Characterisation and in vitro and in vivo evaluation of supercritical-CO2-foamed β-TCP/PLCL composites for bone applications

S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes 35 www.ecmjournal.org Abstract Most synthetic bone grafts are either hard and brittle ceramics or paste-like materials that differ in applicability from the gold standard autologous bone graft, which restricts their widespread use. Therefore, the aim of the study was to develop an elastic, highly porous and biodegradable β-tricalciumphosphate/poly(Llactide-co-ε-caprolactone) (β-TCP/PLCL) composite for bone applications using supercritical CO2 foaming. Ability to support osteogenic differentiation was tested in human adipose stem cell (hASC) culture for 21 d. Biocompatibility was evaluated for 24 weeks in a rabbit femur-defect model. Foamed composites had a high ceramic content (50 wt%) and porosity (65-67 %). After 50 % compression, in an aqueous environment at 37 °C, tested samples returned to 95 % of their original height. Hydrolytic degradation of β-TCP/PLCL composite, during the 24-week follow-up, was very similar to that of porous PLCL scaffold both in vitro and in vivo. Osteogenic differentiation of hASCs was demonstrated by alkaline phosphatase activity analysis, alizarin red staining, soluble collagen analysis, immunocytochemical staining and qRT-PCR. In vitro, hASCs formed a pronounced mineralised collagen matrix. A rabbit femur defect model confirmed biocompatibility of the composite. According to histological Masson-Goldner’s trichrome staining and micro-computed tomography, β-TCP/PLCL composite did not elicit infection, formation of fibrous capsule or cysts. Finally, native bone tissue at 4 weeks was already able to grow on and in the β-TCP/PLCL composite. The elastic and highly porous β-TCP/PLCL composite is a promising bone substitute because it is osteoconductive and easy-to-use and mould intraoperatively. Keywords: Bone substitute, composite, adipose stem cells, osteogenic differentiation, rabbit distal femur defect, β-tricalciumphosphate, poly(L-lactide-co-ε-caprolactone). *Address for correspondence: Sanna Pitkänen, BioMediTech, Faculty of Medicine and Health Technology, FI-33014 Tampere University, Tampere, Finland. Telephone number: +358 401901789 Email: [email protected] Copyright policy: This article is distributed in accordance with Creative Commons Attribution Licence (http://creativecommons.org/licenses/by-sa/4.0/). European Cells and Materials Vol. 38 2019 (pages 35-50) DOI: 10.22203/eCM.v038a04 ISSN 1473-2262 CHARACTERISATION AND IN VITRO AND IN VIVO EVALUATION OF SUPERCRITICAL-CO2-FOAMED Β-TCP/PLCL COMPOSITES FOR BONE APPLICATIONS S. Pitkänen1,2,*, K. Paakinaho1,3,4, H. Pihlman5, N. Ahola3, M. Hannula6, S. Asikainen7, M. Manninen4, M. Morelius5, P. Keränen5, J. Hyttinen6, M. Kellomäki3, O. Laitinen-Vapaavuori5 and S. Miettinen1,2 1 Adult Stem Cell Group, BioMediTech, Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland 2 Research, Development and Innovation Centre, Tampere University Hospital, Tampere, Finland 3 Biomaterials and Tissue Engineering Group, BioMediTech, Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland 4 Orton Orthopaedic Hospital, Helsinki, Finland 5 Department of Equine and Small Animal Medicine, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland 6 Computational Biophysics and Imaging Group, BioMediTech, Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland 7 Polymer Technology Research Group, Department of Chemical and Metallurgical Engineering, Aalto University, Espoo, Finland Introduction Bone is the second most commonly transplanted tissue, with approximately 2.2 million orthopaedic surgeries carried out worldwide annually (Campana et al., 2014; Kinaci et al., 2014; Van der Stok et al., 2011). As people age, this number is expected to rise. Among bone transplants, autologous bone is considered 36 www.ecmjournal.org S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes the gold standard, while allogenic bone and bone substitutes are also used. Important patient groups affected by problems concerning bone transplants are young children, who do not have enough bone tissue to be harvested, and elderly people, whose bone quality is weak. In addition, several problems are related to the use of autologous bone transplants, such as morbidity of the harvesting site, insufficiency and viability of the harvested bone and infection (Freeman and McNamara, 2017; Van der Stok et al., 2011). Furthermore, even though allograft bone may provide enough bone, transplanted tissue might be rejected; also, proteins and osteoinductive factors might be denaturated following sterilisation (Boyce et al., 1999). Moreover, when using allografts, disease transmission and infection are recurring problems, as 18 % of donated femoral heads