3D Scaffolds of Polycaprolactone/Copper-Doped Bioactive Glass : Architecture Engineering with Additive Manufacturing and Cellular Assessments in a Coculture of Bone Marrow Stem Cells and Endothelial Cells

3D Scaffolds of Polycaprolactone/Copper-Doped Bioactive Glass: Architecture Engineering with Additive Manufacturing and Cellular Assessments in a Coculture of Bone Marrow Stem Cells and Endothelial Cells Xiaoju Wang,* ,†,# Binbin Zhang Molino, ‡,# Sanna Pitkanen, §,∥,# Miina Ojansivu, §,∥ Chunlin Xu, † Markus Hannula, ⊥ Jari Hyttinen, ⊥ Susanna Miettinen, §,∥ Leena Hupa, † and Gordon Wallace ‡ † Johan Gadolin Process Chemistry Centre, Åbo Akademi University, Piispankatu 8, 20500 Turku, Finland ‡ ARC Centre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, University of Wollongong, Northfields Avenue, Wollongong, New South Wales 2522, Australia § Adult Stem Cell Group, BioMediTech, Faculty of Medicine and Health Technology, Tampere University, Arvo Ylpon katu 34, P.O. BOX 100, FI-33014 Tampere, Finland ∥ Research, Development and Innovation Centre, Tampere University Hospital, Arvo Ylpon katu 6, P.O. BOX 2000, FI-33521 Tampere, Finland ⊥ Computational Biophysics and Imaging Group, BioMediTech, Faculty of Medicine and Health Technology, Tampere University, FI-33014 Tampere, Finland * SSupporting Information ABSTRACT: The local delivery of Cu2+ from copper-doped bioactive glass (Cu-BaG) was combined with 3D printing of polycaprolactone (PCL) scaffolds for its potent angiogenic effect in bone tissue engineering. PCL and Cu-BaG were, respectively, dissolved and dispersed in acetone to formulate a moderately homogeneous ink. The PCL/Cu-BaG scaffolds were fabricated via direct ink writing into a cold ethanol bath. The architecture of the printed scaffolds, including strut diameter, strut spacing, and porosity, were investigated and characterized. The PCL/Cu-BaG scaffolds showed a Cu-BaG contentdependent mechanical property, as the compressive Young’s modulus ranged from 7 to 13 MPa at an apparent porosity of 60%. The ion dissolution behavior in simulated body fluid was evaluated, and the hydroxyapatite-like precipitation on the strut surface was confirmed. Furthermore, the cytocompatibility of the PCL/Cu-BaG scaffolds was assessed in human bone marrow stem cell (hBMSC) culture, and a dose-dependent cytotoxicity of Cu2+ was observed. Here, the PCL/BaG scaffold induced the higher expression of late osteogenic genes OSTEOCALCIN and DLX5 in comparison to the PCL scaffold. The doping of Cu2+ in BaG elicited higher expression of the early osteogenic marker gene RUNX2a but decreased the expression of late osteogenic marker genes OSTEOCALCIN and DLX5 in comparison to the PCL/ BaG scaffold, demonstrating the suppressing effect of Cu2+ on osteogenic differentiation of hBMSCs. In a coculture of hBMSCs and human umbilical vein endothelial cells, both the PCL/BaG and PCL/Cu-BaG scaffolds stimulated the formation of a denser tubule network, compared to the PCL scaffold. Meanwhile, only slightly higher gene expression of vWF was observed with the PCL/Cu-BaG scaffold than with the PCL/BaG scaffold, indicating the potent angiogenic effect of the released Cu2+. KEYWORDS: 3D printing, tissue engineering scaffold, angiogenesis, polycaprolactone, copper-doped bioactive glass, gradient porosity, coculture of mesenchymal stem cells and endothelial cells ■INTRODUCTION Synthetic bone grafts that resemble the architecture and composition of the bone extracellular matrix are always sought by biomaterial scientists and surgeons to overcome the limitations of bone autografts and allografts in the treatments of large bone defects in patients. 1−3 The application of tissue engineering (TE) scaffolds has emerged as a strategic approach for bone reconstitution, as it combines the engineered biomaterial structure, soluble/mechanical factors (e.g., release of stimulating bio(macro)molecules/ions), and regenerative cells to provide a structural and physiological support to the spatial tissue growth. 4,5 Received: January 23, 2019 Accepted: July 18, 2019 Published: July 18, 2019 Article pubs.acs.org/journal/abseba Cite This: ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 © 2019 American Chemical Society 4496 DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 This is an open access article published under a Creative Commons Attribution (CC-BY) License, which permits unrestricted use, distribution and reproduction in any medium, provided the author and source are cited. Downloaded via TAMPERE UNIV on October 16, 2019 at 08:10:42 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. As the first man-made biomaterial capable of forming chemical bonding to bone, 6 bioactive glasses (BaGs) have been extensively explored during the past several decades in bone TE because of their osteostimulative capability to guide and stimulate the bone growth. 7−9 However, successful neovascularization (the sprouting of blood vessels) in the bone TE constructs remains challenging. 10,11 Several wellestablished compositions of silicate BaGs, such as 45S5 and S53P4, are capable of inducing angiogenic differentiation of mesenchymal stem cells (MSCs). 