UNIVERSITÀ CAMPUS BIO-MEDICO DI ROMA FACOLTÀ DI INGEGNERIA DOTTORATO DI RICERCA IN INGEGNERIA BIOMEDICA XXIV CICLO Bioinspired scaffold for regenerative medicine: production engineering and scaffold characterization Supervisor Prof. Marcella Trombetta Candidate Sara Maria Giannitelli, M Eng. Co-supervisor Alberto Rainer, PhD
Index Index... 2 Abstract... 5 Introduction... 6 Chapter 1: Scaffolds in Tissue Engineering... 8 1.1 Scaffold requirements... 8 1.2 Scaffold materials... 9 1.2.1 Natural versus Synthetic polymers... 10 1.2.2 Poly(ε-caprolactone) (PCL)... 11 1.3 Biomimetic scaffolds... 13 1.4 Fabrication techniques... 15 1.4.1 Conventional fabrication methods... 16 1.4.2 Electrospinning... 17 1.4.3 Solid freeform fabrication... 18 1.5 Scaffold characterization... 20 1.5.1 Morphological... 21 1.5.2 Chemical and thermal... 22 1.5.3 Physical... 24 1.5.4 Biological: in vitro-in vivo... 25 1.5.5 Finite element analysis in scaffolds characterization... 27 Objective and thesis roadmap... 27 Chapter 2: Development of an extrusion based deposition system... 29 Overview... 29 2.1 Extrusion-based techniques... 29 2.2 Home made RP system... 31 2.2.1 Development of a dispensing head... 33 2.3 Printing parameters... 37 2
2.4Development of dedicated software modules... 39 Conclusions... 44 Chapter 3: Rapid prototyping of biopolymeric functionalized scaffolds for bone regeneration.. 45 Overview... 45 3.1 Rapid prototyped functionally graded scaffolds... 45 3.1.1 Biomaterials choice... 46 3.1.2 Implemented design... 47 3.1.3 Functionalization protocol... 48 3.2 Home-made RP system performance... 50 3.3 PCL scaffold fabrication... 53 3.3.1 Morphological study... 54 3.3.2 Porosity... 55 3.3.3 Mechanical characterization... 56 3.4 PCL-HA scaffold fabrication... 57 3.4.1 Morphological study... 58 3.5 Scaffold functionalization... 59 3.6 In-vitro experiments on tissue-engineered constructs... 60 3.6.1 Scaffolds biocompatibility... 60 3.6.2 Cell differentiation... 61 Conclusions... 64 Chapter 4: A bioinspired approach to the design of porous additively manufactured scaffolds with optimized mechanical properties... 65 Overview... 65 4.1 Design of scaffold micro-architecture in extrusion based techniques... 66 4.2 Load Adaptive Scaffold Architecturing (LASA) algorithm... 69 4.2.1 Theory and calculation... 70 4.2.2 Design and Finite Element Simulations... 72 4.2.3 Samples fabrication... 76 3
4.2.4 Mechanical testing... 77 4.3 Application of LASA... 78 Conclusions... 81 Chapter 5: Electrospun fibers with nanostructured surface morphology... 82 Overview... 82 5.1 Electrospinning in drug delivery... 83 5.2 Scaffold fabrication and characterization... 85 5.2.1 Scanning electron microscopy (SEM)... 86 5.2.2 Differential scanning calorimetry (DSC)... 88 5.2.3 Attenuated Total Reflectance (ATR)-Fourier transform infrared spectroscopy (FTIR)... 92 5.3 Release study... 97 5.3.1 Short term study... 97 5.3.2 Long term study... 99 5.3.3 Mechanical characterization... 102 5.3.4 Three-stages release model... 103 5.4 Biocompatibility assay... 104 Conclusions... 106 Chapter 6: Conclusions and future work... 107 6.1 Ongoing research... 108 6.1.1 Biomimetic micro-fibrous PLLA scaffold for tendon regeneration: a preliminary in vitro study... 108 6.1.2 Preparation of a polycaprolactone/bioglass hybrid scaffold through a direct sol-gel method... 111 References... 114 Appendix... 122 4
Abstract Scaffolds play a pivotal role in the tissue engineering paradigm, as they provide an extra-cellular matrices onto which cells can attach, grow, and form new tissues. Cellseeded scaffolds can either be cultured in vitro, or directly implanted in vivo, using the body s own system as a bioreactor. The efficacy of the first approach is often questioned due to requirement for at least two surgical procedures and the delay in treatment while the construct is being cultured in vitro before implantation. Thus, the desired biological scaffold performance has consequently shifted from a passive role, where scaffolds were merely accepted by the body, to an active role in which they instruct their biological surroundings in a predictable and controlled fashion. Due to the primary importance of scaffold design, one of the challenges in tissue engineering is to reproduce an analog of the in vivo scenario, mimicking the microenvironment that promote cell-cell and cell-matrix interactions. Between the main issues that need to be considered in the engineering of functional constructs, material selection, tissue specific micro-architecture design and developing methods for scaffolds fabrication, are often included. Aim of this thesis is to explore alternative strategies for the development of bioactive scaffolds, off-the-shelf available and intended to be used to guide tissue regeneration, exploiting the endogenous regenerative abilities of the body. This challenge has been addressed by combining the synthesis of novel biologically inspired design with innovative production techniques, in order to optimize scaffold architecture, composition and mechanical properties, leading to what would more conventionally be termed a smart scaffold. 5
Introduction Disease, injury and trauma can lead to damage and degeneration of tissues in the human body, which necessitates treatments to facilitate their repair, replacement or regeneration. Treatment typically focuses on transplanting tissue from one site to another in the same patient (autograft) or from one individual to another (transplant or allograft). While these treatments have been revolutionary and lifesaving, major problems exist with both techniques. Harvesting autografts is expensive, painful, constrained by anatomical limitations and associated with donor-site morbidity due to infection and hematoma. Similarly, allografts and transplants also have serious constraints due to problems with accessing enough tissue for all of the patients who require them and the fact that there are risks of rejection by the patient s immune system as well as the possibility of introducing infection or disease from the donor to the patient. Tissue engineering (TE) has emerged in the last decade of the 20th century as an alternative approach to circumvent the existent limitations in the current therapies for organ failure or replacement. It has been defined as the application of scientific principles to the design, construction, modification and growth of living tissues using biomaterials, cells and growth factors, alone or in combination. There are three strategies in tissue engineering; (1) the use of isolated cells or cell substitutes to replace those cells that supply the needed function, including genetic or other manipulations before infusion. (2) The delivery of tissue-inducing substances, such as growth and differentiation factors, to targeted locations. (3) The introduction of three-dimensional (3D) matrices (scaffolds), where cells can be either recruited from the host tissues in vivo or seeded in vitro. Scaffold guided tissue engineering approach applies methods from materials engineering and cell/molecular biology to create artificial constructs for tissue regeneration. Cell-seeded scaffolds are either cultured in vitro, to synthesize tissues which can then be implanted into an injured site, or are directly implanted, using the body s own systems as bioreactor. Many clinicians question the efficacy of in vitro tissue engineering due to the requirement for at least two procedures and the delay in treatment, while the construct is 6
being cultured in vitro. From a commercial perspective, this approach also poses problems due to the prolonged regulatory process required before such construct can be approved for clinical use. However, in tissues such as cartilage, which do not have the ability to regenerate themselves when damaged, long term in vitro tissue engineering is currently the only solution to prevent the requirement of an eventual joint arthroplasty. Other tissues, such as bone for example, have an intrinsic ability to repair, remodel and regenerate. TE in this case aims to harness this innate regenerative capacity. One way to do so might be to engineer the scaffold in such a way that itself provides regenerative signals to the cells which might avoid the requirement for prolonged in vitro culture, prior to implantation. Therefore, much ongoing research is devoted to developing more sophisticated scaffolds that optimally recreate extra cellular matrix (ECM) environment in a temporally coordinated and spatially organized structure. In the following Chapter, a general overview of various aspects of scaffold design will be provided, highlighting the specific topics handled in this thesis. Functional requirements involved in scaffold design, types of materials, fabrication and characterization techniques used in developing state of the art scaffolds, challenges and future research direction, will be presented. 7
Chapter 1: Scaffolds in Tissue Engineering 1.1 Scaffold requirements Within TE, one of the major research themes is scaffold design, which ultimately determines the functionality of the construct and of the grown tissue. It comprehends the material and fabrication method used, as well as other features like scaffold shape, size, architecture, and surface topography. Although the final requirements depend on the specific TE application, several general guidelines can be considered for all the designs. The scaffold should be/have: -biocompatible; it should provoke an appropriate biological response of the host tissue and prevent any adverse reaction. -biodegradable; the scaffold should degrade into non- toxic and easily metabolized products without interfering with the function of the surrounding tissue. Rate of scaffold s degradation must be tuned appropriately so that it retains sufficient structural integrity until the neo-tissue (cells and extracellular matrix without vascularisation) is developed. This property strictly depends on the intrinsic characteristics of biomaterial used for scaffold fabrication, including the chemical structure, the presence of hydrolytically unstable bonds, the level of hydrophilicity/hydrophobicity, crystalline/amorphous morphology, the copolymer ratio, and the molecular weight. Other physical and chemical factors such as the overall porosity, the pore size distribution, the scaffold morphology (fiber, foams, 3D printed structure) and the ph at the implantation site, can also contribute to regulate degradation kinetic. -suitable mechanical properties; engineered tissues must possess the appropriate mechanical properties to fulfill their structural role. The required properties depend on the specific application, in e.g. cardiovascular versus bone prostheses. Particularly, in the reconstruction of hard, load-bearing tissues such as bone and cartilages, the mechanical strength to retain the scaffold s structure after implantation is essential. The degradable scaffold should maintain sufficient mechanical strength to manage any in vivo stresses and physiological loadings imposed on the engineered construct. For 8
tissue-engineered vascular grafts, hemodynamic competence and suturing characteristics are also critical. -appropriate porosity; in term of magnitude of the porosity, pore size, distribution and interconnectivity. A highly porous scaffold with an open, fully interconnected porosity and a large surface-to-area volume ratio is desirable to allow cell in-growth and uniform distribution within the material. Compared to a closed pore structure, an interconnected network enhances the diffusion rates to and from the center of the scaffold and facilitates vascularization, thus improving oxygen and nutrient supply and waste removal. Furthermore, a micro-porosity is also required in order to promote capillary ingrowth. The effects of pore size on tissue regeneration have been testified by literature data demonstrating optimum pore size for neovascularization (5µm), fibroblast ingrowth (5-15µm), hepatocytes ingrowth (20µm), for regeneration of adult mammalian skin (20-125µm), and for bone regeneration (100-700µm). Besides size and porosity, shape and tortuosity of pores can also affect the rate and extent of tissue in-growth. However the mechanical strength of a scaffold tends to decrease as the porosity increases. Thus, there may be a conflict between optimizing the porosity and maximizing mechanical properties. - tunable shape; scaffold must have appropriate geometry and size in order to exactly fits the site of replacement. For this purpose it should be manufactured in a reproducible, controlled and cost effective fashion in a variety of shapes and sizes. The interdisciplinary character of these features, combined with the extensiveness of the application field, yields the high level of complexity involved with TE scaffold design. All pieces of the puzzle need to fall in place before the engineered construct can be translated into clinical application. 1.2 Scaffold materials One of the aspect to which all of the criteria listed above are dependent upon, is the choice of biomaterial. Various materials, including biopolymers, bioceramics and metals, have been used to produce scaffolds for several TE purposes. Due to the variation in properties required in soft versus hard applications, especially in term of mechanical behavior, the constructs for these two sub- categories generally use different 9
classes of biomaterials and different processing techniques. For soft TE scaffold, e.g. skeletal muscle or cardiovascular field, a wide variety of polymers are generally applied. On the other hand, hard tissue replacements, e.g. bone substitutes, are generally based on more rigid polymers, ceramics and metals. It s clear that each class of biomaterial has unique advantages and limitations. Clinical studies and laboratory experiments have a main role in the choice of the right material for each application or in the identification of a successful combination of biomaterials. In the past decades, a tendency toward the synthesis of new composites, or hybrid materials was observed. The rationale for designing composites is to provide many attractive properties that each individual material cannot have. For example, in the case of organic/inorganic composites, like polymers with ceramic particles, inorganic component provides bioactivity and higher biocompatibility whereas polymer matrix will ensure mechanical stability. Furthermore, the addition of ceramic particles will act as a buffer to the degradation of byproducts produced by the synthetic polymer. 1.2.1 Natural versus Synthetic polymers Polymeric biomaterials originate from a wide range of natural as well as synthetic sources. Naturally-derived polymer, such as collagen, fibrin, glycosaminoglycans (GAGs), chitosan, alginates and starch, can be extracted from plants, animals or human tissues. They show excellent biocompatibility, low toxicity and low chronic inflammatory response due to the high similarity with native cellular environments, especially when the compounds are naturally present in ECM. Furthermore they are easily degradable by an enzymatic or hydrolytic mechanism. However, they show poor mechanical performance, difficult processing and they can undergo to batch-to-batch variations. Other concerns are the potential pathogen transmission, antigenecity and lack of constant material supply. Synthetic polymers do not show variation in chemic/physical properties and offer high versatility and good workability. The greatest advantages are the ability to tailor mechanical properties and degradation kinetics on the basis of specific application. However, biocompatibility is generally lower than natural polymers. For acidic degradation products, high local concentrations can affect cell growth on the scaffolds, in vitro and cause inflammatory responses, in vivo. 10
One of the most widely used biomaterials is a group of synthetic polymers known as polyesters. These include aliphatic polysters such as polylactic acid (PLA), polyglycolic acid (PGA), poly-ε-caprolactone (PCL), as well as their copolymers or blends. In addition, other synthetic polymers also have significance especially in bone tissue engineering, including poly(methyl methacrylate), polyethylene, polypropylene, polyurethane, poly(-ethylene terephthalate), polyetherketone, and polysulfone. It is beyond the scope of this chapter to discuss all polymers in detail; particular attention will be paid on PCL because of its relevance in the present work. 1.2.2 Poly(ε-caprolactone) (PCL) PCL is an aliphatic linear polyester, obtained by a ring-opening polymerization of ε - caprolactone using a catalyst such as stannous octanoate (Figure 1). It has a glass transition temperature of -62 C and a melting point of about 60 C, depending on the degree of crystallinity, which in turn is dictated by the molecular weight (normally 3000 100 000 g/mol) and, to some extent, by the scaffold fabrication process. It is biocompatible, bioresorbable and a low-cost synthetic polymer. Due to its semicrystalline and hydrophobic nature, it exhibits a very slow degradation rate (2 4 years depending on the starting molecular weight and fabrication process) and has mechanical properties suitable for a variety of applications. It is a Food and Drug Administration (FDA) approved material and has been clinically used as a slow release drug delivery device and suture material since the 1980s (i.e. Capronor, SynBiosys, Monocryl suture). Recent clinical trials, ranging from plugging cranial burr holes, orbital floor support, craniosynostosis, craniofacial reconstructions, to dental applications, have demonstrated favorable results and drawn positive responses. Schantz and Lim et al.[1] reported on clinical use of PCL scaffolds over a 12-month period, in cranial reconstruction of burr holes. They concluded that the burr plugs have excellent biocompatibility and are well tolerated by patients with no detectable signs of acute local or systemic immune reactions. The implant s strength and fracture-resistant properties enable it to be firmly anchored in the surrounding calvarium, leading to stable reconstruction achieving the functional and aesthetic objectives of cranioplasty. However, these developments thus far have been limited to low load-bearing maxillofacial treatments only. 11
PCL also has some rheological and viscoelastic properties that allow it to be formed from a wide range of scaffold fabrication technologies. The ease unto which a polymeric material is accommodated into different scaffold fabrication technologies is a property that should not be underestimated. PCL and its copolymers have demonstrated this utility by being successfully used in electrospinning, gravity spinning [2], phase separation, solid freeform fabrication [3] and microparticles [4, 5], due to the low melting temperature, very good blend-compatibility, FDA approval and low cost. The wide use of PCL and the increasing interest of TE community toward this polymer is testified by the trend of publications regarding only the electrospun PCL meshes, during the last 10 years (Figure 1). Figure 1: PCL molecular structure and publications regarding PCL electrospun meshes during the last 10 years, until March 2011. Sourced from ISI Web of Knowledge [6] One disadvantage of PCL, however, is its high hydrophobic nature, resulting in poor wettability and uncontrolled biological interactions with the material [7]. Surfaces with moderate hydrophilicity are able to absorb adequate amount of proteins, while preserving their natural conformation, unlike hydrophobic or very hydrophilic surfaces. In order to rectify this issue, surface modification techniques can be adopted to alter 12
chemical and/or physical surface properties. The optimal degree of hydrophilicity, however, depends on the cell type and on the specific surface treatment applied. Four main approaches have been tested on PCL surface [6]: - Plasma treatment, which improves the hydrophilicity by forming oxygen-containing groups on the surface. -Chemical treatment with reagents such as sodium hydroxide (NaOH). Scaffold immersion in NaOH aqueous solution introduces hydroxyl groups and side chain modification on PCL surface by the formation of carboxylate ( COOH) groups. -Coating or adsorbing natural ECM proteins, which introduce cell recognition sites for improved cell biomaterial interaction. -Blending biologically active materials with PCL to provide signals to increase cell affinity. Some of these approaches can be adopted also to encourage common used biomaterials to have, for example, bioinstructive and stimuli-responsive properties. 1.3 Biomimetic scaffolds In order to achieve tissue formation the scaffold does not only have to provide a support for cell adhesion and proliferation but also stimulate the desired differentiation of cells and thus promote the secretion of new ECM constituents. Both physical and chemical surface properties are of major importance for this purpose. Physical factors include the wettability, surface energy and micro-nanotopographic properties, such as roughness. From a chemical point of view, the presence of biological binding sites and the released products are also critical. Recently, several strategies have been developed to enhance tissue regeneration tuning this chemical and physical factors in order to mimic the natural environment in which cells grow. ECM has an instructive role providing a dynamic and spatially heterogeneous constellation of microstructural, compositional and mechanical cues that influence cell behavior. Harnessing the mechanosensitive capacity of cells, for example, provides immense opportunities for tissue regeneration. Scaffold mechanical properties have profound biological consequences in terms of implant bioactivity versus failure, transmission of mechanical stimuli, and for a wide range of processes at the tissue, cell and sub-cellular levels. Key roles in molecular signaling pathways are played by cell adhesion 13
complexes and cellular cytoskeleton, whose contractile forces are transmitted through transcellular structures. Therefore, the mechanical properties of the substrate to which cells are attached are critical to the regulation of cellular mechanotransduction and subsequent cellular behavior (attachment, proliferation and differentiation). It is now clear, for example, that substrate stiffness can regulate both the behavior of mature cells and the differentiation pathway of stem cells. In fact, when MSCs were grown on firm gels that mimic the elasticity of muscle, differentiation down a myogenic lineage was observed, whereas when MSCs were grown on rigid gels that mimic pre-calcified bone, they differentiated down an osteogenic pathway. Therefore, increasing research is now being directed at utilizing the mechanosensitive capacity of cells to develop scaffolds and biomaterials with specific mechanical properties which can be used to direct the behavior of the cells with which they interact. [8, 9] In addition to biomechanical signals, cellular behavior is strongly influenced by biological and biochemical cues. Appreciation of the complexity of the cell response to ECM signaling has helped in the development of 3D scaffolds that imitate its properties. Generally, cells bind to ECM trough the receptors on the cell walls. One class of receptors are integrins, which bind selectively to specific binding sites such as arginineglycine-aspartic acid (RGD) tripeptide found in cell adhesive proteins such as vibronectin, laminin and fibronectin. Beside attachment these connections mediate several intracellular signals that define mobility, cellular shape and regulate cell cycle. Therefore, several attempts have been made to develop biomimetic materials that mimic integrin-binding in various biological systems. In addition, ECM proteins such as collagen I, laminin, fibronectin, vitronectin, and fibrinogen as well as peptides designed from these proteins have been applied onto scaffolds to enhance cell adhesion and proliferation. The use of scaffolds as delivery systems for growth factors, adhesion peptides and cytokines is receiving considerable attention in TE field. Incorporation of angiogenic growth factors in scaffolds in order to improve their vascular potential is one of the most investigated topic. [10, 11] Another area of critical importance is controlling, and understanding, the host immune response and preventing infection following implantation. To this end, the incorporation of drugs (i.e. inflammatory inhibitors and/or antibiotics) into scaffolds has been proposed as a method to reduce the possibility of infection after surgery. Finally, the use 14
of scaffolds as delivery systems for therapeutic genes is undergoing considerable investigation. Gene therapy approaches (viral and non-viral) which utilize DNA encoding for therapeutic genes potentially provide a stable and effective approach to allow sustained and controlled release of therapeutic factors. [8] 1.4 Fabrication techniques Considerable research effort has been devoted to develop and optimize fabrication techniques apt to the production of scaffolds with architectures mimicking the structure of the tissue to be repaired. Conventional methods include: solvent casting and particulate leaching, gas foaming, fiber meshes and fiber bonding, phase separation, melt molding, emulsion freeze drying, solution casting and freeze drying. More innovative and promising approaches concerns the use of electrospinning and rapid prototyping. Yet, also these fabrication methods have some disadvantages if considered alone. On the basis of the selected technique, scaffold design can be divided into two broad categories. The first design incorporates a precise geometrical layout and scaffolds that fall into this category include construct with regular pore (Figure 2a) and woven textile meshes (Figure 2b). Controlled pore structures can be achieved by rapid prototyping, while woven textile meshes via precise weaving techniques or particular electropsinning set-up. The second category involves the formation and deposition of scaffold struts or walls in a non-precise manner. Structures which belong to this category include foams (Figure 2c), obtained via porogen decomposition, gas forming or salt leaching, and random micro-nanofiber meshes (Figure 2d), fabricated by electrospinning or phase separation methods. [12] 15
