The influence of stereolithographic scaffold architecture and composition on osteogenic signal expression with rat bone marrow stromal cells

Abstract

Scaffold design parameters, especially physical construction factors such as mechanical stiffness of substrate materials, pore size of 3D porous scaffolds, and channel geometry, are known to influence the osteogenic signal expression and subsequent differentiation of a transplanted cell population. In this study of photocrosslinked poly(propylene fumarate) (PPF) and diethyl fumarate (DEF) scaffolds, the effect of DEF incorporation ratio and pore size on the osteogenic signal expression of rat bone marrow stromal cells (BMSCs) was investigated. Results demonstrated that DEF concentrations and pore sizes that led to increased scaffold mechanical stiffness also upregulated osteogenic signal expression, including bone morphogenic protein-2 (BMP-2), fibroblast growth factors-2 (FGF-2), transforming growth factor-β1 (TGF-β1), vascular endothelial growth factor (VEGF), and Runx2 transcriptional factor. Similar scaffold fabrication parameters supported rapid BMSC osteoblastic differentiation, as demonstrated by increased alkaline phosphatase (ALP) and osteocalcin expression. When scaffolds with random architecture, fabricated by porogen leaching, were compared to those with controlled architecture, fabricated by stereolithography (SLA), results showed that SLA scaffolds with the highly permeable and porous channels also have significantly higher expression of FGF-2, TGF-β1, and VEGF. Subsequent ALP expression and osteopontin secretion were also significantly increased in SLA scaffolds. Based upon these results, we conclude that scaffold properties provided by additive manufacturing techniques such as SLA fabrication, particularly increased mechanical stiffness and high permeability, may stimulate dramatic BMSC responses that promote rapid bone tissue regeneration.

1. Introduction

Scaffold design parameters are considered important in order to achieve a functional complex of cell/scaffold constructs, including pore size, porosity, interconnectivity, surface properties, mechanical strength, the amounts and types of filler material, cell seeding density, and exogenous growth factors [1]. In general, transplanted cell population may recognize the differences in these physical and mechanical cues and the subsequent cellular functions might be changed. Modulation of scaffold physical properties as well as changes in scaffold design parameters may influence the various cellular functions. Both stiffness (mechanical cues) and pore geometry (architectural cues) among these parameters are of importance to upregulate the endogenous osteogenic signal expression.

It is known that the scaffold stiffness influences adhesion [2], motility [3,4], morphology [5–7], proliferation [3,8,9], and osteoblastic differentiation [8–13] of cells. Cells can recognize scaffold mechanical cues (e.g., stiffness) and respond with secondary signal transduction that can bring about cellematrix interaction [14]. Recent studies have revealed that specific lineage of stem cell differentiation cascades can be directed by matrix elasticity [13,14]. In particular, it has been shown that mechanical properties of extracellular matrix (ECM) may regulate the osteogenic signaling mechanisms of celleECM through the sequential activation of FAK, RhoA/ROCK, MAPK, and Runx2 [3,11,12]. This osteogenic mechanotransduction can be also enhanced by a combination of mechanical cues with other stimuli such as ligand presentations on ECM [7].

In addition to mechanical stimuli, architectural cues for a 3D porous scaffolds including porosity, pore size, interconnectivity, and channel orientation are also important design parameters that can affect osteogenic signal expression of a seeded cell population [1,15]. A study with human mesenchymal stem cells on coralline hydroxyapatite scaffolds has shown that pore size could be a factor controlling both the level of bone morphogenetic protein-2 (BMP-2) mRNA expression and osteoblastic differentiation [16]. Similarly, another in vitro study also demonstrated that porous architecture of 3D silk fibroin scaffolds with the optimized porosity facilitated an increase in osteoblastic phenotypes of BMSCs [17]. An in vivo study with β-tricalcium phosphate scaffolds has shown that significantly higher osteoblastic differentiation was observed in higher porosity (over 65%) groups than in lower porosity groups [18]. Moreover, continuous pore geometry in scaffolds manufactured via solid freeform fabrication (SFF) has shown a higher cell ingrowth depth compared to scaffolds with random pore architecture [19].

In order to investigate the effect of mechanical and architectural cues on the stimulation of osteogenic signal expression, a composite material of poly(propylene fumarate) (PPF) and diethyl fumarate (DEF) was used in this study. This composite has shown unique photo-crosslinking characteristics [20]. By changing the molecular weight of PPF, the amount of photoinitiator and the ratio of PPF/DEF, the crosslinking density and mechanical properties of the PPF/DEF composite can be modulated. Due to this controllability, the mechanical stiffness of PPF/DEF scaffold can easily modulated during fabrication process. Besides of the controllable stiffness, PPF/DEF is a useful resin material for stereolithography (SLA). Incorporation of DEF with PPF reduces the viscosity of this liquidic polymeric mixture making it easier to utilize for SLA. The SLA device uses a laser to initiate the resin photo-crosslinking reaction and fabricate a 3D scaffold by vertical layering. SLA is one of the most versatile SFF techniques due to its accuracy, precision, and computer aided pre-design of the 3D external and internal scaffold geometry. SLA has been found useful for the rendering of patient- and defect-specific bone implants based on a patient's 3D CT scan [21,22]. SLA can also control the scaffold design parameters such as pore architecture and mechanical stiffness by modulating of photo-crosslinking reaction.

The global hypothesis of this study is that the modification in design parameters of 3D PPF/DEF composite scaffolds may facilitate osteogenic signal expression and enhanced level of signal expressions associated with the downstream osteoblastic differentiation of a seeded cell population. Therefore, the first part of this study is an investigation of the effect of DEF contents and pore size on the endogenous osteogenic signal expression and downstream osteoblastic differentiation of seeded bone marrow stromal cells (BMSCs) on 3D PPF/DEF scaffolds. It should be emphasized that changing the DEF incorporation ratio in PPF/DEF composite scaffold alters the crosslinking density and mechanical stiffness of the resulting scaffold. This sequential modulation in properties of the scaffold could stimulate the upregulation of osteogenic signal expression. The second part of this study is to investigate the effect of pore geometry within a scaffold, as another architectural cue, on the early osteogenic signaling profiles. For this second object, the advantage of controlled channel geometry of 3D macroporous PPF/DEF scaffolds, fabricated by SLA, over random pore structure, fabricated by porogen leaching method, on the osteogenic signal expressions has been investigated for the first time.

The specific objective of this study are: (1) to characterize the physical properties of 3D macroporous PPF/DEF composite scaffolds, (2) to investigate the effect of DEF content (subsequent changes in stiffness as a mechanical cue) and pore size (architectural cue) on osteogenic signal expression profiles and downstream osteoblastic differentiation, and (3) to investigate the effect of pore geometry (i.e., continuous channel geometry versus random pore structure) on the osteogenic signal expression of rat BMSCs.