have been shown to be affected by bacterial or fungal infections (Barbour and King, 2003; Boyce et al., 1999; Freeman and McNamara, 2017). Therefore, there is an evident and growing need for safe, synthetic bone transplants and tissue engineering strategies. Even though various biomaterials have been proposed for bone replacement, the study and development of these structures have rarely proceeded to clinical applications (de Misquita et al., 2016). Before a bone substitute can become a clinically used product, its efficacy and safety has to be demonstrated both in vitro and in vivo as well as in a clinical trial. According to the regulation (EU) 2017/745 of the European Parliament and of the Council of 5 April 2017 on medical devices, a product cannot receive a CE mark without clinically proven data (Web ref. 1). An ideal bone substitute should be osteoconductive or osteoinductive, biocompatible, biodegradable (Van der Stok et al., 2011) and highly porous, with a pore size larger than 250 µm to allow cell and tissue ingrowth (Mathieu et al., 2005; Turnbull et al., 2017; Zadpoor, 2015) and vascularisation (Rouwkema et al., 2008). Furthermore, the implant should not induce the formation of fibrotic tissue (Van der Stok et al., 2011) or infection. Cost-effective, large-scale manufacturing is also important when designing a new bone substitute for clinical use. Ceramics, such as hydroxyapatite and other calcium phosphates, are highly compatible with bone tissue and, therefore, the most widely used synthetic bone substitute materials for treating large bone defects. Among calcium phosphates, β-tricalciumphosphate (β-TCP) induces significantly more bone in an ectopic site in vivo as compared to hydroxyapatite (Chatterjea et al., 2013; Yuan et al., 2010) and, therefore, β-TCP was chosen as the calcium phosphate phase of the composite. However, there has been conflicting results concerning the osteoinductive properties of β-TCP and speculation that they might vary according to the surface properties of the material (Bohner and Miron, 2019; Duan et al., 2018). Calcium phosphate materials are hard and brittle, making them difficult to shape intraoperatively and implant. Synthetic polymers such as polylactide, polyglycolide, poly-εcaprolactone and their copolymers are biocompatible and support osteogenic differentiation of stem cells in vitro (Campana et al., 2014; Jeong et al., 2008; Temple et al., 2014; Wang et al., 2016). Poly(L-lactide-co-εcaprolactone) (PLCL) is a copolymer of L-lactide and ε-caprolactone with highly desirable characteristics as an implant material: elasticity, flexibility, high tensile strength, controllable degradation rate and good biocompatibility (Holmbom et al., 2005; Wang et al., 2016). PLCL scaffolds have been used in bone tissue engineering although, as PLCL is inert, hydrophobic and lacks biological recognition sites, the interaction with tissue and cells is not good enough either in vitro or in vivo (Jeong et al., 2008; Wang et al., 2016). By using supercritical CO2 (scCO2) processing, it was possible to create a homogenous, porous composite of the bioactive β-TCP and elastic PLCL. scCO2 foaming was chosen as the processing technique as it is easy, costeffective and non-toxic. Furthermore, the technique enables the production of polymeric structures with different pore sizes and total porosities. Briefly, CO2 becomes a supercritical fluid above its critical temperature and pressure (31.10 °C, 73.9 bar), adapting characteristics between a liquid and a gas. Polymeric materials can be dissolved in scCO2 and, as the pressure is decreased controllably, the CO2 gas nucleation and expansion creates the wanted porous structure inside the polymeric material (Mathieu et al., 2005). Multipotent human adipose stem cells (hASCs) are easily available. Furthermore, hASCs have been widely studied in the field of bone tissue engineering in vitro (Kyllönen et al., 2013a; Kyllönen et al., 2013b; Ojansivu et al., 2015; Tirkkonen et al., 2013; Vanhatupa et al., 2015) and in vivo (Jeon et al., 2008; Wilson et al., 2012). Therefore, hASCs are a relevant cell type for studying both the cytocompatibility of a potential bone substitute and the ability of a scaffold to support osteogenic differentiation in vitro. The aim of the current study was to overcome the major limitations of previous bone-applicationintended materials, to develop an elastic, highly porous, biodegradable and biocompatible β-TCP/ PLCL composite for bone applications by using scCO2 and to test its performance both in vitro and in vivo. Furthermore, the aim was to create a composite mimicking the mechanical properties of cartilaginous soft callus, which serves as the natural scaffold for bone regeneration during bone healing. For the first time, it was shown that scCO2 processing of