Figure 2: SEM pictures o different type of matrices. (a) PCL scaffold with a laydown pattern 0 /90 created by RP technique. (b) PLGA woven textile mesh. (c) Hybrid foams obtained by gas foaming process. (d) Electrospun PCL random fibres 1.4.1 Conventional fabrication methods Conventional fabrication techniques have been broadly used to make 3D scaffolds for tissue-engineering and drug-release applications. However, a number of drawbacks can be outlined in term of control of tissue formation and drug release profile. Although the pore size and shape of these matrices are controllable to some extent, not completely interconnected and tortuous pathways are created. Furthermore, the incorporation of drugs, growth factors, or other biological agents is hampered by the processing conditions. In thermoplastic technologies, the high temperatures involved in the manufacture can compromise the stability of the compound to be integrated, resulting in its denaturation and loss of activity. In solution-based techniques the solvents used can also hinder the stability of the desired biological factor due to a ph change, which will promote aggregation and loss of activity.[13] 16
Tesi di dottorato in Ingegneria Biomedica, di Sara Maria Giannitelli, 1.4.2 Electrospinning Electrospinning has attracted great interest due to its ability to process polymeric solutions into fibrous structures at the micro/nanoscale by simply controlling few process parameters. Such matrices are characterized by a high surface area-to-volume ratio and resemble the physical structure of protein fibrils in native extracellular matrix. Furthermore, its low cost in constructing different set-up, its versatility a wide range of possible polymer solutions can be employed for the obtainment of scaffolds with different features and reproducibility are to be included between major advantages. The electrospinning process is based on the application of an electric field between a polymeric solution and a metal ground collector. When the electric field reaches a critical value, the electrostatic force overcomes the surface tension of the polymeric solution and a charged polymer jet is ejected from a capillary tip or needle. As the jet travels towards the grounded collector, it undergoes a stretching process together with a fast solvent evaporation. This process results in the formation of a random non-woven mesh composed of solid and continuous fibers. Varying the process parameters, e.g. strength of the electric field, distance needle collector, polymer concentration, allows tuning of the fiber diameter. Figure 3: electrospun nanofibers with various morphologies and assembled patterns. [14] 17
Additionally, changing the type of collector or the set-up configuration, different fibers organization can be achieved (Figure 3). For example, while depositing fibers on a static collector plate produces a randomly oriented nonwoven fiber matrix, deposition on a high speed rotating drum or mandrel produces aligned fiber matrices. 1.4.3 Solid freeform fabrication Solid freeform fabrication (SFF) techniques have been explored as an alternative method to improve scaffolds manufacturing. The main advantages are both the capacity to rapidly produce complex 3D models and the ability to process various raw materials. In TE field, SFF have been used to produce scaffolds with customised external shape and predefined internal morphology, allowing the control of pore size and pore distribution [15]. Although there are several commercial variants, the general process involves the use of a computer model derived from CAD or CT/MRI data, converted into.stl file format and sliced into thin cross-sections (Figure 4). Figure 4: Tissue engineering of patient-specific bone grafts. CT scan data of the patient defect (a) are used to generate a computer-based 3D model (b). This model is then imported into RP system software to be sliced into thin horizontal layers, with the tool path specified for each layer (c). The sliced data are used to instruct the RP machine (d) to build a scaffold (e) layer by layer, based on the actual shape of the computer model (c). 18
Utilization of computer-aided technologies in tissue engineering has evolved in the development of a new field of Computer-Aided Tissue Engineering (CATE). It can be defined as the application of enabling computer-aided technologies, including computer-aided design (CAD), image processing, computer-aided manufacturing (CAM), and rapid prototyping (RP) and/or solid freeform fabrication (SFF) for modeling, designing, simulation, and manufacturing of biological tissue and organ substitutes. According to broad diffused classification [16], CATE embraces three major applications: 1) computer-aided tissue modeling, including anatomic, biophysics and CAD-based modeling; 2) computer-aided tissue informatics, including tissue classification applied to tumor detection, morphometric and cytometric study; 3) computer-aided tissue scaffold design and manufacturing. In this latter case, the final architecture can be produced either directly using an additive manufactured process, or indirectly, by producing a sacrificial mould into which a biomaterial is cast. In this alternative route, pioneered by Chu et al. [17], the negative mold, which encompasses both the external shape and the internal porous architecture of the bone scaffold, is designed using CAD software and produced using RP techniques. The mold is, finally, removed chemically by using solvents or thermally by melting or burning the mold. Indirect RP scaffolds manufacturing methods offer the possibility of using biomaterials and design that cannot be processed directly via additive manufacturing but are inappropriate for developing hydrogel scaffolds, because it cannot be removed without damaging both internal and external architecture. Other disadvantages are the longer and more complex production process, and increased toxicity due to solvents presence. Commercially available additive manufacturing techniques may be categorized into three major groups based on the way materials are deposited [18]. The first group includes laser-based machines that either photopolymerize liquid monomer (Figure 5d) or sinter powdered materials (Figure 5c). The second major group actually prints material, including printing a chemical binder onto powdered material (Figure 5a) or directly printing wax. The third major group is of nozzle-based systems, which process material either thermally or chemically as it passes through a nozzle (Figure 5b). According to the type of feedstock (filament, powder, granulate) and to the extrusion 19
mechanism (mechanical or pneumatic), extrusion-based techniques have been further classified as Fused Deposition Modeling (FDM), Precision Extrusion Deposition (PED), Bioplotting and other variants. Figure 5: (a) 3D printing process, (b) fused deposition modelling process, (c) selective laser sintering process, (d) stereolithography system [15] 1.5 Scaffold characterization In vitro and in vivo biological experiments are of major importance but beforehand scaffold mechanical, structural and chemical analysis is required. First of all, scaffolds characterization enables to evaluate advantages and drawbacks of the fabrication methods applied. Secondly, the architecture of a scaffold will influence the mechanical properties and the biological performance. It is thus crucial to understand how these characteristics are correlated. A brief overview of some of the most important scaffold parameters with the corresponding characterization techniques has been reported in Table 1 and discussed in following paragraphs. Particular attention has been devoted to such properties considered in this thesis work. 20
Properties Bulk properties Molecular weight Chemical composition/structure Thermal properties Porosity, pore size Morphology Mechanical properties Degradation properties Surface Properties Surface properties Orientation of groups Texture Surface energy and wettability Techniques Gel permeation cromatography (GPC) Nuclear magnetic resonance (NMR), X-ray diffraction (XRD), Fourier transform infrared (FTIR) and FT-Raman (FTR) spectroscopy Differential scanning calorimetry (DSC) Mercury intrusion porosimetry, gas pycnometry Scanning electron microscopy (SEM) Mechanical testing In vitro, in vivo Electron spectroscopy for chemical analysis (ESCA), static secondary ion mass spectrometry (SIMS) Polarized IR SEM, atomic force microscopy (AFM) Contact angle measurement Table 1: characterization of bioresorbable polymeric scaffolds 1.5.1 Morphological Depending on fabrication technique, scaffold architecture can be defined by different parameters, namely, porosity, pore size, surface area to volume ratio, fiber diameter and alignment/anisotropy (in case of fibrous matrix), interconnectivity and tortuosity. Scanning electron microscopy (SEM) is one of the most commonly used technique because of its fundamental role in the determination of scaffold morphology and in giving feedback for synthesis and processing optimization. Different dimensions can be evaluated by SEM and correlated with processing parameters (such as the nozzle used in the electrospinning or extrusion based process). The programmed structure in term of lay-down pattern, fibers alignment or pores size, can be verified by SEM as well as pores interconnection and uniformity. Information obtained from SEM analysis can also be used for porosity calculation. Mercury porosimetry, gravimetry, liquid intrusion, gas pycnometry, gas adsorption, and capillary flow porosimetry are devoted uniquely to this purpose. 21
Scaffold porosity is intended as the ratio of the volume of pores (Vp) to the apparent volume (Va, overall geometry volume including pores). Represented as a percentage, it can be expressed by the general formula: Vp Porosity% = 100 Va Depending on the method applied to measure or estimate the pore volume, this formula could be modified in other equivalent expressions. For example, if polymer density is known, porosity value can be calculated as suggested by [12]. Vt Porosity% = 1 100, Va where Vt (true volume) is the scaffold actual volume calculated as the ratio between the mass and the density of the material that makes up the scaffold. Other theoretical estimation of porosity include the Archimedes method and the liquid displacement method. In the case of RP scaffolds is also possible to make purely theoretical estimation of the porosity on the basis of scaffold volume or mass measurement and the knowledge of the lay-down pattern. More recently, micro computer tomography, showed great potential in analyzing scaffold both in vitro and in vivo. This fully non destructive technique allows direct quantification of the micro-structural architecture of scaffolds in three-dimensions. µct imaging is based on the attenuation of X-rays in a 2D-cross section. The reduced intensity of emergent X-rays is detected and the 2D section is reconstructed pixel by pixel by the use of selected algorithms. However, this technique is manly used for 3D porous scaffolds analysis due to it resolution typically down to 6 µm which is too large to be useful for micro-nano structural analysis. [12, 19] 1.5.2 Chemical and thermal Information on scaffolds chemical composition and structure can be obtained by nuclear magnetic resonance (NMR) spectroscopy, X-ray diffraction, Fourier transform infrared (FTIR) and FT-Raman (FTR) spectroscopy. In particular, FTIR has been widely used in biomaterial study especially in the attenuated total reflectance sampling mode. It is 22
useful to provide a quick, semi-quantitative method for confirming the presence of additives to the main polymer structure, such as hydroxyapatite and water soluble components, or the presence of residual solvents. The ATR method is normally considered non-destructive, however, good contact between the sample and ATR crystal requires applying significant pressure which can damage delicate scaffold morphologies. The technique is generally considered as being a surface analysis tool as sample depths are typically 0.5 2 µm. Noncontact reflectance mode is also possible. Although bulk analysis techniques are critical for achieve a scaffold complete characterization, they are of more routine nature than surface ones. The use of surface materials different from the bulk is recently becoming a common strategy for regulating biological interactions and cell response. Electron spectroscopy for chemical analysis (ESCA) and static secondary ion mass spectrometry (SIMS) are between most powerful tools for analyzing surface chemistry and composition. Furthermore, information on surface wettability and energy can be obtained by contact angle measurements. Thermal properties such as glass transition temperature, melting temperature and crystallinity can be determined by differential scanning calorimetry (DSC). DSC measures the amount of energy absorbed or released by a sample, as it is heated or cooled, providing quantitative and qualitative data on endothermic and exothermic processes. The heat of fusion is thus used to compute the percentage crystallinity, X C calculated as: X C ΔH =, ΔH m 0 m where, ΔH mis the measured enthalpy of melting and 0 ΔH m is the enthalpy of melting of 100% crystalline polymer. Crystallization process depends on the capacity of polymer chains to move and form a crystalline structure. The presence of crystallites in the polymer usually leads to enhanced mechanical properties, unique thermal behavior, increased fatigue strength and can affect degradation behavior. Processing techniques are known to modify polymer crystallinity. Other techniques, such as wide angle X-ray diffraction (WAXD) and atomic force microscopy (AFM), can be adopted to evaluate scaffold thermal 23
properties as well as they can be derived from polymer viscoelastic behavior detection. [6, 20] 1.5.3 Physical Degradation Scaffolds degradation should be tested in vitro under conditions that best simulate real environment. Degradation kinetics can be affected by fatigue loading or mechanical stress which may results in accelerated degradation of polymer due to the mechanochemistry. Rate degradation can be monitored using changes in mass and molecular weight of the polymer. The mass of each scaffold (W 0 ) should be measured prior to the in-vitro study in a dry condition using a balance with high resolution. For most in vitro degradation studies, scaffolds are immersed in Phosphate Buffered Saline (PBS) at 37 C for different periods of time. After extraction from degradation media, the scaffolds should be first carefully dried in a vacuum and their mass re-measured (W f ). Weight loss can be assed as: Weight loss% = W f W W 0 0 100 Changes in the average molecular weight of a polymer may be determined as a function of degradation time using gel permeation cromathography (GPC). Other techniques to detect molecular changes include measuring intrinsic viscosity. Mechanical properties Accurate measurement of mechanical properties of scaffolds for biomedical applications is essential, to guarantee they can withstand forces during surgical operation and those exerted by physiological activities and/or by tissue growth both at short and long term. Depending on the scaffold target application, different possible mechanical tests can be performed. Conventional quasi- static techniques or fatigue tests are very popular, providing the stiffness and strength of the material, as well as its long term mechanical resistance under cyclic loads. Quasi-static tests are performed in 24
universal testing equipments in which a prescribed displacement rate is selected and controlled. Precise standards have been developed to regulate these testing methods (ISO, ASTM) providing general guidelines to define a suitable number of specimens to be tested and other experimental conditions. Fatigue tests are cyclic dynamical mechanical tests in which a cyclic stress (or strain) is applied to the test specimen and the strain (or stress) response of the material is recorded. Other techniques such as AFM-based nanoindentation have also been introduced for the measurement of local stiffness, hardness and flexural properties of a single electrospun fiber. Mechanical analysis can be performed at different stages of research. It can be used as a strategy to correlate scaffold architecture to its mechanical strength and thus create predictive models. These models make it possible to design a certain scaffold in function of a targeted mechanical strength and porosity. Secondly, it can be implemented as a quality control for the evaluation of mechanical properties of the produced scaffolds. Finally, it can be performed at different stages of in vitro/vivo experiments, in order to evaluate whether the degradation kinetics of the polymer match the tissue formation or how cell colonization affects scaffold mechanical behavior. For a mechanical test to be accurate and repeatable it is important to report macroscopic dimensions (gauge length and cross-sectional area), the strain rate, the applied load, as well as whether they have been performed at room temperature or under physiological conditions (at 37 C, in PBS or culture media). In order to compare scaffold properties with that of target implant site, natural tissue are also tested in the same conditions and parameters.[21] 1.5.4 Biological: in vitro-in vivo The first step for a successful biological characterization is scaffold sterilization. Methods vary greatly between different laboratories: ethylene oxide, UV radiation or soaking in ethanol are the most commonly used. Moreover, given the micro- or nanofeatures of the scaffolds, sterilization treatment has to be chosen not to detrimentally affect the scaffold. Ethanol soaking treatment, which is commonly considered as good laboratory practice for the sterilization of biopolymers, showed severe limitations in terms of bacterial contamination. Autoclave and dry oven cycle, which are other very common sterilization techniques of proven effectiveness even in the clinical setting, 25
induced modifications in the crystallinity of the polymer, which may reflect on other properties, such as degradation time. UV and hydrogen plasma treatments appeared to be the most suitable methods for sterilization in terms of structural preservation, crystallinity and bacterial contamination. [22] In order to evaluate cell-viability, cell distribution, proliferation and differentiation, many techniques are available. SEM, confocal microscopy and cell metabolism assay are between the most known. Another emerging technology is the real-time PCR for absolute quantification of gene expression (aqpcr) instead of reverse transcriptionpolymerase chain reaction (rt-pcr). The improvements in cell culture conditions and the development of bioreactors have greatly enhanced the reliability of in vitro experiments. In recent years, a plethora of bioreactors have been developed in order to improve cell seeding, attachment, proliferation and differentiation by the use of this more biomimetic microenvironment for cell culture. In fact, cell seeding in three-dimensional scaffolds is particularly difficult but essential for the following cellular steps of proliferation and differentiation. Diffusion through culture medium and tissue, typically limits oxygen transport in vitro, leading to hypoxic regions and limiting the viable tissue thickness. For this reason, one of the most common bioreactors nowadays is the perfusion bioreactor, where a fluid flow is forced through the scaffold to give a very high seeding efficiency, remove metabolic waste and transport nutrients towards cells. This type of bioreactors allows the combination of 3D culture and perfusion flow and thus mimics the physiological environment. Stephens et al. recently created such a perfusion flow bioreactor with an upright microscope in order to allow the real time imaging of the cell/material interactions [23]. This might be very helpful in gaining deeper insight in the influence of surface properties, type of biomaterial, presence of flow-induced shear and scaffold architecture on cell response and tissue in-growth. In vivo experiments are performed on the most interesting tissue engineering constructs. As an important step towards clinical application, research in large animal models is essential. Nevertheless, results obtained from in vivo studies cannot directly be extrapolated to their clinical performance. The existing assays, both in vitro and in vivo, need improvements in order to be predictive for clinical application. Briefly, the in vivo testing of a polymeric scaffold involve the surgical creation of a defect in the tissue of 26
interest and the implant of the scaffold. The animal and site of implantation will then be studied for a desired duration of time at designated intervals, in order to evaluate the progress in tissue regeneration in presence and in absence of the scaffold. Histological appearances of implants are analyzed at the selected time points. In this way, not only scaffold biocompatibility but also tissue formation can be assessed. [24, 25] 1.5.5 Finite element analysis in scaffolds characterization Biomaterials science has mainly adopted a trial-and-error approach, with modifications being made to an existing scaffold design on the basis of in vitro or in vivo results. Finite-element analysis (FEA) contributed to the reduction of experimental tests and to shortening scaffolds design process. It has been first used in tissue engineering for a post-hoc investigation of the scaffold mechanical behavior and to predict its interaction with the surrounding tissues. Such analysis can be used to vary several geometrical or material parameters at the same time and to choose the most suitable ones for the replacement of a desired natural tissue. CAD and micro-ct based models of the produced porous scaffolds are normally considered as input geometry. The reasonably accuracy of simulation results, in comparison with experimental data, has led to the recent use of FEA as a predictive tool for the a priori design of scaffold architecture. [25] Objective and thesis roadmap This thesis aims at developing new bioinspired scaffolds for regenerative medicine applications, by merging the synthesis of novel biomimetic materials with innovative production techniques. To reach this objective, research activity will embrace many of the key issues involved in TE, spanning from chemistry of the materials, cellular and molecular biology of the interaction between scaffolds and living cells, to the automatic tools for 3D scaffold fabrication. Due to its good handling properties and consolidated clinical application, PCL has been used, alone, in combination with other biomaterial or biologically active molecules, for scaffolds fabrication. A set of technologies for conveniently process polymers or composites into scaffolds with graded features and composition, have been employed. 27
However, since the huge number of topic involved within scaffold design framework, we will give insight only in some major aspects and in parameters involved to provide their optimization. In particular, selected topics belong to three classes: material properties, surface micro-structural characteristics and scaffold 3D architecture. Concerning these aspects, distinct strategies for functional scaffold design have been provided. In particular: - a rapid prototyping station for the fabrication of 3D composite scaffolds, with external shape tailored on the specific patient needs as well as controlled internal microarchitecture, has been developed. Taking advantage of this machine, 3D smart scaffolds have been produced by exposing bioactive molecules on their surface with the aim to promote the synthesis of neo-ecm and drive osteogenic differentiation of human mesenchymal progenitor cells seeded therein. (Chapter 2 and Chapter 3) -A new bioinspired design approach has been developed for the fabrication of 3D scaffold architecture highly optimized in term of porosity and mechanical properties. This method is based on the a priori finite element analysis (FEA) of the physiological load set at the implant site. Combination with solid free-form fabrication process has led to the implementation of the proposed approach to a clinically relevant example. (Chapter 4) - A methodology for the production of PCL electrospun fibers with surface morphology and degradation rate tunable on the specific application requirements, has been proposed. This methodology has the potential to increase the versatility of electrospun materials in several biomedical applications, ranging from soft tissue regeneration to controlled drug delivery. (Chapter 5) 28