2. Materials and methods

2.1. PPF synthesis and scaffold fabrication

PPF was synthesized according to previously reported methods [23]. Briefly, DEF and propylene glycol were reacted with zinc chloride as a catalyst and hydroquinone as a crosslinking inhibitor to form an intermediate compound. Then, transesterification was occurred to create the final PPF under vacuum condition. The number average molecular weight of the purified PPF (M_n = 1200 Da in this study) was determined by gel permeation chromatography. For PPF/DEF composite scaffold fabrication, PPF was mixed with DEF with various weight ratios (Table 1). 75 wt% of salt porogen crystals (>500 μm for “large” pore size and 180–300 μm for “small” pore size) and 0.5 wt% of photoinitiator bis(2,4,6-trimethylbenzoyl) phenylphosphine oxide (BAPO, Ciba Specialty Chemicals, Tarrytown, NY) were then homogeneously mixed with the PPF/DEF polymer mixture. The resulting paste was packed into a glass cylinder mold and photocrosslinked under UV light (intensity of 2.68 mW/cm²) for 2 h. Crosslinked polymer networks with salt porogens were cut into disks (6 mm in diameter and 3 mm in thickness) and placed in water for 3 days to leach out salt. The resulting macroporous PPF/DEF scaffolds were air-dried for 24 h and then dried again in a vacuum for 24 h. All experimental groups with random pore structure are listed in Table 1.

Table 1.

Experimental groups with random pore architecture: L =“large” pore (i.e., >500 μm), S = “small” pore (i.e., 180–300 μm).

Experimental groups	PPF	DEF	Pore size (μm)	Leached porogen amount (wt%)
L1	100	–	>500	77.41 ± 0.48
L2	90	10	>500	79.51 ± 1.14
L3	75	25	>500	78.43 ± 0.55
L4	66	33	>500	81.67 ± 0.58
S1	100	–	180–300	75.36 ± 0.24
S2	90	10	180–300	77.37 ± 0.52
S3	75	25	180–300	79.77 ± 0.58
S4	66	33	180–300	77.48 ± 0.37

2.2. SEM imaging

The top surface of scaffolds was visualized using a scanning electron microscope (SEM) (SU-70, Hitachi, Tokyo, Japan). Samples were gold sputter-coated. The images were obtained at 3 kV accelerating voltage.

2.3. Sol fraction

To assess the crosslinking density of PPF/DEF scaffolds, sol fraction test was performed by a previous method [20]. Each photocrosslinked scaffold was placed in 20 ml of methylene chloride solvent in a glass vial. The weight of the initial sample before incubation in solvent (W_i) was measured, and the samples were then incubated on a shaker at 75 rpm for 160 h at room temperature. Then, samples with solvent were transferred onto a weighed filter paper (W_p). These were completely dried in an oven at 70 °C for 2 h and weighed again (W_p+s). The sol fraction was calculated from the formula: sol fraction = (W_i − (W_p+s − W_p))/W_i. Five independent samples were assessed (n = 5).

2.4. Mechanical properties

According to the American Society of Testing Materials (ASTM) Standard D695-2a, compressive mechanical testing was performed using an Instron (Norwood, MA) mechanical tester (Instron 5565) to measure the compressive modulus and offset yield strength. The cylindrical porous scaffolds with 6 mm in diameter and 12 mm in length were compressed along its vertical axis. Compression was applied at a speed of 1.3 mm/min until the compressive strain reached 0.5 mm/mm. The compressive modulus and yield strength at 1% offset were calculated using Bluehill 2.16 software (Instron). Four replicates in each experimental group were tested (n = 4).

2.5. Permeability

The water permeability of scaffolds was determined according to the methods previously described based on Darcy's law [24,25]. An apparatus was constructed using a 2-L open container functioning as a water reservoir large enough to keep the pressure across the scaffold nearly constant (i.e., by keeping the height of the water in the apparatus nearly constant). Attached to the bottom of this reservoir was a short tube in which the scaffold, first wrapped in parafilm to create an air-tight seal along the side wall, was held. 1 L of water was added to the reservoir, and the water penetrated through the scaffold vertically was collected. After 120 s, the mass of water collected was recorded and the mass flow rate was calculated. This mass flow rate was converted to volumetric flow rate using the density of water. This procedure was repeated with 5 independent samples (n = 5). Permeability (K) was then calculated using the equation K = (ΔQ × L × μ_H2O)/(ΔP × A_CS), where ΔQ is a volumetric flow rate, L is the length of a scaffold, μ_H2O is the viscosity of water (8.90 × 10⁻⁴ Pa s at 25 °C), ΔP is a hydrostatic pressure difference between top and bottom of water column, and A_CS is the cross-sectional area of a scaffold.

2.6. Surface hydrophilicity

The hydrophilicity of the surface of the composite material was determined by a static contact angle measurement. Composite disks were fabricated by placing the PPF/DEF/BAPO mixture into rectangular molds on glass plates and crosslinked under UV light for 2 h [23]. A 5 μl droplet of water was then dropped onto the disk surface and a picture was then taken after 15 s. Image J software was used to analyze the angles at both sides of each water droplet and the average of both values was utilized for the further statistical analysis. Three measurements were performed for each sample and independent triplicate samples were tested.

2.7. Protein adsorption

To assess the level of adsorbed protein onto the scaffold, dried scaffolds were first completely wetted by a series of pre-socking: 1hr in ethanol, 30 min in PBS (twice), and over night incubation in PBS again. Samples were then placed in culture media with 10% FBS for 4 h at 37 °C on the shaker (25 RPM). After incubation, samples were washed with PBS 3 times, and adsorbed proteins were extracted during 1 h incubation in 250 μl of 1% sodium dodecyl sulfate (SDS) solution (repeated 2 times). The protein concentration was assessed using a BCA protein assay kit (Pierce, Rockford, IL). This test was completed with triplicate samples and three measurements were taken of each sample.

2.8. Rat bone marrow stromal cell isolation

Rat BMSCs were isolated from Wistar Hanover rats (male, 101–125 g, Taconic) following a University of Maryland approved IACUC animal protocol according to the method previously described [23]. Briefly, femora and tibiae were dissected from a rat that had been euthanized by carbon dioxide gas and incubated in a control culture media (α-MEM culture media containing 0.2 mm of ascorbic acid and 10 (v/v)% of penicillin/streptomycin) for 10 min. This incubation step was repeated three times. Then, both sides of explanted bones were clipped off and whole bone marrow inside was flushed out with 10 ml of fresh culture media using a syringe. Collected bone marrow was first suspended and filtered through a cell strainer with 70 μm pores. Filtered bone marrow was centrifuged, resuspended with control media containing 10% of FBS, and plated in a T-25 cell culture flask. Media was changed every 2–3 days, and cells were passaged up to 3 times. All samples were incubated under standard cell culture conditions of 37 and 5% of CO₂ level. Each scaffold was sterilized and presoaked as described previously [26] prior to the cell seeding.