β-TCP/ PLCL composite with a high ceramic content (50 wt%) could be used to produce scaffolds with porosity as high as 65-67 %. Furthermore, β-TCP granules were brought to the surface of the composites by a dynamic compression treatment. In vitro studies with hASCs demonstrated both cytocompatibility and osteogenic capacity of the scaffold. Furthermore, biocompatibility, osteoconductivity and bone tissue ingrowth were shown in vivo in a rabbit femur defect model. S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes 37 www.ecmjournal.org Materials and Methods Composite manufacturing and characterisation Composites were manufactured by melt-mixing PLCL-polymer (70L/30CL; Purasorb PLC7015, Corbion Purac Biomaterials, Gorinchem, the Netherlands) with 50 wt% β-TCP, having a particle size range between 100 and 300 µm (Plasma Biotal Ltd., Buxton, UK). Composites were foamed using scCO2, as described in the granted patent (Web ref. 2), by using a Supercritical Carbon Dioxide Reactor System (SFE250, Waters Ltd., MA, USA). The residual lactide monomer content was measured by gas chromatography (DC8000, CE Instruments, Rodano, Italy) after post-melting. Foamed blocks were cut into discs for in vitro studies [diameter (Ø) = 8 mm, height (h) = 3 mm)] and into cylinders for in vivo studies for cancellous bone (Ø = 3.2 mm, h = 10 mm) and intramuscular implantation (Ø = 4.0 mm, h = 10 mm). All samples were sterilised by γ-irradiation with a minimum dose of 25 kGy. Prior to in vitro and in vivo studies, elastic scaffolds were treated by dynamically pre-compressing them repeatedly in an aqueous environment at 37 °C for a minimum of 20 cycles and compression level of at least 50 %. Composites were imaged by using a scanning electron microscope (SEM; Philips XL-30) and Xradia MicroXCT-400 X-ray imaging system with a voxel size of 5.6 µm (Carl Zeiss X-ray Microscopy Inc.) before and after the compression treatment. Porosities and pore sizes were calculated from micro-computed tomography (µCT) images with Fiji (Schindelin et al., 2012) using BoneJ (Doube et al., 2010) plugin. All the visualisations were conducted with Avizo 9.3.0 Software (Thermo Fisher Scientific). Mechanical testing of composites Mechanical testing was performed for both intact and pre-compressed samples (Ø = 8 mm, h = 3.5 ± 0.6 mm), as dry at room temperature (RT) and in an aqueous environment at 37 °C using the Instron Electropuls E1000 (High Wycombe, UK) by compressing unconfined samples 1 mm/min until a 50 % strain was reached. Used crosshead speed was adapted from the standard ISO 604. Elastic modulus was determined from linear section of the stress-strain curve, between 0 and 20 % strain. Hydrolytic degradation of composites in vitro and in vivo Hydrolytic degradation of the porous composites in vitro was studied at 4, 12 and 24 weeks at 37 °C in Sörensen buffer solution (n = 6) and in vivo as intramuscular implantation on underside of the supraspinatus muscle (n = 6 at 4 and 12 weeks; n = 1 at 24 weeks due to strong scaffold degradation). A porous PLCL (70/30) polymer scaffold was used as a reference. Degradation was monitored by size exclusion chromatography (SEC) utilising a Waters Associates system equipped with a Waters 717Plus Satellite autosampler, a Waters 510 HPLC solvent pump, two linear PL gel 5 µm columns connected in series and a Waters 2414 differential refractometer. The number-average molecular weight (Mn), weightaverage molecular weight (Mw) and polydispersity of the samples were determined against polystyrene standards at RT. Chloroform was used as the eluent and was delivered at a flow rate of 1 mL/min. Samples were dissolved in chloroform at a polymer concentration of 13.5 ppm. The injection volume was 100 mL. Isolation, characterisation and seeding of hASCs in vitro hASCs were isolated from adipose tissue samples obtained from three female donors (40 ± 11 years old) undergoing surgical procedures at the Department of Plastic Surgery, Tampere University Hospital after patients gave their consent. The study was conducted in accordance with the Ethics Committee of the Pirkanmaa Hospital District, Tampere (R15161). Isolation of hASCs was conducted using a mechanical and enzymatic protocol as described previously (Lindroos et al., 2009). Isolated hASCs were expanded in basic medium (BM) consisting of Dulbecco’s modified Eagle’s medium: nutrient mixture F-12 (DMEM/F-12 1 : 1; Thermo Fischer Scientific), 5 % human serum (HS; BioWest, Nuaillé, France), 1 % antibiotics (100 U/mL penicillin; 100 U/ mL streptomycin; Lonza) and 1 % L-glutamine (GlutaMAX I; Thermo Fischer Scientific). hASCs were cultured at 37 °C in 5 % CO2 and medium was changed twice a week. Cells were detached with TrypLE Select (Life Technologies). Experiments were carried out at passage 3. To verify the mesenchymal origin