Chapter 2: Development of an extrusion based deposition system Overview The possibility of producing tissue engineering constructs by means of freeform fabrication techniques, starting from patient specific medical imaging data, has raised relevant scientific interest in the last decade. In this chapter, the development of a robotic printing device for the fabrication of scaffolds with controlled and reproducible microarchitecture, has been described. The first part is focused on the evaluation of different mechanical set-up, particular attention has been dedicated to the optimization of the microfluidic head devoted to the dispensing of biomaterials. The second part mainly pertains to the description of software modules for the control of the printing operations (linear actuator, electrovalve etc), and of algorithms for the generation of toolpath with assigned porosity and design features. Several examples of scaffolds obtained by the implementation of these control routines have been shown. 2.1 Extrusion-based techniques During the last decade, technologies for the fabrication of polymeric scaffolds benefited from the development of several rapid prototyping (RP) techniques to produce freeform porous structures, which geometries can be derived from solid models easily obtained, for example, from diagnostic medical imaging [26, 27]. Advances introduced by rapid prototyping have significantly improved the ability to control scaffold architecture (size, shape, interconnectivity, geometry and orientation), yielding to biomimetic structures varying in design and material composition, thereby enhancing control over mechanical properties, biological effects and degradation kinetics. Integration with imaging techniques has allowed the production of scaffolds that are 29
customized in size and shape to be tailored for specific applications or even for individual patients. Additive manufacturing techniques have been widely investigated for the processing of thermoplastic biopolymers into structures with controlled shape and tailored porosity [28-30] and have been successfully applied to the preparation of several scaffolds for different applications, such as bone [31], osteochondral defects [32-34] and lumbar interbody fusion [35]. In the following section the attention has been focused in the description of extrusion based technique because of their high similarities with our developed system. Fused deposition modeling (FDM), introduced in 1989 and marketed in 1990 by Stratasys Inc., is one of the most widely used technology. This process combines heat and extrusion techniques to create three-dimensional scaffolds layer-by-layer. Parts are fabricated by stacking layers, each of which is produced using an extruded road onto a substrate in a specifically designed pattern or raster. The platform is lowered and subsequent layers are built directly on top of the other. The polymer, in filament form, is guided into the extrusion chamber by two rollers that provide the necessary pressure to force the material through the nozzle (Figure 6a). Due to the costly and time consuming process necessary for the precursor filament fabrication, this feature represents a severe drawback that limit the range of available materials. The filament should also be long and continuous thorough the build process because any breakage will result in the stopping of the system. Furthermore, with even slightly brittle materials, it has been observed that backpressure from the chamber can cause buckling and breakage of the filament. 30
Figure 6: schematic representations of fluid dispensing approaches, (a) rollers with filament, (b) rotary screw, and (c) pressure [36] Due to these limitations, a variation of the FDM process, called precision extruding deposition (PED) system, has been developed by Shor et al. [37] The hallmark difference between PED and conventional FDM is that the scaffolding material can be directly deposited without filament preparation, allowing the processing of ceramic composites. Bulk material in granulated form is fused in a liquefier chamber while pressure created by a rotary-screw mounted on a high-precision positioning system forces the fluid out of the needle (Figure 6b). In this work, a variation of this latter method has been proposed by the introduction of a compressed air extrusion system instead of the precision screw (Figure 6c). 2.2 Home made RP system A 3D printer, suitable for solvent-based and extrusion deposition techniques, was designed and manufactured. The printer was assembled using commercially available precision linear actuators, controlled by a dedicated motion control unit, interfaced to a PC. The device was supposed to have 3 translational degrees of freedom for the motion of the sample/dispensing head, with the additional requirement of allowing the isotropy of the fabricated sample over 2 dimensions. The adopted configuration is composed by a horizontal 2-axis (X-Y) motorized sample holder plate and a motorized vertical axis (Z) for the printing head. The linear guides used for the assembly of the robotic station were chosen so to allow the positioning of the dispensing head in the horizontal plane with an accuracy and repeatability better than 1 µm, a value that allows a sufficiently accurate control of the trabecular geometry of the scaffolds. A coarser repeatability was 31
chosen for the Z axis (above 10 µm), due to the looser positioning requirements in that direction. The motorized stages are connected to a computer-programmable motion controller (model SM-32, Micos) and powered by a dedicated power amplifier (model MPA, Micos). The motion control unit (MCU) was interfaced to a desktop PC via an ISA bus, whose high speed connection (8 MHz) guaranteed the implementation of real time control algorithms. The MCU platform, therefore, was equipped with trajectorybased control routines, in order to provide the maximum flexibility and ease of use for the printing process. As a result of this identification, the final set-up was developed, including: - a 2 axes motorized moving sample holder (Micos LS110 and LS85); - a motorized Z stage, holding the dispensing head (Micos PLS85); - a rotating mandrel actuated by a stepper motor (Astrosyn model 17 PM, Minebea Co., USA); - a PC-based motion controller (Micos, DMC); - an AISI 316L stainless steel heated dispensing head; - a two axes joystick for manually controlling movements of the sample holder; - USB microscope. The final version of the 3D printing device adopts the following configuration: Parameters Value Notes Number of axis 4 Note: Two axes for the movement of the sample plate, one axis for the printing head Working area 305 mm (X axis) 101 mm (Y axis) 190 mm (Z axis) Holding plate size 100 100 mm 2 Samples are dispensed on a PTFE substrate Resolution 1µm (X axis) 0.5 µm (Y axis) 0.5 µm (Z axis) Repeatability 1 µm(x axis) 1 µm(y axis) 10 µm (Z axis) 32
Speed 90 mm/s Maximum speed, actual speed ranges from 1mm/s to 1cm/s. Extrusion chamber volume 65 ml Actual dispensable volume Extrusion pressure up to 50 bar Gas: N 2 Extrusion temperature 200 C Table 2: 3D printing set-up conditions 2.2.1 Development of a dispensing head Particular attention has been dedicated to the optimization of the extrusion system devoted to the dispensing of biomaterials. Two different approaches has been investigated: solved based deposition and fused based deposition. After a preliminary investigation, a single dispensing method has been selected for scaffolds fabrication. First version: solvent based deposition A first version of the printing head was designed and developed with the aim to cast a polymeric solution which is rapidly solidified by solvent evaporation. Casting was performed via extrusion using a pressurized aluminum syringe installed on the Z-axis (Figure 7). The dispensing head was equipped with a high pressure supply line, including one inlet valve and an outlet one. A standard commercial needle was attached at the end of the chamber. 33
Figure 7: 3D printer, preliminary setup (solvent based deposition) The properties of the filaments (e.g. diameter and spacing) were controlled by the sample holder movement speed and the pressure within the extrusion chamber. Several tests were performed, in order to identify the feasible combinations of printing parameters: needles diameter, solutions viscosities, pressure and deposition velocity. One of the identified viable combinations included a solution of Poli(ε-caprolactone) in Dichloromethane (PCL/DCM, 40% w/v) with a 30 gauge commercial needle. However, the 3D microstructure obtained by solvent based deposition was irregular, giving a construct with low attitude to host cells. This problem, added to other manufacturing drawbacks, mainly due to the evaporation of the solvents after 2-3 hours of processing and to the presence of solvent residuals within the scaffold porous structure, suggested the choice of a different printing technique. Second version: heat based extrusion process Since the microstructure of the scaffolds produced by solvent based deposition presented several limitations, a pressure assisted fused deposition technique was investigated. The aim was to create models out of heating thermoplastic material, extruded through a nozzle, positioned over a computer-controlled table. This required a 34
modification on our 3D printing equipment, providing a heating mantle capable of melting the polymer inside the extrusion syringe (Figure 8). For this purpose, the dispensing head was embedded into an aluminum heating mantle, supplied with two heating cartridges and a resistance thermometer (Pt100), connected to a programmable temperature controller (model 400, Gefran, Brescia, Italy). Figure 8: heat based melting setup and detail of the heating chamber The chamber temperature value was chosen according to the rheological properties and the melting temperature of the thermoplastic material. A nozzle was attached at the end of the liquefier with the function to reduce the cross sectional area of the melt as well as to shape the thermoplastic material according to the desired road configuration. All the electronic components, the connections necessary for the heating element and for the electrovalves control system, were fixed and connected together in a step by step procedure which involves wire stripping, soldering of components, cable connection and closing in a rack. A concurrent activity concerned the identification of the best substrate to dispense on. Several supports were tested including: glass plates, plastic Petri dishes or other plastic supports. Because of outstanding chemical inertia and good adhesion properties with respect to the adopted polymeric solutions, a Teflon support was chosen and properly fixed to the moving platform. 35
Manufacturing parameters were optimized in order to cope with the new properties of the mould. Most of them are interdependent and require to be modified and reoptimized each time materials used in the process changed. Preliminary tests identified problems related to radiated heat coming from the heating chamber. As a solution, a cooling fan was adopted to provide rapid solidification of the extruded structures and prevent collapsing of the trabecular elements of the scaffold. After manufacturing parameter optimization, fabrication of reproducible sets of multilayer 3D scaffolds composed by PCL and PCL/HA (hydroxyapatite) blends was possible. However, with this system it was found to be difficult to go beyond a certain fiber resolution and to process composites with high viscosity melt-flow properties. Furthermore, minor drawbacks as the small dimension of the reservoir (4 ml) and the necessity to optimize the heat flow in the extrusion direction suggested to modify the design of the dispensing head, extrusion chamber an heater, both. Optimized version of the heat based dispensing head In the new design, the volume of the aluminum reservoir was incremented up to 65 ml, allowing a remarkable printing autonomy before getting back into refill operations. An heating wire was wrapped around the tip housing to locally reduce the viscosity and minimize the resistance downward by means of a gradient of temperatures (Figure 9b). By this modification we were able not only to reduce fibers diameter but also to extrude composites with higher percentage of ceramic component. To further increase the performance of the improved system a new set of dispensing needles was considered (SUBREX, Figure 9c). Subrex nozzles, with their conically-shaped geometry, offer superior clog-resistant performance with reduced back pressure at increased flow rates. Heating chamber was also modified in its dimension and replaced with bigger and smoother one (Figure 9). 36
a b c Figure 9: extrusion head (a) with detail of wired resistance (b) and precision nozzle (c) The choice of the new design has added the chance to deposit polymers also on concave substrate and hence to reproduce undercut features. At the end of the heating mantle, few millimeters above the threaded opening for the attachment of the nozzle, a thermocouple was placed. This assures that at the extrusion point the temperature of the melt corresponds exactly to the value set by the operator (i.e. the melting temperature). The new set-up allowed us to reach very competitive fiber dimension, printing velocity of 1cm/s, printing autonomy of several hours and to obtain composite scaffolds with high ceramic content. 2.3 Printing parameters There are many parameters that control the printing process and their relationship should be clearly understood to obtain the desired porous scaffold with minimal problems. Some factors that contribute to a successful printing and their relationship are listed in Table 3. 37
Parameters Melt temperature ( C) Nozzle size (G) Rod width (mm) Layer thickness (mm) Pressure (bar) Dispensing speed (mm/s) Effects The temperature in the extrusion chamber. Even if in molten state PCL is extremely viscous. Raising the temperature beyond its melting point was the only way to reduce its viscosity. Diameter of the nozzle. This would affect the amount of the material extruded and the diameter of each filament. Distance between two parallel scan vectors. It determines the porosity of the final scaffold and the stability of the laid filaments. If it is set too large the filament may sag. Distance between each consecutive layer. It influence the binding between each layer and the build time. Determines the speed at which molten material is extruded from the nozzle, namely the flow rate. This parameter works in cooperation with flow rate to enable consistent and uniform laying down of the filaments as they are extruded. It can be also regulated to allow some stretching of the filaments to further reduce their diameters. Table 3: printing parameters and their significance The deposition process in each layer starts with a rod of width and thickness specifically controlled by polymer viscosity, nozzle diameter, flow rate and translation speed. The liquefier temperature control is extremely important, it serves more than just to melt the polymer; even in its molten state, some polymers such as PCL are extremely viscous, thus raising the temperature beyond their melting value was the only way of reducing the viscosity. Needle type, length and diameter are all influenced by the viscosity of the material. Long needles are used to slow down the dispensing process for low viscosity materials; short or conical needles are used for highly viscous materials to speed up the process and to counteract the occurrence of needle blockage. The amount of fluid dispensed is controlled by regulating the magnitude and duration of pressurized air that drives the polymer into the liquefier. The plotting material must flow consistently through the nozzle when pressure is applied to avoid under-dispensing. However it is also essential that it is not too high as this can lead to over-dispensing which ruins the scaffold architecture by over filling the pores with material. A close observation of the dispensed filament and a slight tweaking of the plotting pressure 38
during operation time can circumvent this problem. Thus, a balance and a compromise need to be attained between viscosity and flow rate such that the output of molten polymer would be consistent and uniform. Translation speed can also be used for fiber diameter regulation, as an increase in its value can stretch the filaments and reduce their diameter, and vice versa. Apart from process parameters, also design features can affect the scaffold production. For example, the spacing distance between laid filaments, determining the porosity of the final scaffold, will influence their stability. As this air gap is also the space that the following layer of filaments have to bridge, if it is too large, the filaments may sag, if too short they can collapse. Another important design parameter is the layer thickness, namely distance between each consecutive layer. Continuous layer bonding is essential to build a scaffold; hence the first dispensed layer of the material must adhere to the printing substrate, and it must then bond with the second and so on until the scaffold is complete. A correct choice of this parameter can influence the binding capability as well as the build time. 2.4Development of dedicated software modules The 3 axis linear motion system is controlled through the software DMC Terminal. Predefined programs with tool path information can be uploaded through motion planner. Once the preliminary set-up of the 3D printing device was developed, basic manual control routines were produced, also by interfacing the printing device with 2- axes joystick for the handling of the sample plate. Interfacing routines in DMC-code, were also developed to synchronize sample movements with the actuation of high pressure valves. However, in the case of complex geometries, like irregular anatomical structures, separate applications have been developed to generate the tool path information and interfaced with the control system. A first refinement of standard routines was implemented using CAD applications (AutoCad by AutoDesk, San Rafael, CA) to design simple free form shapes and, then, to convert them into printing primitives for the RP device. More automatized ad-hoc control routines were implemented by the use of a software for scientific computation (Matlab by MathWorks, Inc., Natick, MA). These routines enabled the manufacturing of scaffolds 39
with simple contour geometries (square, rectangular and circular) with an assignable size, distance between consecutive lines and raster filling patterns. A key advantage of the RP technology is its ability to produce complex 3D shape from a given CAD model reconstructing highly detailed medical imaging data. For this purpose, a software interface that allows the fabrication of free-from solids with outer dimension according to CT/MR data was developed. Several kinds of software in the market can act as the visualization medium to capture morphological data on a biological structure and to process such data by a computer to generate the code required to manufacture the structure by a rapid prototyping equipment. In our case, the Materialise s Interactive Medical Image Control System (Mimics) software has been applied for processing medical data. A further improvement of software modules dedicated to the printing technology was obtained by the development of an algorithm to design trabecular elements of the scaffold based on Finite Element Analysis. This software module aims at providing a design tool to fabricate optimized scaffold microstructures on the basis of the physiological loading conditions at the implant site. Optimized software routines enabling an automatic conversion from stress tensor tables to control primitives, for generic geometries, are also developed. Characteristic objects obtained by the most significant previous described routines are introduced in the following section. - Direct control: armoring coil Equipping the developed RP system with a rotating mandrel disposed, as shown in Figure 10a, and opportunely combining its rotation with the linear motion of the Y-axis stage, it s possible to accurately extrude polymeric armoring coils. This system has been used to reinforce tubular electrospun scaffolds in order to ameliorate mechanical properties of tissue engineering vascular graft. Figure 10b represents a PCL armour composed of a single RP filament with mean diameter of 0.3 mm and pitch of 0.9 mm, wrapping the outer layer of the scaffold in a helix arrangement. By acting on RP process parameters, thickness and pitch of the extruded coil can be tailored in order to match mechanical properties of native artery. Regulating the Y-axis movement directions, coil with different arrangement can be achieved. 40
Figure 10: schematic representation of the RP apparatus with the rotating mandrel (a), electrospun tubular scaffold with the PCL monolayered helical reinforcement (b) [38] - Matlab routines (script reported in appendix): grid architecture An example of multilayer structures obtained iterating this routine is presented in following figure. The developed tool led to the fabrication of scaffolds with layers of parallel fibers, alternatively oriented at 0-90. Values assigned to the input parameters were: spacing between two consecutive trabecular elements, scaffold size, distance between consecutive layers and their total number. This kind of grid space filling pattern has been adapted to different external geometry; a schematic representation of the trajectories of the first two construction layers, in the case of circular and rectangular shaped scaffolds, is shown in Figure 11a, while in Figure 11b multilayer RP scaffolds extruded according to this rule are shown. 41
Figure 11: construction patterns used by the RP device (a); rectangular and circular scaffolds fabricated according to these trajectories (b). - Generation of working trajectories from selected CAD geometry: circular scaffold Starting from a CAD geometry (Autocad) it was possible to reproduce scaffolds with controlled external shape and complex internal microarchitecture. Figure 12 shows a circular crown shaped PCL scaffold obtained alternating layers (A and B) with different patterns: layer A is constituted by trabecular elements oriented in circumferential way, while a radial orientation is maintained in the B layer. Structural segments have been organized in such a way that the entire scaffold can be fabricated without any stop of the printing process. A B Figure 12: CAD geometry of two deposition layers (A and B), alone and overlapped. PCL scaffold fabricated according to these trajectories. 42
- Generation of working trajectories starting from medical imaging data: tibial defect The presented scaffold (Figure 13b) represents a selected portion of a tibia sheep defect derived from a CT images. Figure 13: free-form RP scaffold representing a portion of tibia sheep (b) reproduced starting from medical imaging data (a) The imaging data were first loaded into MIMICS (Materialise, NV) software where an appropriate threshold range was fixed to capture only information relevant for 3D geometry reconstruction. The model is then exported in STL format and converted into various layers equally spaced along Z-direction by means of an open-source program (Skeinforge). On the basis of fixed parameters, a filling raster pattern was automatically elaborated by the program, for each layers. The direction of the scan lines is set to intersect that of the preceding layer at a selected angle; hence, the built strands crossed at each layer to form the scaffold. Other construction parameters such as strand distance and layer height have significant effect on the quality of the built part. A G-code was finally generated and converted to printable instructions. The scaffold shown in Figure 13b, based on the model in (a), indicates the potential of the RP developed system to build customized free-form structures derived from medical imaging data. 43
Conclusions A rapid prototyping station for the fabrication of 3D scaffolds, tailored on patient specific needs, has been developed. The system is capable of creating structures of thermoplastic material, with precision comparable to other commercially available layered manufacturing techniques. A major advantage over conventional FDM process lies in that the scaffolding material can be directly deposited without involving filament preparation. This opens the door to a wider range of materials and ceramic composites. In addition, a dedicated set of routines has been developed in order to acquire 3D-CAD solid models and to reproduce layered 3D scaffolds according to the desired external geometry, porosity and filling pattern. In the next two chapters this system and its related algorithms, will be used for the preparation of biomimetic scaffolds possessing the required features to act as bone substitutes. 44
Chapter 3: Rapid prototyping of biopolymeric functionalized scaffolds for bone regeneration Overview The feasibility of using the extrusion based system previously described for the manufacturing of structures with designed micro-architectures has been investigated by means of PCL and PCL-hydroxyapatite (PCL-HA). Microstructure of the obtained materials has been evaluated by Field Emission Scanning Electron Microscopy. After the optimization of printing parameters, multilayer scaffolds have been fabricated and investigated in term of mechanical properties, with the aim to engineer truly bioinspired constructs, able to mimic the complexity and graded morphology of bone tissue. Fabrication of 3D printed biopolymeric scaffolds has been combined with surface functionalization in order to provide correct sequences of signals to both enhance cell adhesion and matrix remodelling. Material-driven control of bone-marrow human mesenchymal stem cells (hmscs) behavior and differentiation could represent a promising tool for bone tissue engineering. Thus, the behavior of hmscs cultured upon surfaces functionalized with osteogenic growth factors, has been evaluated. 3.1 Rapid prototyped functionally graded scaffolds Computer-assisted rapid prototyping is a promising methodology for the production of patient-specific biological substitutes [26]. Here we propose an approach based on the use of RP for the fabrication of hydroxyapatite-rich composite scaffolds and on their successive functionalization. Our aim is to reproduce a microenvironment not only able to assist and guide cell adhesion, proliferation and differentiation, but also mimicking 45
