids, and stained with lead citrate and uranyl acetate to enhance the contrast. The prepared cell samples were observed and photographed with a JEM-2100 TEM (JEOL, Japan).

2.12 Statistical analysis

Each experiment was repeated at least 3 times on different days, and the data were expressed as the mean ± SD. All of the calcium ion release, mechanical properties, cell proliferation, ALP activity and gene expression measurements were collected in triplicate for each group. An unpaired Student's t-test was used to evaluate the significance among the experimental groups. A value of p < 0.05 was considered statistically significant.

3. Results

3.1 Scaffold characterization

β-TCP scaffolds with tailored shapes, pore sizes and porosities were successfully fabricated via robocasting (Fig. 1B). As shown in Fig. 2A and C and Table 1, the average diameter of the pores in the sintered β-TCP scaffolds and collagen gel coated scaffolds were 252.7 ± 9.484 μm and 245.8 ± 7.820 μm, respectively. No statistically significant difference in pore size between the two groups was observed. The volumetric porosity of the sintered β-TCP scaffold was 34.59 ± 0.603% and the pore interconnectivity was 99.99 ± 0.006% (Table 2). Similarly, the porosity and pore interconnectivity of the scaffolds were not changed significantly after being coated with type I collagen gel. Besides macro pores formed by the intersected β-TCP filaments, some micro pores (∼1 μm) were observed on the filaments (Fig. 2B), which may be caused by the volatilization of aqueous water and organics. In contrast, Fig. 2D showed that the type I collagen gel formed a ultrafine nanofibrous network on the sintered β-TCP filaments. The diameter of these collagen nanofibers was approximate 50 μm.

Fig. 2 Representative SEM images of sintered β-TCP scaffolds (a and b) and collagen gel coated scaffolds (c and d).

Table 2 Characterization of the sintered β-TCP scaffolds and the collagen gel coated scaffolds

	Avg. pore diameter (μm)	Porosity (%)	Pore interconnectivity (%)
Sintered β-TCP scaffolds	252.7 ± 9.484	43.27 ± 0.54	99.67 ± 0.57
Collagen gel coated scaffolds	245.0 ± 4.04	44.21 ± 1.07	99.96 ± 0.36

---

3.2 Calcium ion release from scaffolds

As shown in Fig. 3, the concentration of calcium ions in both groups was comparable during the initial 12 hours. Surprisingly, notably more calcium ions were released from the collagen gel coated scaffolds (61.537 ± 0.581 μg mL⁻¹) than the sintered β-TCP scaffolds (50.053 ± 0.307 μg mL⁻¹) after 24 h of incubation. This trend continued for the rest 6 days. After 7 days of incubation, the concentration of calcium ions in the group of collagen gel coated scaffolds reached 111.867 ± 7.118 μg mL⁻¹, which was approximately 129% of that in the group of sintered β-TCP scaffolds (86.809 ± 0.817 μg mL⁻¹).

Fig. 3 Calcium ion concentration in DMEM medium after being incubated with either sintered β-TCP scaffolds or collagen gel coated scaffolds for up to 7 days. *Statistically significant, p < 0.05; **, p < 0.01.

3.3 Mechanical properties of the scaffolds

The mean compressive strength of the sintered β-TCP scaffolds and collagen gel coated scaffolds was 19.009 ± 4.296 MPa and 15.318 ± 3.442 MPa, respectively (Fig. 4A and B). No statistically significant difference in compressive strength between groups was observed. The Young's modulus of the sintered β-TCP scaffolds (1.881 ± 0.574 GPa) and collagen gel coated scaffolds (1.622 ± 0.645 GPa) were comparable (Fig. 4C).

Fig. 4 Mechanical properties of sintered β-TCP scaffolds and collagen gel coated scaffolds. (a) Representative curves for compressive strength. (b and c) Compressive strengths (b) and Young's moduli (c) of scaffolds cultured without or with mMSCs for 4 days. *Statistically significant, p < 0.05.