of the cells, surface marker expression of hASCs at passage 1 was characterised by fluorescent-activated cell sorter (FACSAria; BD Biosciences) as described previously (Lindroos et al., 2009). Monoclonal antibodies against CD14-PE-Cy7, CD19-PE-Cy7, CD45RO-APC, CD73PE, CD90-APC (BD Biosciences), CD34-APC, HLADR-PE (Immunotools, Friesoythe, Germany) and CD105-PE (R&D Systems) were used. The analysis was performed on 10,000 cells per sample and unstained hASC samples were used to compensate for the background autofluorescence levels. The expression of the surface markers CD73, CD90 and CD105 was positive, the expression of CD14, CD19, CD45 and human leukocyte antigen DR isotype (HLA-DR) was negative and that of CD34 was moderate (Table 1). The surface markers’ expression of hASCs confirmed the mesenchymal origin of the cells (Dominici et al., 2006). Before cell seeding, sterile composites were precompressed in BM and incubated for 24 h in BM at 37 °C. hASCs were seeded in a 50 µL medium drop at a density of 510 cells/mm3 into the scaffolds. Cells were allowed to attach for 2-3 h before 500 µL of BM or osteogenic medium (OM) were added. OM consisted of BM with the addition of 250 µM ascorbic acid 2-phosphate (Sigma-Aldrich), 10 mM 38 www.ecmjournal.org S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes β-glycerophosphate (Sigma-Aldrich) and 5 nM dexamethasone (Sigma-Aldrich). As a 2-dimensional (2D) control for alizarin red staining, 500 cells were seeded in 1 mL of BM or OM in a 24-well plate (Nunc, Roskilde, Denmark). Cell viability and proliferation Cell viability was evaluated qualitatively by staining hASCs with fluorescent live/dead-staining probes (Molecular Probes) after 7, 14 and 21 d. hASCs were incubated for 45 min at RT with a mixture of 0.5 µM calcein acetoxymethyl ester (Molecular Probes) and 0.25 µM ethidium homodimer-1 (Molecular Probes). Images of living cells (green fluorescence) and dead cells (red fluorescence) were acquired using an Olympus IX51 phase contrast microscope with fluorescence optics and Olympus DP30BW camera (Olympus). Cell number was analysed quantitatively after at 7, 14 and 21 d by analysing the total amount of DNA by CyQUANT Cell Proliferation Assay Kit (Molecular Probes), according to manufacturer’s protocol, as reported previously (Kyllönen et al., 2013a). Samples were analysed after two freeze-thaw cycles and fluorescence was measured at 480/520 nm with a microplate reader (Victor 1420 Multilabel Counter; Wallac, Turku, Finland). Analysis of osteogenic differentiation in vitro Alkaline phosphatase (ALP) activity was determined after 7, 14 and 21 d, as described previously (Kyllönen et al., 2013a). ALP activity was determined from the same cell lysates as total DNA content. Absorbance was measured at 405 nm (Victor 1420). Soluble total collagen was analysed at 7, 14 and 21 d by Soluble Collagen Assay Sircol™ (Biocolor, Carrickfergus, UK), as described previously (Tirkkonen et al., 2013). Briefly, collagen was dissolved for 2 h at 4 °C with 0.5 M acetic acid (Merck) containing 0.1 mg/mL pepsin (Sigma-Aldrich), while gently shaking. Thereafter, 100 µL samples were dyed with 500 µL of Sircol™ Dye Reagent (Sirius red in picric acid; Biocolor) for 30 min at RT, while gently shaking. Then, samples were centrifuged for 10 min at 13,400 ×g and the dyed collagen pellets were washed with 750 µL of ice-cold Acid-Salt Wash Reagent (Biocolor). After another 10 min centrifugation at 13,400 ×g, the dye was diluted by adding 250 µL of Alkali Reagent (Biocolor) on top of the dyed collagen pellet. Intensity of the dye was measured from two parallel 100 µL samples in a 96-well plate (Nunc) using a microplate reader (Viktor 1420). Mineralisation was studied by alizarin red S staining after 14 and 21 d, as described previously (Kyllönen et al., 2013a). In brief, paraformaldehyde (Sigma-Aldrich)-fixed cell-scaffold constructs were stained with filtered 2 % alizarin red S (pH 4.2; SigmaAldrich) and photographed after several washing steps. Dye was extracted with cetylpyridinium chloride (100 mM, Sigma-Aldrich) and its intensity was determined by measuring the absorbance at 540 nm (Victor 1420). Quantitative real-time reverse transcription polymerase chain reaction (qRT-PCR) analysis was used to compare the relative expression of osteogenic genes in different experimental groups. Total RNA was isolated from hASCs at 14 and 21 d using the NucleoSpin RNA II kit reagent (Macherey-Nagel) according to the manufacturer’s protocol. Singlestrand cDNA was synthesised from total RNA using the High-Capacity cDNA Reverse Transcriptase Kit (Applied Biosystems). Data were normalised to the expression of the housekeeping gene human acidic ribosomal phosphoprotein large P0 (hRPLP0). Primer sequences and accession numbers for RPLP0 and osteogenic genes