the mechanical properties of the native tissue and providing at the same time important signaling upon implantation. The possibility to concentrate in a leading framework biological signals capable of both recruit and guide cells for tissue replacement could have a dramatic impact in osteochondral regeneration as well as in the treatment of many other pathological conditions. Functionalized biomaterials with bioactive proteins together with 3D printing technologies may allow the production of effective bioinspired scaffolds capable of reach this aim [39]. 3.1.1 Biomaterials choice Polymer-based biomaterials are suitable for rapid prototyping technique, and in particular for fused deposition modeling, thanks to their thermoplastic behavior and low melting temperature. On the other hand, an osteoinductive material is fundamental to drive osteogenic differentiation of mesenchymal progenitors, and so far, ceramics (e.g. hydroxyapatite, tricalcium phosphate) provided the most encouraging results from this point of view. However, this kind of scaffolds resulted to be fragile, so definitely unadapt to resemble the elasto-plastic mechanical behavior of the native joint. To this extent, biodegradable nano-structured polymer, in combination with bioceramic particles, can overcome the limitations of conventional polymeric and ceramic bone and osteochondral substitutes. For all these reasons, we hypothesize that composite-based scaffolds, processed by RP technique, might represent a valuable solution when implanting an immature graft for bone repair. Thus, PCL and PCL/HA composites have been selected as scaffolding materials. Polycaprolactone: is a semi-crystalline aliphatic polymer with a melting temperature of approximately 60 C, a high thermal stability and a high decomposition temperature (350 C). It s biocompatible and biodegradable with a degradation rate slower in comparison to other biopolymers. Due to its relatively high mechanical strength and good thermal stability it has been chosen by several Authors for the fabrication of scaffolds [40], especially when processed by RP and devoted to hard tissue regeneration. However, surface modification (physical or chemical) or incorporation of bioactive molecules in PCL scaffolds, is often performed in order to promote quality of cell-biomaterials interaction. Combination of PCL with ceramic component in 3D structure, has been showed to enhance PCL scaffold properties and led to a promising 46
mix of bioactivity, biodegradability, and strength. The main goal of hybrid materials is to take advantage from both organic and inorganic domain in order to improve the final properties by the synergetic effect of the domains [41]. In particular, the inorganic component can provide hardness, strength, and thermal stability, while the organic one provides ductility, hydrophobic character, and chemical reactivity [42]. Hydroxyapatite (HA) (Ca 10 (PO 4 ) 6 (OH) 2 ): has been widely investigated in bone tissue engineering field because of its chemical similarity to the mineral constituent of natural bone, its excellent biocompatibility and high bioactivity. The great results obtained for many years by HA use in tissue engineering field, has suggested the necessity to find the right combination of parameters for its rapid prototyping processing. However, because of its brittle characteristic and chemical properties, it is often difficult to process ceramic materials by means of this class of techniques and in particular by extrusion based ones. Therefore, the use of HA in composite blends with polymer acting as binder, has been demonstrated to considerably improve its processability and also to positively incremented mechanical properties. PCL/HA scaffolds with different content of ceramic component have been fabricated by Precision extrusion technique and tested in vitro with different cell types. [43, 44] Furthermore, selective laser sintering and 3D plotting systems have been used for PCL/HA scaffolds successful production. [45-47] 3.1.2 Implemented design The study of an optimized scaffold architecture is out of the scope of this chapter and will be discussed in the next one. Here, a space filling pattern commonly used in literature for rapid-prototyped structures has been chosen and Matlab routines previously described (Section 2.4) have been implemented for machine code generation. Polymer strands were extruded with a layer pattern of 0 /90 orientation (Figure 14) while fiber diameters, layer thickness, number of layers and, road width were tuned according to the desired features. 47
Figure 14: scheme of the scaffold production process Several layers of trabecular elements have been deposited one atop the other, by coordinating the relative motion of the dispensing head and the sample holder with the extrusion of the molten polymer. Each extruded filament will adhere to the lower layer, therefore obtaining a porous monolith. 3.1.3 Functionalization protocol The selected materials, processed by RP techniques as above described, were functionalized by grafting bioactive molecules on the scaffold surface. This degree of modification was aimed at imparting a bioinstructive facet to this scaffolds in order to further enhance cell-biomaterial interaction promoting the synthesis of neo-ecm and guiding human mesenchymal stem cells toward desired differentiation. Recently, the attention of the scientific community attention has been turning towards the fabrication of more complex structures, integrating living cells and an artificial extracellular matrix into a single rapid-prototyped engineered tissue.[48] Aiming at tissue regeneration and repair, local sustained delivery of various growth factors within the bioabsorbable scaffold has been used to stimulate the differentiation of seeded cells to desirable specific organized tissues. Growth factors are proteins secreted by a broad range of cell types, whose main aim is the transmission of signals able to activate specific development programs that control cell migration, differentiation, and proliferation. They act in a concentration and time-dependent manner, often requiring minute quantities to elicit biologic activity, and their action can depend on a variety of factors, including cell location within a tissue structure and cell 48
cycle state. While trying to identify a certain type of factor who is directly involved into bone formation and remodeling, the TGF-β superfamily has a pivotal role. It is involved in bone development, repair and remodeling after trauma, by means of proliferation and differentiation regulation ability [49]. Of high interest in the group of bone morphogenetic proteins (BMPs), which are members of the TGF-β superfamily, BMP-2 through BMP-8 represent the main candidate osteogenic proteins, playing an important role in embryonic development, generation of the central nervous system, and tissue repair [50, 51]. Due to its pleiotropic functions, BMP-2 has been chosen as a primer differentiation factor for our experiments. Although extrusion based techniques demonstrated to considerably improve the quality of tissue-engineered constructs, the high temperatures involved during fabrication of molten polymers, remain still critical for the direct incorporation of biological factors. In this framework, a chemical route for the grafting of peptides on the surface of the RP scaffolds was fully established and optimized. Both PCL and composite scaffolds were treated in order to increase the surface concentration of carboxyl groups, slightly modifying the method described by Sun et al. [Sun et al, 2006]. Briefly, the scaffolds underwent alkali-catalyzed hydrolysis, obtained by soaking in NaOH solution in order to hydrolyze ester groups (leading to free COO surface groups), followed by protonation into HCl. After extensive washing in PBS, activated samples, bearing carboxylic groups, were immersed in a catalyst solution containing 1-ethyl-3-(3- dimethylaminopropyl) carbodiimide hydrochloride (EDC) and N-hydroxysuccinimide (NHS), in MES buffer (ph=6.20). Finally, samples were transferred into growth factors solutions in order to obtain a covalent bond between carboxylic acid groups exposed their hydrolyzed surfaces and the amine groups of the drugs. A schematic representation of the developed protocol is shown in Figure 15. Figure 15: scheme of the grafting protocol: R indicates the generic growth factor (i.e. BMP) 49
Tesi di dottorato in Ingegneria Biomedica, di Sara Maria Giannitelli, 3.2 Home-made RP system performance Solvent based deposition The first bioprinting technique evaluated for the fabrication of scaffolds was the solvent based deposition. With this apparatus it has been possible to produce multi-layered PCL and PCL-HA scaffolds, presenting filaments with a diameter below 100 µm, orthogonally superimposed, and spaced by 500 µm. A 40% w/v solution of PCL in dichloromethane (DCM, Sigma) and a 40% w/v solution of PCL in DCM with 5wt.% of hydroxyapatite nanopowders, were manufactured by the following parameters (Table 4): Procedure Working Temperature Needle Dispensing speed Solvent based Ambient 30 Gauge 1 mm/s Table 4: manufacturing parameters for PCL scaffolds: solvent based extrusion process FE-SEM images (Figure 16), revealed that the microstructure obtained by solvent based deposition was an irregular 3D construct with low attitude to host cells. Figure 16: SEM micrographs of bioprinted PCL scaffold This problem, added to other manufacturing drawbacks, mainly due to the evaporation of the solvents after 2-3 hours of processing and to the presence of solvent residuals within the porous structure, suggested to move toward a different printing technique. 50
Pressure assisted fused deposition The new additive manufacturing apparatus, based on pressure assisted deposition of melt polymeric materials, has been tested. Briefly, the dispensing equipment comprise: a reservoir embedded into an aluminum heating mantle, supplied of two heating cartridges and a resistance thermometer (Pt100), connected to a programmable temperature controller. The dispenser was heated up in order to melt the polymer and the extrusion process is performed by pressurizing the chamber with argon gas by means of a control electrovalve. After manufacturing parameter optimization, fabrication of a reproducible sets of multilayer 3D scaffolds composed by PCL and PCL/HA blends, presenting filaments orthogonally superimposed and equally spaced was possible. Such materials were charged into the reservoir in pellet form (PCL), or in "sheets" (PCL/HA) and melted at 80 C. In particular, to prepare PCL/HA composite scaffolds with a good degree of homogeneity, a masterbatch was prepared by dispersing HA into a PCL solution in DCM. The slurry was cast on a glass and solvent evaporated. The composite was then placed into the extrusion chamber. The manufacturing process was characterized by the following parameters (Table 5): Procedure Working Temperature Needle Dispensing speed Heat based 80 C 21 Gauge 1 mm/s Table 5: manufacturing parameters for PCL scaffolds: heat based extrusion process Obtained microstructures were evaluated by Field Emission Scanning Electron Microscopy (FE-SEM). Micrographs showed regular trabecular-like fibers with a mean diameter of 300 µm and a porosity of 350 µm (Figure 17). These physical properties are adjustable modifying process parameters. Furthermore, EDS (energy dispersive spectroscopy) characterization performed on PCL/HA samples confirms the presence of the ceramic phase upon processing. 51
Figure 17: SEM micrographs of PCL multilayer scaffolds On the basis of these results, we attested the feasibility of our heat based extrusion process to obtain scaffolds suitable for tissue engineering aims and we will select this fabrication method for all further scaffolds production. Thus, a dispensing head with the same working principle but with the inclusion of some advantageous features (described in detail in the previous Chapter 2.2.1) has been fabricated. This optimized system allowed us to reach printing conditions more useful in term of dispensing velocity and autonomy as well as to increase the resolution of the extruded fibers and the weight percentage of ceramic component (up to 50 wt% of HA). Figure 18 shows a four layers scaffold with the best resolution achievable, obtained with the processing parameters summarized in Table 6: Procedure Working Temperature Needle Dispensing speed Heat based 90 C 30 Gauge 8 cm/s Table 6: manufacturing parameters relative to PCL scaffold fabrication The following images confirmed the efficacy of the modifications introduced to the extrusion chamber, in achieving trabecular elements with thinner and more regular geometry, respect to the previous ones. 52
Tesi di dottorato in Ingegneria Biomedica, di Sara Maria Giannitelli, Figure 18: picture and SEM micrograph of PCL multilayered scaffolds with the best resolution achievable 3.3 PCL scaffold fabrication Poly(ε-caprolactone) (MW 80.000, Sigma-Aldrich, Milwaukee, WI) was used in its standard pellet form and directly poured into the stainless steel reservoir. The thermal couple was set to a temperature of 90 C to melt the polymer and sufficiently reduce its viscosity. The extrusion system delivered PCL through a precision deposition nozzle (30G) by means of a pressurized gas line. The in-house developed control software provided functions for 3D visualization and real-time monitoring during the fabrication process. 3D cylindrical and cubic scaffolds (Figure 19) were fabricated with the developed RP system, equipped with the optimized version of the dispensing head. Figure 19:cubic and cylindrical shaped scaffolds 53
The spacing between consecutive fibers was set to 300µm, the distance between consecutive layers to 350µm and the total number of deposited layers to 25. Such scaffolds have been analyzed by SEM microscope to verify the 3D pores dimension and the coherent disposition of the extruded filaments to form the desired textures. 3.3.1 Morphological study Scanning Electron Microscopy (FE-SEM, Supra 1535, Leo Electron Microscopy, Cambridge, UK) was used to evaluate both the micro-structure and the internal morphology of the fabricated constructs. Scaffold surfaces were gold-sputtered and observed using 15 kv accelerating voltage. In order to collect micrographs of the sample cross sections, once solidified and cooled, specimens were cut with a straight razor. This method allowed us to obtain a sharp fracture without affecting or deforming single road shape. Figure 20: scanning electron microscope image of scaffold cross section SEM image clearly demonstrate the success of the proposed printing technology in the production of 3D PCL scaffolds with 0 /90 layered pattern. In particular, Figure 20 shows that the use of a round shape nozzle orifice, in combination with a correct choice of printing parameters, has led to the formation of roads with circular cross-section. In 54
the same image, we can also detect the presence of an interconnected porosity with pores that continue through the depth of the scaffold. Figure 21: scanning electron micrographs of PCL bioprinted samples. Top view From an higher magnification image of the top surface (Figure 21), we can also appreciate that the pores created are also regular shaped and coherent to the designed texture. Digital-image processing software (Image Tool Pro 3.0, developed by the National Institute of Health, USA) was used for calculating mean diameter and mean pore size of bioprinted scaffolds. Measured values were about 160 µm and 100 µm, respectively. SEM analysis, demonstrates the success of our extrusion based system to fabricate three-dimensional PCL scaffolds with structure controlled at the micro-scale level. 3.3.2 Porosity Pore structure is a very important scaffold feature because it determines permeability, mechanical properties and cell growth. Literature data reports that the optimal pore size and porosity of a scaffold for bone tissue engineering are 200 500 µm and between 50 and 75% respectively [52]. For simple scaffold geometry and filling patterns, a theoretical porosity value can be determined on the basis of the amount of polymer deposited in raster roads of known diameter RW, according to the tip size used. For each cubic block, assuming that the roads (laid filaments) had a circular cross-section of uniform diameter, the amount of material deposited (Vt: true volume) could be estimated by: 55
Vt = L N V RW where L is the number of roads per layer (determined from the average number of roads created in the 0 raster direction), N is the number of layers per block, and V RW is the volume of each road (with the geometry of a cylindrical tube). The derivation of theoretical porosity for a cylindrical shaped scaffold created by an extrusion based machine is illustrated elsewhere.[53] Using the general expression introduced in Section 1.5.1 and measuring the apparent scaffold volume Va by means of a caliper, is possible to estimate scaffold porosity. Furthermore, an experimental quantification has been achieved by correlating scaffold geometrical dimensions and material density as suggested in ASTM F 2450-04. The apparent scaffold volume (Va) was determined from caliper measurements, the total specimen mass (m) by a micro-balance. Using a PCL density value of 1.145g/cm 3, a porosity level of 60% was calculated. 3.3.3 Mechanical characterization Mechanical properties were evaluated on cubic samples (in triplicate, dimensions: 6mm) using a tensile tester (model 3365, Instron, Norwood, MA) equipped with a 500N cell load. All the tests were performed in accordance with the ASTM D695-96 guidelines. The samples were sandwiched between compression plates and tested at a speed of 1 mm/minute without any pre-load. One hysteresis cycle was performed for each sample: load was applied until a stress reduction of 40% was detected and then removed. Stress-strain data were computed from load-displacement measurements. Elasticity modulus (E) was determined as the slope of the curve in the elastic region, compressive yield strength as the stress after which the initial linear region deviated from linearity. Scaffold fabricated using an extrusion based process, exhibit mechanical properties strongly affected by road shape, road to road interaction as well as deposition trajectory. As a natural consequence of the evolutionary building configuration, scaffolds will have different mechanical behavior depending on the loading direction. It is quite evident that properties evaluated along the roads direction will differ from that evaluated across the 56
roads, where the strength is determined only by the inter-road adhesion. To determine anisotropic influences on scaffold mechanical properties, compression tests have been repeated in the same above described conditions, in two different loading orientations (Figure 22a). Figure 22b reports average curves of mechanical behavior in both the analyzed conditions. a b Figure 22: (a) layer orientation relative to compressive load, transverse on the left and axial on the right. (b) Resulting stress strain curves Scaffolds demonstrated the typical behavior of a porous material undergoing deformation. Material shows an initial linear-elastic phase, followed by a densification phase. The structure then finally reaches the critical stress level when channels start to collapse and elastic buckling of the column-like structure occurs at the buckling point. Furthermore, results confirmed that layer orientation affected scaffold mechanical properties. A reduction in stiffness value from 95,5 to 28,4 MPa has been evaluated in the case of axial loading direction. 3.4 PCL-HA scaffold fabrication PCL-HA composite scaffolds, with different percent of ceramic components have been fabricated with the developed RP system. PCL was mixed to nanometric hydroxyapatite (Sigma Aldrich, Milwaukee, WI) powders at selected weight ratios and dissolved in a dichloromethane/methanol mixture. The slurry was cast on a glass and, after solvent evaporation, was placed into the extrusion chamber in form of membrane (Figure 23a). 57
For each ceramic percentage, desired pore size and porosity were obtained by varying temperature, pressure and nozzle diameter. Each layer was filled with the designed scaffold pattern of a 0/90 orientation to generate the porous structure (Figure 23b). Figure 23: bulk PCL-HA membrane (a) and scaffold (b) 3.4.1 Morphological study Scanning Electron Microscopy (FE-SEM, Supra 1535, Leo Electron Microscopy, Cambridge, UK) was used to characterize morphology and microstructure of the fabricated scaffolds. Regularity and uniformity of the obtained fibers demonstrate the applicability of our RP system to the fabrication of 3D hydroxyapatite-rich composite scaffolds. Figure 24 evidences how HA presence affected surface topography, resulting in rougher regions, with HA protruding from the polymeric matrix (arrows in figure). 58
Figure 24: scanning electron micrographs of PCL/HA samples. Arrows indicate HA nanopowders dispersed in PCL matrix 3.5 Scaffold functionalization PCL and PCL/HA (25 wt%) printed scaffolds, have been chosen as optimal substrates for functionalization and, thus produced according to the above described conditions. Rectangular shaped scaffold with a trabecular inter-axis spacing of 350 µm have been fabricated depositing layers oriented at 90 to each other. From this geometry, small circular patches of 6 mm diameter have been retailed utilizing a dermatologic punch. In this way we can accurately standardize surface, calculate the area available for the grafting and consequently the amount of growth factors. At this point, scaffolds were functionalized as reported above ( 3.1.3) with an appropriate growth factor (BMP-2), in order to obtain constructs able to recapitulate a multilayered structure equal to the native histoarchitecture of bone tissue. Both PCL and PCL/Hap scaffolds were immersed in a 0.1 M sodium hydroxide (NaOH) solution and reacted for 90 min at 40 C. After hydrolysis, scaffolds were protonated with 0.01 mol/l HCl (30 min) to give polymer surfaces bearing carboxylic groups. Surface carboxylic group density was measured by inverse titration. Hydrolyzed samples were immersed into an EDC/NHS catalyzing mixture containing 0.4 mol/l EDC (Sigma) and 0.1 mol/l NHS (Sigma) in MES buffer, for 10 min. After appropriate washing, they were transferred into growth factors solutions in MES buffer (BMP- 2/MES 1000 ng/ml) for 1 h in order to obtain a covalent bond between the hydrolyzed 59
polymeric surface and the peptide. Yield of growth factor immobilization reaction was assessed by ELISA method performed directly on the scaffolds (Figure 25). Figure 25: scheme of the ELISA procedure performed on the functionalized scaffolds An average growth factor surface concentration of 0.3 pmol/cm 2 was measured. This value is in the biologically active range to provide correct signaling to cells cultured on the scaffolds. 3.6 In-vitro experiments on tissue-engineered constructs 3.6.1 Scaffolds biocompatibility Bioprinted scaffolds were assayed for their biocompatibility with bone marrow-derived human mesenchymal stem cells (hmscs, Lonza), in terms of cytotoxicity and cell proliferation. 1x10 4 hmscs were separately plated in a 96-well plate (BD, Falcon, San Jose, CA) in growing media and cultured for 24 hours. Then, PCL, PCL/Hap, PCL/BMP2, PCL/Hap/BMP2 scaffolds were added to the wells. Cell toxicity due to the scaffolds was determined at 4, 8 and 24 hours by using a Vybrant Cytotoxicity Assay Kit (Molecular Probes, Invitrogen) in accordance with the manufacturer's instructions. This kit monitors the release of the cytosolic enzyme glucose 6- phosphate dehydrogenase (G6PD) from damaged cells into the surrounding medium. Plate was read on a fluorescence microplate reader (Tecan, Männedorf, CH) at 540 nm with correction at 595 nm. 60
To determine the cell proliferation, 5 10 3 hmscs were separately plated in a 96-well plate (BD, Falcon, San Jose, CA) in growing media and cultured for 24 hours. All the types of scaffolds were added to the wells. Cell proliferation was determined at 1, 3 and 7 days by using a MTT Assay (Molecular Probes, Invitrogen) in accordance with the manufacturer's instructions. Briefly, MTT was added to the cell culture media and incubated at 37 C for 3 hours. Cells were then washed three times and dimethyl sulfoxide (DMSO) added to each culture and incubated for 10 minutes. Plate was read on a fluorescence microplate reader (Tecan, Männedorf, CH) at 540 nm with correction at 595 nm. Above-mentioned assays showed a viability upper than 87% (Figure 26a) and a proliferation rate upper than 45% with respect to negative control (Figure 26b), respectively. Figure 26: cytotoxicity (a) and proliferation assay (b) Adequate cell viability and proliferation confirmed the generation of a non hostile environment. 3.6.2 Cell differentiation Fourth passage hmscs (Lonza) were cultured in contact with the disk-shaped scaffolds at a concentration of 5 10 5 cells/cm 2. Cells were suspended in 30 µl of basal medium and the suspension droplet was left onto the scaffold for 8 hours for adhesion to occur. Then, basal media was added and tissue engineering constructs (TECs) were cultured 61
for a period of 21 days. Functionalized scaffold were maintained in basal medium (αmem + 10% FBS, 1% PS and 1% L-glutamine), while control scaffolds were also cultured with ostoegenic media (Lonza, DMEM+10% FCS, 100mM ascorbate-2- phosphate, and 10mM (β-glycerol phosphate, 100nM dexamethasone) as a positive control. After 3 weeks, scaffolds were snap frozen for mrna extraction and RealTime PCR analysis. 1 µg of total RNA was retrotranscribed using High Capacity cdna Reverse Transcription Kit (Applied Biosystems) according to the manufacturer s instructions. Realtime quantitative PCR was performed on 100 ng of DNA to detect the expression of osteogenic genes (PHALK, COL1A, GLA protein) and 18S rrna as housekeeping gene, using specific primers (TaqMan Gene Expression Assay, Applied Biosystems) and TaqMan Universal MasterMix II (Applied Biosystems) in a total volume of 20 µl, according to the protocol and thermal profile suggested by the manufacturer. Preliminary PCR results (Figure 27) indicate that functionalization led to an enhancement of bioactivity. In particular, the addition of hydroxyapatite and BMP-2 resulted in higher expression of osteogenic genes (particularly alkaline phosphatase and osteocalcin). Figure 27: gene expression 62