2.9. Initial metabolic activity assay

A MTT assay (Sigma–Aldrich) was used to determine the metabolic activity of the seeded BMSC population influenced by (1) dissolved DEF in aqueous cell culture media and (2) the PPF:DEF ratio in composite scaffolds. In order to demonstrate the cytotoxic effect of different amount of DEF dissolved in cell culture media on cell monolayers, 0.1 million cells were seeded in each well of a 24 well plate and allowed to attach for 24 h. 0 (control group), 5, 10, and 20 mm of DEF in media with 10% FBS were applied in a well and the plate was incubated for 1, 2, and 4 h in an incubator. 50 (v/v)% of methanol in same culture media was used as a negative control. To determine the effect of the PPF:DEF ratio in 3D composite scaffolds on the metabolic activity of seeded BMSC population, 0.3 million cells (in 25 μl of control media containing 10% FBS) were seeded onto a sterilized scaffold with 100:0 (control group), 90:10, 75:25, and 66:33 of PPF:DEF ratio. After 4 h of incubation to allow the cells to attach to the surface of scaffolds, cell/scaffold constructs were washed with PBS. Then, 200 μl of 5 mg/ml of reconstituted MTT was added with 2 ml of 10% FBS media to each well for both assays, and incubated 2.5 h to allow the formation of MTT formazan crystals. Resulting crystals were dissolved with 200 μl of solublization solution and 200 μl of supernatant was used measure the optical density at 570 nm using M5 SpectraMax microplate reader (Molecular Devices, Sunnyvale, CA). Optical density was normalized to that of the control group (0 mm DEF group for 2D study and 100% PPF scaffold for 3D study) and % activity (=optical density of each experimental group/optical density of the control × 100) was plotted.

2.10. Cell seeding and culture

0.5 million cells in 25 μl of osteogenic supplemented (OS) media (i.e., control media supplemented with 10 mm Na–β-glycerophophate and 10⁻⁸ m dexamethasone) containing 10% FBS, were seeded onto a sterilized scaffold for further assays after typsinization from the culture flask. After 3 h of incubation in an incubator, 2.5 ml of OS media with 10% FBS was supplied to each sample. The media was first changed 24 h after seeding (Day 1) and replaced every 2 days thereafter until day 8.

2.11. Osteogenic signal expression

The total RNA was isolated from each cell/scaffold construct with Trisol (Sigma–Aldrich) following the protocol provided by the manufacturer. Isolated RNA was reverse-transcribed using a High Capacity cDNA Archive kit (Applied Biosystems, Foster City, CA). For the pre-amplification, cDNA samples were first mixed with pooled 0.2× Taqman Gene Expression assay (Applied Biosystems) including four osteogenic growth factor genes including BMP-2 (Taqman ID: Rn00567818_m1), fibroblast growth factors-2 (FGF-2, Rn00570809_m1), transforming growth factor-β1 (TGF-β1, Rn00572010_m1), and vascular endothelial growth factor (VEGF, Rn00582935_m1), two osteogenic marker genes including alkaline phosphatase (ALP, Rn00564931_m1) and osteocalcin (OC, Rn00566386_g1), and a transcriptional factor of Runx2 (Rn01512296_m1). This mixture was then incorporated with PreAmp master mix. Thermal condition for the PCR pre-amplification reaction was 10 min at 95 °C and 10 cycles of 15 s at 95 °C, and 4 min at 60 °C. The pre-amplified cDNA sample was diluted with 1× TE buffer (1:5) and utilized to investigate the relative expression level of target genes. A house-keeping gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an endogenous control gene. qRT-PCR was conducted on the ABI Prism 7000 sequence detector (Applied Biosystems), using thermal condition of 2 min at 50 °C, 10 min at 95 °C, and 50 cycles of 15 s at 95 °C and 1 min at 60 °C. Relative gene expression level of genes of interest was normalized the GAPDH control gene. Mean of fold changes compared to the calibrator group was analyzed using the ΔΔCt method and its standard deviations are reported (n = 3).

2.12. Osteoblastic differentiation

Early osteoblastic differentiation of rat BMSCs was first determined by ALP protein activity using using p-nitrophenyl phosphate (pNPP) assay (Sigma–Aldrich). Cell lysate was obtained from cell/scaffold constructs through three cycles of freeze (30 min at −80 °C), thaw (30 min at 37 °C), and sonication in the bath sonicator (30 min). The supernatant containing protein and DNA was mixed with pNPP liquid substrate and incubated for 1 h at 37 °C. After stopping the reaction by 2 m NaOH, the absorbance was recorded at 405 nm using a microplate reader. ALP activity was normalized by DNA amount from the same supernatant, which was measured using the PicoGreen assay kit. mm ALP/μg DNA was reported with triplicate.

2.13. Design, fabrication, and processing of stereolithographic scaffolds

A scaffold solid model was designed using SolidWorks^® CAD software (Dassault Systèmes SolidWorks Corp, Concord, MA). Based on a previously established design for use in bone tissue engineering [27], a modified scaffold model with controlled channel geometry was developed. The 3D Lightyear™ software (3D Systems Corporation, Rock Hill, SC) was used to add supports to the model and to generate model slice data from the CAD design. A 50 μm slice thickness was used. The slice data were then transferred to a Viper™ HA SLA system (3D Systems Corporation), which uses the slice information to fabricate the 3D model in a serial fashion, layer-upon-layer, by “drawing” each slice with a UV laser to produce spatially-controlled photo-crosslinking of a polymer resin. The initial layers are drawn such that the support structure of the solid model attaches to a perforated elevator platform. Thereafter, following the exposure of each slice, the elevator descends the distance of one layer thickness into a vat containing the polymer resin. This additive manufacturing technique allows for the serial addition of layers atop one another. In order to account for over-curing errors, which result when the UV-initiated photo-crosslinking produces a cured (solid) polymer depth which is larger than intended in the direction of the incident laser irradiation, 7 layers of z-correction were used. The z-correction parameter removes layers from any down-facing surface. For example, a 400 μm thick feature divided into 50 μm slices would yield 8 layers, but with 7 layers of z-correction the feature would be drawn using only a single exposure. The polymer resin for stereolithographic scaffold fabrication was prepared using a 66:33 ratio of PPF:DEF with 0.5 wt% of BAPO. Scaffolds were then fabricated in the Viper™ HA system. The laser parameters used for the build process were calibrated such that a critical exposure (E_c) of 15 mJ/cm² and a depth of penetration (D_p) of 4.574 mils (1 mil = 0.001 inch) were used. These values are used by the SLA system to define the relationship between exposure and cured polymer depth. Following the layer-by-layer build process, any liquid resin remaining in the scaffold pores was removed by alternating washes with acetone and ethanol. The scaffolds were allowed to dry in a fume hood over night, and were then transferred to a ProCure™ 350 ultraviolet post-curing apparatus (3D Systems Corporation) for 2 h. Finally, the scaffold support structures, which had provided attachment to the perforated elevator platform during SLA-based fabrication, were removed using a scalpel.