RUNX2a, OSTERIX and DLX5 are listed in Table 2. The qRT-PCR mixture contained cDNA, primers and SYBR Green PCR Master Mix (Applied Biosystems). Reactions were conducted with ABI PRISM® 7300 Sequence Detection System as reported previously (Kyllönen et al., 2013a). Results were processed with ABI PRISM® 7300 Sequence Detection System-software (Applied Biosystems). Osteogenic marker proteins collagen type I (COL-I) and osteocalcin (OCN) were detected with an indirect immunocytochemical staining method. At 7, 14 and Table 1. Surface marker expression of hASCs at passage 1. Positive > 98 %; negative < 2 %; moderate < 50 % > 2 %. SD: standard deviation. Surface markermean ± SD expression CD14 Serum lipopolysaccharide binding protein 0.6 ± 0.6 negative CD19 B lymphocyte-lineage differentiation antigen 0.4 ± 0.2 negative CD34 Sialomucin-like adhesion molecule 34.8 ± 32.2 moderate CD45 Leukocyte common antigen 1.6 ± 0.3 negative CD73 Ecto-50-nucleotidase 98.2 ± 1.3 positive CD90 Thy-1 (T-cell surface glycoprotein) 99.8 ± 0.1 positive CD105 SH-2, endoglin 98.3 ± 1.2 positive HLA-DR Major histocompatibility class II antigens 0.6 ± 0.1 negative S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes 39 www.ecmjournal.org buprenorphine (0.03 mg/kg, Bupaq® 0.3 mg/mL, Richter Pharma) were used as a pain medication. A 20 mm skin incision was made on the lateral aspect of the right femoral condyle and a 3.2 mm diameter and 10 mm deep drilling hole was made above the lateral collateral ligament. Composites were moistened and pre-compressed in blood collected from the drilling hole prior to implantation. The drilling hole was filled with 3.2 × 10 mm cylinder-shaped β-TCP/PLCL composite implant and periosteum and skin were closed. For the degradation study, a skin incision was made midline over the spine between the scapulae and two 4 × 10 mm β-TCP/PLCL composites were inserted through an applicator tube to the right supraspinatus muscle and two 4 × 10 mm polymer implants to the left supraspinatus muscle. Subcutaneous atipametzole (0.75 mg/kg, Antisedan® 5 mg/mL, OrionPharma) was used to reverse the sedative effect of the medetomidine after the procedure. For postoperative pain, subcutaneous buprenorphine (0.03 mg/kg) and carprophen (4 mg/ kg) were used. Subcutaneous metoclopramide (0.2 mg/kg, Primperan® 5 mg/mL, Sanofi Oy, Espoo, Finland) was used to increase the intestinal motility. Rabbits were in a cage rest for 2 weeks after the surgery and, then, in a large-group housing area. They had access to hay and water ad libitum. One rabbit died during the procedure due to anaesthesiarelated causes and one died 3 d after the operation as a result of a complication unrelated to the femoral defect. The 18 rabbits were randomly divided in groups of 6 animals and euthanised 4, 12 and 24 weeks after the procedure. Subcutaneous injection of medetomidine (0.3 mg/kg) and ketamine (35 mg/kg) was followed by intracardial injection of pentobarbital (300 mg/rabbit, Mebunat® vet 60 mg/mL, Orion Pharma Oy, Espoo, Finland). The intramuscular implants were collected for further analysis. µCT analysis (MicroXCT-400, Zeiss) was performed on all the harvested femoral condyles 21 d, hASCs were fixed with 4 % paraformaldehyde (Sigma Aldrich) in Dulbecco’s phosphate-buffered saline (DPBS) with 0.05 % Triton-X 100 (Sigma Aldrich) for 10 min followed by washing steps. Cells were blocked in 1 % bovine serum albumin (BSA) in DPBS for 1 h at 4 °C. Primary antibodies mouse monoclonal anti-COL-I (dilution 1 : 2,000; Abcam) and mouse monoclonal anti-OCN (dilution 1 : 100; Abcam) were diluted in 1 % BSA and incubated over night at 4 °C. As a negative control, 1 % BSA without primary antibody was used. After washing steps, secondary antibody donkey anti-mouse AlexaFluor 488 IgG (dilution 1 : 1,000; Invitrogen) diluted in 1 % BSA was added and incubated for 45 min at RT. Cells were washed repeatedly and treated with 0.1 % 4, 6-diamidino-2-phenylindole (DAPI) (Sigma Aldrich) in DPBS for 5 min at RT. Finally, cells were washed and imaged using an Olympus IX51 phase contrast microscope with fluorescence optics and Olympus DP30BW camera (Olympus). Images were edited with Adobe Photoshop version CS4. Animal experiments The Finnish Animal Experiment Board (ESAVI/5398/04.10.07/2014) approved the animal study and used protocols. 