For each experimental condition, bioprinted scaffolds were characterized also by micro- CT in order to verify the formation of mineralization throughout neo-formed ECM. Scaffolds were fixed with 4% formaldehyde and analyzed by micro-ct, and a scaffold was embedded in Tissue Tek O.C.T. and rapidly frozen in liquid nitrogen. Discs were sectioned with cryotome in transversal direction to make 15 µm thick sections and fixed in acetone. Slides were stained with Rodamine-Phalloidine for F-Actine and nuclei were stained using TOTO (Molecular Probes, invitrogen). Figure 28: PCL bioprinted scaffold with adherent hmsc under culture. (B) Section of the TEC after 21 days of culture showing the layer of cells around the surfaces of the scaffold fibers (F-actin: red, Nuclear staining TOTO3: blue) Microscopic images of the TEC during cell culturing and on the F-Actin and nuclear staining (Figure 28) show cells adherent on the scaffold surfaces and growing in the pore of the scaffold. It s possible to appreciate how pores are bridged by proliferating cell sheets forming cellular cords between roads. Microstructure characterization by micro-ct revealed mineralization area on the surface of the scaffold functionalized with BMP2 indicating an osteogenic MSCs differentiation (arrows in Figure 29c) with respect to the control scaffold (Figure 29). 63
Figure 29: representative image of the Micro-CT of the TEC cultured under osteogenic conditions showing high radio-dense area on the surface of the scaffold (arrows in c) Figure 30: representative image of the Micro-CT of the control TEC showing homogenous radio-dense scaffold. Conclusions These results highlight the efficacy of functionalized scaffolds to drive differentiation of hmscs toward bone. The ability to modify biomaterials surface in order to elicit specific biological response from cells cultured therein, might represent a valid strategy targeting tissue regeneration. 64
Chapter 4: A bioinspired approach to the design of porous additively manufactured scaffolds with optimized mechanical properties Overview When approaching the fabrication of porous scaffolds, two main issues have to be simultaneously addressed: one regards the optimal porosity, which depends on the ability of hosted cells to migrate and proliferate within the scaffold; the other one regards the capability of the produced scaffold to bear physiological loads once implanted, without dramatic collapse. The latter issue is particularly urging in the case of scaffolds intended for the regeneration of mineralized tissue, where the scaffold implant should provide a leading framework for the repair of the damage. An optimization problem therefore emerges: on the one hand, the ideal scaffold should exhibit a sufficiently large porosity; on the other hand, it should be stiff and robust enough to withstand physiological load. Such requirements are evidently antagonistic, since a large porosity negatively impacts robustness. For this reason, the development of design strategies capable to optimally trade-off between the two opposite needs is one of the objectives of tissue engineering. In this chapter an innovative strategy for the fabrication of highly optimized structures, based on the a priori finite element analysis (FEA) of the physiological load set at the implant site, is presented. The resulting scaffold micro architecture does not follow a regular geometrical pattern; on the contrary, it is based on the results of a numerical study. The algorithm was applied to our solid free-form fabrication process, using poly(ε-caprolactone) as the starting material for the processing of additive manufactured structures. A simple and intuitive geometry was chosen as a proof-of-principle application, on which finite element simulations and mechanical testing were performed. Then, to demonstrate the capability in creating mechanically biomimetic 65
structures, the proximal femur subjected to physiological loading conditions was considered and a construct fitting a femur head portion was designed and manufactured. 4.1 Design of scaffold micro-architecture in extrusion based techniques The physiological structure of native tissue is inherently heterogeneous and complex. Instead of try to reproduce the exact internal microarchitecture, literature is mainly focused on the creation of simplified model equivalent in term of porosity and mechanical properties to the tissue to be repaired. On this tendency, additive manufactured scaffolds are made of unit cells with well-characterized mechanical and transport properties. Depending on the fabrication technique, different scaffold design methods have been proposed. For Solid Free Form systems based on laser technology (e.g. SLA, SLS) and printing techniques (3DP, TheriForm ), design approaches that mimic tissue architecture using 3D unit cell repetition have been applied. One way to achieve this objective is to create libraries of unit cells at different physical scales that can be assembled to form whole scaffold architecture. Such libraries may be created either using, CAD-based approaches, image-based design approaches or, especially in recent years, triply periodic minimal surfaces (TPMS). [54] Figure 31 reports some examples of 3D unit cells obtained by their implementation. a b c Figure 31: common used unit cell elements derived from three different approaches: CAD-based [54] (a), image-based [55] (b) and TPMS [56] (c) 66
All these methods cannot be easily combined with extrusion based systems (e.g. FDM, PED) due to the nature of the process that avoid the possibility of a rapid response rate of the start-stop extrusion mechanism. In this latter case, interior designs manly consisted in regular continuous road patterns whose parameters such as road width, road gap, layer thickness and orientation, were changed to achieve the desired pore structures, porosity and mechanical strength. Some of the most famous patterns implemented by extrusion based techniques for TE scaffolds fabrication are shown in Figure 32. The simpler structure has cube-shaped pores formed by orthogonal rasters, more complex interconnected one can be achieved by changing the angle of between subsequent layers deposition. These structures are highly similar to the honeycomb, with its regular array of identical pores, when viewed in the direction of the fabrication process. The main difference lies in pores morphology: the bee s honeycomb comprises hexagonal pores surrounded by solid faces/walls that nest together to fill a plane, while the extrusion based scaffold structure is built from intercrossing filaments stacked in horizontal planes and comprises pores surrounded by solid edges/struts. Furthermore, depending on raster angle value, man-made honeycombs pores can also be triangles, squares, or other polygonal shapes. [57] Figure 32: various lay-down patterns selected to process TE scaffolds (a) 0/90, (b) 0/60/120, (c) 0/45/90/135 and (d) 0/30/60/90/120/150 [57] 67
Nevertheless, a low cell seeding efficiency and heterogeneous cells distribution are often considered between major drawbacks of such scaffold architectures. In a recent study, Sobral et al. hypothesized that the presence of tortuous conduit inside the scaffold, increasing the residence time of cells in the construct and the likelihood of contact between them, should enhance seeding efficiency. They fabricated scaffolds with pore size gradient along one direction and demonstrated that they significantly improve seeding efficiency as well as cell distribution respect to homogeneous scaffolds. The gradient porosity was obtained printing layers oriented at 90 each other and varying fiber spacing from the outer layers to the middle and vice versa. [58] An extension of the general use of road patterns (0 /90, 0 /60 ) for the design of scaffold internal architecture has been developed by the use of space filling curves. The continuous functions that define these curves are associated with a finite domain in a Euclidean space (plane curve); in 2-dimension this space is bounded by a square and without loss of generality the domain interval [0,1] can be considered. Due to the features of extrusion techniques, continuous, self-similar, non-intersecting curves that satisfy the following constraints (valid in each layer) are particularly attractive. a) Do not intersect: allows patterns to be laid out in layer by layer fashion; b) Constant distance between points: helps in the individuation of a repetition unit cell that make up the pattern; c) Only one start and stop point: avoid multiple closures of gas valve in a single layer and thus the agglomeration of material due to the delayed response time. This particular class of space-filling curves, named fractal space-filling curves, can be mathematically generated by starting with a very simple pattern that grows through the application of rules (Figure 33). Recursive techniques based on well-defined grammar (Lindenmayer system) are used for their generation. 68
Figure 33: recursive generation process (Sierpinsky and Hilbert space filling curves) [59] A proof of principle of how space filling fractal curves can be used as pore structural elements in extrusion based techniques, has been proposed by Sun et al. In particular, two fractal space filling curves (Hilbert and Sierpinsky) were separately used for the fabrication of circular scaffolds by Precision Extrusion Deposition with the aim of provide a better substrate for cell organization. [59] A more complicate extension of this concept was developed by Panditevan et al [60] and applied to the design of bone scaffold with subject-specific external shapes and complex internal microarchitectures. Six different space-filling fractal curves, have been investigated for the fabrication of location-controlled density raster tool paths in order to vary the amount of material being deposited in a unit area and thereby to control the porosity. By changing the type of fractal curve and it level, between different scaffold regions, the desired path density and thus scaffold porosity was obtained. 4.2 Load Adaptive Scaffold Architecturing (LASA) algorithm In this contest, a new approach for the fabrication of polymeric scaffolds with a microarchitecture predicted through an a priori FEA of the implant site geometry, under a physiologically derived load system, has been introduced [61]. To the authors knowledge, a good piece of literature currently focuses on the use of Finite Element (FE) methods to define the scaffold internal micro-architecture, assessing their biomechanical properties both computationally and/or experimentally by post hoc testing [62]. Computer tomography (CT) data have been used as input geometries for design and optimization of scaffolds micro-architecture within a FE environment, for 69
creating bone substitutes with subject-specific external geometry and internal architecture (porosity and pore distribution) [63-65]. Another class of FE methods is based on the homogenization theory [66, 67]. Such methods use asymptotic expansion of relevant physical variables to generate multiscale equilibrium equations that can be used to determine the mechanical properties of each of the unit cells composing the scaffold domain. The last category of FE approaches is represented by CAD-based methods, which are based on libraries of CAD generated polyhedral shapes, with known geometric, mechanical, and fluid-flow properties, that can be assembled together according to the application needs. Once a cell unit is selected and sized, an automatic algorithm for generating complex polyhedral scaffolds can be employed to assemble the micro-architecture of the scaffold while maintaining unaltered the external geometry of the patient's anatomy as generated from medical imaging data [68-70]. More recent was the use of FEA as a predictive tool with the aim to optimize scaffolds providing an a priori control of the mechanical properties as a function of porosity and scaffold architecture. Instead of starting form a predefined unit cell, this tool starts form a dense non-porous block of material, searching for a topologically optimized scaffold unit according to porosity and stiffness requirements. However, also in this application, a unit cell based approach has been considered. The proposed method, instead of using repeating units to build the scaffold, proposes to start with a dense solid model corresponding to the whole scaffold s outer shape and to find the best trabecular architecture to withstand the imposed physiological load system, by engineering the material arrangement within the scaffold volume. It is therefore both geometry specific and load specific, and establishes the basis for successfully achieving fabrication of scaffolds with a micro-architecture predicted through an a priori Finite Element Analysis (FEA) of the tissue portion to fit, to which a site-specific load system is applied. 4.2.1 Theory and calculation Given an elastic continuum C, to which a load system is applied, the stress at every point P in the domain C is characterized by a stress tensor σ ( R 3x3 ). Being σ symmetrical for statics reasons, for every P there exists an orthonormal reference frame R, which makes σ diagonal. The three components of the diagonalized stress tensor, σ 1, 70
σ 2, and σ 3 are the so-called principal stresses, which correspond to pure compression/tension occurring along the three axes (x 1, x 2, x 3 ) of R [71]. A technique for ideally carving (i.e. creating a porosity) within C, while minimizing the reduction of stiffness, consists in placing the residual material along the principal directions, thus minimizing the distortion of the structure due to shear stresses. With this consideration, the best solution for the definition of the scaffold design, could be to drive the position and direction of the solid material according to the vector field of the principal stresses. In particular, the principal stress directions can be evaluated numerically using Finite Element Analysis (FEA), and the corresponding principal stress lines can be postprocessed in a suitable environment in order to obtain the toolpath for a addidive manufacturing system. For the validation of the proposed approach, a simple geometry for the domain C was considered. The geometry consisted of a semicircle, with a radius of 2 cm, clamped at the bottom along its diameter, and loaded with a compressive force (10 N) applied on the top, perpendicularly to the abovementioned diameter (Figure 34a). Figure 34: steps for the obtaining of the bioinspired load-adaptive architecture on a test semicircular geometry: (a) definition of the geometry, constraints, and the load systems; (b) plot of the envelope of the two principal stress directions for the plane stress problem. (c, d) CAD geometries of the different design rules applied to the semicircular geometry: LASA pattern (c); GRID pattern (d) 71
Material properties were chosen according to the values for poly(ε-caprolactone) (PCL, Young s modulus E = 230 MPa, Poisson s ratio ν = 0.35, as experimentally derived from tensile tests on PCL filaments), and a thickness of 1 mm was associated to the 2D model for calculus purposes. The FEA problem was solved using COMSOL Multiphysics (COMSOL, Inc., Burlington, MA), in the hypothesis of plane stress. In order to view directional information, streamlines for principal stresses (σ 1 and σ 2 ) were traced as intersecting orthogonal lines originating from each seed-point (horizontal lines for σ 1 and vertical lines for σ 2 ) (Figure 34b). The seed points were chosen on a regular grid in order to optimize density of the streamlines and consequently distance between the trabecular structures. 4.2.2 Design and Finite Element Simulations A load-adaptive architecture (LASA model) was derived from the solution of the above described FE problem. The geometry of a reticular structure was designed, in which orientation of the trabecular elements was obtained as the envelope of the two principal stress directions, with seed points regularly spaced along the diameter (Figure 34c). As a control, a second architecture (rectangular grid, GRID) was designed with linear struts alternatively oriented at 0 and 90, to create a trabecular structure (Figure 34d). This architecture was chosen as a control because of its common usage for additive manufactured scaffolds. Thickness of the trabecular elements was set at 400 µm for the two models; the number and the spacing of the trabecular element for the two models were arranged to keep the total porosity, intended as the ratio between the area of the trabeculae over the total area of the semicircular domain, constant. FE simulation of the mechanical performance of the two architectures (LASA and GRID, Figure 34 c and d), obtained as previously described, was also performed. 2D models representing the scaffold architectures were reproduced using CAD software (AutoCad, Autodesk, San Rafael, CA). The two geometries showed outer dimensions identical to the continuum C. CAD data were then imported into COMSOL Structural Mechanics Module with the above described load system and boundary conditions and solved in terms of strain and stress fields, including von Mises stress and elastic strain 72
energy density. Materials properties were chosen according to the above cited values for PCL (E = 230 MPa, ν = 0.35), and a thickness of 1 mm was associated to the 2D model for calculus purposes. Solution data were exported in term of nodes, elements and values matrices, and imported in MATLAB (The MathWorks, Inc., Natick, MA) for further post-processing. Area-weighted mean stress was calculated according to: ( σ vm ) i wi * i σ vm =, (1) w i i where ( σ vm) i is the value of von Mises stress for the i-th element, w i is the area of the ith element, and i w is the total area of the mesh domain. i Stiffness k [N/m] was calculated starting from the set of equations (2) F = kδ U e = 1 2 k δ 2 (2) Giving 2 F k =, (3) 2 U e where F [N] is the imposed load, δ [m] is the displacement and U e [J] is the elastic strain energy, calculated as the integral of the strain energy density u [J/m 2 ] over the domain (s) as in the following: U u ds. (4) e = S 73
Figure 35 and Figure 36 show the results of FEA on the 2D CAD models obtained by the implementation of the two different design rules LASA and GRID applied to the simple benchmark geometry. The two models presented almost identical surface area, corresponding to a porosity of 61%, but showed different behavior in terms of load bearing. This phenomenon is represented in Figure 35, showing the distribution of von Mises stress for the two models. For the sake of clarity, the same color scale was used in Figure 35a and b. In order to exclude singularities, a circular portion of the geometry (1 mm radius) centered on the point of application of the concentrated load, was omitted from representation. As it can be observed by the color plots (Figure 35a,b) and the histograms (Figure 35c,d) representing the magnitude of von Mises stress over all the elements, the maximum stress value is lower for LASA model. Additionally, mean stress was found to be lower for LASA architecture with respect to GRID architecture (1.02 vs. 1.21 MPa). Figure 35: stress distribution. (a, b) von Mises stress for LASA (a) and GRID (b) design. (c, d) Distribution histograms for LASA (c) and GRID (d) models. 74
To provide a better understanding of how LASA architecture is effective in providing a more uniform stress distribution over the trabecular elements, in Figure 36 local von Mises stress values are presented as a ratio with respect to their mean value, using a logarithmic scale. By the analysis of the color plot (Figure 36a,b) and the corresponding histograms (Figure 36c,d), it is possible to observe how the stress distribution is peaked around the mean value for the LASA model. In the GRID model, on the contrary, most of the structure is stressed at values two to three orders of magnitude lower than the mean stress, with a small portion of the structure bearing the highest stresses (arrow in Figure 36d). Figure 36: stress distribution. (a, b) ratio between local von Mises stress value and mean stress value (log color scale) for LASA (a) and GRID (b) models. (c, d) Distribution histograms for LASA (c) and GRID (d) models. Arrow in (d) indicate highly stressed elements in GRID model. Stiffness values are presented in Table 7 for the two structures. In addition, data for a solid PCL body with outer dimensions corresponding to the continuum C (SOLID model) were added for comparison reasons. It can be observed how LASA architecture presented a higher stiffness with respect to GRID model. Interestingly, its value had the 75
same order of magnitude of the solid model ( k 0. 44), despite its porosity of over 60%. LASA k SOLID Model stiffness [N/m] porosity [%] LASA 5.50 10 5 61 GRID 1.40 10 5 61 BULK 1.26 10 6 0 Table 7: stiffness data (as derived from FEA) for LASA and GRID structures 4.2.3 Samples fabrication The two proposed geometries (LASA and GRID) were implemented using the previously described additive manufacturing equipment. The apparatus comprises an AISI 316L stainless steel heated dispensing head ended with a 21G nozzle, an X-Y motorized stage assembled from two cross-mounted linear stages for the positioning of the dispensing head, and a Z-axis for controlling its distance from the stage. The motorized stages are connected to a computer-programmable motion controller and powered by a dedicated power amplifier. The dispensing head is embedded into an aluminum heating mantle, supplied of two heating cartridges and a resistance thermometer (Pt100), connected to the programmable temperature controller. Dispensation process is pressure-assisted, and is performed by pressurizing the extrusion head with argon gas by means of a control electrovalve. A custom-developed control software generates the process toolpath and controls the actuation of all the system components. Solids were fabricated using PCL (M n = 80,000 g/mol, Sigma, Milwaukee, WI), temperature of the extrusion head was set to 85 C, and relative speed between the nozzle and the XY table was set to 5 mm/s. A total of 8 layers were deposited for each scaffold. The two scaffolds showed identical external geometry and dimensions, and were approximately of same weight (2 g). Performances of the two different scaffold architectures were evaluated by means of mechanical testing. Figure 37 shows a detail 76
of the first two deposited layers for the two architectures obtained according to LASA and GRID models. Figure 37: PCL structures produced according to the LASA (a) and GRID (b) models. 4.2.4 Mechanical testing The obtained PCL structures were also empirically assessed for mechanical properties, in a load condition close to the one used for FEA. Scaffolds were clamped along the diameter to a rigid aluminum plate, positioned on the basement of a mechanical test machine (model 3365, Instron, Norwood, MA), equipped with a 500 N load cell. Samples were loaded on top with a 10 mm diameter cylindrical puncher under a displacement-controlled mode at a deformation rate of 1 mm/min, until a displacement of 1 mm was reached. Figure 38: results of the compression test performed on the PCL structures, in a loading condition close to the FE model 77
Figure 38 shows the resulting force-displacement curve. For every displacement, the corresponding load on the LASA structure was higher than its value for the GRID structure. This is in close agreement with the simulation results, confirming that LASA architecture presented a higher stiffness to compressive loading with respect to the GRID one. These results demonstrate how design rules chosen for the fabrication of additive manufactured scaffolds for tissue engineering truly affected the mechanical properties of produced implants. The overall porosity value and the average pore size, as well as the presence of open porosity, represent constraints that must be respected to obtain a material capable of promoting a suitable response from the host tissue, and the proliferation of newly formed tissue within the scaffold, resulting in a good integration. At the same time, for a given porosity value, the scaffold must be able to withstand physiological loads, to temporarily substitute tissue function during regenerative processes. 4.3 Application of LASA In view of a possible application to the fabrication of scaffolds for tissue engineering, the proposed technique has been evaluated on a clinically relevant anatomic portion. The femur head was chosen, subject to physiological loading conditions. In principle, using clinical CT scans and extracting the region of interest with commercial software (e.g. Mimics, Materialise NV, Leuven, Belgium), subject-specific geometries can be processed. For the purposes of the present work, however, a widely accepted 3D model of an adult human femur was used (Figure 39, 3 rd generation composite femur, Dept. Mechanical and Industrial Engineering, Ryerson University, Ontario, Canada) [72, 73]. The model presented two surfaces, describing the cortical and cancellous bone regions. It was further simplified by considering the whole femur as constituted by an isotropic linear elastic material with homogeneous Young s modulus and constant Poisson's ratio. Although bone is actually heterogeneous, non linear and anisotropic, these simplifications are often made when modeling bone using FEA [74, 75]. The femur bone domain was discretized by 19,079 tetrahedral 4-nodes 78
