6 Additive Manufacturing of Star Poly(ε-caprolactone) Wet-spun Scaffolds for Tissue Engineering
Applications
Carlos Mota1, Dario Puppi1, Dinuccio Dinucci1, Matteo Gazzarri1, Federica Chiellini1and Emo Chiellini1
1 Laboratory of Bioactive Polymeric Materials for Biomedical and Environmental Applications (BIOLab), Department of Chemistry and Industrial Chemistry, University of Pisa, via Vecchia Livornese 1291, 56010 San Piero a Grado (Pi), Italy.
Abstract
Three-dimensional wet-spun scaffolds were produced with a layer-by-layer principle using a computer-controlled wet-spinning apparatus. Three-arm star-shape poly(ε-caprolactone) (*PCL) solutions or hydroxyapatite (HA) nanoparticles suspensions in *PCL solutions were extruded directly into a coagulation bath to obtain a solidifying continuous polymeric filament that was deposited with a predefined pattern. The processing parameters for the production of *PCL and
*PCL/HA scaffolds were optimised by investigating their influence on fibre and scaffold morphology. Scaffolds with porosity in the range of 20%-60%, average pore size in the X and Y direction varying from 190 to 297 µm and in the Z direction from 54 to 126 µm were developed.
The fibre constituting the scaffold showed a mean diameter varying in the range 189 to 274 µm and a porous morphology both in the outer surface and in the cross-section. Cell culture experiments carried out by using a murine preosteoblasts cell line showed good cell response in terms of viability and proliferation. The obtained results suggest the suitability of the proposed technique for the production of customised scaffolds for bone tissue engineering.
Keywords: tissue engineering, scaffolds, wet-spinning, additive manufacturing, star polycaprolactone.
6.1 Introduction
Tissue engineering (TE) and regenerative medicine play an important role in the development of tissue and organ substitutes. One of the most common approach in the field of TE involves the development of three-dimensional (3D) matrices (scaffolds) combined with cells and/or bioactive agents, that serve as temporary platform for tissue regeneration at the affected site until the normal physiological functions are restored [Hutmacher and Cool, 2007; Stock and Vacanti, 2001]. A temporary 3D structure should be biocompatible, biodegradable (ideally, the degradation rate should match the rate of formation of the new tissue) and should support of cell activity while the new extracellular matrix (ECM) is formed [Kim and Mooney, 1998]. A complex 3D scaffold structure with controlled porosity, pore size and distribution, and external shape is necessary for a successful regeneration process [Karageorgiou and Kaplan, 2005].
Several techniques have been proposed in the last two decades for the production of different scaffold architectures and they are typically classified into conventional and additive manufacturing (AM) techniques. Conventional techniques, such as freeze drying, solvent casting combined with particulate leaching, phase inversion techniques, fibre bonding, electrospinning, have been widely studied for the development of bone TE constructs giving promising results [Hutmacher, 2000;
Sachlos and Czernuszka, 2003; Yoshimoto et al., 2003]. However, most of them present major drawbacks, namely the low control over pore size, distribution and interconnection, and external scaffold shape and size [Sachlos and Czernuszka, 2003].
AM techniques, also called solid freeform fabrication techniques, represent a promising alternative to conventional techniques due to the possibility of producing scaffolds with tuned morphology and mechanical properties. During the last decade, various AM techniques, such as fused deposition modeling [Hutmacher, 2000; Zein et al., 2002], stereolithography [Lee et al., 2007], selective laser sintering [Eosoly et al., 2010; Williams et al., 2005] and 3D printing [Leukers et al., 2005; Seitz et al., 2005], were studied for the development of complex 3D templates for bone TE.
Wet-spinning is a phase inversion process induced by immersion precipitation of a polymeric solution into a non-solvent (coagulation bath), originally developed for the production of continuous polymeric fibres for diverse industrial applications. Thanks to the possibility of processing a wide range of natural and synthetic polymers, wet-spinning was recently proposed for the production of scaffolds from continuous biodegradable fibres. For instance, wet spun non- woven meshes [Puppi et al., 2011a; Puppi et al., 2011b; Yilgor et al., 2009] made of chitosan [Heinemann et al., 2009; Yilgor et al., 2009], starch-based materials [Leonor et al., 2011;
Pashkuleva et al., 2010; Tuzlakoglu et al., 2010] and three-arm star-shape poly(ε-caprolactone) (*PCL) [Puppi et al., 2011a; Puppi et al., 2011b] were investigated for TE applications. However, wet-spinning presents a lack of reproducibility of the scaffold pore architecture, and the external shape of the 3D scaffold is limited to simple geometries, generally defined by the collection container.
Star polymers consist of linear polymer chains attached to a smaller central moiety. Due to their small size, spherical structure and limited interaction between molecules, star polymers have different properties compared to the linear polymers with equivalent molecular weight. Generally, they show lower crystallinity as well as melting and solution viscosity resulting in different processing and degradation properties [Burchard, 1999; Celik et al., 2009; McKee et al., 2005;
McKee et al., 2004]. The polymer investigated in this study is *PCL consisting of three linear PCL chains, roughly with the same molecular weight, attached to a central aluminium core [Balakrishnan et al., 2004; Balakrishnan et al., 2006]. *PCL was already investigated by our group for TE purposes in the form of electrospun or wet-spun non-woven meshes showing promising results when tested in cell culture experiments using preosteoblast cells [Puppi et al., 2010; Puppi et al., 2011b].
Hydroxyapatite (HA) is a synthetic osteoconductive ceramic with a chemical composition resembling that of bone apatite extensively studied for bone TE applications [Sopyan et al., 2007].
However, despite its good biocompatibility properties, HA is a brittle material and for this reason is often combined with polymeric materials in order to enhance its toughness [Ma, 2008]. In addition, the incorporation of HA particles into polymer matrices has been showed to improve the osteoconductivity and mechanical properties of polymeric scaffolds and to create a pH buffer against the acidic degradation products of the polymeric matrices [Koh et al., 2006; Rezwan et al., 2006; Ural et al., 2000].