2.14. Random pore architecture (RPA) scaffold fabrication

Based on the CAD file for SLA scaffold fabrication, random pore architecture (RPA) scaffolds were also fabricated with 6 mm in diameter and >500 μm in pore size by the porogen leaching method. A PPF/DEF/BAPO mixture was prepared with the same weight ratio for SLA fabrication (i.e., 66:33 wt% of PPF:DEF). The overall pore volume of RPA scaffolds was controlled by the amount of salt porogen. Briefly, using the value of total volume and pore volume of SLA scaffold (0.0953 and 0.0603 cm³, respectively), the required porogen volume of a RP scaffold was assumed same as the pore volume of a SLA scaffold. The required weight of salt porogen (0.1302 g per scaffold) was obtained from the density of salt (2.16 g/cm³) and the pore volume from SLA scaffold design. In addition, the required weight of the PPF/DEF/BAPO mixture (0.0434 g per scaffold) was obtained from the density of mixture (1.24 g/cm³ by the triplicate measurement) and the solid volume (0.0350 cm³). Therefore, the weight ratio of mixture and salt porogen (0.0434:0.1302 = 25:75) was used for further RPA scaffold fabrication. This 75 wt% of salt ensured the interconnectivity of RPA scaffolds. PPF/DEF/BAPO mixture was homogeneously mixed with salt porogen, packed into a cylindrical glass vial, and crosslinked by 2 h of UV radiation. Crosslinked samples were retrieved from the glass mold and cut into disks to have the same dimension of SLA scaffolds (i.e., 6 mm in diameter and 5.2 mm in thickness).

2.15. Scaffold characterization

After gold sputter coating, scanning electron microscopic (SEM) images were obtained to verify the 3D structure of both SLA and RP scaffolds. In order to determine the permeability of both scaffolds, the same permeability test described in the previous section was performed (n = 4). Microcomputed tomography (μCT) imaging system (SkyScan 1172, Aartselaar, Belgium) was used to quantify porosity (vol%), surface to volume ratio, and interconnectivity (n = 3). Each scaffold was scanned at 40 kV of X-ray voltage and 250 mA of current. NRecon and CTAn software (SkyScan) were used for volumetric reconstruction and image analysis. A cylindrical volume of interest (VOI) with 4.5 mm in diameter and 5 mm in height was selected to remove the edge effect. Porosity and interconnectivity were calculated from the equations [28], porosity (%) = (VOI – volume of scaffold object)/VOI × 100 and pore interconnectivity (%) = (V – V_{shrink–wrap})/(V – V_m) × 100 where V is the total volume of VOI, V_{shrink–wrap} is the volume of VOI after shrink–wrap process, and V_m is the volume of the scaffold object. In addition, to determine the initial cell loading efficiency, 0.5 million cells with 25 μl of OS media were dropped onto the top of the scaffolds. Cell suspension around the bottom of the scaffold was collected and re-dropped onto the scaffolds after 30 min of cell seeding. Samples were incubated for 3 h and 2.5 ml of OS media with 10% FBS was provided in each well. After 24 h, unattached remaining cells in the media were counted (duplicate measurement with four samples). The cell loading efficiency was determined by (0.5 million – the total number of unattached cells per well)/(0.5 million) × 100 (%).

2.16. Protein secretion

BMP-2 and osteopontin (OP) protein secretion from the transplanted BMSC population to the culture medium were determined by enzyme-linked immunosorbent assay (ELISA) Media was collected, frozen at −20 °C, and stored until the analysis was performed. Samples were thawed, vortexed briefly, and assessed for the protein level based on the manufacturer's protocol. A quantikine BMP-2 ELISA kit (RnD systems, Minneapolis, MN) and OP (rodent) kit (Assay Design, Ann Arbor, MI) were used. The absorbance at 450 nm was measured using a microplate reader and the samples were run in biological duplicate with technical duplicate.

2.17. Statistical analysis

The data from all studies were analyzed by one-way analysis of variance (ANOVA) and Turkey's multiple-comparison test. p < 0.05 was considered to indicate a significant difference between experimental groups. The means and the standard deviations were reported in each figure.

3. Results

3.1. Characterization of 3D PPF/DEF scaffolds

3D macroporous composite PPF:DEF scaffolds were fabricated by a simple salt porogen leaching (Fig. 1). SEM images showed the porous structure of the scaffolds with two different pore sizes (Fig. 1A and B). The physical properties of PPF/DEF scaffolds were modified by varying the scaffold design parameters including the PPF:DEF ratio and the pore size. In order to investigate the photo-crosslinking characteristics of these composite scaffolds, sol fraction was determined (Fig. 2). DEF incorporated scaffold groups exhibited significantly lower sol fractions in both pore sizes compared to the PPF control group (p = 2.38 × 10⁻⁸ for large pore groups and 2.36 × 10⁻⁵ for small pore groups). Moreover, 25 and 33% DEF groups with small pore size (S3 and S4) showed significantly higher sol fractions (p = 2.81 × 10⁻² and 4.80 × 10⁻², respectively). Decrease in a sol fraction is correlated with increasing DEF content. As DEF content increases so does crosslinking density. Mechanical testing (Fig. 3) also indicated this relationship. Compressive modulus in Fig. 3A demonstrated that increasing DEF content increased the Young's modulus of the resulting scaffolds by up to 25%. All DEF incorporating groups with both pore sizes exhibited higher modulus than the PPF control (0% DEF group). In addition, increasing the mechanical strength of PPF:DEF scaffold by increasing its DEF content was also found in offset yield strength data (Fig. 3B). All DEF incorporated groups with both pore sizes showed significantly higher yield strength than the PPF control (p = 1.11 ×10⁻⁴ for large pore groups and 1.04 ×10⁻⁴ for small pore groups). These observations in 3D macroporous scaffolds were also found in non-porous photocrosslinked composites [20].

Fig. 1.

SEM images of scaffolds with a 66:33 ratio of PPF/DEF. (A) Large pore size (>500 μm) scaffold and (B) small pore size (180–300 μm) scaffold.

Fig. 2.

Sol fraction test. # indicates a significant difference between different DEF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05).

Fig. 3.

Mechanical testing. (A) compressive modulus and (B) offset yield strength. # indicates a significant difference between different DEF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05).

Water permeability from the top to the bottom of porous scaffolds was assessed (Fig. 4A). Due to the hydrophobicity of PPF, the 0 and 10% DEF groups showed a low level of water permeability. However, 25 and 33% DEF groups exhibited increased permeability. 75% of PPF and 25% of DEF incorporation resulted in significantly higher water permeability in both pore size groups (p = 1.73 × 10⁻³ for L3 and 2.23 × 10⁻² for S3). Large pore groups showed higher permeability than small pore scaffolds in more than 25% DEF incorporated groups. The contact angle data appear to suggest that increasing DEF content increased hydrophilicity (Fig. 4B). Both 25 and 33% DEF groups showed a significantly higher hydrophilicity (lower hydrophobicity) than both 0 and 10% of DEF groups (p = 1.15 ×10⁻¹²). Passive protein adsorption to PPF:DEF composite surfaces was also determined (Fig. 4C). In small pore groups, DEF incorporating groups showed significantly higher protein adsorption compared to the PPF control (p = 1.92 × 10⁻⁴) although the levels of adsorption in large pore groups were not significantly different. DEF incorporating groups with small pores exhibited significantly higher protein adsorption levels than those with large pores. Overall, increasing DEF content in a PPF/DEF composite scaffold increased crosslinking density, mechanical strength, permeability, and hydrophilicity without changing the level of protein adsorption to the scaffold surface.