12,13 This is attributed to their ionic dissolution products of soluble Ca, Si, and P that can upregulate the genes related to angiogenesis, such as those encoding vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF). 13,14 Moreover, the essential participation of Cu2+ in angiogenesis has been acknowledged: Cu2+ in a proper dosage has a clear proangiogenic function by activating a group of proangiogenic growth factors, including VEGF and platelet-derived growth factor (PDGF), therefore promoting endothelial cell proliferation and tubule formation. 15−17 At present, the local delivery of Cu2+ from the Cudoped BaG to induce angiogenesis has been investigated in various forms, e.g., porous scaffold of Cu-doped mesoporous BaG, 18 composite hydrogel containing Cu-doped mesoporous BaG, 19 functional coating of Cu-doped BaG nanoparticles and electrospun nanofibers of Cu-doped BaG. 20−22 Strategically, the incorporation of Cu-doped BaG in bone TE scaffolds may offer a straightforward approach to address the neovascularization challenge in bone tissue regeneration. The critical aspects to be taken into consideration when designing a bone TE scaffold are the selection of suitable biomaterials to support cell adhesion and proliferation as well as the definitive structural parameters of the scaffold to mimic the native bone tissue. Polycaprolactone (PCL), a bioresorbable polymer extensively used in various biomedical applications, offers excellent material properties and cytocompatibility. 23 The extrusion-based 3D printing (3DP) techniques offer a high-level control of the scaffold architecture, including pore size, distribution, and interconnectivity. 24 Previously, the TE scaffolds for bone regeneration manufactured by the 3DP using PCL or its composites with inorganic minerals such as the microparticles of tricalcium phosphate, 25 hydroxyapatite, and calcium polyphosphate have been studied. 26,27 Conventionally, the melted PCL is used as the carrier phase for the mineral microparticles, and the printed constructs keep the shape fidelity through fast cooling after being extruded through the nozzle in 3DP. In the current study, a solvent-based approach was proposed to prepare the well-dispersed BaG microparticles of Cu-doped S53P4 (S53P4−Cu1) in PCL solution (in acetone) as a homogeneous ink fed in the direct ink writing (DIW), and the printed struts solidified rapidly in the cold ethanol due to the solubility change of PCL, thus facilitating the fabrication of the PCL/Cu-BaG composite scaffolds. This method offers a facile printing as no heating is needed and also extends the possibility to incorporate other bioactive molecules, such as growth factors, within the scaffold during the printing process. Then, the ion dissolution behavior and bioactivity (in terms of supporting the hydroxyapatite precipitation) of the composite scaffolds were evaluated in simulated body fluid (SBF). Furthermore, the cytocompatibility of the composite scaffolds as well as the impact of the ion release on the early osteogenic differentiation of human bone marrow stem cells (hBMSCs) were evaluated in vitro. Finally, we studied whether the Cu2+ released from the 3D scaffolds could promote the angiogenesis in a coculture of hBMSCs and human umbilical vein endothelial cells (HUVECs). The observations are indicative for utilizing the Cudoped BaG aiming at enhancing the vascularization in TE scaffolds. ■MATERIALS AND EXPERIMENTAL Glass Melting of BaGs (S53P4 and S53P4−Cu1) and Preparation of BaG Microparticles. Two BaGs, S53P4 (53% SiO2−4%P2O5−20%CaO−23%Na2O wt %) and S53P4−Cu1 (53% SiO2−4%P2O5−19%CaO−23%Na2O−1%CuO wt %), were prepared using the melt-quenching method. The batches consisted of analytical grade reagents Na2CO3, CaCO3, CaHPO4·2H2O, Cu(NO3)2·2.5H2O (all purchased from Sigma-Aldrich), and Belgian glass quality quartz sand (0.32 mm, Varnia Oy). The batches were melted in a Pt crucible at 1360 °C for 3 h, cast, annealed, crushed, and remelted to ensure the homogeneity. The annealed glass block was crushed and gradually fractionated with a set of woven wire mesh sieves with #mm from 500 μm down to 45 μm (laboratory test sieves, Retsch GmbH) to give several powdered fractions. The finest powder fraction, which could pass through the #mm = 45 μm sieve, was further milled for 10 min using a benchtop planetary ball mill (MinMill, Philips) to obtain the fine particles of BaG for the DIW ink formulation. Scaffolds Fabrication via DIW. A 30 g portion of PCL (Mw= 80 000, Aldrich) was dissolved in 100 mL of acetone overnight in a 50 °C water bath to obtain a viscous solution, which was then used as the carrier phase for the BaG particles in the ink formulation. The BaG particles were homogeneously dispersed into the viscose PCL solution using a planetary centrifugal mixer (ARE-250,Thinky Corporation) at a rotation speed of 2000 rpm. The compositional ratio between the PCL and BaG was adjusted to 4:1, 2:1, or 1:1, according to the content of PCL and BaG in wt %. The scaffold fabrication via DIW was carried out using an adapted Korean Institute for Machinery and Materials (KIMM) Bioplotter. The KIMM Bioplotter software was used to produce a G-code tooling path for printing. Square block models (x= 10 mm, y= 10 mm, and z=1,2, or 4 mm) were