The aim of the present study was to develop 3D *PCL and *PCL/HA composite wet-spun scaffolds with a controlled and predefined internal structure and external shape by means of a computer- controlled wet-spinning apparatus. The influence of different processing parameters, such as polymer concentration (C*PCL), solution feed rate (F) and deposition velocity (Vdep), on the scaffold architecture and porosity, and on the filament morphology was evaluated by means of scanning electron microscopy (SEM). Thermal properties of the raw *PCL and of the developed scaffolds were studied by thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC).
The in vitro cytocompatibility of the constructs was carried out by using a murine preosteoblast cell line.
6.2 Materials and Methods
6.2.1 Materials
Three-arm star branched poly(ε-caprolactone) (*PCL, Mw = 189000 g/mol), was provided by Michigan Biotechnology Institute (Lansing, MI, USA), hydroxyapatite nanoparticles (HA, size <
200 nm) were bought from Sigma-Aldrich (Italy). Acetone and ethanol were purchased from Sigma-Aldrich (Italy) and used as received.
6.2.2 Computer-aided wet-spinning apparatus
Scaffolds were produced using an in-house modified subtractive Rapid Prototyping (RP) system (MDX-40A, ROLAND DG Mid Europe Srl, Italy) allowing for the deposition of polymeric solutions with a controlled 3D pattern (Figure 1).
Figure 1 - Scheme of computer-aided wet-spinning apparatus
The milling head unit of the Roland machine was replaced by a syringe pump system (NE-1000, New Era Pump Systems Inc., Wantagh, NY, USA). The lay-down pattern for the production of the scaffolds was calculated using an algorithm developed in Matlab software (The Mathworks, Inc.).
The generated pattern was uploaded into the Roland equipment through the software Vpanel for MDX-40A.
6.2.3 Solution preparation
*PCL pellets were dissolved in acetone at 35°C for 3 h under gentle stirring, to obtain homogeneous solutions at various concentrations (15 to 30% w/v). For the production of *PCL/HA composites scaffolds, HA nanoparticles were added to the polymeric solution (HA/*PCL weight ratio of 20%) and left under vigorous stirring at 35°C for 1 h until a homogeneous dispersion of the nanoparticles was achieved.
6.2.4 Scaffolds production and design
The prepared solution was loaded into a 5 ml syringe, fitted with a blunt tip stainless steel needle (23 Gauge, I.D. of 0.34 mm and 15 mm of length), and placed into the syringe pump controlling F.
The extruded solution from the tip of the needle was collected on a glass beaker containing ethanol.
The initial distance between the tip of the needle and the bottom of the beaker (Z0) was 1 mm. 3D scaffolds with a predefined architecture were fabricated with a layer-by-layer process using the previous described computer-aided wet-spinning system (Figure 2).
Figure 2 – Design of 3D scaffolds with a cylindrical geometry characterized by a 0-90° lay-down pattern, a diameter of 15 mm and either 10 or 25 overlapped layers (~2 and 5 mm thickness, respectively).
Samples of cylindrical geometry with a diameter of 15 mm and a thickness of around 2 mm (10 layers) and with around 5 mm of thickness (25 layers) were produced for biological investigations.
The deposition trajectory was calculated for the production of scaffolds with 0-90º lay-down pattern, fibre spacing (d2, distance between fibres axis) of 500 µm and layer thickness (d3, z-axis needle translation between each layer) of 200 µm. The effect of three processing parameters was evaluated: the influence of Vdep was evaluated in the range 180 to 300 mm·min-1, of C*PCL in the range 15 to 30% w/v and that of F in the range 0.6 to 1.2 ml·h-1. After the production, the
constructs were removed from the coagulation bath and kept under fume hood overnight. A further removal of the residual solvents was performed by placing the constructs in a vacuum chamber for 48 h.
6.2.5 Morphological analysis
The macro and micro morphology of the 3D scaffolds were assessed using SEM (JEOL LSM 5600LV, Japan) under backscattered electron imaging. SEM micrographs were acquired from the top view and the cross-section of the scaffolds at different magnifications. In order to perform cross-section analysis, scaffolds were fractured in liquid nitrogen.
The average values for fibre diameter (d1), d3 and pore size were determined by means of ImageJ 1.43u software (National Institutes of Health, USA) on SEM micrographs with a 35X magnification. Data was calculated over 20 measurements per scaffold.
The scaffold porosity was evaluated using the theoretical approach proposed by Landers et al [Landers et al., 2002]:
4 100 1 100 1
3 2
2 1 ×
− ⋅
=
×
−
=
d d
d Vcube
Vscaffold
P π
where P is the porosity of the scaffold, d1 is the fibre diameter, d2 the fibre spacing and d3 layer thickness. The porosity inside the filaments was not considered for this calculation.
6.2.6 Thermal properties
Thermal properties of both raw materials and scaffolds were evaluated by means of thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC). TGA was performed using a TGA Q500 instrument (TA Instruments, Italy) in the temperature range 30–
600ºC, at heating rates of 10 and 20ºC·min-1 and under a nitrogen flow of 60 ml·min-1.
DSC was performed in the range -100 to 100ºC, at a rate of 10ºC·min-1, using a Mettler DSC-822 (Mettler Toledo, Italy). Samples with a weight of 5-10 mg were scanned including first and second heating, and first cooling under nitrogen atmosphere with a flow rate of 80 ml·min-1. Glass transition temperature (Tg) was evaluated by considering the inflection point of the heating part of the thermogram; the melting temperature (Tm) and the melting enthalpy (∆Hm) were acquired from
the endothermic part of the thermogram. In the case of raw materials, values were obtained from the second heating curve, while in the case of scaffolds from the first heating curve.
6.2.7 Biological evaluation
6.2.7.1 Sterilization of the scaffolds
Scaffolds were placed in a 24 wells plate and sterilized under UV light for half an hour for each side prior to wash with 70% ethanol/water solution for three hours. After ethanol removal, scaffolds were extensively washed with Phosphate Buffer Saline (PBS) 0.01 M pH 7.4, (PBS 1X) containing a penicillin/streptomycin solution (1%) and were then left overnight at 37 °C in a 5%
CO2. To facilitate cell attachment samples were then immersed in complete culture medium and incubated for other twenty-four hours before cell seeding.