Fig. 4.

(A) Permeability (B) Contact angle measurement. (C) Protein adsorption. For (A) and (C), # indicates a significant difference between different DEF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05). For (B), significant differences between groups (p < 0.05) were indicated by #.

3.2. Initial metabolic activity of BMSCs on 3D PPF/DEF scaffolds

The effect of free DEF molecules in aqueous cell culture media on the metabolic activity of a cell monolayer was investigated (Fig. 5A). Increasing DEF content in the media resulted in a decrease in initial metabolic activity of cell monolayers for up to 4 h. Use of 20 mm of DEF showed a similar level of the metabolic activity as the negative control (50% methanol group) after 1 h. All DEF incorporating groups up to 20 mm exhibited the same level as the negative control after 4 h of incubation. Then, the effect of DEF incorporation on the activity of seeded BMSCs onto photocrosslinked 3D PPF/DEF composite scaffolds was also assessed (Fig. 5B). All DEF incorporated groups showed a level of cellular metabolic activity similar to the PPF control after 24 h of incubation. Therefore, once the DEF molecules were participated in photo-crosslinking reactions with PPF polymer chains, the negative effect on cellular metabolic activity by increasing DEF amounts was not observed in 3D scaffolds.

Fig. 5.

Initial metabolic activity of rat BMSCs exposed to DEF molecules in aqueous culture media (A) and seeded onto photocrosslinked PPF/DEF scaffolds (B). # indicates a significant difference between groups (p < 0.05).

3.3. Osteogenic signal expression profiles of rat BMSCs cultured in 3D PPF/DEF scaffolds

The effect of DEF incorporation and pore size on the early osteogenic signal expression was investigated by using quantitative RT-PCR on day 8 (Fig. 6). For BMP-2 (Fig. 6A), DEF incorporating groups with large pores exhibited significantly higher expression levels compared to the 0% DEF control (p = 2.74 × 10⁻⁶). 25% DEF group showed a more than 4.4 fold increase over the calibrator. These DEF incorporating groups also showed significantly higher BMP-2 expression than the small pore groups while the 0% DEF control with small pore group showed higher expression than any experimental groups.

Fig. 6.

Osteogenic signal expression profiles of rat BMSCs population on day 8 including (A) BMP-2, (B) FGF-2, (C) TGF-β1, (D) VEGF, and (E) Runx2. Fold change in each group was analyzed by the level of calibrator (100 wt% PPF group with large pore size, L1). # indicates a significant difference between different PPF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05).

FGF-2, TGF-β1, and the angiogenic factor VEGF showed similar expression profiles. For FGF-2 (Fig. 6B), a clear trend of increasing expression was observed in large pore groups by increasing DEF content. 66% of PPF and 33% of DEF scaffolds with large pore (group L4) exhibited an approximately 4 fold increase compared the calibrator group (group L1). In large pore groups, significantly higher FGF-2 expression was observed in DEF incorporating groups when compared to the 0% DEF control group (p =6.62×10⁻⁶). These groups with a large pore showed significantly higher expression levels than groups with small pores. In small pore groups, DEF incorporating scaffolds did not show higher expression than the 0% DEF control. This relationship between increasing DEF content and enhanced signal expression was also found in TGF-β1 (Fig. 6C) and VEGF (Fig. 6D). For both signals, a similar trend of increasing expression with increasing DEF amount was observed in the large pore groups. 66% of PPF and 33% of DEF scaffolds with large pores (group L4) exhibited about a 4.4 fold increase in TGF-β1 expression and 5.4 fold increase in VEGF expression compared to the 0% DEF control. Two higher DEF incorporation groups with large pores (group L3 and L4) showed significantly higher TGF-β1 expression (p = 1.07 × 10⁻⁵) and all DEF incorporating groups with large pores showed significantly higher VEGF expression (p = 4.74 × 10⁻⁸) than the 0% DEF control. The pattern of higher signal expression in large pore groups than in a small pore groups was also found in both TGF-β1 and VEGF. In Fig.6E, Runx2 transcription factor expression in large pore showed the same clear trend of significantly increasing expression level by increasing DEF contents (p = 2.01 × 10⁻⁵) while small pore groups did not exhibit the significance between groups with different DEF content. In large pore groups, the 33% DEF group (L4) showed more than 3 fold increase over the 0% DEF control (group L1). Both 25 and 33% DEF groups with large pores (L3 and L4) showed significantly higher expression compared to those with small pores (S3 and S4) (p = 9.55 × 10⁻³ and 7.33 × 10⁻⁴, respectively).

3.4. Osteoblastic differentiation

ALP and OC mRNA expressions were assessed on day 8 (Fig. 7) to determine the relation between the stimulated osteogenic signal expression by changing the scaffold parameters (DEF contents and pore size) and downstream osteoblastic differentiation of BMSCs in 3D PPF/DEF scaffolds. As an early osteoblastic differentiation marker, ALP expression also showed a related trend of increasing expression as more DEF was incorporated in composite scaffolds with large pores (Fig. 7A). DEF incorporating groups with large pores (L2, L3, and L4) showed significantly higher ALP expression compared to the 0% DEF control (L1) (p = 3.42 × 10⁻⁷). The DEF incorporating groups also exhibited significantly higher ALP expression than small pore groups with the same DEF content. In small pore groups, the upper limit of DEF incorporation (S4) showed significantly higher levels of ALP expression than 0% control (S1) (p = 2.02 × 10⁻⁸). Fig. 7B indicates that DEF incorporation would also enhance the OC expression, as a late osteoblastic differentiation marker, in both pore sizes compared to 0% DEF control (p = 9.73 × 10⁻⁶ in large pores and 8.81 × 10⁻⁷ in small pores) (Fig. 7B). To confirm the influence of enhanced expression of osteogenic signals into osteoblastic differentiation, ALP protein activity was also measured over the 8 days of culture (Fig. 8). The result demonstrated that all experimental groups showed generally increasing intracellular ALP production (normalized by DNA level) up to day 8. 66% of PPF and 33% of DEF in both pore sizes (L4 and S4) showed significantly higher ALP protein levels when compared to the 0% DEF control groups (L1 and S1, respectively) at each time point.

Fig. 7.

Osteogenic differentiation was determined by ALP and OC mRNA expression on day 8. Fold change in each group was analyzed by the level of calibrator (100 wt% PPF group with large pore size, L1). # indicates a significant difference between different DEF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05).

Fig. 8.

Normalized ALP protein expression over 8 days of culture. # indicates a significant difference between different DEF contents within a large pore size groups while + indicates a significant difference within a small pore size groups (p < 0.05). & indicates a significant difference between two pore size groups in the same PPF:DEF ratio (p < 0.05).