designed with the Bioplotter software, and the strut spacing (SS) was defined as either 400 or 800 μm between the center of the struts. For the scaffolds with a gradient porosity, the SS was decreased from the center to the edge of the scaffold, from 800 to 200 μm. The viscous ink as formulated was loaded into a fluid dispensing system (Optimum by Nordson EFD), with the syringe barrel installed onto the KIMM printer and connected to an air pressure regulator. A precision tip made of stainless steel (25GA, Nordson EFD) was used as the dispensing nozzle. The scaffolds were printed layer-by-layer through extruding the ink into a cold ethanol bath. The feed rate was kept constant at 2 mm s−1, and the extrusion pressure was then manually adjusted in the range 1.5−3.0 bar according to the viscosity of the ink for an optimal printing process. The scaffolds were then soaked in absolute ethanol for 24 h, and the ethanol was replaced every 8 h to complete the phase exchange of PCL from acetone to ethanol. Finally, the scaffolds were collected and dried in air before further characterizations. Scaffold Imaging with Optical Microscopy, Scanning Electron Microscopy (SEM), and Micro Computed Tomography (μ-CT). The microscopic features of scaffolds were observed by using a LEICA M205A optical microscope. The reported strut diameter and strut spacing were assessed in triplicate (n= 3). The morphological features of the scaffold strut were characterized by using SEM. To assess the macroporosity, μ-CT was used to analyze the inner structure of the scaffolds. The imaging was performed with an Xradia MicroXCT-400 (Zeiss, Pleasanton, CA) device: 1600 projections were taken with a pixel size of 19.70 μm. The source voltage was set to 80 kV, and the source current was 125 μA. To achieve the desired image quality, a 3 s exposure time was used. The device manufacturer’s XMReconstructor software was used to reconstruct the 3D volumes. Image segmentation was done with Avizo software (Thermo Fisher Scientific, Waltham, MA) using manual thresholding. Approximately 8.2 ×8.2 ×2.4 mm ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4497 volumes were selected, and the porosity, strut size, and pore size of the 3D scaffolds were calculated with the BoneJ plugin in the Fiji program. Thermogravimetric (TG) Analysis. TG analysis of the scaffolds was performed using a differential thermal analysis instrument (Netzsch STA 449F1) in the gas flow of synthetic air at a speed of 100 mL min−1. The temperature ramped up to 850 °C at an elevating rate of 10 °C min−1. The BaG wt % of the scaffold was determined as relative to the residue content in the TG analysis. Mechanical Tests. The compressive response of the printed scaffolds was measured with a Shimadzu EZ-L Universal Mechanical Tester. Scaffolds were compressed at 1 mm min−1and a maximum loading of 500 N was applied. When calculating the strength, the dimensions along the x,y, and z-axes measured with a digital caliper were used. Strain was registered as a function of stress for an applied and increasing force and the measurements were conducted in triplicate to elaborate the compressive modulus based on the stress vs strain curves. Ion Release Study of PCL/Cu-BaG Scaffolds in SBF. The SBF was prepared according to the protocol of Kokubo, and the exact composition is presented in the Supporting Information S1. 28 The 3D scaffolds were immersed in 10 mL of SBF in airtight polyethylene containers that were placed in an incubating orbital shaker held at 37 °C and agitated at 100 rpm. The samples were incubated for a total period of 30 days and sampled at intervals of 6 h, 24 h, 3 days (3 d), 7 d, 14 d, 22 d, and 30 d. At each time point, 0.5 mL of the immersion solution was sampled, and 0.5 mL of fresh SBF was additionally replenished for a continued immersion. The ionic concentrations of Ca, P, Si, and Cu ions in the sampled solution were analyzed with an inductively coupled plasma optical emission spectrometer (ICP-OES) (Optima 5300 DV, PerkinElmer, Shelton, CT). For the ICP-OES measurements, the sampled aliquots were diluted 10 times with deionized H2O. At the end of the immersion test, the scaffolds were carefully collected, washed extensively with ethanol, and dried in the air. The surface morphology and elemental analysis of the scaffolds were characterized with an SEMEDXA instrument (EDXA, LEO Gemini 1530 with a Thermo Scientific UltraDry Silicon Drift Detector, X-ray detector by Thermo Scientific). Isolation and Characterization of hBMSCs and HUVECs. The hBMSCs were isolated from a bone marrow aspirate sample obtained from a surgical procedure at the Department of Orthopedics and Traumatology, Tampere University Hospital, with the patient’s consent. The hBMSCs used in the study were harvested from a female donor of 80 years of age. The study was conducted in accordance with the Ethics Committee of the Pirkanmaa Hospital District, Tampere (R15174). hBMSCs were isolated using centrifugation through a Ficoll gradient. First, the bone marrow aspirate was suspended in Dulbecco's phosphate buffered saline (DPBS), and the solution was pushed through a 100 μm cell strainer. The bone marrow aspirate solution was centrifuged, and the fat layer