6.2.7.2 Cell culturing and cell seeding
Mouse calvaria-derived preosteoblastic cells (MC3T3-E1) subclone 4 were obtained from the American Type Culture Collection (ATCC CRL-2593) and cultured as monolayer in alpha minimum essential medium (α-MEM, Sigma), containing ribonucleosides, deoxyribonucleosides and sodium bicarbonate, supplemented with L-glutamine (2 mM), fetal bovine serum (10%), penicillin:streptomycin solution (100 U/ml:100µg/ml) (1%) and antimycotic. The cultures were mantained at 37 °C in a 5% CO2.
Confluent MC3T3-E1 cells at passage 25 were trypsinized (0.25% trypsin-EDTA), detached from the flask and seeded directly on the scaffolds. The cells, at an appropriate density (0.5 x 104), were initially dispersed in a small amount of complete culture medium (100 µl) and dropped onto the scaffolds. After one hour of incubation at 37°C in humidified atmosphere containing 5% CO2, scaffolds were covered with other 600 µl of complete medium. For this study, the MC3T3-E1 seeded cells were allowed to attach on the *PCL and *PCL/HA scaffolds using the complete α- MEM medium for 24 hours. In order to induce and promote the osteoblastic phenotype expression of MC3T3-E1, where scheduled, scaffolds were cultured in the osteogenic medium by supplementing the α-MEM with ascorbic acid (0.3 mM) [Quarles et al., 1992] and β- glycerophosphate (10 mM) [Wang and Yu, 2010]. The medium was replaced every 48 hours.
6.2.7.3 Cell viability and cell proliferation
Cell viability and proliferation were measured by using the cell proliferation reagent WST-1 (Roche) on days 7, 14, 21 and 28, after cell seeding. The test is based on the mitochondrial enzymatic conversion of the tetrazolium salt WST-1 into formazan, the soluble product. The assay was performed by incubating for four hours cell-seeded scaffolds with the WST-1 reagent, diluted 1:10, at 37°C and 5% CO2. Measurements of formazan dye absorbance were carried out with a Biorad Microplate Reader (Biorad) at 450 nm, with the reference wavelength at 655 nm.
6.2.7.4 Quantification of collagen production
Assessment of collagen production by MC3T3-E1 cultured on *PCL based scaffolds, as indicator of osteoblastic marker, was carried out at days 7 and 28. At each time points the medium was removed and samples were washed twice with PBS. Scaffolds were then incubated with Direct Red 80 dye (Sigma) dissolved in picric acid (0.1%), for 1 hour at room temperature. After the incubation time the dye was removed and samples were washed three times with HCl 10 mM to remove dye excess. The elution of the bound stain was performed with NaOH 0.1 N for 30 minutes at 37°C. Supernatants were plated in 96 well plate and the absorbance was read at 540 nm.
Scaffolds with no cells were treated following the same protocol and considered as blank.
The photometric quantification of the collagen was obtained by mean of a calibration curve (Figure 3) prepared with collagen type I, derived from calf’s skin (Sigma).
Figure 3 – Collagen calibration.
Briefly, known concentrations of collagen in acetic acid (0.1 M) were filmed and dried on glass slides by solvent casting technique, in order to obtain a thin film. Films were then fixed and treated as the samples [Junquiera et al., 1979].
6.2.7.5 Alkaline phosphatase (ALP) activity
Alkaline phosphatase (ALP) activity was determined in cultured MC3T3-E1-*PCL constructs on days 7, 14, 21, and 28. The measurement was assessed with a colorimetric method that is based on the conversion of p-nitrophenyl phosphate into p-nitrophenol by the ALP enzymatic activity.
Scaffolds were washed three times with PBS and then placed into 1 ml of a lysis buffer, containing Triton X-100 (0.2%), magnesium chloride (5 mM) and trizma base (10 mM) at pH 10. Samples underwent freezing-thawing cycles by keeping at –20°C and subsequently at room temperature [Wutticharoenmongkol et al., 2007]. This process was repeated three times in order to extract the intracellular ALP [Wang and Yu, 2010]. Following this step, a volume of 20 µl of supernatant was taken from the samples and added into 100 µl of p-nitrophenyl phosphate substrate (Sigma). A standard calibration, prepared dissolving ALP from bovine kidney (Sigma) in the same lysis buffer, was added to the substrate and the reaction was left to take place at 37°C for 30 minutes. The reaction was stopped by adding 50 µl of 2 M NaOH solution and after 5 minutes waiting absorbance was measured at 405 nm in a spectrophotometer. The molar concentration of ALP was normalized with the total protein content of each sample, which was measured using Bradford protein assay (Pierce). The amount of the proteins was calculated against a standard curve. The results for ALP activity assay were reported as nano-moles (nmol) of converted substrate *(mg of protein·minute)-1.
6.2.7.6 Mineralized matrix deposition
The mineralized ECM deposition of MC3T3-E1 cultured on *PCL, was analysed qualitatively by using the Alizarin Red Staining (ARS) method [Stanford et al., 1995] at day 28. In fact, as confirmed from the literature [Whited et al., 2011] (see chapter 5), ECM mineralization is considered a late stage indicator of osteoblastic phenotype. Samples, fixed with 3.8% p- formaldehyde at room temperature for 30 minutes, were stained with 2% Alizarin Red solution pH 4 for 10 minutes. After the dye incubation, samples were extensively rinsed with de-mineralized water 0.2 µm filtered, in order to remove the dye excess. Because of the known properties of alizarin-mineral complexes to emit red fluorescence [Kagayama et al., 1999], microscopic
qualitative evaluation of ECM mineralization was achieved with confocal laser scanning microscopy (CLSM), by exciting cultured scaffolds with a 561 nm solid-state laser emission.