To evaluate the effects of mMSCs on the mechanical properties of the scaffolds, mechanical testing was also performed after coculturing both scaffolds with mMSCs for 4 days. While the compressive strength changed little in both groups (Fig. 4A and B), the Young's modulus of collagen gel coated β-TCP scaffolds remarkably decreased to 0.752 ± 0.230 GPa (Fig. 4C). Because the Young's modulus indicated the stiffness of the scaffold, our results demonstrated that mMSCs improved the elasticity of collagen coated β-TCP scaffold.

3.4 Cell proliferation

The immunostaining of the cell cytoskeletal protein F-actin showed that mMSCs on the sintered β-TCP scaffolds and the collagen gel coated scaffolds were confluent and covered the whole surface of the filaments of the scaffolds after 4 days of culture (Fig. 5). With the accumulated ECM secreted by the mMSCs, some MSCs on both scaffolds also began to migrate towards the pores (Fig. 5C and F, white arrow). The proliferation of the mMSCs on both sintered β-TCP scaffolds and collagen gel coated scaffolds was evaluated by a CCK-8 assay for one week. A continuous increase in cell number was observed for mMSCs on both scaffolds, indicating that both sintered β-TCP scaffolds and collagen gel coated scaffolds had good biocompatibility. Nevertheless, the mMSCs on the collagen gel coated scaffolds proliferated faster than those on the sintered β-TCP scaffolds since the 3^rd day (Fig. 5G).

Fig. 5 Cell proliferation on either sintered β-TCP scaffolds or collagen gel coated scaffolds. (a–f) Immunofluorescent staining of the intracellular cytoskeleton protein of F-actin (red) and nuclei with DAPI (blue) after 4 days of culture. (g) DNA assay for up to 7 days. *Statistically significant, p < 0.05.

3.5 Osteogenic differentiation of mMSCs

The time-dependent osteogenic gene expression in mMSCs on the sintered β-TCP scaffolds and collagen gel coated scaffolds, including early differentiation markers (Runx-2, ALP and type I collagen), late markers (OPN and BSP) and the key osteoblastic growth factor BMP-2, were measured via quantitative real-time PCR. As shown in Fig. 6, collagen gel coated scaffolds up-regulated the expression of ALP, type I collagen, OPN, BSP and BMP-2 in mMSCs for all time points investigated. Significantly higher expression level of Runx-2 was also observed on collagen gel coated scaffolds on day 4 and day 7. But on day 14, Runx-2 expression in MSCs on collagen gel coated scaffolds was decreased greatly to a similar level of that in mMSCs on the sintered β-TCP scaffolds. As expected, MSCs on collagen gel coated scaffolds expressed significantly higher OPN and BSP, both of which mediated the matrix mineralization, on day 14.

Fig. 6 The expression of genes related to osteoblastic differentiation and biomineralization relative to the housekeeping gene GAPDH by mMSCs on either sintered β-TCP scaffolds or collagen gel coated scaffolds. *Statistically significant, p < 0.05; **, p < 0.01.

Consistent with the gene expression results of ALP, more intensive ALP staining was found in the mMSCs cultured on the collagen gel coated scaffolds compared to those on the sintered β-TCP scaffolds (Fig. 7A). Moreover, ALP quantitative analysis confirmed that higher ALP activity was achieved by the mMSCs on the collagen gel coated scaffolds (Fig. 7B). These results indicated that collagen gel coated scaffolds promoted the osteoblastic differentiation of mMSCs.

Fig. 7 (a) ALP staining after culturing mMSCs on either sintered β-TCP scaffolds or collagen gel coated scaffolds for 7 days. (b) Quantitative analysis of alp activity after culturing mMSCs on the scaffolds for 7 days. **Statistically significant, p < 0.01.