was removed. Thereafter, the bone marrow aspirate solution was pipetted carefully on top of a Ficoll gradient (Histopaque-1077; 1.077 g mL−1; Sigma-Aldrich; St. Louis, MO). A 2.6 mL portion of Ficoll (Histopaque; Sigma-Aldrich) per 1 mL of bone marrow aspirate sample was used. The bone marrow aspirate solution was centrifuged for 20 min at 800g, and the mononucleated hBMSCs were harvested from an interphase between the Ficoll and plasma phases. The cells were washed twice with 5 mL of MEM Alpha medium (Thermo Fisher Scientific, Waltham, MA) per 1 mL of collected interphase, and the suspension was centrifuged for 15 min at 400g. The cell pellet was suspended in basic medium (BM) consisting of MEM Alpha medium (Thermo Fisher Scientific), 5% human serum (HS; BioWest, Nuaille, France), and 1% antibiotics (100 UmL −1penicillin; 100 U mL−1streptomycin; Lonza, Basel, Switzerland) with 5 ng mL−1human FGF-2 (Miltenyi Biotec; Bergisch Gladbach, Germany). Isolated hBMSCs were expanded in BM at 37 °C in 5% CO2, and medium was changed twice per week. Cells were detached with TrypLE Select (Thermo Fisher Scientific). The experiments were carried out at passage 3. HUVECs were extracted from the umbilical cord acquired from scheduled Cesarean section at the Department of Obstetrics and Gynecology, Tampere University Hospital, with the donor’s consent according to Hamilton et al. 29 The study was conducted in accordance with the Ethics Committee of the Pirkanmaa Hospital District, Tampere (R13019). Briefly, the cord was separated from the placenta; the umbilical vein was cannulated with a 20G needle, and the needle was secured by clamping the cord over the needle with a clamp. The vein was perfused with PBS to wash out blood, and then, the opposing end of the umbilical vein was clamped. Subsequently, the vein was infused with collagenase II (Sigma). The umbilical cord was incubated in a water bath at 37 °C for 15 min. After incubation, the collagenase solution containing HUVECs was flushed from the cord into a 50 mL polypropylene tube. The cells were centrifuged at 1200 rpm for 6 min and resuspended in EGM-2 BulletKit (Lonza) medium supplemented with 2% HS and seeded into 25 cm3flasks. The HUVECs were cultured at 37 °Cin5%CO 2, and the medium was changed twice per week. Cells were detached with TrypLE Select (Thermo Fisher Scientific). The experiments were carried out at passage 3. To verify the mesenchymal origin of the hBMSCs and the endothelial phenotype of HUVECs, surface marker expression was characterized by flow cytometry (FACSAria; BD Biosciences, Erembodegem, Belgium) as described previously, as documented in Supporting Information S2. 30 Cell Seeding and Culture on Scaffolds. In the cell culture study, the PCL scaffolds were used as a control to the composite scaffolds; the scaffolds PCL/S53P4 were also included as a Cu-free control to the scaffolds of PCL/S53P4−Cu1. All the tested scaffolds (10 mm ×10 mm ×1 mm) were sterilized by incubating the scaffolds 2 ×10 min in 70% ethanol, and then, the scaffolds were left to dry for 2 h in a biosafety cabinet. Thereafter, the scaffolds were incubated in BM for 48 h prior to cell seeding. All scaffolds’ability to support hBMSC viability, proliferation, and early osteogenic differentiation was evaluated in hBMSC culture. hBMSCs were seeded, 50 000 per scaffold, in a 50 μL OM (BM with 200 μM ascorbic acid 2-phosphate (Sigma-Aldrich), 10 mM βglycerophosphate (Sigma-Aldrich), and 5 nM dexamethasone (SigmaAldrich)) drop on to the scaffolds. Cells were allowed to attach for 3 h before adding 2 mL of OM per well. The medium was changed to a fresh one twice per week during the experiment. The scaffolds’ability to support vascularization was assessed in the coculture experiment with hBMSCs and HUVECs. Only PCL, PCL/ S53P4 = 4:1, and PCL/S53P4−Cu1 = 4:1 scaffolds were used in the coculture experiment due to the observed cytotoxicity of higher S53P4−Cu1 content with the other compositional ratios. The originally printed scaffolds (10 mm ×10 mm ×1 mm and d= 400 μm) were cut into four quarters for further use. hBMSCs were seeded, 20 000 per scaffold, and cultured for 6 d before seeding an equal amount of HUVECs. After seeding the HUVECs, the medium was changed to EGM-2 (Lonza, Basel, Switzerland). hBMSCs alone were used as a control group and cultured in OM throughout the experiment. Cell Viability and Proliferation. Cell viability was evaluated qualitatively by staining the hBMSCs with fluorescent live/deadstaining probes (Thermo Fisher Scientific) after 1 d and 14 d of culture. The samples were incubated for 45 min at room temperature in a mixture of 0.5 μM calcein-AM and 0.25 μM ethidium homodimer-1. Images of viable cells (green fluorescence) and dead cells (red fluorescence) were taken using an Olympus IX51 phase contrast microscope with fluorescence optics and Olympus DP30BW camera (Olympus, Tokyo, Japan). Cell viability was quantitatively analyzed by a lactate dehydrogenase (LDH) activity assay (Abcam, Cambridge, UK). The LDH reduces NAD to NADH, which then interacts with a specific probe to produce a color. The medium samples were collected at every medium changeand stored at −20 °C until analysis. The analysis was conducted according to the manufacturer’s protocol. After 30 min of incubation at room temperature, the absorbance, i.e., the intensity of the color, was determined at 450 nm with a microplate reader (Victor 1420 Multilabel Counter; Wallac; Turku, Finland). Cell number was determined quantitatively after 7 d and 14 d of hBMSC culture by analyzing the total amount of DNA by a CyQUANT cell proliferation assay kit (Thermo Fisher Scientific) as reported previously. 