6.2.7.7 Cell morphology investigation by confocal laser scanning microscopy (CLSM)
Morphology of MC3T3-E1 cells grown on *PCL constructs and 3D culture organization were investigated by means of CLSM on days 7, 14, 21, and 28. Cells were fixed with 3.8% p- formaldehyde for 30 minutes in PBS 1X, permeabilized with a PBS 1X/Triton X-100 solution (0.2%) for 15 min and incubated with a solution of 4’-6-diamidino-2-phenylindole (DAPI) (Invitrogen) and phalloidin-AlexaFluor488 (Invitrogen) in PBS for 45 min at room temperature in the dark. After dye incubation, samples were extensively washed with PBS 1X and observed by including specimens between two glass coverslips. A Nikon Eclipse TE2000 inverted microscope equipped with an EZ-C1 confocal laser and Differential Interference Contrast apparatus was used to analyse the samples (Nikon). A 405 nm laser diode (405 nm emission) and an Argon Ion Laser (488 nm emission) were used to excite respectively DAPI and FITC fluorophores. Images were captured with Nikon EZ-C1 software with identical instrumental settings for each sample. Images were further processed with The GIMP (GNU Free Software Foundation) image manipulation software and merged with Nikon ACT-2U software.
6.2.8 Statistical evaluation
Quantitative data were presented as mean ± standard deviation (SD). Data sets were screened by one-way ANOVA and a Tukey test was used for post hoc analysis; significance was defined at p <
0.05.
6.3 Results and Discussion
A number of studies published over the last years have shown the possibility to produce non-woven micrometre meshes composed of wet-spun fibres and suitable as TE scaffolds. The proposed fabrication techniques involve a physical bonding of prefabricated wet-spun fibres or a continuous, randomly-oriented deposition of the solidifying fibre in the coagulation bath, However, these techniques suffer from lack of control overpore architecture and the external shape [Leonor et al., 2011; Pashkuleva et al., 2010; Tuzlakoglu et al., 2010]. Moreover, they require a dedicated operator for long time to perform the collection of the wet-spun fibres by a manual motion of the
collection tank [Puppi et al., 2011a; Puppi et al., 2011b]. The present research activity was focused on the development of 3D *PCL and *PCL/HA composite wet-spun scaffolds with a controlled internal network of pores and a predesigned external shape by means of a new computer-controlled wet-spinning apparatus.
6.3.1 Investigation of production parameters
Scaffolds were produced using an innovative AM process based on a computer–aided technique that allows controlling the deposition conditions of wet-spun fibres. The production process and the resulting scaffold architectures influenced by various processing parameters (Table 1).
Table 1 - AM wet-spinning technique parameters influencing scaffold production process and scaffold structure.
Studied parameters Polymer concentration (C*PCL) Deposition velocity (Vdep) Solution feed rate (F) Other parameters involved in the process Solvent/non-solvent system
Lay-down pattern Layer thickness (d3) Needle internal diameter
Needle tip-to-beaker bottom initial distance (Z0) Fibre spacing (d2)
The processing parameters for the production of 3D microfibre scaffolds made of *PCL or
*PCL/HA composite were investigated during the present activity. In particular, the influence of C*PCL, Vdep and F were investigated. The solvent/non-solvent (acetone/ethanol) system was selected due to the successful results attained during previously reported works by our group [Puppi et al., 2011a; Puppi et al., 2011b].
As it will be shown in the next paragraphs, by changing the processing parameters it was possible to obtain scaffolds with different d1, pore size and porosity. Table 2 summarizes the scaffold structure parameters obtained for the different fabrication conditions investigated.
Table 2 - Fibres, pore size and porosity of *PCL and *PCL/HA scaffolds at different processing conditions.
*PCL
Polymer concentration (C*PCL) [% w/v]
Fibre diameter (d1)
[µm]
Pore size (X and Y axis)
[µm]
Pore size (Z axis)
[µm]
Porosity [%]
20 246±19 225±40 68±21 (*) 36
25 242±18 235±23 95±14 (*) 43
30 232±4 232±22 126±12 (*) 49
Deposition Velocity (Vdep) [mm·min-1]
200 266±8 (*) 190±19 (*) 54±8 (*) 24
240 232±4 (*) 232±22 (*) 126±12 49
280 201±8 (*) 279±40 (*) 118±15 59
Feed Rate (F) [ml·h-1]
0.8 189±4 (*) 256±33
(*vs.1.0 and 1.1 ml·h-1) 122±9
(*vs. 0.9 and 1.1 1ml·h-1) 60
0.9 195±3 (*) 271±31
(*vs.1.0 and 1.1 1ml·h-1) 109±9 (*) 58
1.0 232±4 (*) 232±22 (*) 126±12
(*vs. 0.9 and 1.1 1ml·h-1) 49
1.1 274±8 (*) 297±38 (*) 62±12 (*) 26
*PCL/HA
Polymer concentration (C*PCL) [% w/v]
Vdep = 240 mm·min-1; F = 1 ml·h-1;
20 194±6 (*,**) 280±23 (**) 99±8 (**) 58
30 202±2 (*,**) 275±25 (**) 89±17 (**) 54
Architecture parameters values are presented as mean±standard deviation. (*) parameter significantly different (p < 0.05) when compared to those of other scaffolds developed within each independent study of processing variable effect (C*PCL, Vdep and F), (**) parameter significantly different (p < 0.05) when compared to that of the analogous plain *PCL scaffold obtained applying the same processing conditions.
6.3.1.1 Effect of polymer concentration (C*PCL)
The influence of C*PCL on the morphology of the fabricated scaffolds was studied in the range 15- 30% w/v, keeping constant Vdep (240 mm·min-1) and F (1 ml·h-1). When C*PCL was lower than 20%
w/v, the filament solidification process was too slow to obtain a fibrous structure because of fibre- fibre merging at the contact points. An increase in C*PCL from 20 to 30% w/v resulted in enhanced fibre morphology and thus in a well-defined 3D fibrous structure (Figure 4).
Figure 4 – Backscatter SEM micrographs of the top view and cross-section of *PCL scaffolds obtained for C*PCL in the range 20 – 30% w/v (Vdep=240 mm·min-1 and F=1 ml·h-1).