3.6 Biomineralization

SEM images showed many dot-like structures on the surface of mMSCs grown on the collagen gel coated scaffolds, while only a few on those grown on the sintered β-TCP scaffolds (Fig. 8C and D). These dot-like structures were actually either clubbed or globular in morphology, as observed at a higher magnification (Fig. 8F). The average diameter of the globular structures was 108.6 ± 11.0 nm. A further analysis via energy disperse spectroscopy showed that these nano structures were mainly composed of carbon (C), phosphorus (Pi) and calcium (Ca) (Fig. 8G and H). Considering the morphology and size together with the chemical composition, the observed structures may be matrix vesicles that are associated with small crystals of calcium phosphate minerals, which occur in the pre-mineralized matrix of bone.²⁶ To ascertain this assumption, the gene expression of specific MV proteins, including annexin V, calnexin and calreticulin, were measured.^27–30 As expected, they were both markedly up-regulated in the mMSCs on the collagen gel coated scaffold versus the sintered β-TCP scaffold (Fig. 9), confirming a more active production of matrix vesicles and the associated biomineralization. TEM also showed that cells cultured on the collagen gel coated scaffold had more matrix vesicles in the cytoplasm (Fig. 10). In addition, significantly more vesicular membrane structures were observed in cells on the collagen gel coated scaffolds. Because matrix vesicles arise from the plasma membrane, an adequate supply of membrane materials from inside the cells also indicate an active budding process of the matrix vesicles.

Fig. 8 (a–f) representative SEM images of mMSCs on either sintered β-TCP scaffolds (a, c and e) or collagen gel coated scaffolds (b, d and f). (g and h) chemical elements of clubbed (g) or globular nano structures (h) as determined by energy disperse spectroscopy.

Fig. 9 Gene expression of the major components of matrix vesicles relative to the housekeeping gene GAPDH by mMSCs on either sintered β-TCP scaffolds or collagen gel coated scaffolds. **Statistically significant, p < 0.01.

Fig. 10 Representative TEM images of mMSCs grown on either sintered β-TCP scaffolds or collagen gel coated scaffolds.

4. Discussion

Robocasting enables the rapid fabrication of scaffolds with customized 3D geometries and porous structures. Hence, it has become an attractive scaffold manufacturing technology for clinically favored synthetic bone scaffolds. In this study, we successfully fabricated β-TCP based scaffolds with precisely controlled and interconnected porous structures via robocasting. With regard to the properties of the internal pores of the scaffolds, optimal pore diameters ranging from 200 μm to 1200 μm were suggested in the literature to guarantee potential space for tissue ingrowth, good diffusion of nutrients to cells within local pores and enhanced angiogenesis.^5,31 Therefore, β-TCP scaffolds with a pore diameter of 250 μm and a filament diameter of 250 μm were used in our study, which enables cell migration and nutrients exchange while avoiding the compromised mechanical strength of a larger pore size. The pore size of the fabricated scaffolds was comparable to the setting size (Table 2). Apparently, the sintering process only caused minor shrinkage of the β-TCP filaments. In addition to the pore size, porosity and pore interconnectivity are the other two major criteria for porous structures. The porosity of native cortical bone ranges from 5% to 30%, while native cancellous bone porosity ranges from 30% to 90%. The porous structures of the sintered β-TCP scaffolds with a porosity of 43.27 ± 0.54% and a pore interconnectivity of 99.67 ± 0.57% were similar to those of native cancellous bones. Meanwhile, both the compressive strength (19.009 ± 4.296 MPa, Fig. 4A and B) and the Young's modulus (1.881 ± 0.574 GPa, Fig. 4C) of the sintered β-TCP scaffolds were significantly higher than those of native cancellous bone, whose compressive strength was 2 to 12 MPa and Young's modulus was 0.05 to 0.5 GPa.³² In native bone, type I collagen assembles into fibers, which is one of the key structural component of the extracellular matrix. Although both collagen hydrogel and the composite of collagen/calcium phosphate have been printed into scaffolds using low temperature 3D printing technology,^33,34 it failed to mimic the nanofibrous structures of collagen in native bone. Collagen fibers are known to mediate mineralization by inducing heterogeneous nucleation of hydroxyapatite at the hole zones in the collagen structure.^35–37 Therefore, the lack of a collagen nanofibrous structure may impede the mineralization process. Additionally, the lack of a post sintering process also resulted in insufficient mechanical strength. Hence, to better mimic the native extracellular matrix of bone without sacrificing the mechanical strength, we coated a thin layer of collagen gel on the sintered β-TCP scaffold to form a collagen nanofibrous network that was similar to the natural collagen nanofibers in bone. The coating of collagen gel on the surface of the β-TCP filaments did not significantly change the overall porous structures of the scaffold, including the pore size, porosity and interconnectivity (Fig. 2A and C, Table 2). Additionally, the effects of collagen gel coating on the mechanical properties were minimal (Fig. 4). The mechanical properties of the scaffolds in both group were mainly attributed to the sintered β-TCP filaments. But it should be noted that collagen gel coated scaffolds with mMSCs exhibited a decreased Young's modulus than those without mMSCs (Fig. 4C). The increased elasticity may be caused by the elasticity of the seeded mMSCs themselves and the newly ECM secreted by mMSCs. Meanwhile, sintered β-TCP scaffolds with and without mMSCs exhibited similar elasticity, indicating a less active cell/material interaction they induced. It is also surprising that notably more calcium ions were released from the collagen gel coated scaffolds than the sintered β-TCP scaffolds. A possible explanation for the increased calcium ion release on collagen gel coated scaffolds may be the forming of calcium chelate complex in the group of collagen gel coated scaffolds. The coating of collagen gel hydrolyzed slowly into soluble collagen peptide in DMEM medium. These soluble collagen peptides then chelated calcium ions and therefore enhanced the release of calcium ions from scaffolds into the solution.