elements. Previous studies have confirmed that tetrahedron is the best choice for meshing human femur and that it is well suited to model irregular geometries, due to its quadratic displacement behavior [76]. Figure 39: 3 rd generation composite femur, Dept. Mechanical and Industrial Engineering, Ryerson University, Ontario, Canada For the purposes of the present work, a force of 250 N parallel to the shaft axis was applied on the top surface of the femur head, distributed on a circular area of 1.5 cm radius. In the pioneering work by Koch [77], describing the laws of bone architecture following the early studies by Wolff and Cullman load on the femur was assumed approximately 30% of the body weight in the bilateral standing position. The line of action of the force was defined as the line joining the center of the head of the femur to the center of gravity of the lower end of this bone. Further studies have demonstrated the importance of taking into account forces exerted by muscles on the femur head, especially for describing single-limb stand [78]. However, simplified loading conditions have often been used, such as concentrated loads directed along the femur shaft direction [79] or, alternatively, at an angle of 20 to the shaft axis in the coronal plane [65]. The FE problem was solved in COMSOL Multiphysics under the hypotheses of static linear analysis. Pointwise stress tensor and principal stress directions were calculated 79
for the entire femur volume. A text file containing the direction of principal stresses as a function of the position was generated and exported into MATLAB. Curves following the principal stress directions were traced and further post processing was necessary to automatically generate control primitives compatible to standard computer-aided manufacturing (CAM) language. However, only data relative to a specific region of interest were post-processed and visualized. Figure 40: (a) representation of the loading scheme for the application of the algorithm to the femur head. A portion of the head has been removed in the figure to visualize the ROI used for further processing. (b) calculation of the envelope curves defining the pattern of the LASA architecture of the ROI. Coordinates are reported in machine units. (c) Side and top view of the model obtained by additive manufacturing. A small cubic volume within the femur head was selected for the scaffold fabrication (Figure 40a). In particular, the orientation was chosen so that its faces were aligned to standard anatomical planes. With reference to conventional nomenclature, the printing process occurred on a coronal plane (i.e. the coronal plane represents the printer XY plane). A slicing distance of 500 µm was used for the definition of the deposition layers. For each layer, the projections on the coronal plane of two principal stress directions were calculated using MATLAB and used for the determination of envelope curves for the ROI. Care was taken in selecting the seed point of each principal stress trajectory in order to achieve the specified distance between trabecular structures. The 80
result of this process was a series of equally spaced 2D plots, corresponding to different cross-sections, each at a slightly different Z-coordinate value. Figure 40b shows a detail of the plot trajectories for the first 3 deposition layers, which are obtained as envelopes of the projections of two principal stress directions on the coronal plane. The obtained slices were used to design and manufacture a small scaffold fitting a portion of the femur head. MATLAB routines were developed to convert such plots into control primitives for the scaffold fabrication. Scaffolds were printed in PCL with the above described additive manufacturing equipment. The resulting PCL structure is represented in Figure 40c. These results represent a proof-of-principle demonstration of the possibility to produce mechanically biomimetic structures for tissue engineering applications, with a loadadaptive scaffold architecturing algorithm. Conclusions This study establishes the basis for successfully achieving the design of scaffolds with microarchitecture predicted through a priori analysis of the implant site geometry under the physiologically derived load system. This approach defined as load-adaptive scaffold architecturing: LASA is based on a finite element analysis (FEA) for obtaining the principal stress directions under a physiologically derived load system, and can be easily coupled to computer-assisted fabrication technologies for the controlled deposition of biopolymer layers according to the calculated pattern. A protocol for the implementation of such algorithms was presented in the case of a simple and intuitive geometry. Furthermore, the feasibility of this bioinspired approach was evaluated in the case of a scaffold fitting a portion of femur head. Numerical modeling and mechanical testing contributed to validate the proposed model as effective in obtaining optimized materials able to enhance mechanical stiffness while providing a porosity distribution which is suitable for tissue engineering applications. This work highlights the potential of this combination for assisting in the ad hoc scaffold design with tailored properties, while the engineering potential of scaffolds architecture is mostly evaluated only in the post hoc testing. 81
Chapter 5: Electrospun fibers with nanostructured surface morphology Overview Considerable research effort has been devoted to develop and optimize fabrication techniques apt to the production of scaffold with architectures mimicking the structure of the tissue to be repaired. Within the past decade, electrospinning has attracted great interest due to its ability to process polymeric solutions into fibrous structures at the micro/nanoscale by simply controlling few process parameters. Such matrices are characterized by a high surface area-to-volume ratio and resemble the physical structure of protein fibrils in native extracellular matrix. This processing technique is mainly applied to the fabrication of scaffolds for soft tissues regeneration as well as of matrices for controlled drug delivery. To further enhance biocompatibility of fibrous scaffolds, chemical/physical surface modifications (via plasma treatment or protein coating) and/or microstructural modifications (by the creation of microporosity or surface roughness), have been tested. In this chapter, composite scaffolds were synthesized with the aim of designing PCL matrices to be tailored to application-specific requirements, like controllable degradation rate and bioactive molecules delivery, tailor-made mechanical properties, improved cell adhesion and proliferation. For this purpose, a wide range of compositional blends have been tested, and the resulting structures have been chemically, morphologically and biologically evaluated. The experimental activities reported in this chapter have been designed and performed at the CEIT (Centro de estudios e investigaciones tecnicas de Gipuzkoa, University of 82
Navarra, San Sebastian, Spain), during the candidate s PhD visiting period, under the direct supervision of Prof. Gyeong-Man Kim. 5.1 Electrospinning in drug delivery In the last years, various polymers [80, 81], ceramics [82] and composites [83, 84] have been successfully electrospun into micro- and nanofibrous structures for tissue engineering purposes. Controlled delivery of drugs by their incorporation into polymer fibers represents one of the most investigated applications of electrospinning. Indeed, biocompatible water-soluble polymers such as polyethylene oxide (PEO), polyacrylamide (PAM) and polyvinylpyrrolidone (PVP) are effective in improving dissolution rates of poorly water-soluble drugs, avoiding the use of organic solvents [85, 86]. The incorporation of water-soluble biocompatible polymers during electrospinning process has also been successfully applied for the improvement of chemical, structural and topological features of several polymeric meshes, in order to increase hydrophilicity and biocompatibility [87, 88]. Achievable scaffold morphology strictly depends on the electrospinning set-up used in the fabrication process (Figure 41). In particular, Bogntizki et al. have taken advantage from the co-continuous phase morphologies resulted from electrospinning of two different polymers in the same solution to modify fiber surface topology. The selective dissolution of the water-soluble component led to fibers with controlled porosity and microarchitecture [89]. More recently, other authors have succeeded in overcoming limitation arising from low cell infiltration capability, typical of electrospun nanofiber membranes, by the use of dual-spinneret electrospinning set-up [90, 91]. They obtained a homogeneous, interpenetrated mixture of two nanofibers types; after selective removal of the water-soluble phase, the remaining mesh showed a significant increase in pore size and resulted in a better cell colonization. Finally, co-axial electrospinning, combined with the use of water-soluble polymers, has allowed the production of polymeric hollow fibers [92]. 83
a b c Figure 41: schematic diagrams of electrospinning process in different configurations: (a)single-nozzle [87] (b) dual-spinneret [14], (c) co-axial set-up [15] In the present chapter, simultaneous electrospinning of poly(ε-caprolactone) (PCL) blended with a biocompatible water-soluble polymer was used to produce scaffolds with tunable surface morphology and degradation rates. As a sacrificial component, PVP was chosen for its excellent biocompatibility, remarkable solubility in water and most organic solvents, and capability to interact with a wide range of hydrophilic and hydrophobic materials [93]. Thanks to its electrospinnability, it has been widely used alone [94-96] or in combination with other polymers [88, 97] or nanoparticles [98, 99] to produce micro/nano electrospun fibers for tissue engineering. PVP has recently been used in combination with PCL [100, 101]; however, to the Authors knowledge, no systematic investigation on the effect of PVP on the resulting scaffold topology after its extraction has been performed. Providing a better understanding of how to forge fibers with the desired microarchitectures has a great relevance when trying to gather control over scaffold mechanical properties, degradation rate and cell attachment. Although PCL is an established absorbable biopolymer, its hydrophobic nature and slow degradation kinetics may hamper its use in several biomedical applications requiring faster absorption rates. In fact, even if PCL degradation in vivo results in the formation of non-toxic and readily metabolized by-products, the process is too long (up to 5 years [102, 103] and its rate is difficult to control. The interest in trying to overcome this limitation is testified by the number of studies that have attempted to investigate PCL dissolution mechanisms and find possible solutions [104, 105]. The influence of factors such as molecular weight, morphology and chemical structure has 84
been investigated, and the effect of polymer crystallinity reduction over an accelerated degradation rate has also been demonstrated [106]. In the following sections, the possibility of blending PCL to PVP in a simultaneous electrospinning process achieving control over scaffold degradation, and ameliorating the cellular response of the biomaterial, is demonstrated. 5.2 Scaffold fabrication and characterization PCL (Mn=80,000) and PVP (Mn=360,000), were purchased from Sigma Aldrich (US) and used without further treatment or purification. Chloroform and Methanol were obtained from Merck Chemical Co. (Germany). Blend mixtures of PCL with PVP (100/0, 90/10, 80/20, 60/40, 50/50, 40/60 and 0/100 wt.%) were dissolved in chloroform/methanol (4/1 in vol.%) at different concentrations. The resulting solutions were vigorously stirred with a magnetic stir bar for at least 24h at room temperature to ensure homogeneity. In order to characterize bulk properties of PCL/PVP blends, mixture solutions were film-casted on the slide glasses. These membranes will be used as control for the evaluation of the effects induced by the electrospinning process. Figure 42: electrospinning set-up 85
Electrospinning was carried out at room temperature in a vertical configuration (Figure 42), using a 1 mm inner diameter flat-end needle with a 10 cm working distance. The applied voltages were in the range from 10 kv to 20 kv, driven by a high-voltage power supply (Knürr-Heizinger PNC, Germany). The flow rates were adjusted with a programmable syringe pump (model 210, KD Scientific Inc., USA). Electrospun fibers were collected in a non-woven fashion on a grounded 10 15 cm metal sheet covered with an aluminum foil under ambient conditions. The optimal process parameters are listed in Table 8. PCL/PVP Concentration Voltage Working distance (cm) Flow rate Average diameter (µm) FWHM (wt.%) (% w/v) (kv) (ml/hr) 100/0 20 10 10 0.75 1.20 0.50 90/10 20 10 10 0.75 1.48 1.20 80/20 15 10 10 0.75 1.44 1.32 60/40 15 10 10 0.75 1.88 1.90 50/50 15 10 10 0.75 1.56 1.37 40/60 15 10 10 0.75 1.27 0.48 0/100 15 20 10 0.75 0.95 0.26 Table 8: electrospinning parameters and their effect on fiber size and size distribution 5.2.1 Scanning electron microscopy (SEM) Size and morphology of the electrospun fibers were investigated by a scanning electron microscopy (Phenom G2 Pro Desktop Scanning Electron Microscope, Lambda Photometrics Limited, UK). Fibers diameter and size distribution were analyzed by measuring over 200 fibers in randomly recorded SEM micrographs using an image analyzer (Image J, developed by the National Institute of Health, USA). All samples were attached to SEM mounts using carbon film, and then sputtered with gold to avoid overcharging. 86
Figure 43: SEM micrographs of PCL/PVP electrospun fibers Figure 43 shows representative SEM images of the PCL and PCL/PVP blended electrospun fibers. Using the combination of electrospinning parameters in Table 8, electrospun fibers were successfully produced without any beads in the entire range of blend compositions. Figure 44: fiber diameters and distribution 87
Figure 44 shows the average fiber diameter, their distribution and the full width half maximum (FWHM) as a function of PVP concentration. Both, pure PCL and PVP, exhibited relatively smaller fiber diameter with uniform size distribution. The average fiber diameter of pure PCL (1.20 µm) increased with the amount of PVP up to 40 wt.% of PVP (1.88 µm) and then started to decrease. It is interesting to note that there was the same tendency of broadening in the size distribution depending on the PVP concentration, as in fiber diameter. These results were attributed to the impediment in the jet thinning during the electrospinning process. With increasing asymmetric blend composition the jet extruded from the nozzle becomes unstable and will be hardily stretched out during the flight to the counter electrode. As a consequence, fibers with larger diameter will be produced. The pure PVP fibers exhibited an average diameter of 0.95 µm and a narrowest size distribution. 5.2.2 Differential scanning calorimetry (DSC) DSC study on pure PCL, blend casted films (hereafter referred as bulk) and electrospun fibers, was performed by Perkin Elmer Differential Scanning Calorimeter (DSC 6000) in a temperature range of 15 100 C with a controlled heating and cooling rate of 5 C/min. The DSC temperature and heat flow values were calibrated with indium as standard. The weight of each sample was approximately 0.5 mg. Test was carried out in nitrogen atmosphere. The melting point (T m ), crystallization temperature (T c ) and % crystallinity of bulk and electrospun fibers, were determined from the heating and cooling scans. The degree of crystallinity (X c ) was assessed using DSC 6000 software by analyzing the endothermic area and calculated as reported in Section 1.5.2. Thermal and crystalline properties of bulk blends and blend ES-membranes were investigated as a function of the PVP weight fraction. The resulting DSC thermograms are shown in the Figure 45. 88
Figure 45: DSC thermograms for bulk and ES-membranes The most interesting feature in our DSC traces is that all samples, both bulk blends and ES-membranes, show a single endo- and exothermal peak over the entire composition range. This indicates that PCL in both cases existed in a semi-crystalline state. For bulk blends, an addition of 10 wt% of PVP into PCL determined a drastically decrease of the endothermic peak from 58.8 C to 54.0 C. However, despite further mixing with higher amount of PVP, T m remained almost constant. In comparison, the melting temperatures for blend ES-membranes were nearly invariant over the entire compositional range of PVP content. Interestingly, electrospinning affected the T m only for pure PCL, decreasing its value from 59 C to 54 C. The significant decrease in the melting temperature of pure PCL upon electrospinning could be attributed to the decrease in the overall crystallinity of the electrospun scaffold, which will be more discussed later in this section. Nevertheless, in the whole range of blend compositions, 89
melting peaks for bulk blends were always slightly higher than those for blend ESmembranes. In other words, electrospinning does not remarkably affect the melting behavior of fibers as compared to the bulk. Figure 45c) and d) show the exotherms upon cooling the melt samples. There is a discernable trend in the overall increase of the exothermic peaks after PVP addition. The crystallization temperature (T c ) increased with amount of PVP, indicating a progressively more difficult crystallization of PCL in the blend systems. In addition, it is noteworthy that T c was significantly shifted to higher values upon electrospinning, corresponding to the promoted rapid crystallization. This might be caused by the confined crystallization through the dimensional reduction of the material during the electrospinning process. From these results it can be concluded that PVP domains dispersed in the blend systems do not act as nucleation sites for PCL crystallization. Figure 46 plots T m, T c and X c of PCL/PVP blends against PVP weight fraction. The corresponding value of T m, T c and heat of fusion (ΔH m ) for all samples were summarized in Table 9, together with the calculated crystallinity. Figure 46: thermal and crystalline properties of bulk and electrospun PCL/PVP membranes 90
PCL/PVP T m ( C) T c ( C) ΔH m (J/g) Crystallinity (%) (wt.%) Bulk Fiber Bulk Fiber Bulk Fiber Bulk Fiber 100/0 58.8 54.0 22.1 29.7 76.7 68.4 55 49 90/10 54.2 54.3 24.8 33.5 56.5 55.2 45 44 80/20 53.0 53.6 21.7 31.1 46.9 46.9 42 42 60/40 53.9 53.1 27.3 32.8 33.5 32.6 40 39 50/50 53.8 52.9 25.9 35.1 27.2 25.1 39 36 40/60 54.0 53.1 26.5 32.3 20.6 19.0 37 34 Table 9: thermal and crystalline properties of bulk and electrospun PCL/PVP membranes The degree of crystallinity (DOC, i.e. volume of crystalline fraction) was calculated from the value of melting enthalpy (ΔH m ) of each blend, which is the integral of the endothermal peak in the DSC curve, using equation from Section 1.5.2 Hereby we used a reference enthalpy of fusion (ΔH f for totally crystalline PCL) of 139.5 J/g [107]. The overall crystallinity, in both systems decreased significantly as the amount of amorphous PVP increased, as clearly shown in Figure 46. This trend is characteristic of a crystalline-amorphous blend where the amorphous component merely dilutes the crystalline one. From microscopic point of view, the amorphous PVP chains can actually disrupt the formation of crystallites in semicrystalline PCL, by preventing close packing of the polymer chains. Consequently, as the content of PVP increases in the blend, the nucleation and growth of the PCL crystals are reduced significantly. It is interesting to note that the overall DOC in the blend ES-membranes was relatively lower compared to the bulk blends, indicating that the electrospinning caused a slightly decrease in crystallinity. However, the reduction of crystallinity for electrospun fibers, compared with the bulk polymer, is a general occurrence. In fact, the electrospinning process is associated with a high shear stress and a very rapid structure formation of the polymer material. Thus, it gives rise to a less appropriate environment to form perfect crystallites as in bulk. Further suppression of PCL crystallinity in the blend ESmembrane upon introducing PVP, is caused by the destruction of the orientation order of the macromolecule chains. This means that the amorphous PVP molecules act as 91
steric hindrances within the PCL-rich domains in the electrospun fibers, and thus the crystallization is restricted increasing the amount of PVP. 5.2.3 Attenuated Total Reflectance (ATR)-Fourier transform infrared spectroscopy (FTIR) IR spectroscopy is sensitive to polymer microstructure, and thus FTIR spectroscopic technique has proven to be a powerful tool in analyzing molecular specific interactions in the polymer blends. For the electrospun fibers, attenuated total reflectance (ATR) FTIR is ideally suited as a tool to obtain microstructural information [108], since clear transmission spectra are difficult to collect due to light scattering and surface reflection from the ultrafine electrospun fibers. In this study, the molecular structures formed in the polymer blends and electrospun nanofibers were investigated in detail by ATR- FTIR. Measurements were performed using a Perkin Elmer Spectrum 100 FT-IR spectrometer equipped with Universal Diamond/ZnSe ATR sampling accessory. Prior to sampling, the diamond crystal was cleaned with methanol and a background reading collected. To ensure good optical contact between the sample and ATR cell, which is essential to achieve strong spectral bands with an adequate signal-to-noise ratio, the samples were slightly compressed with a pressure arm. The corresponding ATR-FTIR spectra in the wavenumber range of 650 4000 cm 1 with a resolution of 2 cm 1 were collected in order to observe the conformational changes caused by blending of PCL with PVP and by electrospinning process. Figure 47 shows FTIR spectra of pure PCL and PVP in the 3000 1000 cm 1 region before and after electrospinning. 92
Figure 47: FTIR spectra of pure PCL and PVP bulk and ES-membranes As shown in Figure 47, characteristic bands of pure bulk PCL are observed and the corresponding assignments are listed above [109]: - CH stretching region (2800 3100cm 1 ) presents two bands, centered at 2944 and 2867cm 1 which are assigned to the asymmetric and symmetric stretching of methylene (CH 2 ) groups, respectively. - The strong peak at 1720 cm 1 is attributed to the C=O ester carbonyl group stretching mode. - Bands at: -1471 and 1419 cm 1 are assigned to methyl group (C-H) stretching, -1397 and 1365 cm 1 to symmetric and asymmetric C-O stretching, -1239 and 1164 cm 1 to asymmetric and symmetric COC stretching, -1294 cm 1 corresponds to ester group C-O and C-C stretching, -1107 cm 1 to the skeletal stretching for C-O and C-C, -1046 cm 1 to symmetric COC stretching. 93
The FTIR spectrum for bulk PVP (Figure 47) showed its characteristic bands. The CH stretching region (2800 3100cm 1 ) presents three bands, centered at 2955, 2924 cm 1 for C-H asymmetric and 2894 cm 1 symmetric stretching due to the CH 2 groups of the long aliphatic alkyl groups in PVP. The most intense absorption band is found at 1646 cm 1 due to the stretching vibration of the C=O group in amide. The absorption band at 1494 cm 1 corresponds to C-N stretching vibrations, bands at 1461 and 1422 cm 1 are assigned to CH deformation of cyclic CH 2 group. Bands at 1373 and 1318 cm 1 are attributed to C-H bending, those at 1287 and 1274 cm 1 to the characteristic absorption of C-N stretching in N-vinylpyrrolidone ring. Other weak bands are also found at 1229 and 1170 cm 1 which are assigned to CH 2 twist. Finally, the band at 1075 cm 1 is due to C H bending and ring deformation. As seen from Figure 47, the pure PCL and its electrospun fibers had almost identical spectra, indicating that no significant chemical modification was induced by electrospinning. In contrast to the PCL, the FTIR spectra for PVP revealed the shift of bands at 2894-2955 cm -1 and at 1646 cm -1, whereas the other bands were not or rarely affected by the electrospinning. The only peak at 1646 cm -1 for the carbonyl group had a blue-shift, i.e., toward a higher wavenumber of 1652 cm -1 after the electrospinning. For a given same solvent mixture, the reason why appeared changes in FTIR bands only in PVP and not in PCL could be explained in terms of the intrinsic dipole moment of polymers. It has been established that the dipole moment of polymer plays a crucial role in the productivity via electrospinning; higher values, correspond to smaller and more uniform electrospun fibers. From the literature, PCL has an electrical dipole moment of 0.64D, whereas PVP has a larger dipole moment of pyrrolidone groups (~3.3D)[110]. Under certain high electrostatic potential during the electrospinning process, such larger dipole moment in PVP makes its polymer chains easily to stretch out compared to PCL molecules. This ultra-high shear stress upon electrospinning not only induces an orientation of macromolecules but also, together with steric hindrance caused by the PVP pendent side groups, gives rise to the disassociation of PVP chains in the electrospun fibers. As a result, the carbonyl stretching vibration of PVP will shift to a higher wavenumber, i.e. appearing the blue-shift of the C=O group in FTIR spectra. The 94