Analysing the top view micrographs, a more uniform morphology along the single fibre and better fibres alignment was evident at higher concentrations. In particular, fibres look flattened at the fibre-fibre intersection points for lower concentrations, and the bonding area in the conjunction region decreased with the increasing of C*PCL. The pore size in the Z-axis increased with the increase of C*PCL from 68±21 (C*PCL=20% w/v) to 126±12 µm (C*PCL=30% w/v) (Figure 5). The obtained porosity was 36%, 43% and 49% for C*PCL of 20, 25 and 30% w/v, respectively. Statistical analysis of the d1 and of the pore size in X and Y axis revealed non-significant differences among scaffolds obtained from different C*PCL (Table 2). Inversely, statistical analysis of the Z-axis pore size revealed significant differences among the scaffolds obtained at different C*PCL. The best results, in terms of fibre morphology and alignment, and scaffold porosity, was achieved for C*PCL
of 30% w/v, making this concentration ideal for further processing parameters studies.
Figure 5 – Effect of the polymer concentration on: a) pore size on the X, Y and Z axis, b) fibre diameter and scaffold porosity.
6.3.1.2 Effect of deposition velocity (Vdep)
The effect of Vdep on scaffold morphological parameters was evaluated in the range 180 to 300 mm·min-1, keeping constant the other investigated parameters (C*PCL= 30% w/v, F=1 ml·h-1). When Vdep was either below 200 mm·min-1 or above 280 mm·min-1, a 3D fibrous structure was not achieved. For low Vdep the porous structure was compromised due to the “fusion” between adjacent fibres, while in the case of high Vdep the adhesion between adjacent layers was limited compromising the structural integrity of the scaffold. However, in the range 200 to 280 mm·min-1 a 3D fully interconnected porous structure, characterized by a good degree of fibre alignment, was obtained (Figure 6).
Figure 6 - Backscatter SEM micrographs of the top view and cross-section of the *PCL 3D scaffolds produced for Vdep studies with constant C*PCL=30% w/v and F=1 ml·h-1.
By increasing Vdep in this range, the d1 decreased from 266±8 to 201±8 µm and the porosity increased from 24 to 59% (Figure 7). Statistical analysis of the d1 and X and Y pore size revealed significant differences among scaffolds obtained employing different Vdep. Moreover, the Z axis pore size obtained for Vdep =200 mm·min-1 was significantly lower when compared to the other Vdep. Scaffolds obtained at the highest Vdep showed delamination between adjacent layers, likely because of the limited merging at the fibre-fibre contact points. On the base of the analysis of the micro and macro-morphology, the optimal Vdep selected to study the influence of F variation was 240 mm·min-1.
Figure 7 – Effect of deposition velocity on: (a) pore size on the X, Y and Z axis, (b) fibre diameter and scaffold porosity.
6.3.1.3 Effect of solution feed rate (F)
The influence of F on d1 and 3D scaffold morphology was investigated in the range 0.6 to 1.2 ml·h-
1, keeping constant the optimised conditions of C*PCL and Vdep (30% w/v and 240 mm·min-1, respectively). When F was either lower than 0.8 ml·h-1 or higher than 1.1 ml·h-1, it was not possible to obtain a homogenous scaffold structure. However, when F was in the range 0.8 – 1.1 ml·h-1, a good fibres morphology and alignment was achieved. By increasing F in this range, the d1
increased from 189±4 to 274±8 µm and porosity decreased from 60 to 26%, respectively (Figure 8). Statistical analysis of the d1 revealed significant differences among values obtained at different F. The X and Y axis pore size differences among scaffolds obtained at different F were statistically significant, with the exception of scaffolds obtained at F of 0.8 and 0.9 ml·h-1. The differences of poresize in Z axis were significantly different, with the exception of that between scaffolds obtained at 0.8 and 1.0 ml·h-1. By analysing the fibre morphology (Figure 9) and the results obtained on the base of the above presented investigations, the optimised processing conditions for
*PCL scaffolds fabrication was selected to be C*PCL = 30% w/v, Vdep = 240 mm·min-1 and F = 1.0 ml·h-1.
Figure 8 – Effect of F on (a) pore size on the X, Y and Z axis, (b) fibre diameter and scaffold porosity.
Figure 9 - Backscatter SEM micrographs of the top view and cross-section of *PCL 3D scaffolds produced with different F (C*PCL=30% w/v and Vdep=240 mm·min-1)
6.3.2 Fusion between filaments
The final scaffold macroscopic architecture, pore size and porosity depend on the processing conditions, as previously reported. One important factor influencing scaffold integrity is the adhesion of subsequent layers. The partial fusion that takes place in the filaments contact points of successive layers prevents the delamination effect, thus enhancing the scaffold integrity. The fusion between the filaments is influenced by factors like C*PCL, F, Vdep, the initial needle tip-to-beaker bottom (Z0) offset and then the distance between the nozzle tip and the layer that is being built. If the distance that the extruded solution covers to reach the bottom of the beaker is too large, the fibre is almost completely solidified when it settles onto the previously deposited layer and, as consequence, the adhesion at the contact point is limited. Inversely, if such distance is too small, the fibre is not enough solidified when reaches the previous layer and a fibre morphology is not achieved causing limited porosity on the Z direction. Similar effects are obviously given by the velocity of the filament during its travelling that is influenced by F and Vdep. The optimised values of initial offset between the needle tip and the bottom of the beaker (1 mm), inter-layer needle translation (0.2 mm), F (1 mL·h-1) and Vdep (240 mm·min-1) allowed to obtain a 3D cohesive architecture characterized by good scaffold integrity. Moreover, the distance between two adjacent filaments of the same layer also influences directly the fusion in the contact point due to the density of the material supported by each filament as observed in the study by Li et al [Li et al., 2009]. The obtained scaffold thickness is strongly dependent on the phenomena described.