3.5. Design and fabrication of stereolithographic PPF/DEF scaffolds

The CAD design for the SLA scaffold fabrication has a continuous channel geometry (Fig. 9A). A representative example of the resulting SLA scaffolds is shown in the SEM images (Fig. 9B) and μCT images (Fig. 9C). SLA scaffolds exhibited a controlled open channel geometry as originally designed in the CAD file. Simultaneous scaffold fabrication using a commercial SLA system was suitable for constructing the controlled architecture of 3D porous scaffold, though some manufacturing defects are shown (indicated by white arrows). Some of these minor defects, such as the erosion of edge features, might also be related to post-rendering polymerization shrinkage. RPA scaffolds fabricated by the porogen leaching have shown the irregular pattern of porous architecture (Fig. 9D and E). Additionally, a series of properties of both architectures were determined by μCT image analysis, including porosity, surface to volume ratio, and interconnectivity (Table 2). Theoretical (designed) volume% porosity of the SLA scaffold was 58% based on the CAD file. Measured volume% porosity of both scaffolds was 63.4% for RPA and 60.5% for SLA scaffolds. Comparing the porogen leaching and stereolithographical processes, the RPA and SLA scaffolds in this study possessed a similar level of total pore volume. The tortuous pore architecture of the RPA scaffolds resulted in a higher surface to volume ratio than what was seen in the SLA scaffolds (1.63 ×10⁻² for RPA and 0.89 × 10⁻² for SLA, 1/μm). Interconnectivity of the SLA scaffold was significantly higher than that of RPA scaffold at all pore connection sizes due to the formation of continuous open channels (Table 2, Fig. 9B and C). Interconnectivity is assumed to be a fraction of the porous volume in the scaffold object that was accessible from the outside through any opening of a certain minimum size (here, 200–800 μm) [28]. At lower pore connection sizes of 200 and 400 μm, interconnectivity of the SLA scaffold is over 90% since these minimum connection pores were smaller than the actual pore size (800 μm as designed). Since all the pore channels were connected and open to the outside in SLA scaffolds, high interconnectivity could be achieved. Therefore, as expected, permeability of the SLA scaffolds (K_SLA = 82.17 × 10⁻¹² m²) is significantly higher than that of RPA scaffolds (K_RPA = 3.15 × 10⁻¹² m²). These quantitative data indicated a highly permeable architectural morphology in the SLA scaffold. The cell loading efficiency was observed to be 82.66% of initially seeded cells remained after 24 h in the RPA scaffolds while 74.38% were found in SLA scaffold. This higher cell loading efficiency in RPA could be explained by the tortuous structural morphology of RPA scaffolds and subsequent higher surface to volume ratio, than SLA scaffolds.

Fig. 9.

SLA scaffolds with a controlled architecture. (A) A computer aided design (CAD) file to represent the controlled pore geometry of SLA scaffold (unit: mm), (B) the morphology of a fabricated SLA scaffold by SEM (Diagonal view: top, side view: bottom left, and top surface: bottom right), (C) μCT visualization of a SLA scaffold (full side view: left and continuous pore geometry: right), (D) SEM morphology of RPA scaffolds fabricated by the porogen leaching method (Diagonal view: left and top view: right), and (E) μCT visualization of a RPA scaffold (full side view: left and tortuous pore morphology: right). Scale bar in (C) and (E) represents 2 mm.

Table 2.

Characterization of RPA and SLA scaffolds. Porosity (vol%), surface to volume ratio, and interconnectivity were assessed by μCT image analysis.

Architectural scaffold properties	RPA	SLA
Porosity (designed)	–	58 (vol%)
Porosity (measured, vol%)	63.4 ± 3.26	60.50 ± 1.16
Surface area to volume (× 10−2, 1/μm)	1.63 ± 0.04 (*)	0.89 ± 0.05
Interconnectivity (%)
Minimum connection size of 200 μm	84.50 ± 7.19	98.15 ± 0.76 (*)
Minimum connection size of 400 μm	54.51 ± 11.78	93.87 ± 1.02 (*)
Minimum connection size of 800 μm	15.76 ± 3.73	45.20 ± 3.58 (*)
Permeability (×10−12, m2)	3.16 ± 0.87	82.17 ± 50.03 (*)
Initial cell loading efficiency (%)	82.66 ± 6.70 (*)	74.38 ± 5.90

^*indicates a significantly higher value between two different scaffold architectures (p < 0.05).

3.6. Osteogenic signal expression and osteoblastic differentiation in SLA scaffolds

In order to investigate the effect of pore geometry as an architectural cue on osteogenic signal expression, BMP-2 secretion from the cell population was assessed by ELISA while FGF-2, TGF-β1 and VEGF mRNA expression were determined by qRT-PCR (Fig. 10). In the first part of this research, we have seen that osteogenic signal expression was influenced by the mechanical stiffness and structural parameters including permeability and pore size of a 3D scaffold. In the second part, it has been observed that pore geometry of the 3D scaffold could also affect osteogenic signal expression. Cumulative BMP-2 protein secretion in SLA scaffolds was higher than in RPA scaffolds (Fig. 10A). FGF–2mRNA expression level on day 4 (Fig. 10B) was significantly higher in SLA scaffolds than in RPA scaffolds (p = 9.99 × 10⁻⁶) while the expression levels on day 8 were similar in different architectural scaffolds. TGF-β1 and VEGF signal expression exhibited a similar pattern (Fig.10C and D). For both signals, the expression levels in SLA scaffolds on day 4 were significantly higher (about 9 fold changes) than in RPA scaffolds (p = 3.59 × 10⁻³ for TGF-β1 and 2.47 × 10⁻⁶ for VEGF) and this level was the highest in all comparisons. For all three osteogenic signals, the expression level in the SLA was peaked on day 4 and thereafter decreased till day 8 while these levels in the RPA scaffolds increased over time.

Fig. 10.

Osteogenic signal expression for comparison between RPA scaffolds and SLA scaffolds. (A) Cumulative BMP-2 protein secretion over 8 days. Osteogenic signal expression profiles of BMSCs including (B) FGF-2, (C) TGF-β1, and (D) VEGF. Fold change in each group was analyzed by the level of calibrator (RPA on day 4). * indicates a significantly higher expression in the SLA scaffold compared to RPA at each time point (p < 0.05). # indicates a significant difference between time points within RPA scaffolds while + indicates a significant difference within SLA scaffolds (p < 0.05).