above mentioned results were well reflected in the fiber morphology studied by SEM: the electrospun PVP fibers were more fine and uniform compared to the PCL ones. FTIR spectra of PCL/PVP blends before and after electrospinning are shown in Figure 48. Spectra are given between 2000 1000 cm -1, because most important spectral changes occurred in this range. In Figure 48a) and b), the top spectra exhibit the bands for pure PCL, while the bottom are relative to the PVP. As seen clearly, the strong absorption band at 1722 cm -1 is assigned to the carbonyl stretching vibration of PCL, and that centered at 1646 cm -1 to the C=O stretching vibration of PVP. Figure 48: FTIR spectra of PCL/PVP blends before and after electrospinning It is quite interesting to notice that there is no/little obvious variation in the location of the absorption IR peaks for the blends with respect to the pure polymers. Furthermore the appearance of new bands in the spectra of the mixtures was not detected, while the corresponding peak intensities were changeable with blend composition, in some extent. If PCL and PVP interact, e.g., in miscible or partially miscible blends, then it should be seen in the FTIR spectra a significant change in the intensity and/or frequency bands, due to interactions between different groups, as well as the emergence of new signals, if chemical reactions occurred between the components. However, the spectra for our blend systems in both cases, before and after electrospinning, were just resulted by a 95
superposition of the spectra of the pure polymers. This result strongly suggests that the specific interaction between the two components can be negligible. Although PCL and PVP both exhibited a carbonyl peak, as we confirmed by DSC, the PCL domains in the blends were still in semicrystalline phase. Carbonyl peak centered at 1722 cm -1 is sensitive to the conformation and packing of its molecular chains upon blending and even electrospinning. Therefore, we have more deeply inspected the PCL and PVP carbonyl stretching bands as a function of the blend composition, with the aim to measure the extent of such interactions and to explore about any structural changes upon electrospinning. Figure 49 shows scale-expanded FTIR spectra in the carbonyl stretching region (1740-1620 cm -1 ) for PCL/PVP blends and blend ES-membranes. Figure 49: scale-expanded carbonyl stretching region of FTIR spectra. (a) bulk, (b) ES-membranes. Ratio between carbonyl peak intensities (c) In this figure we have calibrated both peaks using the PCL band (1722 cm -1 ) as reference. As the PVP content increases in the blends, the intensity of PVP band (1652 cm -1 ) increases gradually and broadens simultaneously. It is worthy to note, that the electrospinning amplifies the C=O peaks of PVP drastically compared to those in PCL/PVP bulk blends (Figure 49b). The ratios of C=O peak intensity between PCL and PVP are shown in Figure 49c. It is clearly recognized, that the C=O peak ratios for the electrospun blend membranes were more than four times higher than in the case of bulk blends. This is one of nano-effects: for the electrospun fibers much more PVP domains are available on the surface to be monitored with FTIR spectroscopy than on the surface of bulk materials, in other words providing much larger specific areas. The dropping of 96
this ratio is much more rapid in the ES-membranes than in bulk blends, indicating that, due to the increase in the dynamical disorder of the blend components, the electrospinning enforces blends material readily to become amorphous. This result are in agreement with the trends observed from DSC measurements. From the summarized spectral features listed above, it can be concluded as followings; I) the presence of the crystalline phase throughout the entire composition range indicated that PVP existed as a heterogeneous phase. II) The dispersed PVP phase exhibits weak interaction with the PCL matrix, e.g., via van der Waals bonding. III) The trend of the spectra indicates that all polymer blends investigated were more amorphous than the PCL starting material. IV) The PCL/PVP blends have very low degree of miscibility. V) The electrospinning facilitates PCL/PVP blends to be amorphous. 5.3 Release study In vitro release tests have been performed on ES-membranes of a selected set of samples (PCL/PVP: 90/10, 80/20, 50/50). A controlled amount of blend electrospun meshes (2 mg) was incubated in 4 ml of Phosphate Buffer Solution (PBS, ph 7.4) at 37 C under static condition. At predetermined time intervals, the blend ES-membranes were retrieved from the incubation buffer and the PVP release was detected by: UVvis., FT-IR and SEM analysis. 5.3.1 Short term study Incubation buffers were collected and the amounts of released PVP were determined spectrophotometrically using a Perkin Elmer Lambda 35 UV vis spectrometer. The characteristic UV absorbance peak of PVP in buffer solution was determined at λmax = 207.60 nm and converted to the PVP concentration according to the calibration curve of this polymer in the same buffer. The percentage of PVP released from the matrix was calculated according to the following equation: [ C] VPBS P[%] = 100 Q 0 97
where [C] is the unknown PVP concentration value (mg/ml), determined on the basis of calibration curve; V PBS is the volume of dissolution medium (ml of PBS), Q 0 is the total amount of PVP in 2mg of electrospun matrix. All spectra were recorded at room temperature and the results were averaged on three runs. Figure 50: short-term PVP release profile detected by Uv-visible spectroscopy Figure 50 shows the results of these studies. PVP release profiles are given in percentage, over the incubation time. All graphs are characterized by an initial fast (burst) release and a subsequent prolonged release. The initial burst observed for the 90/10, 80/20 and 50/50 PCL/PVP was 67%, 75% and 97%, after 1h incubation. At the end of incubation time (24h), the remaining weight percentages of PVP were around 25, 10, and almost 0%, respectively. It is interesting to note that the PVP in 50/50 blend system was almost completely leached out within a day. The period of our experiment by UV-vis was too short to expect significant release of PVP by PCL degradation, as PCL degradation is slow in PBS medium due to its semi-crystalline and hydrophobic nature. Thus, PVP release study from PCL electrospun nanofibers over long periods of time has been considered. 98
5.3.2 Long term study The long term study of in vitro degradation was performed by FTIR spectroscopy and changes in surfaces morphologies at different time periods were also investigated by SEM. Characteristic peaks of 1722 cm -1 for PCL and 1652 cm -1 for PVP were used to analyze the PVP release. Peak ratios before degradation experiment were normalized with the measured ratios as a function of incubation time. Figure 51 shows release profiles determined by FTIR measurements. Figure 51: long-term PVP release profile detected by quantitative FT-IR spectroscopy By these long term release studies, it clearly appears that the release kinetics are characterized by three stages: at short times, there is an initial fast (burst) release (Stage I), while after about 1 day, the release rate becomes essentially constant (Stage II) until 10-30 days. In the Stage III, the release rate increases with increasing the content of PVP in the fibers. It should be emphasized here that within our experimental period (60 days) the incomplete release was observed in the 80/20 and 90/10 blend ES-membranes. While the PVP domains were almost interconnected in the 50/50 blend, poor interconnectivity was shown in the 80/20 and 90/10 blend ES-membranes which restrict the diffusion of PVP entrapped deep inside the fibers. More than 5 and 10% were remained 99
within the ES-membranes (in 80/20 and 90/10, respectively) at the end of the incubation period. The initial burst (Stage I) was attributed to the amount of PVP phase present on the ES fibers. This might be due to the effect of electrospinning on the degree of phase separation between the PVP and PCL domains. Electrospinning of a polymer blend solution might kinetically impose the extensive phase separation between PVP and PCL, due to the quick solvent evaporation during the electrospinning process, which results in the ultra-fine domain within the electrospun fibers. Furthermore, it is believed that at higher PVP concentrations, due to the fact that PCL/PVP blends are immiscible, the PVP phase had more tendency to migrate to the surface or near the surface of nanofibers during the processing. It means that at higher concentration of PVP, a higher amount of polymer was located on the ES-membrane surface. Therefore, it can dissolve in PBS medium and be easily washed out, leading to a faster PVP release rate. Stage II is a linear stage mainly influenced by the hydrolytic stability of the skeletal PCL fiber matrix. It is characterized by a relatively slow release due to the difficulty in PVP diffusion out of the PCL matrix. In fact, the release of PVP entrapped in the ES fibers occurs only after the medium has penetrated into the channels and reached the entrapped PVP domains. Stage III is a phase of an increased release, caused by polymer degradation which leads to an higher permeability of the buffer solution into the PCL fiber matrix. The retrieved PCL/PVP blend electrospun meshes were further investigated with SEM in order to characterize their morphological changes caused by degradation after PBS incubation. Before the characterization, samples were rinsed several times with distilled water and then dried in vacuum at room temperature. 100
Figure 52: SEM micrographs of blend ES-membranes as function of incubation time Figure 52 shows SEM micrographs of blend ES-membranes as a function of incubation time. As expected, the ES-membrane of pure PCL didn t affected by the buffer solution, due to its relative high crystallinity and hydrophobicity. After 5h, it can be clearly seen that the blend ES-fibers showed a rugged surface morphology, in which the roughness became higher with increasing the PVP content. In one day incubation, the number of ultra-fine pores were observed on the surface of every single fibers, obviously. These results indicate that the PVP domains dispersed on the fiber surface were dissolved by contact with the incubation medium, making submicron-sized porosity within the individual fibers. As the duration of incubation increased, the number of pores on the fiber surface increased and became more clear. It is also interesting to note that the 101
number of fibers decreased with increasing incubation time; as seen in the last row, only a few of thicker ES-fibers were endured at the end of incubation period (60 days). This result indicates that the fiber diameter significantly influenced in vitro degradation of blend ES-Membranes. This effect was most likely attributed to the differences in the surface to volume ratio of the fibers; the thinner fibers exhibit much higher surface to volume ratio in comparison to the thicker ones. A high surface to volume ratio means high water penetration, which leads to high degradation rates. It is emphasized that the blend ES-fibers maintained structural integrity in the original dimension till the end of the experiments, although they were massively eroded due to dissolution and extraction of the hydrophilic PVP part from the fibers upon incubation. 5.3.3 Mechanical characterization Mechanical tests were performed on PCL/PVP electrospun samples with different PVP weight percentages (0, 20, 50). After one day of immersion in PBS, samples were retrieved, completely dried and tested using an Instron machine with a 10N cell load. A paper frame with central square window was used as a support for specimen preparation. Fiber mesh was placed in the central square window and glued to the paper frame. After clamping, lateral sides were cut and the tensile test was started. In accordance with [111], a speed of 30 mm/min was selected and the test was extended up to failure. Specimens were 3 cm long, 0.7 cm wide and 50 µm thick (width and thickness were measured prior to the experiments with a flat tip micrometer). At least five samples for each scaffold type were tested. 102
Figure 53: tensile tests on PCL/PVP ES-membranes Histogram in Figure 53 summarizes values of ultimate stress and displacement at ultimate stress for three different PVP contents. As expected, changes in fiber morphology due to PVP release affect scaffolds mechanical behavior. In particular, with the increase of PVP content, a reduction of both parameters was detected. 5.3.4 Three-stages release model On the basis of the collected results a three-stage release kinetics for the PCL/PVP blend ES-membranes has been proposed. Figure 54 shows the plot of PVP concentration versus release time and a schematic representation of the three-stage release kinetics. The dissolution of PVP domains in Stage I results in the formation of pore in submicron-size on the electrospun fibers. Once the initial burst is overcome, the gradual formation of channels between the entrapped PVP domains is initiated by the hydrolysis of PCL skeletal phase that increase the specific surface areas of the ES fibers. The fine channels are reached to the entrapped PVP domains, which allows to diffuse and permeate the buffer solution through the PCL skeletal phase and then to wash out PVP from the ES fibers. Noted that the release rate in the Stage I is higher for the ES fiber with higher content of PVP. The duration of the stationary release in the Stage II will be extended with the amount of PVP. It is most likely attributed to the fact that less PVP in the ES fibers correspond to larger distances between the entrapped PVP domains. The diffusion and permeation of the buffer solution to the PVP domains in larger distance 103
need more time than in short distance. The trend of the release rate in Stage III is just the same as in Stage I, i.e., the higher PVP content, the higher the release rate. Figure 54: three-stage release kinetics for blend ES-membranes with different PVP contents Stage I: burst release by formation of pores on the electrospun fibers Stage II: stationary release by formation of channels within the electrospun fibers Stage III: slow release of the entrapped PVP in the electrospun fibers 5.4 Biocompatibility assay Cell viability and proliferation were determined using the Vybrant Cytotoxicity Assay Kit and the MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide] assays, respectively. The first monitors the release of the cytosolic enzyme glucose 6-phosphate dehydrogenase (G6PD) from damaged cells into the surrounding medium. The second is based on the principle that the mitochondrial dehydrogenases of viable cells cleave the yellow tetrazolium salt MTT to produce purple formazan crystals. Three different experimental groups have been considered: 104
- pure PCL ES-membranes (as control); - PCL/PVP (80/20 wt.%) ES-membranes before PBS immersion; -PCL/PVP (80/20 wt.%) ES-membranes after PBS immersion (at stage II of release profile). Disk-shaped scaffolds (3 for each groups, Ø = 6 mm) were UV sterilized and seeded with human mesenchymal stem cells (hmsc passage 6, Lonza) at a concentration of 1 10 4 cells/cm 2. Cells were cultured in direct contact with the surfaces in α-mem supplemented with 1% penicillin/streptomycin, 10% FBS (fetal bovine serum), 1% L- glutamine. Proliferation was determined at 1, 3 and 7 days by using the above mentioned Assay in accordance with the manufacturer's instructions. Briefly, constructs were incubated with 5 mg/ml MTT at 37 C, 5% CO 2 for 3h, and then the absorbance at 595 nm was measured with a microplate reader (Tecan, Männedorf, CH). Figures summarize the obtained results. Figure 55: cytotoxicity and proliferation assay 105
Above-mentioned cytotoxicity assay showed a 100% viability in all the experimental groups. Proliferation results evidenced a positive trend over time; however, at day 7, significant differences were observed between the two composite scaffolds. Furthermore, since the fibers porosity produced by the PVP extraction in the composite membranes at Stage 2 of the release profile, hmscs have been more easily grown on these scaffolds than on the pure PCL ones. Conclusions Defect-free PCL/PVP fibers were successfully produced by electrospinning and the effect of fabrication process on internal microstructure was evaluated in comparison with cast films. Selective removal of PVP led to fibers with specific surface topology and nanosized intrafibrillar porosity, demonstrating the efficacy of simultaneous electrospinning of PCL with a sacrificial, water-soluble polymer in forging the ultrastructural features of electrospun materials. The PCL/PVP ratio demonstrated to have great influence on scaffold morphology, as well as on the degradation rate. Presence of residual PVP was reported in scaffolds with higher PCL content (80/20 and 90/10 samples), reasonably due to the formation of PVP domains entrapped into the PCL-rich fibers. A three-stage model of long-term PVP release kinetics was proposed on the basis of the obtained results. The porous nanostructure deriving from PVP dissolution resulted in scaffolds with improved cell adhesion. The present work envisages the possibility of designing electrospun PCL scaffolds to be tailored to application-specific requirements, by regulating the compositional ration with a sacrificial templating phase. This methodology has the potential to increase the versatility of electrospun materials in several biomedical applications, ranging from tissue regeneration to controlled drug delivery. 106
Chapter 6: Conclusions and future work This thesis aimed at developing scaffolds for tissue engineering and regenerative medicine, combining bio-inspired designs with the use of novel biomimetic materials processed by advanced fabrication techniques. Although, one target application has been selected for each presented engineering solution, the proposed approaches can be adopted in a wide range of TE fields. Hereinafter, together with concluding remarks, an outlook to the ongoing and future works, is reported. - A versatile and facile rapid prototyping station for the production of 3D scaffolds, tailored on specific design, has been developed. The feasibility of using this system for the manufacturing of structures with controlled micro-architectures has been demonstrated in case of PCL and PCL-hydroxyapatite (PCL-HA) composites. The possibility to operate under sterile conditions makes this instrument the ideal choice for organ printing approaches, intended as the rapid prototyping of engineered biological tissues obtained by the 3D printing of a biomaterial gel containing a suspension of living cells. Furthermore, using the different available head types, the deposition of many other materials, such as single- or two-component gels as well as cell suspensions, is allowed. Thus, future aim will be the use of this system for the fabrication of cellularised and non-cellularised scaffolds for several purposes. To perform this activity, the whole RP equipment will be inserted within a laminar flow hood and all the surfaces in contact with biological materials will be sterilized by autoclaving or UV irradiation. Material gelling will occur immediately after its deposition due to the thermal-induced transition obtainable increasing the temperature of the thermo-stated holding stage to the physiological value (37 C). - Soft scaffolds with surface morphology and degradation rate tunable on the basis of pore-forming content, have been developed by the use of electrospinning process. The addition of a water soluble polymer to the starting solutions, led to a nanostructured porosity into otherwise smooth PCL fibers. This methodology has the potential to improve cell proliferation and increase the versatility of electrospun matrices in several 107
biomedical applications, ranging from soft tissue regeneration to controlled drug delivery. Further research concerning the improvement of grown tissue functionality by varying scaffold topology, is under study. Preliminary results regarding the application of electrospun biomimetic matrices for tendon regeneration, are shown in the following section ( 6.1.1). - After assessing the beneficial effect of hybrid organic/inorganic scaffolds on hard tissues replacement, we are now moving toward the development of Class II hybrid materials. While class I have been produced by dispersing bioactive HA nanopowders in a polymeric matrix, class II materials require the formation of chemical bonds between organic and inorganic moieties. A protocol for the fabrication of bioactive hybrid foams has been recently developed slightly modifying the sol-gel reactions usually applied in the bioglass synthesis. Preliminary results are shown in the following section ( 6.1.2). 6.1 Ongoing research 6.1.1 Biomimetic micro-fibrous PLLA scaffold for tendon regeneration: a preliminary in vitro study Management of tendon lesions accompanied by big loss of substance represents a serious issue in orthopaedic surgery. Such lesions, which cannot be repaired with an end-to-end suture, demand for tendon augmentation procedures. It is known that tendon niche comprises primarily parallel collagen fibers, and plays an important role in regulating their function and differentiation. Aim of this study is the synthesis of a biomimetic micro-fibrous PLLA scaffold with aligned fibres to be used in combination with tenocytes and tendon stem cells, to produce a construct for tendon augmentation. Preliminary in vitro results were obtained with tenocytes. [Ruzzini L, Abbruzzese A, Giannitelli SM et al, Histology and Histophatology 2011] 108
Materials and Methods Aligned fibers were synthesized by electrospinning adopting a spinning disk configuration, starting from a poly-l-lactide (inherent viscosity 0.9-1.2 dl/g) 13 wt% solution in a 10:1 dichloromethane/methanol mixture. The solution was then fed into a 5 ml glass syringe, which was controlled by a syringe pump with a rate of 2.5 ml/h. A high voltage (15 kv) was applied to the needle tip, placed 15 cm above the collector. The disk rotating speed was set at 6000 rpm. Scanning electron microscopy (SEM) was adopted to evaluate samples morphology. Two different human cell populations were isolated for the study and seeded on the scaffolds: tenocytes from intact tendons (TI) and from ruptured tendons (TR). Scaffolds were investigated in terms of cell viability and proliferation. Quantization and typing (type I and III) of synthesized collagen was performed after 1 and 3 weeks of in vitro culturing. Preliminary results SEM analysis confirmed that we were able to obtain fibrous PLLA scaffolds with a high degree of fibres alignment. Constructs revealed good engraftment, with cell in-growth and proliferation within the scaffolds, and a specific organization following the direction of PLLA microfibers orientation (Figure 56). 109
Figure 56: cell alignment on PLLA electrospun fibers, after 7 and 21 days TI and TR showed comparable results in terms of total collagen production. In particular, TIs showed the expression of Collagen I, while Collagen III was not expressed; at the same timepoint, TRs showed a shift in collagen expression from Type III to Type I. Conclusions PLLA electrospun scaffolds with aligned fibres, demonstrated compatibility with both tenocytes isolated from ruptured and intact tendons. Scaffolds promoted the aligned orientation of cells, and induced type I collagen production of tenocytes. As type I collagen is normally expressed in intact and healthy tendons, while type III collagen is 110
expressed in ruptured or tendinopathic tendons, we think that such results candidate this scaffold as promising for tendon augmentation strategies. Main future purpose will be to determine the effects of aligned scaffold topography on the differentiation of human tendon stem cells (htscs). The multi-differentiation potential of htscs toward osteogenesis, adipogenesis, and chondrogenesis is well established. To evaluate the effect of this biomimetic niche toward tendon-lineage differentiation, the expression of specific markers at the mrna level will be examined by means of real time PCR. 6.1.2 Preparation of a polycaprolactone/bioglass hybrid scaffold through a direct sol-gel method Over the last decade, many efforts have been made towards the development of new bone replacement materials. Class II hybrid scaffolds can represent a valuable solution to overcome the limitations of conventional ceramic and polymeric bone and osteochondral substitutes, due to the potential derived from the strong bonding between organic and inorganic domains. In particular, a bioactive poly(ε-caprolactone) (PCL)/bioglass hybrid II scaffold was successfully synthesized stating from a polyurethane foam. Materials and Methods The polymer precursor, PCL diol (Sigma, MW=2000) was mixed with 3- isocyanatopropyl triethoxysilane (IPTS, Sigma) using 1,4 diazabicyclo[2,2,2]octane (DABCO; Sigma) as a catalyst. The molar ratio of PCL, IPTS, and DABCO was 1:3:2 and the reaction was performed in flowing dry N 2 gas at 50 C for 10 minutes in a teflon batch. Sol-gel process was employed in order to obtain chemical bonds between the two components, instead of a simple dispersion of the ceramic powders in a polymer matrix. Thus, tetraethyl-orthosilicate (TEOS, Sigma), ethanol (Sigma) and nitric acid (1M), previously mixed in a different batch, were added to the prepared mixture. The polyurethane hybrid scaffolds were obtained by dispersing poly(methylenephenyldiisocyanate) (PMDI, Voranate M220, Dow Chemicals, industrial 111
Tesi di dottorato in Ingegneria Biomedica, di Sara Maria Giannitelli, grade) in the polyol solution and adding a proper surfactant. Reaction continued until the increasing viscosity prevented continuation of stirring. Polymerization occurred in 8h into a desiccator at room temperature. Hybrid materials underwent to a complete characterization by field emission scanning electron microscopy (FE-SEM), ATR/FTIR spectroscopy, mechanical compression tests, in vitro degradation, and cytotoxicity. To further improve the potential of the fabricated composites conferring bioactive function to the construct, calcium and phosphorus precursors were added in the following stoichiometric ratio: SiO2:CaO:P2O5 = 70:26:4 [Rainer A, Giannitelli SM et al, Acta Biomaterialia 2008]. Preliminary results FE-SEM micrographs of polyurethane hybrid foams (Figure 57), showed a quite homogenous and highly porous structure, with pore features (size and interconnectivity) suitable for cell penetration. Figure 57: SEM micrographs of polyurethane hybrid foams The addition of the inorganic domain resulted in an increase of degradation rate respect to the polyurethane control foams, while didn t affect their mechanical behavior. In vitro biocompatibility results (hmscs p7, Lonza), evaluated by means of Vybrant Citotoxicity Assay Kit (Invitrogen), revealed a cell viability higher than 98%. 112