20 female New Zealand white rabbits were used. Anaesthesia was induced by subcutaneous injection of ketamine (35 mg/kg, Ketador vet® 100 mg/ mL, Richter Pharma, Wels, Austria) and medetomidine (0.3 mg/kg, Domitor® 1 mg/mL, OrionPharma, Espoo, Finland). Intravenous propofol boluses (2-5 mg/ rabbit, Vetofol® 10 mg/mL, Norbrook Laboratories, Newry, Ireland), ketamine bolus (10 mg/kg) or mask anaesthesia with 1.5 % isoflurane (IsoFlo® vet 100%, Abbott Laboratories, Chicago, IL, USA) were used during the surgical procedure if needed. Intravenous trimethoprim/sulfamethoxazole (15 mg/ kg, Duoprim® 200/40 mg/mL, Intervet International, Boxmeer, the Netherlands) was used as an antibiotic prophylaxis. Intravenous carprophen (4 mg/kg, Norocarp® 50 mg/mL, Norbrook Laboratories) and Table 2. Primer sequences and accession numbers of genes analysed by qRT-PCR. Name Full name Accession number Sequence Product size (bp) hRPLP0 Ribosomal protein, large, P0 NM_001002 Forward 5’-AAT CTC CAG GGG CAC CAT T-3’ 70 Reverse 5’-CGC TGG CTC CCA CTT TGT-3’ hALPAlkaline phosphatase NM_000478.4 Forward 5’-CCC CCG TGG CAA CTC TAT CT-3’ 73 Reverse 5’-GAT GGC AGT GAA GGG CTT CTT-3’ hRUNX2A Runx2A, variant 1 NM_001024630.3 Forward 5’-CTT CAT TCG CCT CAC AAA CAA C-3’ 62 Reverse 5’-TCC TCC TGG AGA AAG TTT GCA-3’ hOSX Osterix AF_477981 Forward 5’-TGA GCT GGA GCG TCA TGT G-3’ 79 Reverse 5’-TCG GGT AAA GCG CTT GGA-3’ hDLX5 Distal-less homeobox 5 NM_005221.5 Forward 5’-ACC ATC CGT CTC AGG AAT CG-3’ 75 Reverse 5’-CCC CCG TAG GGC TGT AGT AGT-3’ 40 www.ecmjournal.org S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes before histological preparation. Tube voltage of 120 kV and tube current of 83 µA were selected. From each sample, 1,600 projections were taken with a 13.4 µm voxel size. Exposure time was 4 s. Projections were reconstructed with the manufacturer’s XMReconstructor software. Image processing and analysis were done with Avizo Software (Thermo Fisher Scientific). Thereafter, the femoral condyles were fixed in 10 % buffered formalin solution followed by routine ascending ethanol series and methyl methacrylate embedding. A hard-tissue microtome (Leica, SM2500) was used to cut 5 µmthick slice. Masson-Goldner’s Trichrome (MGT) staining was performed. Statistical analysis Statistical analysis was performed with SPSS version 23 (IBM) using a non-parametric test, due to the small Fig. 1. β-TCP/PLCL composite characteristics before and after pre-compression treatment. (a) Representative SEM (scale bar: 200 µm; PLCL coloured violet; β-TCP coloured pink) and µCT (arrows indicate the granules released from under the polymer film; scale bar: 1,000 µm) images. (b) Compressive stress at 20 and 50 % strain when dry and 1 h, 7 or 14 d at 37 °C in aqueous environment. (c) Modulus when dry and 1 h, 7 or 14 d at 37 °C in aqueous environment. Statistical significances indicated as a p ≤ 0.05 (n = 6). S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes 41 www.ecmjournal.org sample size. The effects of stress and strain (n = 6) on intact and pre-compressed composites and of precompressing and incubation at 37 °C on modulus (n = 6) of composites were compared using MannWhitney U-test with Bonferroni correction. Effects of BM or OM in combination with the composite on cell number, ALP activity, mineralisation, soluble collagen amount and gene expression were compared using Mann-Whitney U-test with Bonferroni correction. Results were considered significant when p < 0.05. Experiments for cell number, ALP activity, mineralisation and soluble collagen amount were repeated with 3 donor lines each with 3 parallel samples (n = 9). Experiments for gene expression were repeated with 3 donor lines each with 2 parallel samples (n = 6). Results Characteristics of the composite Scaffold porosity was 65-67 %, with an average pore size of 380 ± 130 µm. In addition, compressive stress at 20 and 50 % strain and modulus were analysed at different time points. SEM images (Fig. 1a) showed that before pre-compression treatment, β-TCP (coloured pink) was mainly on the cutting surface of the composite, while the surfaces of the pores were smooth and a PLCL film (coloured violet) mainly covered the β-TCP granules. However, the effect of the pre-compression was evident, as after the protocol the PLCL surface of the pores had ruptured and β-TCP granules protruded through it. Furthermore, SEM images demonstrated that the treatment tore additional holes in the pores of the composite. µCT images agreed with SEM images as they revealed that after the treatment more β-TCP granules were on display on the pore surfaces or even detached from the polymer phase in contrast to intact composite (Fig. 1a). Both the compression treatment and warming at 37 °C had a significant effect on the mechanical properties of the composite. The compression treatment of a dry composite decreased the modulus significantly in contrast to an intact composite, but 1 h at 37 °C aqueous solution treatment erased such a difference (Fig. 1c). The same phenomenon was seen in the compressive stress analysis at both 20 and 50 % strain (Fig. 1b). Moreover, warming and wetting of the intact composite significantly decreased the modulus in comparison to dry intact composite (Fig. 1c). Intriguingly, both modulus and strength increased between 1 h and 7 d in an aqueous environment at 37 °C (Fig. 1b,c). No samples were broken during the compression analysis and samples tested in an aqueous environment at 37 °C returned to 95 % of their original height after the 50 % compression. Hydrolytic degradation of the composite in vitro and in vivo Molecular weight (Mn and Mw) change, caused by hydrolytic degradation of the polymer chains in Fig. 2. Hydrolytic degradation analysis results. Mn and Mw of (a) β-TCP/PLCL composites and (b) PLCL scaffolds were analysed in vitro and in vivo after 4, 12 and 24 weeks. 