To investigate the relationship between the enhanced expression level of osteogenic signals and osteoblastic differentiation of BMSCs in different pore architectures, ALP mRNA and OP protein secretion were assessed (Fig. 11). ALP expression in SLA scaffolds was significantly higher than in RPA scaffolds at both time points (p = 4.19 × 10⁻⁵ for day 4 and 1.68 × 10⁻⁶ for day 8) (Fig. 11A). The expression level in SLA on day 8 was 4.2 fold changes compared to the calibrator group (RPA on day 4). OP protein secretion from BMSCs in SLA scaffolds increased and peaked on day 5 while the level in the RPA scaffolds was stagnant over time (Fig. 11B). OP secretion in both types of scaffolds after day 3 was significantly higher than at the initial time point of day 1 (marked by “#”). The secretion level in SLA scaffolds was significantly higher than that in RPA scaffolds at all time points (marked by “+”). Cumulative OP secretion (Fig.11C) confirms that the rate of OP production from the seeded cell population was greater in SLA scaffolds than in RPA scaffolds over time.

Fig. 11.

(A) ALP mRNA expression over 8 days. Fold change in each group was analyzed by the level of calibrator (RPA on day 4). (B) The amount of osteopontin (OP) secretion per scaffold at each time point over 8 days. (C) Cumulative OP secretion. * indicates a significantly higher expression in the SLA scaffold compared to RPA at each time point (p < 0.05). # indicates a significant difference between time points within RPA scaffolds while + indicates a significant difference within SLA scaffolds (p < 0.05).

4. Discussion

The main objective of this study was to determine if the mechanical and architectural parameters of 3D PPF/DEF composite scaffolds facilitate endogenous osteogenic signal expressions and downstream osteoblastic differentiation in a seeded BMSC population. Two scaffold design parameters on the first study are the DEF content and the pore size of PPF/DEF porous scaffolds. It has been shown that DEF incorporation in PPF polymer increased crosslinking density and decreased the sol fraction of cylindrical composite material [20]. Therefore, incorporating DEF into PPF increased the mechanical strength including compressive modulus and fracture strength. Based on these characteristics, the mechanical and architectural properties of photocrosslinked 3D macroporous PPF/DEF scaffolds were examined in this study. Sol fraction (Fig. 2) and mechanical properties (Fig. 3) of porous PPF/DEF scaffolds have exhibited a trend similar to what was seen in the non-porous PPF/DEF constructs in the previous study [20]. Lower sol fraction levels in DEF incorporated groups in both pore size groups indicated higher crosslinked fraction in PPF/DEF composites (Fig. 2). 25 and 33% DEF groups exhibited the highest level of crosslinked fractions, indicated by the lowest sol fraction. In addition, higher crosslinking density with lower sol fraction was observed in large pore size groups although the penetration of solvent into the scaffold inner region and subsequent contact to dissolve the polymeric surface would be expected to be more efficient in large pore groups. DEF incorporation could also increase the mechanical strength of PPF/DEF scaffolds (Fig. 3). 25% of DEF could achieve the highest compressive modulus and offset yield strength. The compressive modulus and yield strength of PPF/DEF composite scaffolds are within a range of those of native human cancellous bone (10–900 MPa of compressive modulus and 0.2–14 MPa of yield strength) [29–31].

Increasing DEF content resulted in the higher water permeability of PPF/DEF scaffolds in both pore sizes (Fig. 4A). In addition, DEF incorporation also increased the hydrophilicity as shown by decreasing contact angles (Fig. 4B). It might be speculated that increasing surface hydrophilicity by increasing DEF content in the composite scaffold could be related to increasing permeability.

In addition to DEF incorporation and subsequent changes in surface properties, particularly an increase in hydrophilicity, pore size could also increase permeability. As a structural cue, permeability could be one of the stimuli to enhance the signal expression of progenitor cells. It has been demonstrated that increasing permeability induced chondrogenic differentiation of BMSCs on 3D poly(ε-caprolactone) scaffolds since permeability could affect diffusion, oxygen tension, and nutrient exchange [32]. A scaffold with 860 μm pore size showed higher permeability as well as increased ratio of type II collagen to type I collagen expression, which was not shown in scaffolds with lower permeability. Therefore, the result in the present study demonstrates that modulation in DEF contents and pore size in PPF/DEF composite scaffolds could effectively alter surface hydrophilicity and permeability.

Although the free DEF molecules in aqueous media had a negative effect on the metabolic activity of cell monolayers (Fig. 5A), cross-linked PPF/DEF scaffolds did not show any significant decrease in the activity of seeded BMSCs compared to the 100% PPF control group up to 24 h (Fig. 5B). This result demonstrated that the crosslinking reaction with the aid of a photoinitiator and UV light is suitable to allow DEF molecules to participate in the formation of additional cross-links between polymer chains [20] and subsequently increased the mechanical strength of the composite (Fig. 3). The crosslinked PPF/DEF scaffolds could be a proper platform for bone tissue engineering in terms of both mechanical strength and biocompatibility.

In order to investigate the effect of increased mechanical properties and permeability by altering the DEF content and pore size on the osteogenic signal expression, qRT-PCR was performed on day 8 (Fig. 6). It is demonstrated that DEF incorporated scaffolds with a large pore size might facilitate osteogenic signal expressions including BMP-2, FGF-2 and TGF-β1 as well as the expression of both the angiogenic signal VEGF and a transcriptional factor Runx2. Within the large pore groups, these signal expressions were dependent on DEF content, subsequent changes in crosslinking density, and stiffness of composite scaffolds. Increasing expression of these signals in large pore groups might be explained by the Runx2-mediated intracellular signaling transduction. The binding and stimulation of BMP-2 and TGF-β1 might be associated with Runx2 transcription factor expression via Smad and MAPK pathway [33–36] while FGF-2/MAPK/Runx2 pathway might promote the osteoblast maturation [33,37,38]. Hence, the increasing Runx2 expression pattern by increasing DEF content in the large pore groups (Fig. 6E) might be associated with an enhanced level of BMP-2, TGF-β1, and FGF-2 expression. It might be speculated that modulation of PPF/DEF properties could upregulate of these growth factor genes and the enhanced expression might increase the possible binding of these proteins to the cells. Finally, the bindings could stimulate the downstream signaling pathway to induce the Runx2 expression [33] and further expression of osteoblastic differentiation markers such as ALP and OC [39,40]. This similar expression pattern of osteogenic signals in Fig. 8 demonstrates that altering the scaffold design parameters, specifically increasing the DEF content within scaffold with large pore geometry in this study, influenced the stimulation of endogenous signal expression through the intracellular signaling mechanisms. In addition, upregulated VEGF expression also appears to increasing DEF amount in the large pore groups (Fig. 6D). Since proper osteointegration of implant materials and bone tissue regeneration must be closely related with neovascularization and angiogenesis, it is of importance to investigate the VEGF expression profile along with other osteogenic signal expressions, especially FGF. FGF is involved in both osteogenesis and angiogenesis [41–43] and VEGF expression is also implicated in osteoblastic differentiation and the regulation of bone remodeling [44,45]. In addition, FGF-2 may regulate VEGF expression [46] and VEGF release from osteoblasts was stimulated by FGF-2 through a series of signaling mechanisms [47–50]. Therefore, the similar expression profiles of FGF-2 and VEGF in Fig. 6B and D could explain the relationship between those signal expressions during osteoblastic differentiation. Increasing the DEF content in composite scaffolds appears to facilitate both FGF-2 and VEGF expression (4 and 5.3 fold increase, respectively) in the L4 group compared to L1 calibrator. From this observation, it might be speculated that the modulation of a PPF/DEF scaffold's mechanical properties by varying the DEF content could stimulate the Runx2-mediated osteogenic signal expressions as well as VEGF expression.