As expected, mMSCs proliferated faster on the collagen gel coated scaffolds compared to the sintered β-TCP scaffolds (Fig. 5), which was consistent with previous reports that type I collagen improved cell proliferation.²² Along with proliferation, the osteoinductivity is critical for bone scaffolds. β-TCP is widely used in bone repair because of its excellent biocompatibility and osteoconductivity, but the osteogenesis effect of β-TCP is very limited.^38,39 Similarly, our results showed that β-TCP scaffolds could not activate the expression of ALP in mMSCs without the aid of osteoinductive agents (Fig. 7A). However, mMSCs cultured on the collagen gel coated scaffolds exhibited an enhanced osteogenic phenotype, as measured by the expression of osteoblastic genes together with the ALP activity (Fig. 6 and 7), compared with cells cultured on sintered β-TCP scaffolds. Especially, the expression pattern of Runx-2 in MSCs on collage gel coated scaffolds may be beneficial for the differentiation of MSCs into fully mature osteoblasts, considering that the down-regulation of Runx-2 in late-stage osteoblastic differentiation is necessary for the transition from immature osteoblasts to mature osteoblasts.⁴⁰ Although type I collagen, either in solubilized form or gel form, has been reported to promote osteoblastic differentiation of mesenchymal stem cells (MSCs),^41–43 it was not usually considered as an osteoinductive material. The enhanced osteoinductivity of collagen gel coated scaffolds may result from the synergistic effect of the combination of collagen and β-TCP. In bone, the differentiation of MSCs into osteoblasts will result in the deposition of mineral into the matrix and ultimately bone formation.⁴⁴ The upregulated gene expression of OPN and BSP, both of which promote mineralization, suggests that the collagen gel coated scaffold may stimulate matrix mineralization.^45–47 The first step of matrix mineralization is the formation of hydroxyapatite crystals within matrix vesicles (MVs) that bud from the surface membrane of osteoblasts.^26,44 Undoubtedly, matrix vesicles play a central role in regulating the initiation of matrix mineralization. Thus, a more active biogenesis of MV observed on collagen gel coated scaffolds demonstrated a promoted matrix mineralization (Fig. 8–10). Additionally, annexin V, calnexin and calreticulin, all play key roles in managing mineral nucleation, transporting Ca²⁺, and controlling calcium homeostasis in the MVs.^{28–30,48,49} The elevated expression of these genes therefore further exhibited enhanced MV-mediated mineralization.