31 CyQUANT GR dye emits fluorescence when bound to nucleic acids. Samples were analyzed after two freeze−thaw cycles, and ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4498 fluorescence was measured at 480/520 nm with a microplate reader (Victor 1420, Wallac). Cu2+ Concentration in Culture Medium of hBMSCs. In the culture of hBMSCs, the medium samples were collected from 3 parallel wells at 2 d, 6 d, and 13 d while the medium was changed and pooled as one sample per time point. The concentration of Cu2+ in the sample was analyzed by the ICP-OES analysis using the same protocol as described in the earlier Ion Release Study of PCL/Cu-BaG Scaffolds in SBF section. Alkaline Phosphatase Activity. Alkaline phosphatase (ALP) activity was determined after 7 d and 14 d of hBMSC culture as described previously. 31 The ALP activity was determined from the same cell lysates as the total DNA content. Absorbance was measured at 405 nm (Victor 1420, Wallac). Quantitative Real-Time PCR. The relative expression of endothelial marker genes PECAM and vWF was evaluated with quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) at 6 d (hBMSCs only), 11 d, and 20 d (hBMSCs only and hBMSCs+HUVECs coculture) time points as previously described. 32 The relative expression of osteogenic genes RUNX2a,OSTEOCALCIN, and DLX5 was analyzed with qRT-PCR at 11 d and 20 d time points (hBMSCs only and hBMSCs+HUVECs coculture). The data was normalized to the expression of a housekeeping gene RPLP0 (human acidic ribosomal phosphoprotein P0), and the calculations were conducted using a previously described mathematical model. 33 For vWF and PECAM, the QuantiTect primer assays were used (Qiagen, Hilden, Germany). The primer sequences and accession numbers for RPLP0 and osteogenic genes RUNX2a,OSTERIX, and DLX5 are listed in Table 1. The qRT-PCR mixture contained cDNA, primers, and SYBR Green PCR Master Mix (Applied Biosystems). The reactions were carried out with an ABIPRISM 7300 sequence detection system, and the results were analyzed with AbiPrism 7300 sequence detection system software (Applied Biosystems). Immunocytochemical Staining. The expression of endothelial marker proteins vWF and CD31 (the product of the PECAM gene) was characterized with immunocytochemical staining after 13 d and 20 d of culture (hBMSCs only and hBMSCs+HUVECs coculture). The protocol was conducted as previously described. 32 The following primary antibodies were used: CD31 monoclonal mouse antihuman antibody (dilution 1:20, Dako, Agilent, Santa Clara, CA) and rabbit polyclonal anti-vWF antibody (dilution 1:100, Abcam). The primary antibodies were incubated overnight at +4 °C. The secondary antibodies goat antimouse Alexa Fluor 488 (for CD31 staining; dilution 1:200; Thermo Fisher Scientific) and goat antirabbit Alexa Fluor 568 (for vWF staining; dilution 1:500; Thermo Fisher Scientific) were incubated for 1 h at room temperature. The images were acquired with an Olympus IX51 phase contrast microscope with fluorescence optics and Olympus DP30BW camera. Statistical Testing. Statistical testing was conducted with SPSS version 23 (IBM, Armonk, NY) using a nonparametric test due to small sample size. The effects of the scaffolds on cell viability, cell amount, ALP activity, and gene expression were compared using the Kruskal− Wallis test with Mann−Whitney U post hoc test and Bonferroni correction. The results were considered significant when p< 0.05. The cell culture experiments were repeated with 1 donor line with 3 or 4 parallel samples (n=3orn= 4). However, most likely due to the small sample size, no significant differences between the test groups were detected. ■RESULTS Scaffolds Fabrication via DIW and Morphological Features. In this work, we used a viscous solution of PCL dissolved in acetone as the carrier phase to disperse the BaG microparticles to formulate a homogeneous ink. Up to 50% of BaGs calculated as the dry weight in PCL/BaG could be mixed into the ink with the aid of orbital mixing. In DIW, the printing nozzle was immersed in a cold ethanol bath, and the printed struts solidified as extruded due to the solubility change of PCL in acetone (soluble) and in ethanol (insoluble). Figure 1 presents the optical and morphological images of the PCL/S53P4−Cu1 scaffolds with various compositional ratios at 4:1, 2:1, or 1:1 (wt %). Overall, the well-defined layouts of the struts were obtained in all the scaffolds, as seen in the first two columns. After the solvent exchange and drying, the printed scaffolds shrank compared to the designed model, more in the zaxis than the xand y-axes (Table 2). The gravitational force was Table 1. Primer Sequences and Accession Numbers of Genes Analyzed by qRT-PCR name 5′-sequences-3′product size (bp) accession number hRPLP0 Frw AATCTCCAGGGGCACCATT 70 NM_001002 Rev CGCTGGCTCCCACTTTGT 70 NM_001002 hOSTEOCALCIN Frw AGCAAAGGTGCAGCCTTTGT 63 NM_000711 Rev GCGCCTGGGTCTCTTCACT 63 NM_000711 hRUNX2a Frw CTTCATTCGCCTCACAAACAAC 62 NM_001024630.3 Rev TCCTCCTGGAGAAAGTTTGCA 62 NM_001024630.3 hDLX5 Frw ACCATCCGTCTCAGGAATCG 75 NM_005221.5 Rev CCCCCGTAGGGCTGTAGTAGT 75 NM_005221.5 Figure 1. Optical images (column i) and SEM images of the scaffolds (column i and column ii) of PCL (d= 400 μm) (row a), PCL:S53P4− Cu1 = 4:1 (d= 400 μm) (row b), PCL:S53P4−Cu1 = 2:1 (d= 400 μm) (row c), and PCL:S53P4−Cu1 = 1:1 (d= 400 μm) (row d). Scale bars: 2 mm (i), 200 μm (ii), 2 μm (iii). ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4499 assumed to cause the higher compaction along the z-axis during the solidification. Also, the content of S53P4−Cu1 in the composite affected the shrinkage rate. As seen from Table 2, less shrinkage occurred in the composite containing more BaG microparticles since the shrinkage is mainly caused by the condensation of PCL polymer after the removal of organic solvent. The shrinkage decreased from 13% to 14% for the PCL scaffold to 4% for the composite scaffold of PCL/S53P4−Cu1 = 1:1 on xand y-axes while for the z-axis the difference in the shrinkage was from 20% compared to 12%. The strut diameter and strut spacing were confined within the range 200−230 μm and 300−340 μm, respectively, after the shrinkage upon solidification. As the content of S53P4−Cu1 in the composites increased, the composites showed a coarser surface morphology as revealed by the SEM images (Figure 1). The increased surface roughness in the composites containing more S53P4−Cu1 microparticles was probably due to the presence of more BaG microparticles that protruded the PCL matrix to a larger extent. In the scaffolds of PCL/S53P4−Cu1 = 2:1 and 1:1, small cavities and even microcracks were seen on the surface, indicating a disruption to the continuous phase of PCL in the struts. TG analysis confirmed the compositional ratio of S53P4− Cu1 in the composite scaffolds, as displayed in Supporting Information S3 as well as listed in Table 2. The residual content was highly consistent with the inorganic content of BaG in the ink dispersion. Also, the cross-section image of the PCL/ S53P4−Cu1 = 2:1 scaffold revealed a moderately homogeneous distribution of BaG microparticles through the PCL matrix (as seen in the SEM image in Supporting Information S3). This indicates that the dispersion of BaG microparticles in viscous PCL solution was rather homogeneous, which consequently facilitated the DIW fabrication of the scaffolds with good textural integrity and high resolution. Compressive Response of the 3D Scaffolds of PCL and PCL/S53P4−Cu1. In a compressive mode, the mechanical responses of the PCL and PCL/S53P4−Cu1 scaffolds were analyzed. The stress−strain response curves for the scaffolds of PCL and PCL/S53P4−Cu1 = 4:1 and 2:1 with the SS = 400 μm are displayed in Figure 2a. The observed stress−strain response of the scaffolds is typical for the highly porous PCL-based scaffolds reported elsewhere. 26,34,35 Their stress−strain curves are characterized by three different regions: a linear region at lower strain values, suggesting an initial rigid mechanical response, associated with elastic behavior of the scaffolds; a region with lower stiffness; and lastly, a region where a rise of stress with increasing strain is recorded, which is related to densification of the porous structure. Compared with the PCL scaffold, all the composite scaffolds exhibited reinforced mechanical properties as the strains were yielded at higher stress values. The compressive Young’s moduli Eof the scaffolds calculated from the slope of the stress−strain curves are displayed in Figure 2b with respect to the compositional ratio between the PCL and S53P4−Cu1. The PCL scaffold showed an Evalue of 5.9 MPa and a yield stress σy(at 0.2% strain) of 0.64 MPa. Among the compositional ratios at 4:1, 2:1, and 1:1 for PCL/S53P4−Cu1, the scaffold of PCL/S53P4−Cu1 = 2:1 exhibited the highest Evalue of 12.5 MPa and yield stress σy(at Table 2. Structural Parameters of 3D Printed Scaffolds of PCL/S53P4−Cu1 scaffold dimension (mm ×mm ×mm) mean ±SEM axial shrinkage (%) strut diameter (μm) mean ±SEM spacing between struts (μm) mean ±SEM BaG content in composite as revealed by TG a PCL 8.65 (±0.03) ×8.72 (±0.02) ×3.20 (±0.03) 13−14% on xand y-axes; ∼20% on z-axis 229 (±16) 297 (±9) PCL:S53P4−Cu1 = 4:1 8.86 (±0.07) ×8.88 (±0.05) ×3.28 (±0.06) ∼11% on xand y-axes; ∼18% on z-axis 207 (±8) 339 (±6) 20.2% PCL:S53P4−Cu1 = 2:1 9.20 (±0.03) ×9.21 (±0.03) ×3.41 (±0.14) ∼8% on xand y-axes; ∼15% on z-axis 232 (±20) 323 (±12) 32.0% PCL:S53P4−Cu1 = 1:1 9.58 (±0.01) ×9.61 (±0.01) ×3.49 (±0.01) ∼4% on xand y-axes; ∼12% on z-axis 231 (±3) 337 (±4) 57.4% a Determined from the weight loss in the TG analysis curves as shown in Supporting Information S3. Figure 2. (a) Stress−strain curves for the 3D scaffolds. (b) Compressive Young’s modulus vs the compositional ratio of PCL/ S53P4−Cu1 in the scaffold. Error bar: SEM. ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4500 0.2% strain) of 1.54 MPa, more than 2 times higher than the PCL scaffold of similar porosity. As revealed by the μ-CT measurements, all the scaffolds printed with SS = 400 μm gave a volume fraction of approximately 0.39 as well as an apparent porosity of 60%. The scaffold of PCL/S53P4−Cu1 = 4:1 demonstrated a similar compressive modulus to the PCL scaffold with an Evalue of 6.29 MPa, while the scaffold of PCL/ S53P4−Cu1 = 1:1 demonstrated an Evalue of 7.19 MPa, which is lower than that of the scaffold of PCL/S53P4−Cu1 = 2:1. The reduction in mechanical properties of the scaffold of PCL/ S53P4−Cu1 = 1:1 can be attributed to the formation of BaG agglomerates, which may result in a heterogeneous dispersion of the inorganic fillers in the PCL matrix, as observed in the SEM images (Figure 1d,iii). Porosity Control on the Printed Scaffold in DIW. As revealed by the μ-CT measurements, the scaffolds printed with SS = 400 μm gave a volume fraction of about 0.393 as well as an apparent porosity of 60.64%. To demonstrate the precision control of the DIW process over the scaffold porosity with the developed ink system, the scaffolds with the SS = 800 μm (shown in Figure 3A) or with a dense outward gradient porosity (shown in Figure 3B) were, respectively, printed using the ink of the compositional ratio at PCL/S53P4−Cu1 = 2:1. It showed that, after solidification, the scaffolds with SS = 800 μm had the same strut diameters of 232 μm as the scaffolds with SS = 400 μm, but the SS remained around 610 μm. The μ-CT measurements gave a volume fraction of 0.274 with an apparent porosity of 72.53%. In the design of dense outward gradient porosity, the strut spacing was set with a linear change from edge to center on both xand y-axes. As displayed in Figure 3B, this resulted in an almost linear change of the SS value from 165 to 663 μm in the scaffold after shrinkage. Figure 3. (A) Optical images of the scaffold of PCL/S53P4−Cu1 = 2:1 with strut spacing at d= 800 μm. (B) Optical images of the scaffold of PCL/ S53P4−Cu1 = 2:1 with a dense outward gradient porosity. (C) μ-CT images of the scaffolds with different porosity parameter. (D) Young’s modulus with respect to various porosity in scaffolds of PCL and PCL/S53P4−Cu1 = 2:1. ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4501 For the scaffolds with different porosity parameters, the μ-CT image on the left in Figure 3C demonstrates the 3D spatial space inside the printed scaffolds, and the image on the right presents the 2D distribution of struts in the scaffolds. In general, the 3D reconstruction of the scaffolds displayed as the “replicates”in accordance to the CAD designs, indicating good controllability of the developed DIW process. The pore architecture parameters in these scaffolds are summarized with respect to the varied porosity in Table 3. The Young’s moduli of these scaffolds were also compared with respect to the porosity in Table 3. Pore Architecture Parameters in the Printed Scaffold of PCL/S53P4−Cu1 = 2:1 scaffold PCL:S53P4−Cu1 = 2:1 strut diameter (μm) mean ±SEM spacing between struts (μm) mean ±SEM volume fraction measured by μ-CT porosity measured by μ-CT (%) d= 400 μm 232 (±20) 323 (±12) 0.393 60.64 d= 800 μm 232 (±20) 610 0.274 72.53 gradient porosity 232 (±20) gradient change 0.261 73.92 Figure 4. (A) SEM image of the strut surface in the scaffold of PCL/S53P4−Cu1 = 4:1 after 30 days of immersion in SBF in various resolutions. (B) SEM image of the strut surface in scaffold PCL/S53P4−Cu1 = 2:1 after 30 days of immersion in SBF in various resolutions. (C) SEM image of the strut surface in scaffold of PCL/S53P4−Cu1 = 1:1 after 30 days of immersion in SBF in various resolutions. (D) SEM image of the strut surface in scaffold of PCL/S53P4−Cu1 = 2:1 at various immersion time points. (E) EDXA performed on the precipitates on the strut surface in scaffold of PCL/S53P4− Cu1 = 2:1 after 30 days of immersion in SBF. ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4502 Figure 3D. Compared with the scaffold with SS = 400 μm, the compressive modulus decreased for both the scaffold with SS = 800 μm and the one with a gradient porosity, as the porosity in both types of scaffolds increased to ∼73%. An increase of the total porous volume from 10% to 20% may result in a factor of 4 decrease in the mechanical strength. 36 Ion Dissolution Profiles and Bioactivity of PCL/Cu-BaG Scaffolds in SBF in Vitro.The bioactivity and ion dissolution behaviors of the scaffolds of PCL/S53P4−Cu1 (SS = 400 μm) at various compositional ratios were evaluated in SBF for a total duration of 30 d. Figure 4A−C, respectively, displays the strut surface morphology in the scaffolds of PCL/S53P4−Cu1 = 4:1, 2:1, and 1:1 after static immersion in SBF for 30 d. In the scaffold of PCL/S53P4−Cu1 = 4:1, only small aggregates of calcium phosphate phase (CaP) decorated on the strut surface (Figure 4A), whereas heavy precipitation of CaP formed layers covered the strut surface in the scaffolds of both PCL/S53P4−Cu1 = 2:1 and PCL/S53P4−Cu1=1:1(Figure 4B,C). The CaP precipitate demonstrated a characteristic cauliflower-like morphology consisting of numerous nanoflakes and gave a Ca/P = 1.56 as measured by the EDXA (Figure 4E), which was close to the Ca/P ratio of 1.67 in natural bone minerals, hydroxyapatite. Figure 4D reveals the morphological