Preliminary bioactivity test (Figure 58), assessed by soaking bioglass hybrid constructs (containing Ca and P) in simulated body fluid (Kokubo test), showed some roundish micron-sized particles on the surface, each formed by nanometric needles. Figure 58: SEM micrograph of bioactive hybrid foam after 1week in Kokubo solution Conclusions This new class of hybrid scaffolds is likely to have the potential to be used as a bone repairing materials. Therefore, a protocol optimization together with a more systematic characterization of bioactive constructs, will be necessary to fully understand the potential of this innovative synthesis in providing cell bio-instruction toward the desired phenotype. 113
References 1. Schantz, J.-T., et al., Cranioplasty after Trephination using a Novel Biodegradable Burr Hole Cover: Technical Case Report. Neurosurgery, 2006. 58(1): p. ONS-E176 10.1227/01.NEU.0000193533.54580.3F. 2. Williamson, M.R. and A.G.A. Coombes, Gravity spinning of polycaprolactone fibres for applications in tissue engineering. Biomaterials, 2004. 25(3): p. 459-465. 3. Hutmacher, D.W., M. Sittinger, and M.V. Risbud, Scaffold-based tissue engineering: rationale for computer-aided design and solid free-form fabrication systems. TRENDS in Biotechnology, 2004. 22(7): p. 354-362. 4. Coccoli, V., et al., Engineering of poly(ε-caprolactone) microcarriers to modulate protein encapsulation capability and release kinetic. Journal of Materials Science: Materials in Medicine, 2008. 19(4): p. 1703-1711. 5. Luciani, A., et al., PCL microspheres based functional scaffolds by bottom-up approach with predefined microstructural properties and release profiles. Biomaterials, 2008. 29(36): p. 4800-4807. 6. Cipitria, A., et al., Design, fabrication and characterization of PCL electrospun scaffolds-a review. Journal of Materials Chemistry, 2011. 21(26): p. 9419-9453. 7. Place, E.S., et al., Synthetic polymer scaffolds for tissue engineering. Chemical Society Reviews, 2009. 38(4): p. 1139-1151. 8. Fergal J, O.B., Biomaterials & scaffolds for tissue engineering. Materials Today, 2011. 14(3): p. 88-95. 9. Boccaccio A, et al., Finite Element Method (FEM), Mechanobiology and Biomimetic Scaffolds in Bone Tissue Engineering. Int J Biol Sci 2011. 7(1): p. 112-132. 10. Owen, S.C. and M.S. Shoichet, Design of three-dimensional biomimetic scaffolds. Journal of Biomedical Materials Research Part A, 2010. 94A(4): p. 1321-1331. 11. Stevens, B., et al., A review of materials, fabrication methods, and strategies used to enhance bone regeneration in engineered bone tissues. Journal of Biomedical Materials Research Part B: Applied Biomaterials, 2008. 85B(2): p. 573-582. 12. Ho, S.T. and D.W. Hutmacher, A comparison of micro CT with other techniques used in the characterization of scaffolds. Biomaterials, 2006. 27(8): p. 1362-1376. 13. Moroni, L., J.R. de Wijn, and C.A. van Blitterswijk, Integrating novel technologies to fabricate smart scaffolds. Journal of Biomaterials Science, Polymer Edition, 2008. 19(5): p. 543-572. 14. Liao, S., C. Chan, and S. Ramakrishna, Electrospun nanofibers: Work for medicine? Frontiers of Materials Science in China, 2010. 4(1): p. 29-33. 114
15. Bártolo, P.J.S., H. Almeida, and T. Laoui, Rapid prototyping and manufacturing for tissue engineering scaffolds. Int. J. Computer Applications in Technology, 2009. 36(1): p. 1-9. 16. Sun, W., et al., Bio-CAD modeling and its applications in computer-aided tissue engineering. Comput. Aided Des., 2005. 37(11): p. 1097-1114. 17. Chu, T.M.G., et al., Manufacturing and Characterization of 3-D Hydroxyapatite Bone Tissue Engineering Scaffolds. Annals of the New York Academy of Sciences, 2002. 961(1): p. 114-117. 18. Hollister, S.J., Porous scaffold design for tissue engineering. Nature Materials, 2005. 4(7): p. 518-524. 19. Jones, A.C., et al., Assessment of bone ingrowth into porous biomaterials using MICRO-CT. Biomaterials, 2007. 28(15): p. 2491-2504. 20. Ratner, B.D., Biomaterials science : an introduction to materials in medicine. 2004, Amsterdam; Boston: Elsevier Academic Press. 21. Reis, R.L. and J.S. Román, Biodegradable systems in tissue engineering and regenerative medicine. 2005: CRC Press. 22. Rainer, A., et al., Comparative study of different techniques for the sterilization of poly- L-lactide electrospun microfibers: effectiveness vs. material degradation. Int J Artif Organs, 2010. 33(2): p. 76-85. 23. Stephens, J.S., et al., Perfusion flow bioreactor for 3D in situ imaging: Investigating cell/biomaterials interactions. Biotechnology and Bioengineering, 2007. 97(4): p. 952-961. 24. Barnes, S.J. and L.P. Harris, Tissue Engineering: Roles, Materials and Applications. 2008: Nova Science Publishers. 25. Lacroix, D., Computer-aided design and finite-element modelling of biomaterial scaffolds for bone tissue engineering. Phil. Trans. R. Soc. A, 2009. 367: p. 1993-2009. 26. Sun, W., Computer aided tissue engineering application to biomimetic modeling and design of tissue scaffold. Biotechnol. Appl. Biochemistry, 2004. 39(1): p. 49-58. 27. Hutmacher, D.W., M. Sittinger, and M.V. Risbud, Scaffold-based tissue engineering: rationale for computer-aided design and solid free-form fabrication systems. TRENDS in Biotechnology 2004. 22(7): p. 132. 28. Tellis, B.C., Trabecular scaffolds created using micro CT guided fused deposition modeling. Materials Science and Engineering: C, 2008. 28(1): p. 171-178 29. Hutmacher, D.W., Mechanical properties and cell cultural response of polycaprolactone scaffolds designed and fabricated via fused deposition modeling. J. Biomed. Mater. Res., 2001. 55(2): p. 203 216. 115
30. Kalita, S.J., Development of controlled porosity polymer-ceramic composite scaffolds via fused deposition modeling. Materials Science and Engineering C 2003. 23(5): p. 611 620. 31. Williams, J.M., Bone tissue engineering using polycaprolactone scaffolds fabricated via selective laser sintering. Biomaterials, 2005. 26: p. 4817-4827. 32. Swieszkowski, W., Repair and regeneration of osteochondral defects in the articular joints Biomolecular Engineering, 2007. 24(5): p. 489-495. 33. Shao, X., Repair of Large Articular Osteochondral Defects Using Hybrid Scaffolds and Bone Marrow-Derived Mesenchymal Stem Cells in a Rabbit Model. Tissue Engineering, 2006. 12( 6): p. 1539-1551. 34. Lee, C.H., et al., Regeneration of the articular surface of the rabbit synovial joint by cell homing: a proof of concept study. Lancet, 2010. 376: p. 440-448. 35. Abbah, S.A., Biological performance of a polycaprolactone-based scaffold used as fusion cage device in a large animal model of spinal reconstructive surgery. Biomaterials, 2009. 30(28): p. 5086-5093 36. Gibson, I., et al., Towards a Medium/High Load-Bearing Scaffold Fabrication System. Tsinghua Science and Technology, 2009. 14, Supplement 1(0): p. 13-19. 37. Shor, L., Precision extruding deposition (PED) fabrication of polycaprolactone (PCL) scaffolds for bone tissue engineering Biofabrication, 2009. 1: p. 015003. 38. Centola, M., et al., Combining electrospinning and fused deposition modeling for the fabrication of a hybrid vascular graft Biofabrication, 2010. 2: p. 014102. 39. Centola, M., et al., Free-form fabrication of biopolymeric scaffolds for the regeneration of osteochondral bone layer Tissue Engineering Part A., 2010. 16: p. A1-A29. 40. Gunatillake, P.A. and R. Adhikari, Biodegradable synthetic polymers for tissue engineering. Eur Cell Mater, 2003. 5(1). 41. Vallet-Regi, M. and D. Arcos, Nanostructured Hybrid Materials for Bone Tissue Regeneration. Current Nanoscience, 2006. 2(3): p. 179-189. 42. Arcos, D. and M. Vallet-Regí, Sol gel silica-based biomaterials and bone tissue regeneration. Acta Biomaterialia, 2010. 6(8): p. 2874-2888. 43. Shor, L., et al., Precision Extruding Deposition for Freeform Fabrication of PCL and PCL-HA Tissue Scaffolds Printed Biomaterials, R. Narayan, T. Boland, and Y.-S. Lee, Editors. 2010, Springer New York. p. 91-110. 44. Shor, L., et al., Fabrication of three-dimensional polycaprolactone/hydroxyapatite tissue scaffolds and osteoblast-scaffold interactions in vitro. Biomaterials, 2007. 28(35): p. 5291-5297. 45. Eosoly, S., et al., Selective laser sintering of hydroxyapatite/poly-ε-caprolactone scaffolds. Acta Biomaterialia, 2010. 6(7): p. 2511-2517. 116
46. Wiria, F.E., et al., Poly-ε-caprolactone/hydroxyapatite for tissue engineering scaffold fabrication via selective laser sintering. Acta Biomaterialia, 2007. 3(1): p. 1-12. 47. Park, S., S. Lee, and W. Kim, Fabrication of porous polycaprolactone/hydroxyapatite (PCL/HA) blend scaffolds using a 3D plotting system for bone tissue engineering. Bioprocess and Biosystems Engineering, 2011. 34(4): p. 505-513. 48. Mironov, V., et al., Organ printing: computer-aided jet-based 3D tissue engineering. TRENDS in Biotechnology, 2003. 21(4): p. 157-161. 49. Centrella, M., T.L. McCarthy, and E. Canalis, Effects of Transforming Growth Factors on Bone Cells. Connective Tissue Research, 1989. 20(1-4): p. 267-275. 50. Cowan, C.M., et al., Evolving Concepts in Bone Tissue Engineering, in Current Topics in Developmental Biology. 2005, Academic Press. p. 239-285. 51. Yilgor, P., et al., Effect of scaffold architecture and BMP-2/BMP-7 delivery on in vitro bone regeneration. Journal of Materials Science: Materials in Medicine, 2010. 21(11): p. 2999-3008. 52. Yang, S., et al., The Design of Scaffolds for Use in Tissue Engineering. Part II. Rapid Prototyping Techniques. Tissue Engineering, 2002. 8(1): p. 1-11. 53. Masood, S.H., J.P. Singh, and Y. Morsi, The design and manufacturing of porous scaffolds for tissue engineering using rapid prototyping. The International Journal of Advanced Manufacturing Technology, 2005. 27(3): p. 415-420. 54. Cheah, C.M., et al., Development of a Tissue Engineering Scaffold Structure Library for Rapid Prototyping. Part 1: Investigation and Classification. The International Journal of Advanced Manufacturing Technology, 2003. 21(4): p. 291-301. 55. Hollister, S.J., R.D. Maddox, and J.M. Taboas, Optimal design and fabrication of scaffolds to mimic tissue properties and satisfy biological constraints. Biomaterials 2002. 23: p. 4095-4103 56. Rajagopalan, S. and R.A. Robb, Schwarz meets Schwann: Design and fabrication of biomorphic and durataxic tissue engineering scaffolds. Medical Image Analysis, 2006. 10(5): p. 693-712. 57. Hoque, M.E., et al., Fabrication using a rapid prototyping system and in vitro characterization of PEG-PCL-PLA scaffolds for tissue engineering. Journal of Biomaterials Science, Polymer Edition, 2005. 16(12): p. 1595-1610. 58. Sobral, J.M., et al., Three-dimensional plotted scaffolds with controlled pore size gradients: Effect of scaffold geometry on mechanical performance and cell seeding efficiency. Acta Biomaterialia, 2011. 7(3): p. 1009-1018. 59. Starly, B. and W. Sun, Internal scaffold architecture designs using lindenmayer systems. Computer-Aided Design and Applications, 2007. 4(1-6): p. 395-403. 60. Pandithevan, P. and G. Saravana Kumar, Personalised bone tissue engineering scaffold with controlled architecture using fractal tool paths in layered manufacturing. Virtual and Physical Prototyping, 2009. 4(3): p. 165-180. 117
61. Rainer, A., et al., Load-Adaptive Scaffold Architecturing: A Bioinspired Approach to the Design of Porous Additively Manufactured Scaffolds with Optimized Mechanical Properties. Annals of Biomedical Engineering: p. 1-10. 62. Eshraghi, S. and S. Das, Mechanical and microstructural properties of polycaprolactone scaffolds with one-dimensional, two-dimensional, and threedimensional orthogonally oriented porous architectures produced by selective laser sintering. Acta Biomaterialia, 2010. 6(7): p. 2467-2476. 63. Pandithevan, P., Reconstruction of subject-specific human femoral bone model with cortical porosity data using macro-ct. Virtual and Physical Prototyping, 2009. 4(3): p. 115-129. 64. Chen, Z., Biomimetic modeling and three-dimension reconstruction of the artificial bone. Computer methods and programs in biomedicine 2007. 88(2): p. 123 130. 65. Pandithevan, P. and G. Saravana Kumar, Finite element analysis of a personalized femoral scaffold with designed microarchitecture. Proc. IMechE Part H: J. Engineering in Medicine, 2010. 224(7): p. 877-889. 66. Shipley, R.J., et al., Design criteria for a printed tissue engineering construct: A mathematical homogenization approach. Journal of Theoretical Biology, 2009. 259(3): p. 489-502. 67. Lin, C.Y., A novel method for biomaterial scaffold internal architecture design to match bone elastic properties with desired porosity Journal of Biomechanics, 2004. 37(5): p. 623 636. 68. Cheah CM, C.C., Leong KF, Automatic algorithm for generating complex polyhedral scaffolds for tissue engineering. Tissue Engineering, 2004. 10: p. 595-610. 69. Melchels, F.P.W., Mathematically defined tissue engineering scaffold architectures prepared by stereolithography. Biomaterials, 2010. 31: p. 6909-6916. 70. Starly B, L.W., Bradbury T, Sun W., Internal architecture design and freeform fabrication of tissue replacement structures. Computer-Aided Design, 2006: p. 115-24. 71. Landau, D.L., Theory of Elasticity. Oxford: Butterworth Heinemann, 1986: p. 187. 72. Cheung, G., et al., Finite element analysis of a femoral retrograde intramedullary nail subject to gait loading. Medical Engineering & Physics, 2004. 26: p. 93 108. 73. Papini, M., et al., The biomechanics of human femurs in axial and torsional loading: comparison of finite element analysis, human cadaveric femurs, and synthetic femurs. Journal of Biomechanical Engineering, 2007. 129: p. 12-19. 74. Keaveny, T.M., et al., Trabecular bone exhibits fully linear elastic behavior and yields at low strains. Journal of Biomechanics, 1994. 27: p. 1127-1136. 75. McIntosh, L., J.M. Cordell, and A.J. Wagoner Johnson, Impact of bone geometry on effective properties of bone scaffolds. Acta Biomaterialia, 2009. 5: p. 680 692. 118
76. Viceconti, M., A comparative study on different methods of automatic mesh generation on human femurs. Med Eng Phys 1998. 20: p. 1-10. 77. Koch, J.C., The laws of bone architecture. American Journal of Anatomy, 1917. 21(2): p. 177-298. 78. Hobbie, R.K. and B.J. Roth, Intermediate Physics for Medicine and Biology. 2007, New York: Springer Science+Business Media. 575. 79. Pálfi, P., Locally orthotropic femur model. J. Comput. Appl. Mech., 2002. 5: p. 103-115. 80. Ruckh, T.T., et al., Osteogenic differentiation of bone marrow stromal cells on poly(εcaprolactone) nanofiber scaffolds. Acta Biomaterialia, 2010. 6(8): p. 2949-2959. 81. Huang, N.F., et al., Myotube Assembly on Nanofibrous and Micropatterned Polymers. Nano Letters, 2006. 6(3): p. 537-542. 82. Lu, H., et al., Electrospun submicron bioactive glass fibers for bone tissue scaffold. Journal of Materials Science: Materials in Medicine, 2009. 20(3): p. 793-798. 83. Spadaccio, C., et al., Poly-L-lactic acid/hydroxyapatite electrospun nanocomposites induce chondrogenic differentiation of human MSC. Annals of Biomedical Engineering, 2009. 37(7): p. 1376-1389. 84. Kim, G.M., G.H. Michler, and P. Pötschke, Deformation processes of ultrahigh porous multiwalled carbon nanotubes/polycarbonate composite fibers prepared by electrospinning. Polymer, 2005. 46(18): p. 7346-7351. 85. Kim, T.G., D.S. Lee, and T.G. Park, Controlled protein release from electrospun biodegradable fiber mesh composed of poly(ɛ-caprolactone) and poly(ethylene oxide). International Journal of Pharmaceutics, 2007. 338(1-2): p. 276-283. 86. Kim, G., H. Yoon, and Y. Park, Drug release from various thicknesses of layered mats consisting of electrospun polycaprolactone and polyethylene oxide micro/nanofibers. Applied Physics A: Materials Science & Processing, 2010. 100(4): p. 1197-1204. 87. Kim, G., J. Park, and S. Park, Surface-treated and multilayered poly(ε-caprolactone) nanofiber webs exhibiting enhanced hydrophilicity. Journal of Polymer Science Part B: Polymer Physics, 2007. 45(15): p. 2038-2045. 88. Xu, F., et al., Improvement of cytocompatibility of electrospinning PLLA microfibers by blending PVP. Journal of Materials Science: Materials in Medicine, 2009. 20(6): p. 1331-1338. 89. Bognitzki, M., et al., Preparation of fibers with nanoscaled morphologies: Electrospinning of polymer blends. Polymer Engineering & Science, 2001. 41(6): p. 982-989. 90. Guimarães, A., et al., Solving cell infiltration limitations of electrospun nanofiber meshes for tissue engineering applications. Nanomedicine, 2010. 5(4): p. 539-554. 119
91. Baker, B.M., et al., The potential to improve cell infiltration in composite fiber-aligned electrospun scaffolds by the selective removal of sacrificial fibers. Biomaterials, 2008. 29(15): p. 2348-2358. 92. Zhao, Q., et al., Using poly[2-methoxy-5-(2 -ethyl-hexyloxy)-1,4-phenylene vinylene] as shell to fabricate the highly fluorescent nanofibers by coaxial electrospinning. Polymer, 2007. 48(15): p. 4311-4315. 93. Yang, Q., et al., Influence of solvents on the formation of ultrathin uniform poly(vinyl pyrrolidone) nanofibers with electrospinning. Journal of Polymer Science Part B: Polymer Physics, 2004. 42(20): p. 3721-3726. 94. Cui, X., et al., Fabrication of continuous aligned polyvinylpyrrolidone fibers via electrospinning by elimination of the jet bending instability. Journal of Applied Polymer Science, 2010. 116(6): p. 3676-3681. 95. Yu, D.-G., et al., Ultrafine ibuprofen-loaded polyvinylpyrrolidone fiber mats using electrospinning. Polymer International, 2009. 58(9): p. 1010-1013. 96. Lee, S.-W. and A.M. Belcher, Virus-Based Fabrication of Micro- and Nanofibers Using Electrospinning. Nano Letters, 2004. 4(3): p. 387-390. 97. Buruaga, L., et al., Role of specific interactions on fiber formation in the electrospinning of poly(vinyl phenol)/poly(vinyl pyrrolidone) blend solutions. Journal of Applied Polymer Science, 2009. 114(5): p. 2922-2928. 98. Bai, J., et al., Preparing AgBr nanoparticles in poly(vinyl pyrrolidone) (PVP) nanofibers. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2008. 329(3): p. 165-168. 99. Jin, W.-J., et al., Preparation of Polymer Nanofibers Containing Silver Nanoparticles by Using Poly(N-vinylpyrrolidone). Macromolecular Rapid Communications, 2005. 26(24): p. 1903-1907. 100. Suganya, S., et al., Herbal drug incorporated antibacterial nanofibrous mat fabricated by electrospinning: An excellent matrix for wound dressings. Journal of Applied Polymer Science, 2011. 121(5): p. 2893-2899. 101. Yong Tang Jia, X.Y.Z., Qing Qing Liu, In Vitro Degradation of Electrospun Fiber Membranes of PCL/PVP Blends. Advanced Materials Research 2011. 332-334: p. 1330-1334. 102. Ye, W.P., et al., In vitro degradation of poly(caprolactone), poly(lactide) and their block copolymers: influence of composition, temperature and morphology. Reactive and Functional Polymers, 1997. 32(2): p. 161-168. 103. Boland, E.D., et al., Electrospinning of Bioresorbable Polymers for Tissue Engineering Scaffolds, in Polymeric Nanofibers. 2006, American Chemical Society. p. 188-204. 104. Bosworth, L.A. and S. Downes, Physicochemical characterisation of degrading polycaprolactone scaffolds. Polymer Degradation and Stability, 2010. 95(12): p. 2269-2276. 120
105. Meng, Z.X., et al., Fabrication and characterization of three-dimensional nanofiber membrance of PCL MWCNTs by electrospinning. Materials Science and Engineering: C, 2010. 30(7): p. 1014-1021. 106. Sekosan, G. and N. Vasanthan, Morphological changes of annealed poly-εcaprolactone by enzymatic degradation with lipase. Journal of Polymer Science Part B: Polymer Physics, 2010. 48(2): p. 202-211. 107. Hartman, O., et al., Biofunctionalization of electrospun PCL-based scaffolds with perlecan domain IV peptide to create a 3-D pharmacokinetic cancer model. Biomaterials, 2010. 31(21): p. 5700-5718. 108. Ren, Z., et al., A study on the hydrogen bonding interaction of the electrospun ladder polyphenylsilsesquioxane/polyisophthalamide composite fibers by ATR FT-IR. Polymer Chemistry, 2011. 2(3): p. 608-613. 109. Elzein, T., et al., FTIR study of polycaprolactone chain organization at interfaces. Journal of Colloid and Interface Science, 2004. 273(2): p. 381-387. 110. Evlampieva, N.P., T.A. Dmitrieva, and E.I. Ryumtsev, Dipole Structure and Electrooptical Properties of Poly-N-vinylpyrrolydone in Nonaqueous Solvents. Russian Journal of Applied Chemistry, 2003. 76(10): p. 1637-1642. 111. Vaquette, C. and J.J. Cooper-White, Increasing electrospun scaffold pore size with tailored collectors for improved cell penetration. Acta Biomaterialia, 2011. 121
Appendix Matlab routine: grid architecture (square boundary) function CreaTraiettoria(lato, spacing) num_avanzamenti= round(lato/spacing); fid=fopen('c:\\nomefile.txt','w'); fprintf(fid,'%s %d%s%d\n','vp',lato,',',0); for i=0:1:(num_avanzamenti/2-1) end fprintf(fid,'%s %d%s%d\n','vp',lato,',',(2*i+1)*spacing); fprintf(fid,'%s %d%s%d\n','vp',0,',',(2*i+1)*spacing); fprintf(fid,'%s %d%s%d\n','vp',0,',',(2*i+2)*spacing); fprintf(fid,'%s %d%s%d\n','vp',lato,',',(2*i+2)*spacing); fprintf(fid, '\n'); fprintf(fid,'%s %d%s%d\n','vp',0,',',-lato); for i=0:1:(num_avanzamenti/2-1) fprintf(fid,'%s %d%s%d\n','vp',-(2*i+1)*spacing,',',-lato); fprintf(fid,'%s %d%s%d\n','vp',-(2*i+1)*spacing,',',0); fprintf(fid,'%s %d%s%d\n','vp',-(2*i+2)*spacing,',',0); fprintf(fid,'%s %d%s%d\n','vp',-(2*i+2)*spacing,',',-lato); end fclose(fid); Machine Code: grid architecture (square boundary) SH CS % clear sequence PRZ=-9000 % move Z-axis BGZ AMZ CS % clear sequence OP1 % open electrovalve ES 1,2 VMXY VS 1300 % vector speed VP 16000,0 % vector position, X-Y movements VP 16000,800 VP 0,800 VP 0,1600 VP 16000,1600 VP 16000,2400 VP 0,2400 VP 0,3200 VP 16000,3200 VP 16000,4000 VP 0,4000 VP 0,4800 VP 16000,4800 122
VP 16000,5600 VP 0,5600 VP 0,6400 VP 16000,6400 VP 16000,7200 VP 0,7200 VP 0,8000 VP 16000,8000 VP 16000,8800 VP 0,8800 VP 0,9600 VP 16000,9600 VP 16000,10400 VP 0,10400 VP 0,11200 VP 16000,11200 VP 16000,12000 VP 0,12000 VP 0,12800 VP 16000,12800 VP 16000,13600 VP 0,13600 VP 0,14400 VP 16000,14400 VP 16000,15200 VP 0,15200 VP 0,16000 VP 16000,16000 VE % end of the vectors sequence BGS % second layer AMS CS PRZ=-5000 BGZ AMZ CS ES 1,2 VMXY VS 1300 VP 0,-16000 VP -800,-16000 VP -800,0 VP -1600,0 VP -1600,-16000 VP -2400,-16000 VP -2400,0 VP -3200,0 VP -3200,-16000 VP -4000,-16000 VP -4000,0 VP -4800,0 VP -4800,-16000 VP -5600,-16000 VP -5600,0 VP -6400,0 VP -6400,-16000 VP -7200,-16000 VP -7200,0 123
VP -8000,0 VP -8000,-16000 VP -8800,-16000 VP -8800,0 VP -9600,0 VP -9600,-16000 VP -10400,-16000 VP -10400,0 VP -11200,0 VP -11200,-16000 VP -12000,-16000 VP -12000,0 VP -12800,0 VP -12800,-16000 VP -13600,-16000 VP -13600,0 VP -14400,0 VP -14400,-16000 VP -15200,-16000 VP -15200,0 VP -16000,0 VP -16000,-16000 VE BGS AMS CS PRZ=-3000 BGZ AMZ CS OP2 % close electrovalve MO % motor off Figure 59: DMC smart terminal interface with the uploaded program Machine Code: direct control (armoring coil) 124
#TRIP n=0 CS SH VMXW SPS=5000 % speed rate VP -43000,1500000 VE BGS AM WT 200 % pause(ms) OP2 EN Matlab routine: Hilbert architecture (square boundary) This Matlab routine fills a square boundary with a Hilbert curve and prints a text file containing the correspondent machine code. function [x,y]= hilbert_curve(xt, yt, L, marg_thick); % function in the L-system for drawing a Hilbert curve in 2D space % usage: [x,y]=hilbert(xt,yt,l,marg_thick) % example: [x,y]=hilbert_curve(0.1,0.9,1,0.1) % starting points: xt = 0.1; yt = 0.9 (upper left) % L: square side (L=1) % marg_thick: left and right distance from the square side (0.1) % Rules-- Cell array {1,x} is the xth string to be replaced % -- {2,x} rule(1).before = 'Y'; rule(1).after = '+XF-YFY-FX+'; rule(2).before = 'X'; rule(2).after = '-YF+XFX+FY-'; nrules = length(rule); %angle: +operator means turn left; -operator means turn right delta = 90; %degrees %starting seed axiom = 'X'; %number of repetitions nreps = input('number of repetitions=\n') %length of the line segment corresponding to the symbol F width=(l-2*marg_thick); %(L less left-right boundary thickness) lenf = width/(2^(nreps)-1);%length of segment at the forward step F for i=1:nreps %one character/cell, with indexes the same as original axiom string 125
end axiomincells = cellstr(axiom'); for j=1:nrules %the indexes of each 'before' string hit = strfind(axiom, rule(j).before); if (length(hit)>=1) for k=hit axiomincells{k} = rule(j).after; end end end %now convert individual cells back to a string axiom=[]; for j=1:length(axiomincells) axiom = [axiom, axiomincells{j}]; end % Now draw the string as turtle graphics % Upper case (F) causes a line to be drawn in the current direction of the turtle angle +operator means turn % left; -operator means turn right %Init the turtle at = 0; da = deg2rad(delta) ; %convert to radians hold on x=[],y=[]; newx=[],newy=[]; for i=1:length(axiom) cmdt = axiom(i); switch cmdt case 'F' newxt = xt + lenf*cos(at); newyt = yt + lenf*sin(at); x=[x; xt]; y=[y; yt]; newx=[newx; newxt]; newy=[newy; newyt]; end line([xt newxt], [yt newyt],'color',[0.3 0.3 0], 'linewidth',2);%line draws one line per column. xt = newxt; yt = newyt; case '+' at = at + da; case '-' at = at - da; otherwise i=i+1; end %drawnow % print a txt file with the machine code for i=1:1:length(x) fprintf(fid,'%s%d%s%d\n','vp',round(x(i)*1000),',',round(y(i)*1000)); 126
end fclose(fid); Figure 60: Hilbert pattern (number of repetitions 5, 4, 3) and PCL filament 127