Although the specific mechanisms underlying the promoted osteogenesis and initiation of mineralization by the collagen gel coated scaffolds needs to be further studied, several factors may be involved. First, collagen may induce the osteoblastic differentiation of mMSCs and matrix mineralization by activating the integrin signaling pathway. Specifically, integrin α2β1, which is the major collagen receptor of cells, has been reported to mediate type I collagen-induced osteoblastic differentiation of bone marrow cells by further activating MAPK or focal adhesion kinase (FAK).⁴¹ Our results also showed that the expression of integrin subunits α2 and β1 in MSCs were up-regulated on the collagen gel coated scaffolds, indicating the involvement of integrin receptors in regulating osteoblastic differentiation (Fig. S1†). Second, the nano-fibrous architecture of collagen on sintered β-TCP scaffolds, may be crucial for the matrix mineralization. It has also been reported synthetic nano-fibrous scaffolds promoted osteoblastic differentiation and biomineralization by mimic a morphological function of collagen fibers.⁵⁰ Third, a higher extracellular concentration of calcium ions in the culturing system of the collagen gel coated scaffolds may contribute to promoting osteogenesis and mineralization. It has been shown that elevated Ca²⁺ acts as a signal for osteoblasts and their precursors and stimulates their proliferation, differentiation and mineral production through the MAPK/ERK pathway.^51,52 Ca²⁺ is also involved in modulating mineralization. Several studies have demonstrated that elevated extracellular Ca²⁺ accelerated vascular calcification by stimulating vascular smooth muscle cells undergoing matrix vesicle mediated calcification.^53,54 Because vascular calcification is a regulated process and is similar to bone formation,⁵⁵ elevated Ca²⁺ may also stimulate the matrix vesicle mediated mineralization in MSCs.

5. Conclusion

In summary, we fabricated β-TCP bone scaffold with well-defined inter-connective pores via robocasting. After hardening the scaffold with sintering process, recombinant type I collagen gel were coated on the surface of the β-TCP scaffold to form a ultrafine fibrous network similar to the natural collagen nanofibers in bone. The thin layer of biomimetic collagen nanofibers on sintered β-TCP scaffold significantly enhanced the osteoblastic differentiation in mMSCs and the matrix mineralization. Our findings demonstrated that collagen nanofiber modified β-TCP scaffolds may have a good potential in bone regeneration.

Conflict of interest

No competing financial interest.

Acknowledgements

This work was financially supported by the National Basic Research Program of China (Grant No. 2012CB619100), the National Natural Science Foundation of China (Grant No. 51402108, 51372085), Guangdong Province Public Interest Research and Capacity Building Special Fund (Grant No. 2015A020212003), the Science and Technology Program of Guangdong Province (Grant No. 2012A061100002), the 111 project (B13039) and Fundamental Research Funds for the Central Universities.