42 www.ecmjournal.org S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes both β-TCP/PLCL composites (Fig. 2a) and PLCL polymer scaffolds (Fig. 2b), was analysed both in vitro and in vivo. Mn and Mw of the samples were determined at 4, 12 and 24 weeks and the results showed that the degradation in vitro was similar to that in vivo for both materials. Furthermore, scaffold molecular weight had already decreased significantly during the first 4 weeks. In addition, Mn and Mw of the composite scaffold decreased by approximately 90 % during the 24-week follow-up. The neat polymer scaffold had initial Mn and Mw values higher than the composite but the difference in the degradation profile was evened out in 4 weeks, after which no clear differences were seen between the materials or the in vitro and in vivo environments. The residual L-lactide content after melt-extrusion and scCO2foaming for the PLCL scaffold was 0.09 wt% and for the β-TCP/PLCL composite 0.06 wt%. Viability and proliferation of hASCs in vitro Cell viability was very good since only single dispersed dead cells were observed in cultures. In BM, cell number was clearly less than that in OM at 7 d (Fig. 3) and the difference was evident also at 14 and 21 d. Cell number was similar in OM at 7 d and BM at 21 d. Cell proliferation was analysed by fluorescent CyQUANT proliferation assay after 7, 14 and 21 d in culture. In concordance with the live/dead staining, cell number was significantly larger in OM in contrast to BM at all time points (Fig. 4a). Moreover, cell number as indicated by live/dead staining was similar in OM at 7 d and BM at 21 d. Variation among the different donor cell lines was evident in proliferation analysis, as OM did not increase the proliferation of hASCs in comparison to BM from one of the donors as it did for the other two. Osteogenic differentiation of hASCs in vitro ALP activity (Fig. 4b) was significantly greater in OM, in contrast to BM at 7 and 14 d. Furthermore, ALP activity in both BM and OM increased significantly from 7 to 14 d. However, ALP activity did not rise significantly from 14 to 21 d in either BM or OM. ALP activity varied between different donor cell lines and especially one donor cell line had higher ALP activity in comparison to the others. Total collagen amount did not vary notably among BM and OM groups at 7, 14 or 21 d (Fig. 4c). Collagen amount increased in both study groups slightly from 7 to 14 d; however, collagen amount seemed to decline from 14 to 21 d. hASCs produced a pronounced mineralised matrix on the composite in both BM and OM at 14 and 21 d (Fig. 4e). Furthermore, they produced notably more mineralised matrix on the scaffold in comparison to hASCs seeded on cell culture plastic (2D) at 21 d (Fig. 4e). At 14 d there was no difference between BM and OM. However, at 21 d, the mineral amount in OM was significantly larger than that in BM (Fig. 4d). Immunocytochemical staining of COL-I was conducted at 7, 14 and 21 d. In BM, COL-I expression was modest at all time points (Fig. 5a). However, in OM, a clear development of a collagenous matrix was seen during the 3-week experiment: at 7 d, cells were producing COL-I; at 14 d, cells were still producing COL-I and had also excreted COL-I to the extracellular matrix (ECM); at 21 d, there was a strong extracellular COL-I matrix seen in the representative image (Fig. 5a). In addition to COL-I, OCN was also stained at 14 and 21 d. OCN was expressed in both BM and OM at both time points; however, the staining was slightly stronger in OM at both time points (Fig. 5b). Fig. 3. Viability of hASCs in β-TCP/PLCL composites. Representative images of live/dead stained hASCs cultured in β-TCP/PLCL composites in BM or OM after 7, 14 and 21 d in culture. Scale bar: 500 µm. S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes 43 www.ecmjournal.org qRT-PCR analysis for osteogenic genes ALP, RUNX2a, OSX and DLX5 was conducted at 7, 14 and 21 d. Relative gene expression did not significantly differ in BM in comparison to OM during the experiment, although, the expression of RUNX2a (p values between groups at 7 d: 0.394; 14 d: 0.078; 21 d: 0.310) and OSTERIX (p values between groups at 7 d: 0.394; 14 d:0.123; 21 d: 0.240) was higher in OM as compared to BM at