In the case of fusion of filaments composing the same layer the overall porosity and reproducibility of the 3D scaffold architecture is compromised. The solvent/non-solvent system strongly influences the phase-inversion process. It has been reported that its incorrect selection might limit or slow the phase inversion of the solution filament and a consequent collapse of the scaffold can occur [Puppi et al., 2011b; Yun et al., 2007].
In our experiments, by increasing C*PCL a less fusion between fibres of subsequent layers was observed. This factor could be related to the change in polymer viscosity and/or in the solvent/non- solvent demixing. Indeed, previously reported studies suggested that increasing polymer concentration a slower solvent/non-solvent mixing takes place [Arbab et al., 2011; Hou et al., 2005; Hou et al., 2006; Kim et al., 1996].
6.3.3 Fibre morphology
The fibres constituting the developed scaffolds showed a highly porous morphology both in the cross-section and in the outer surface (Figure 10).
Figure 10 - SEM micrographs showing fibre porosity: outer surface (a-c) and internal porosity (e-g).
Fabrication parameters: Vdep = 240 mm·min-1, F=1 ml·h-1 and C*PCL from 20 to 30% w/v.
The formation of pores in the fibres, as consequence of the phase inversion process governing solution filament solidification, is related to solvent/non-solvent demixing [Smolders et al., 1992].
In the case of a slow demixing, a uniform porous structure, in terms of size, geometry and distribution, is achieved, inversely a fast demixing originates a dense “skin” on the surface interfacing with the coagulant fluid and an irregular pore distribution typically presenting finger- like pores underneath [Smolders et al., 1992]. An increase of polymer concentration has been proved to lead to a more uniform structure and to inhibit the formation of finger-like voids [Arbab et al., 2008]. Indeed, as shown in Figure 10, an increase of C*PCL from 20 to 30% w/v led to a more porous outer fibre surface.
6.3.4 HA loaded scaffolds
After the optimization of the processing parameters for plain *PCL scaffolds, the fabrication of composite *PCL/HA scaffolds was investigated. By applying the optimised Vdep and F (240 mm·min-1 and 1 ml·h-1, respectively), HA-loaded *PCL scaffolds (HA/*PCL weight ratio of 25%) were fabricated by processing solutions with different C*PCL in the range of 20-30% w/v. *PCL/HA composite scaffolds revealed a morphology close to that of plain PCL scaffolds with a good degree of fibre alignment (Figure 11a,b). For the different HA-loaded *PCL scaffolds, a change of C*PCL
had a significant effect on the obtained d1 but not on the pore size in X, Y and Z axis. Moreover, d1
was significantly smaller (194±6 and 202±2 µm for C*PCL of 20 and 30% w/v, respectively), while the X, Y and Z pore sizes were significantly larger when compared to plain *PCL scaffolds (Table 2). High magnification SEM micrographs showed a good dispersion of HA nanoparticles microaggregates both in the outer surface and in the cross-section of the fibres (Figure 11c,d).
Figure 11 - SEM micrographs of *PCL/HA scaffolds produced with a C*PCL of 30% w/v (=F=1 ml·h-1and Vdep=240mm·min-1): a) top view, b) scaffold cross-section, b) fibre surface, c) fibre cross-section.
As previously reported by Puppi et al [Puppi et al., 2011b], the surface and cross-section porosity of HA-loaded scaffolds was lower in comparison to plain *PCL scaffolds obtained by applying the same processing conditions (Figure 11c,d vs Figure 10c,f). The presence of additives in the polymer solution can affect the kinetics and thermodynamics of the phase-inversion process and therefor the resulting fibre morphology [Kim and Lee, 1998; Mulder, 1996; Puppi et al., 2011b;
Reuvers and Smolders, 1987; Strathmann et al., 1975]. The lower porosity may be due to a faster solidification and correlated to an interaction between polymer and additive which affects polymer solvation and consequently the coagulation process [Puppi et al., 2011b].
6.3.5 Thermal properties
TGA thermograms of both raw polymer and scaffolds show a slight difference in degradation temperature that was around 310ºC (Figure 12). . The degradation temperature of this polymer was previously reported to be 335.86ºC [Balakrishnan et al., 2006]. The amount of HA contained in the composite scaffold can be estimated by residue analysis at 550°C that was 22.92%, nearly the HA weight percentage in the polymeric solution (20% wt).
Figure 12 – TGA thermograms of the raw *PCL and the produced scaffolds (*PCL and *PCL/HA).
The thermal properties of both raw *PCL and produced scaffolds were further characterized by DSC (Figure 13).
Figure 13 – Representative DSC thermograms of raw materials, *PCL scaffolds and *PCL/HA scaffolds.
Results show that no changes on the melting temperature (Tm) and the glass transition temperature (Tg) of the material was induced by processing and HA loading (Table 3). Indeed, the obtained values (Tg ~ -64ºC and Tm ~ -56ºC) were close to those reported by Balakrishnan et al.
[Balakrishnan et al., 2006]. However, the ∆Hm of plain *PCL scaffolds is lower in comparison with
*PCL raw material, suggesting a smaller crystallinity degree. This is in accordance with previous investigations on biodegradable polyesters fibres reporting that the wet-spinning process decreases the crystallinity degree [Gao et al., 2007; Rissanen et al., 2009]. This effect was not observed in HA loaded scaffolds showing a ∆Hm comparable to that of raw materials.
Table 3 – Thermal properties of the scan *PCL and *PCL/HA scaffolds obtained using DSC analyses under the second heating.
Tg (ºC) Tm (ºC) ∆∆∆∆Hm (J·g-1)
*PCLraw material (2nd heating) -64.85 56.28 87.99
*PCL scaffold (C*PCL=30% w/v. Vdep=240
mm·min-1, F=1ml·h-1) (1st heating) -63.52 56.33 73.39
*PCL/HA raw material (physical blending; 2nd
heating) -64.52 55.32 59.34
*PCL/HA scaffold (C*PCL=30% w/v.