References

1. M. Jarcho, Clin. Orthop. Relat. Res., 1981, 157, 259–278.
2. R. Z. LeGeros, Clin. Orthop. Relat. Res., 2002, 395, 81–98.
3. S. Samavedi, A. R. Whittington and A. S. Goldstein, Acta Biomater., 2013, 9, 8037–8045.
4. V. Karageorgiou and D. Kaplan, Biomaterials, 2005, 26, 5474–5491.
5. S. J. Hollister, Nat. Mater., 2005, 4, 518–524.
6. M. Mastrogiacomo, S. Scaglione, R. Martinetti, L. Dolcini, F. Beltrame, R. Cancedda and R. Quarto, Biomaterials, 2006, 27, 3230–3237.
7. S. Bose, S. Vahabzadeh and A. Bandyopadhyay, Mater. Today, 2013, 16, 496–504.
8. S. Bose, M. Roy and A. Bandyopadhyay, Trends Biotechnol., 2012, 30, 546–554.
9. B. Leukers, H. Gülkan, S. Irsen, S. Milz, C. Tille, M. Schieker and H. Seitz, J. Mater. Sci.: Mater. Med., 2005, 16, 1121–1124.
10. H. Seitz, W. Rieder, S. Irsen, B. Leukers and C. Tille, J. Biomed. Mater. Res., Part B, 2005, 74, 782–788.
11. M. O. Wang, C. E. Vorwald, M. L. Dreher, E. J. Mott, M.-H. Cheng, A. Cinar, H. Mehdizadeh, S. Somo, D. Dean, E. M. Brey and J. P. Fisher, Adv. Mater., 2015, 27, 138–144.
12. N. Travitzky, A. Bonet, B. Dermeik, T. Fey, I. Filbert-Demut, L. Schlier, T. Schlordt and P. Greil, Adv. Eng. Mater., 2014, 16, 729–754.
13. U. Gbureck, E. Vorndran, F. A. Müller and J. E. Barralet, J. Controlled Release, 2007, 122, 173–180.
14. T. Serra, J. A. Planell and M. Navarro, Acta Biomater., 2013, 9, 5521–5530.
15. H. Seyednejad, D. Gawlitta, R. V. Kuiper, A. de Bruin, C. F. van Nostrum, T. Vermonden, W. J. A. Dhert and W. E. Hennink, Biomaterials, 2012, 33, 4309–4318.
16. H. N. Chia and B. M. Wu, J. Biol. Eng., 2015, 9 DOI:10.1186/s13036-015-0001-4.
17. A. Khalyfa, S. Vogt, J. Weisser, G. Grimm, A. Rechtenbach, W. Meyer and M. Schnabelrauch, J. Mater. Sci.: Mater. Med., 2007, 18, 909–916.
18. C. Wu, W. Fan, Y. Zhou, Y. Luo, M. Gelinsky, J. Chang and Y. Xiao, J. Mater. Chem., 2012, 22, 12288–12295.
19. M. Ngiam, S. Liao, A. J. Patil, Z. Cheng, C. K. Chan and S. Ramakrishna, Bone, 2009, 45, 4–16.
20. Y. S. Pek, S. Gao, M. S. M. Arshad, K.-J. Leck and J. Y. Ying, Biomaterials, 2008, 29, 4300–4305.
21. A. M. Ferreira, P. Gentile, V. Chiono and G. Ciardelli, Acta Biomater., 2012, 8, 3191–3200.
22. D. W. a. J. Czernuszka, Eur. Cells Mater., 2006, 11, 14.
23. S. H. Kwon, Y. K. Jun, S. H. Hong and H. E. Kim, J. Eur. Ceram. Soc., 2003, 23, 1039–1045.
24. X. Fu, M. Xu, J. Liu, Y. Qi, S. Li and H. Wang, Biomaterials, 2014, 35, 1496–1506.
25. S. T. Ho and D. W. Hutmacher, Biomaterials, 2006, 27, 1362–1376.
26. E. E. Golub, Biochim. Biophys. Acta, Gen. Subj., 2009, 1790, 1592–1598.
27. H. C. Anderson, Clin. Orthop. Relat. Res., 1995, 314, 266–280.
28. X. Zhou, Y. Cui, J. Luan, X. Zhou, G. Zhang, X. Zhang and J. Han, BioSci. Trends, 2013, 7, 144–151.
29. E. Somogyi, U. Petersson, K. Hultenby and M. Wendel, Matrix Biol., 2003, 22, 179–191.
30. Z. Xiao, C. E. Camalier, K. Nagashima, K. C. Chan, D. A. Lucas, M. J. d. l. Cruz, M. Gignac, S. Lockett, H. J. Issaq, T. D. Veenstra, T. P. Conrads and G. R. Beck, J. Cell. Physiol., 2007, 210, 325–335.