However, this close relation between the DEF incorporation and the stimulated signal expression was not observed in the small pore groups (Fig. 6). Although increasing signal expression by increasing DEF content in S2, S3, and S4 was seen within small pore scaffolds, the fold changes in these groups were lower than in the S1 control group. This observation might suggest the importance of architectural cues to facilitate osteogenic signal expression. Small pore scaffolds possessed a low permeable microenvironment (Fig. 4A) and cells could not recognize the effect of mechanical strength (stiffness) in this unfavorable environment although mechanical strength of the small pore scaffolds was similar to that of the large pore scaffolds (Fig. 3). We speculated that large pore geometry is more advantageous for transport of nutrient and waste removal in an aqueous in vitro environment. Permeability data in Fig. 4A confirm that large pore groups allow higher level of transport than small pore groups. Other studies have also suggested that large pore sizes (325 μm) would help cell migration and reduce the formation of cell aggregation that might inhibit the diffusion [51] while in vivo bone ingrowth at 4 weeks post-implantation was increased as the pore size was increased from 350 to 800 μm [52]. Therefore, in the present study, it could be suggested that a large pore size (>500 μm) might accelerate the enhancement of signal expression, which was influenced by the stiffness of PPF/DEF scaffolds. Moreover, the signal expression levels in DEF incorporating scaffolds with large pores (group L2, L3, and L4) are higher than that of DEF incorporating scaffolds with small pores (group S2, S3, and S4) for FGF-2, TGF-β1, VEGF, and Runx2. Therefore, we also speculated that both scaffold design parameters including DEF contents and pore size could influence the upregulation of signal expression. DEF content and subsequent changes in stiffness are not the only factor to induce osteogenic signal expressions, and the combined effect with pore geometry design could affect the signal expressions of seeded BMSC population.

To validate the hypothesis that enhanced signal expression is related to the downstream osteoblastic differentiation, both ALP and OC mRNA expression on day 8 (Fig. 7) and ALP protein activity over 8 days (Fig. 8) were determined. The highest expression of ALP in L4 group was more than a 7 fold change compared to the L1 calibrator group (Fig. 7A). The enhanced level of ALP expression in this group might indicate the combined effect of DEF incorporation and pore size. In addition, it might be anticipated that stimulated signal expressions could be associated with osteoblastic differentiation of BMSCs. Upregulated endogenous expression of BMP-2, TGF-β1, and FGF-2 might stimulate the activation of Runx2 expression, and activated Runx2 could express the downstream ALP expression through the Smad pathway [34,53,54]. Fig. 7B also indicated the combined effect of DEF incorporation and pore size on OC expression although the expression level was higher in small pore groups. ALP gene expression was confirmed by ALP protein activity (Fig. 8). The L4 and S4 groups on day 8 showed statistically higher ALP activity than other formulations within the same pore size. Therefore, the osteoblastic differentiation of seeded BMSCs was also facilitated in 33% of DEF incorporated composite scaffold with a large pore size (i.e., over 500 μm).

The best formulation to facilitate the osteogenic signal expression was the subject of the first part of this study (i.e., 33% of DEF content and large pore size), whereas the effect of pore architecture in 3D scaffolds on the osteogenic signal expression was investigated in the second part of this study. SLA scaffolds with controlled pore geometry were compared to RPA scaffolds with irregular pore architecture. In the first part, we observed that architectural scaffold design parameters including pore size and permeability effectively influence osteogenic signal expression. It is also anticipated that continuous channel geometry in the internal scaffold architecture could facilitate nutrient transport and waste removal due to the higher permeability. The SLA technique facilitates accurate control the internal geometry of the scaffolds including pore size, porosity, and specific surface area, and the permeability [55]. Cell attachment occurs on the pore walls in a porous scaffold and cells grow through the pores via channel guidance to finally occupy the inner pore region [19]. Therefore, the channel geometry in 3D scaffolds is considered an important architectural cue to achieve the functional cell/scaffold complex. In the present study, we observed that SLA scaffolds showed a significantly higher permeability and interconnectivity than RPA scaffolds. Initial cell attachment was higher in the tortuous structure of RPA scaffolds due to the larger surface area than SLA scaffolds [56]. Therefore, SLA scaffolds with continuous channels and higher permeability could be more favorable to the cell population than RPA scaffolds due to the easier transport of various molecules in aqueous cell culture condition, irrespective a higher level of initial cell occupation in RPA.

qRT-PCR data in Fig.10 showed that osteogenic signals including FGF-2, TGF-β1, and VEGF could be upregulated when the BMSC population was cultured in the controlled architecture with continuous pore geometry rather than a randomly oriented architecture. Cumulative BMP-2 secretion was also higher in the SLA scaffolds over 8 days. Specifically, these signal expressions were facilitated on day 4 after seeding. This early enhancement of the osteogenic signals could affect the osteoblastic differentiation of seeded BMSCs, which was assessed by ALP mRNA expression and OP secretion. Although signal expression levels in SLA scaffolds on day 8 were lower than those in RPA scaffolds, ALP expression and OP secretion level were significantly higher in SLA scaffolds. This observation might suggest that upregulation of osteogenic signals observed at the early time point of day 4 could also affect the osteoblastic differentiation and further matrix maturation of the cell population. Hence, the architectural condition with a continuous channel orientation in the 3D porous scaffold might be more effective to enhance the osteogenic signal expression and differentiation. It should be emphasized that enhanced expression of osteogenic signals including FGF-2, TGF-β1, and VEGF at this early time point (day 4) in osteogenic supplemented culture condition could facilitate the osteoblastic differentiation of BMSCs in 3D PPF/DEF scaffolds. In addition, a highly permeable microenvironment with continuous pore geometry could be a useful scaffold design parameters for the enhancement of osteogenic signal expression and subsequent osteoblastic differentiation.

5. Conclusions

This investigation of optimal scaffold design parameters including DEF content and pore size in 3D macroporous PPF/DEF composite scaffolds found that osteogenic signal expression of rat BMSCs cultured on the scaffolds could be enhanced by modulating both parameters. A transplanted BMSC population appeared to recognize the changes in the stiffness, permeability, and hydrophilicity of PPF/DEF scaffolds. Osteogenic signal expressions could be stimulated by the modulation of these scaffold design parameters, and downstream osteoblastic differentiation might be enhanced by DEF incorporation with a large pore size (over 500 μm). In addition to the DEF content and pore size, the channel geometry of internal scaffold architecture could also affect the upregulation of osteogenic signal expression. The observations in this study suggest that a continuous open pore geometry in 3D PPF/DEF scaffolds rendered via SLA fabrication may be a more favorable environment to facilitate early osteogenic signal expression and subsequent osteoblastic differentiation