Vdep=240mm·min-1, F=1ml·h-1) (1st heating) -64.19 56.45 59.18
6.3.6 Biological evaluation
6.3.6.1 Cell viability and proliferation on scaffolds
Biological investigations of the prepared *PCL scaffolds were carried out in order to evaluate cell viability and proliferation by using MC3T3-E1 cell line. Quantitative evaluation of cell proliferation onto scaffolds was weekly performed by means of WST-1 assay. Results highlighted good values of cell proliferation in all investigated samples and growing conditions (Figure 14).
Figure 14 - MC3T3-E1 cell proliferation cultured on *PCL and *PCL/HA based scaffolds investigated by mean of WST-1 reagent, non-osteogenic (n.o.) and osteogenic (ost.) medium.
Nevertheless, at the very beginning the cell-material interactions and the cell seeding density played a crucial role on cell attachment and influenced the cell proliferation in the early culturing times [Kommareddy et al., 2010]. In fact, during the first two weeks of culture a delay in cell proliferation was observed for both culture conditions and scaffold typologies. This behaviour was especially evident for MC3T3-E1 cultured on *PCL/HA constructs. The HA presence in the construct could positively influence the switch between proliferation and differentiation process [Quarles et al., 1992], by promoting the expression of cell osteoblastic phenotype and causing a delay in cell proliferation. However, despite the initial hard proliferation, as observed since the third week of culture, scaffolds were able to support the proliferation of the MC3T3-E1 for longer culturing times, as confirmed from cytocompatibility assays previously performed on *PCL electrospun meshes [Puppi et al., 2010]. Values of cell viability and proliferation were also
compared to those observed for wet-electrospun linear PCL constructs (see chapter 5), which confirmed the good cytocompatibility as well as the slight delay in MC3T3-E1 cell proliferation when cells were grown in the presence of synthetic HA in the polymeric matrix.
6.3.6.2 Collagen production
Direct red 80 dye or analogues have been used since Sixties for staining collagen in histologic specimens [Lopez-De Leon and Rojkind, 1985], for measuring soluble collagen fixed on glass slide [Junquiera et al., 1979] and as a system to bind and quantify triple helical collagen molecules produced from fibroblast [Lee et al., 1998] and osteoblast cells [Franceschi and Iyer, 1992;
Kudelska-Mazur et al., 2005], as well as to evaluate the phagocytosis of bovine collagen by human monocytes/macrophages [Ciapetti et al., 1996]. Direct red is specific for many collagen types (I, III, IV and V) and gives similar calibration curves for each one [Taskiran et al., 1999]. In the present study, type I collagen was selected for the calibration of the assay because of his predominance in early organic matrix of neo-formed bone tissues. According to this method, collagen production of preosteoblast cells, cultured on *PCL and *PCL/HA scaffolds, was measured and quantified, as reported in the Figure 15.
Figure 15- Collagen production from MC3T3-E1 cells cultured on *PCL and *PCL/HA based scaffolds investigated by mean of Direct Red 80 staining method.
3D constructs displayed the capability to induce MC3T3-E1 cell to produce high levels of collagen proteins as confirmed from previous studies with *PCL electrospun meshes [Puppi et al., 2010].
Both typologies of scaffolds and culturing conditions showed analogues values of collagen with a
sharp increasing time-dependent trend, confirming the biocompatibility and the suitability of the prepared supports as bone substitutes.
6.3.6.3 ALP activity measurement
The ALP bone isoform is a glycosylated membrane-bound enzyme that catalyses the hydrolysis of phosphomonoester bonds and may also play a physiological role in the metabolism of phosphoethanolamine, inorganic pyrophosphate, and pyridoxal 5’-phosphate [Beck et al., 1998].
This enzyme is considered an early indicator of osteoblastic phenotype expression and its activity was measured to determine the MC3T3-E1 preosteoblast cell differentiation.
In the first three weeks of culture MC3T3-E1 grown on *PCL and *PCL/HA scaffolds produced comparable ALP levels with no visible increase during the culturing period (Figure. 16).
Figure 16 - Alkaline phosphatase activity detected on MC3T3-E1 cells cultured on *PCL and *PCL/HA based scaffolds, non-osteogenic (n.o.) and osteogenic (ost.) medium.
Nevertheless, a considerable increase in ALP levels was observed for the samples treated with osteogenic medium. This phenomenon confirmed that ascorbic acid and β-glycerophosphate positively act toward the phenotypic differentiation, triggering off enzyme’s expression [Alcaín and Burón, 1994; Wang and Yu, 2010]. Instead, the presence of a synthetic calcium phosphate ceramic (HA), selected as a reinforcing bioactive agent in the *PCL constructs, did not improve the scaffolds’ osteoconductivity during the first 21 days of culture. Increasing values of ALP was detected at the last endpoint of analysis (day 28) for all the investigated samples. In particular, the
osteogenic medium and the synergy between the modified medium and the HA polymeric matrix, that mimics the natural apatite composition of bones and teeth, seemed to induce MC3T3-E1 cell differentiation by confirming data from the literature [Calvert et al., 2005; Wutticharoenmongkol et al., 2007]. Moreover, this behaviour confirmed the need for the MC3T3-E1 to reach an adequate cell confluence on the 3D constructs and an ample production of early matrix proteins as type I collagen, fibronectin and transforming growth factor beta (TGF-β1), prior to the expression of high levels of ALP, a marker of mature osteoblast function [Park, 2010; Quarles et al., 1992;
Wutticharoenmongkol et al., 2007].
6.3.6.4 Matrix mineralization analysis
In studies using osteogenic cultures, mineralization of the ECM is considered a functional in vitro endpoint reflecting advanced cell differentiation [Hu et al., 2008; Orimo, 2010]. The mineral phase generally deposited in osteoblast cultures is a calcium-phosphate substitute, similar to in vivo HA.
ARS is commonly used to detect calcium content present into ECM [Hoemann et al., 2009]. The ARS forms an alizarin red-calcium complex in a chelation process that emits red fluorescence when opportunely excited. At day 28, data showed the presence of rod shaped bodies emitting red fluorescence, diffusion and dimensions of the observed bodies suggested that neo-formed HA deposition was started for all types of scaffolds and culture conditions (Figure 17).