31. F. M. Klenke, Y. Liu, H. Yuan, E. B. Hunziker, K. A. Siebenrock and W. Hofstetter, J. Biomed. Mater. Res., Part A, 2008, 85, 777–786.
32. J. Henkel, M. A. Woodruff, D. R. Epari, R. Steck, V. Glatt, I. C. Dickinson, P. F. M. Choong, M. A. Schuetz and D. W. Hutmacher, Bone Res., 2013, 1, 216.
33. J. A. Inzana, D. Olvera, S. M. Fuller, J. P. Kelly, O. A. Graeve, E. M. Schwarz, S. L. Kates and H. A. Awad, Biomaterials, 2014, 35, 4026–4034.
34. J. Y. Park, J.-H. Shim, S.-A. Choi, J. Jang, M. Kim, S. H. Lee and D.-W. Cho, J. Mater. Chem. B, 2015, 3, 5415–5425.
35. Z. Xu, Y. Yang, W. Zhao, Z. Wang, W. J. Landis, Q. Cui and N. Sahai, Biomaterials, 2015, 39, 59–66.
36. W. J. Landis, F. H. Silver and J. W. Freeman, J. Mater. Chem., 2006, 16, 1495–1503.
37. B. Yang and F. Z. Cui, Curr. Appl. Phys., 2007, 7(1), e2–e5.
38. B. Liu and D.-x. Lun, Orthop. Surg., 2012, 4, 139–144.
39. M. C. Gupta, T. Theerajunyaporn, S. Maitra, M. B. Schmidt, C. E. Holy, S. Kadiyala and S. P. Bruder, Spine, 2007, 32, 720–726.
40. T. Komori, Cell Tissue Res., 2009, 339, 189–195.
41. M. Mizuno, R. Fujisawa and Y. Kuboki, J. Cell. Physiol., 2000, 184, 207–213.
42. T. Kihara, M. Hirose, A. Oshima and H. Ohgushi, Biochem. Biophys. Res. Commun., 2006, 341, 1029–1035.
43. R. M. Salasznyk, W. A. Williams, A. Boskey, A. Batorsky and G. E. Plopper, J. Biomed. Biotechnol., 2004, 2004, 24–34.
44. H. Orimo, J. Nippon Med. Sch., 2010, 77, 4–12.
45. C. M. Giachelli and S. Steitz, Matrix Biol., 2000, 19, 615–622.
46. J. Chen, H. S. Shapiro and J. Sodek, J. Bone Miner. Res., 1992, 7, 987–997.
47. A. L. Boskey, Connect. Tissue Res., 1996, 35, 357–363.
48. W. Wang and T. Kirsch, J. Cell Biol., 2002, 157, 1061–1070.
49. T. Kirsch, G. Harrison, E. E. Golub and H.-D. Nah, J. Biol. Chem., 2000, 275, 35577–35583.
50. K. M. Woo, J.-H. Jun, V. J. Chen, J. Seo, J.-H. Baek, H.-M. Ryoo, G.-S. Kim, M. J. Somerman and P. X. Ma, Biomaterials, 2007, 28, 335–343.
51. S. Nakamura, T. Matsumoto, J.-I. Sasaki, H. Egusa, K. Y. Lee, T. Nakano, T. Sohmura and A. Nakahira, Tissue Eng., Part A, 2010, 16, 2467–2473.
52. J. Ortiz and L. L. Chou, J. Biomed. Mater. Res., Part A, 2012, 100, 1770–1776.
53. H. Yang, G. Curinga and C. M. Giachelli, Kidney Int., 2004, 66, 2293–2299.
54. J. L. Reynolds, A. J. Joannides, J. N. Skepper, R. McNair, L. J. Schurgers, D. Proudfoot, W. Jahnen-Dechent, P. L. Weissberg and C. M. Shanahan, J. Am. Soc. Nephrol., 2004, 15, 2857–2867.
55. C. M. Shanahan, N. R. B. Cary, J. R. Salisbury, D. Proudfoot, P. L. Weissberg and M. E. Edmonds, Circulation, 1999, 100, 2168–2176.

---

Footnote

† Electronic supplementary information (ESI) available: The gene expression of integrin β1 and α2 relative to the housekeeping gene GAPDH by mMSCs on either the sintered β-TCP scaffolds or collagen gel coated scaffolds. See DOI: 10.1039/c5ra26670j

---