change of the strut surface in the scaffold of PCL/S53P4−Cu1 = 2:1 at varioustimepointsduringtheSBFimmersion.Notable precipitation of the CaP phase on the strut surface was only observed after 7 d of immersion. By analyzing the ion concentrations in SBF at various time points, the ion dissolution profiles of the scaffolds were acquired. Figure 5 displays the ion dissolution profiles for PCL/S53P4− Cu1 scaffolds with compositional ratios at 4:1, 2:1, and 1:1, all with dimensions of 10 mm (x)×10 mm (y)×2mm(z) and SS = 400 μmdefined in DIW. With respect to the dissolution of Si, in all three types of scaffolds a burst release was initially detected within 3 d of immersion, which was most likely associated with the initial dissolution of Cu-BaG particles exposed on the most outer surface. After 3 d, the Cu-BaG dissolution slowed down when the ion diffusion from the inner of the particles might have been hindered by the formed silica-rich layer. Overall, a rather low amount of Si was detected for the scaffold of PCL/S53P4− Cu1 = 4:1 due to the low content of S53P4−Cu1. For the scaffold of PCL/S53P4−Cu1 = 2:1, a continuous release of Si at an almost constant rate was detected for the rest of the immersion period, whereas the Si released from the scaffold of PCL/S53P4−Cu1 = 1:1 reached a saturation level at ∼60 mg L−1in SBF after 7 d of immersion. With respect to the dissolution of Cu, in general the release trend was similar to that of Si. A higher concentration of Cu2+ was observed in SBF from the scaffold of PCL/S53P4−Cu1 = 1:1 due to the highest burst release in the initial dissolution. The ion concentration profiles of Ca and P in SBF are reflected both by the ion dissolution from the scaffold and by the precipitation of CaP phase on the strut surface. For the scaffold of PCL/S53P4−Cu1 = 4:1, apparent precipitation of CaP was only seen after 14 d of immersion due to the low concentration of Ca2+ accumulated in SBF, as indicated by the steep ramp in P concentration. For the scaffolds of PCL/S53P4−Cu1 = 2:1 and PCL/S53P4−Cu1 = 1:1, the CaP precipitation was readily indicated after 1 d of immersion by the steep ramp of P concentration in SBF as shown in Figure 5. Figure 5. Ion concentration profiles of Ca, P, Si, and Cu ions in SBF with static immersion of PCL/S43P4-Cu1 scaffolds at various compositional ratios. ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4503 hBMSC Culture on the Scaffolds. To determine the viability of hBMSCs on the scaffolds, a qualitative live/dead staining as well as a quantitative LDH activity assay were conducted. As made evident in Figure 6A, the cells were viable on all the materials after 14 d of culture, and the number of dead cells was negligible. Increasing the content of S53P4 in the composite seemed to induce hBMSC proliferation more in comparison to the PCL control. However, apart from the scaffold of PCL/S53P4−Cu1 = 4:1, the high content of S53P4− Cu1 had a clear negative effect on the cell amount. Cells release LDH enzyme to the culture medium upon cell death associated membrane rupture, making LDH a good quantitative indicator of cell viability. Figure 6B shows the LDH activity levels of the culture medium after 2 d, 6 d, 9 d, and 13 d of culture. Since the medium was changed at each of these time points, the LDH activity values are not cumulative. In general, the LDH activities remained at a relatively low and constant level with all the scaffolds until 9 d, indicating low cytotoxicity of the studied scaffolds. At 13 d the values increased slightly, which is likely due to the high cell density at this point, as observed in the live/dead staining (Figure 6A). Unexpectedly, the amount of Cu2+ released from the scaffolds of PCL/S53P4−Cu1 = 2:1 and PCL/S53P4−Cu1 = 1:1 did not induce elevated LDH levels in the culture medium, as expected based on the live/dead staining. Still, cell amount in the S53P4−Cu1-containing scaffolds was clearly less than that in the other samples, thus possibly explaining this observation. The cell proliferation was quantitatively evaluated at time points of 7 d and 14 d as shown in Figure 6C. The integration of S53P4 in the composite showed an increasing effect on the hBMSC proliferation at both time points and with all the studied BaG contents. Regarding the role of Cu2+ released from S53P4− Cu1, the cell proliferation assay was in line with the live/dead staining, indicating the inhibitory effect of the increased concentration of Cu2+ on the growth of hBMSCs. Still, despite the very low cell amount on the scaffold of PCL/S53P4−Cu1 = Figure 6. Results concerning viability and proliferation of hBMSCs in scaffolds. (A) Representative images of live/dead staining at the 14 d time point. Live cells are stained green and dead cells red.Scale bars: 500 μm. (B) LDH activity results at 2 d, 6 d, 9 d, and 13 d time points (n= 3, mean + SD). (C) Total DNA amount in samples at 7 d and 14 d time points, indicating hBMSC proliferation (n= 4, mean + SD). (D) ICP-OES analysis results on Cu2+ concentration in culture medium at 0 d, 2 d, 6 d, and 13 d time points (n= 1). ACS Biomaterials Science & Engineering Article DOI: 10.1021/acsbiomaterials.9b00105 ACS Biomater. Sci. Eng. 2019, 5, 4496−4510 4504