Figure 17 - Micrographs illustrating red fluorescence emission of calcium bounded ARS on MC3T3-E1 cells cultured on *PCL no (A), *PCL ost (B), *PCL/HA no (C) and *PCL/HA ost (D) at day 28.
Where present in the polymer structure, HA yielded a significant fluorescent background but not comparable with the alizarin specific red light signal. However, calcium deposits were not yet consistent, probably because of the early stage of cell differentiation, as demonstrated from the still increasing values of ALP at day 28 [Beck et al., 1998; Calvert et al., 2005; Wutticharoenmongkol et al., 2007]. Finally, MC3T3-E1 cell line grown on *PCL and *PCL/HA 3D constructs displayed the ability to sustain an appreciable production of mineralized matrix, a marker of osteoblastic differentiation process.
6.3.6.5 Cell culture organization on scaffolds
Fluorescent staining of cytoskeleton and nuclei showed colonization of the cells grown on the
*PCL scaffolds. At early stages of culture (day 7 and 14) microscopic observation showed a poor presence of MC3T3-E1 preosteoblasts adherent on fibres surface (Figure 18a,b), confirming previous data of cell proliferation. Furthermore, cells displayed a variable shape and spreading with consistent F-actin organization comparable with early stages of cell adaptation to the material [Hutmacher et al., 2001], exhibiting great stress fibres stretched along the cytoplasm (Figure 18b).
At days 14 and 21 was rather evident the lower cell number of MC3T3-E1 grown on *PCL/HA constructs instead of those one present on plain *PCL based scaffolds. In accordance with the in vitro differentiation mechanism [Quarles et al., 1992; Wutticharoenmongkol et al., 2007], after an initial phase characterized by replication of undifferentiated cells, preosteoblasts started the differentiation process. In fact, longer culture times (day 28) showed a progressive increase in cells colonizing polymeric meshes for both typologies of scaffolds (*PCL and *PCL/HA) and culture conditions (non osteogenic and osteogenic medium). At this endpoint, beside to undifferentiated spindle-like preosteoblasts, MC3T3-E1 started to exhibit polygonal morphology with a coherent actin structure similar to the osteoblast phenotype, as confirmed from the levels of ALP, collagen and matrix mineralization previously showed. Moreover, higher magnification micrograph (Figure 18b) displayed the beginning of cell penetration and colonization of the lower layers of the meshes.
Finally, microscopic observation suggested that all the scaffolds allowed preosteobalsts growth and colonization, revealing similar results in terms of early stages of cell adhesion. The growing progression was delayed in the *PCL/HA scaffolds, but longer culture times confirmed the suitability of the selected scaffolds to sustain cell colonization and differentiation towards osteoblast phenotype.
(a)
(b)
Figure 18 - CLSM micrographs of MC3T3-E1 cells cultured on *PCL and *PCL/HA based scaffolds at different end-points and growing conditions: 10X (a) and 20X (b) magnifications.
6.4 Conclusions
An innovative computer-assisted wet-spinning technique was developed and employed to produce with a layer-by-layer approach 3D *PCL-based scaffolds. The possibility of producing a wide range of scaffold internal morphology and external shape makes this technique interesting for the preparation of complex scaffolds for TE applications. A proper tuning of C*PCL, F and Vdep enabled to vary in a certain range fibres diameter and its distribution, as well as scaffold porosity and pore size. The fabrication process showed high reproducibility, accuracy and low production time, thus enabling to obtain in automated way highly porous 3D scaffolds with good degree of fibre alignment. The developed *PCL and *PCL/HA scaffolds demonstrated to be fully cytocompatible and able to sustain murine preosteoblast MC3T3-E1 cell adhesion, proliferation and colonization.
Moreover, high levels of early and late osteogenic markers confirmed the suitability of the prepared scaffolds to promote the osteoblastic differentiation process. The in vivo evaluation of these types of scaffolds (*PCL and *PCL/HA) on animal model using New Zealand rabbits is part of an ongoing research that will be published in a forthcoming paper.
References
Alcaín, F.J., Burón, M.I., 1994. Ascorbate on cell growth and differentiation. Journal of bioenergetics and biomembranes 26, 393-398.
Arbab, S., Noorpanah, P., Mohammadi, N., Soleimani, M., 2008. Designing index of void structure and tensile properties in wet-spun polyacrylonitrile (PAN) fiber. I. Effect of dope polymer or nonsolvent concentration. Journal of Applied Polymer Science 109, 3461-3469.
Arbab, S., Zeinolebadi, A., Noorpanah, P., 2011. Exploring the Thermodynamic Aspects of Structure Formation During Wet-Spinning of Polyacrylonitrile Fibres. International Journal of Chemoinformatics and Chemical Engineering (IJCCE) 1, 36-52.
Balakrishnan, S., Krishnan, M., Dubois, P., Narayan, R., 2004. Kinetic and thermodynamic considerations in the synthesis of a new three-arm poly(ε-caprolactone). Polymer Engineering and Science 44, 1491-1497.
Balakrishnan, S., Krishnan, M., Narayan, R., Dubois, P., 2006. Three-arm poly (ε-caprolactone) by extrusion polymerization. Polymer Engineering & Science 46, 235-240.
Beck, G.R., Sullivan, E.C., Moran, E., Zerler, B., 1998. Relationship between alkaline phosphatase levels, osteopontin expression, and mineralization in differentiating MC3T3-E1 osteoblasts. Journal of Cellular Biochemistry 68, 269-280.
Burchard, W., 1999. Solution Properties of Branched Macromolecules. Advances in Polymer Science 143, 113-194.
Calvert, J.W., Chua, W.C., Gharibjanian, N.A., Dhar, S., Evans, G.R.D., 2005. Osteoblastic Phenotype Expression of MC3T3-E1 Cells Cultured on Polymer Surfaces. Plastic and Reconstructive Surgery 116, 567-576.
Celik, A., Kemikli, N., Ozturk, R., Muftuoglu, A.E., Yilmaz, F., 2009. Synthesis, characterization and thermal properties of a novel star polymer consisting of poly([epsilon]-caprolactone) arms