all time points (Fig. 6). Bone regeneration within composites in rabbit femur defects Bone was already able to grow on the surface as well as form inside the composite at 4 weeks (Fig. 7). The increased bone growth was observed on the surface and inside the composite at 12 and 24 weeks. In the 4-week µCT and histological staining images, some bone formation could be seen beyond the original bone margins, which was most likely a periosteuminduced reaction and not bone formation induced by the implant. Histological staining showed no signs of fibrous tissue or formation of cysts during the 24 weeks. µCT imaging confirmed the histological staining results. In addition, especially the representative µCT images at 24 weeks showed that the composite had degraded as compared to 4and 12-week images (Fig. 7). Discussion Composites consisting of calcium phosphates and synthetic polymers have presented promising results in bone engineering applications and it has been Fig. 4. hASC culture and differentiation in β-TCP/PLCL composites. (a) Cell numbers at 7, 14 and 21 d. (b) ALP activity at 7, 14 and 21 d. (c) Total soluble collagen amount at 7, 14 and 21 d. (d) Quantitative mineralisation results after 14 and 21 d. (e) Representative images of mineral-stained cell-scaffold constructs after 14 and 21 d and of hASC cultures on cell culture plastic (2D) after 21 d. Stained blank scaffold without seeded cells is presented on the left of 21 d OM scaffold (scaffold Ø = 8 mm; well Ø = 15.5 mm). Statistical significance indicated as a p ≤ 0.001 or b p < 0.05 (n = 9). 50 www.ecmjournal.org S Pitkänen et al. ScCO2-foamed β-TCP/PLCL composite bone substitutes substitutes. However, existing composite scaffolds with the porosity needed for osteoconduction do not approach the modulus required to replace lost bone function and avoid the need for bone fixation – particularly for lower extremity defects. If immediate restoration of bone mechanical properties is not a design goal, what factors determine the strength needed of candidate scaffolds for bone repair? Would this differ for grafts used in maxillofacial versus long bone defects? Authors: The main aims in composite development are optimal porosity, biocompatibility and osteoconductivity. In addition, easy intra-operative tailoring is highly important, as, for instance, ceramics are not easy to handle or tailor during a surgery, which hinders surgeons’ work. With that said, future efforts aim to improve composite mechanical properties to create a composite with similar modulus to subchondral spongiosa. The requirements for mechanical strength of the scaffold are, of course, lower in maxillofacial defects in comparison to long bone defects and, therefore, the composite described in the study is suitable for non-load bearing sites. Ryan Porter: Supercritical CO2 has also been used for biomaterial sterilisation processes, including bone allografts. Was the γ-irradiation step included for precautionary measures? Would both steps be required in a good manufacturing practice (GMP) manufacturing protocol for the proposed composite? Authors: It is true that the processing method can also be used as a sterilisation method. γ-irradiation sterilisation was used for precautionary measures and because it is the commonly used and highly trusted sterilisation method among clinical products. The composite is classified as a medical device and, therefore, there is no need for a GMPlevel manufacturing protocol but an ISO13485 manufacturing protocol is sufficient. Ryan Porter: What might be the challenges to clinically translate bone graft alternatives manufactured using supercritical CO2? Authors: These are highly important matters when developing a clinical product. The challenges in clinical translation of scCO2-processed structures are related to producing a homogenous pore structure in a reproducible manner. Furthermore, the scalability of the processing is another challenge to overcome in case of a large-scale production. Pamela Habibovic: A problem of the in vitro experiments is the (partially strong) osteogenic differentiation in BM samples. This could indicate pre-differentiation of the used cells, pointing towards an activation of differentiation by the material itself or problems in the assays. This could be clarified by including data of suitable control groups in the results section, e.g. cells cultured in BM and OM on tissue culture plastic etc.. Authors: The reason for not using a 2D well plate control was the challenge related to the differences between experimental setups in 3D and 2D. Due to these differences, e.g. notable differences in cell seeding density and cell attachment, the results between the two setups would not be fully comparable. Editor’s note: The Scientific Editor responsible for this paper was Joost de Bruijn.